Mercurial > projects > dil
annotate trunk/src/dil/lexer/Lexer.d @ 798:c24be8d4f6ab
Added documentation comments.
author | Aziz K?ksal <aziz.koeksal@gmail.com> |
---|---|
date | Sat, 01 Mar 2008 02:53:06 +0100 |
parents | cf2ad5df025c |
children | cb8040538772 |
rev | line source |
---|---|
0 | 1 /++ |
2 Author: Aziz Köksal | |
249 | 3 License: GPL3 |
0 | 4 +/ |
576
0df647660e76
Moved Lexer.d to new package 'lexer'.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
552
diff
changeset
|
5 module dil.lexer.Lexer; |
577
9e811db780a6
Moved LexerFuncs.d to package 'lexer'.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
576
diff
changeset
|
6 |
596
39fac5531b85
Moved dil.Token to dil.lexer.Token.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
577
diff
changeset
|
7 import dil.lexer.Token; |
599
c4cdea3a65ba
Moved dil.Keywords to dil.lexer.Keywords.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
596
diff
changeset
|
8 import dil.lexer.Keywords; |
600
041eae272362
Moved dil.Identifier to dil.lexer.Identifier.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
599
diff
changeset
|
9 import dil.lexer.Identifier; |
601
9f96fd51cb41
Moved dil.IdTable to dil.lexer.IdTable.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
600
diff
changeset
|
10 import dil.lexer.IdTable; |
327 | 11 import dil.Information; |
12 import dil.Messages; | |
13 import dil.HtmlEntities; | |
464
325714d8aa6c
Added new module with compiler constants to resolve import cycle.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
434
diff
changeset
|
14 import dil.CompilerInfo; |
510
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
15 import dil.Unicode; |
755
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
16 import dil.SourceText; |
776
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
17 import dil.Time; |
737
f88b5285b86b
Implemented DDocEmitter.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
715
diff
changeset
|
18 import common; |
f88b5285b86b
Implemented DDocEmitter.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
715
diff
changeset
|
19 |
391
33b566df6af4
Migrated project to Tango.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
390
diff
changeset
|
20 import tango.stdc.stdlib : strtof, strtod, strtold; |
33b566df6af4
Migrated project to Tango.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
390
diff
changeset
|
21 import tango.stdc.errno : errno, ERANGE; |
0 | 22 |
577
9e811db780a6
Moved LexerFuncs.d to package 'lexer'.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
576
diff
changeset
|
23 public import dil.lexer.Funcs; |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
24 |
776
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
25 /// The Lexer analyzes the characters of a source text and |
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
26 /// produces a doubly-linked list of tokens. |
0 | 27 class Lexer |
28 { | |
755
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
29 SourceText srcText; /// The source text. |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
30 char* p; /// Points to the current character in the source text. |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
31 char* end; /// Points one character past the end of the source text. |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
32 |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
33 Token* head; /// The head of the doubly linked token list. |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
34 Token* tail; /// The tail of the linked list. Set in scan(). |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
35 Token* token; /// Points to the current token in the token list. |
4 | 36 |
434
cac9c8d637ad
Fix: members related to error info weren't updated correctly.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
433
diff
changeset
|
37 // Members used for error messages: |
532
50e64bab9c7a
Renamed InformationManager to InfoManager.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
520
diff
changeset
|
38 InfoManager infoMan; |
514
6ddff941862a
Added new error classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
510
diff
changeset
|
39 LexerError[] errors; |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
40 /// Always points to the first character of the current line. |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
41 char* lineBegin; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
42 // Token* newline; /// Current newline token. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
43 uint lineNum = 1; /// Current, actual source text line number. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
44 uint lineNum_hline; /// Line number set by #line. |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
45 uint inTokenString; /// > 0 if inside q{ } |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
46 /// Holds the original file path and the modified one (by #line.) |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
47 NewlineData.FilePaths* filePaths; |
103
511a1aa25896
- Added reportErrors member to Lexer. Moved peek() down and rewrote it a bit making use of reportErrors. error() uses reportErrors too.
aziz
parents:
94
diff
changeset
|
48 |
776
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
49 /// Construct a Lexer object. |
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
50 /// Params: |
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
51 /// srcText = the UTF-8 source code. |
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
52 /// infoMan = used for collecting error messages. |
755
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
53 this(SourceText srcText, InfoManager infoMan = null) |
4 | 54 { |
755
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
55 this.srcText = srcText; |
520
f203c5248d0b
Added 'compile' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
519
diff
changeset
|
56 this.infoMan = infoMan; |
30 | 57 |
755
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
58 assert(text.length && text[$-1] == 0, "source text has no sentinel character"); |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
59 this.p = text.ptr; |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
60 this.end = this.p + text.length; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
61 this.lineBegin = this.p; |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
223
diff
changeset
|
62 |
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
223
diff
changeset
|
63 this.head = new Token; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
64 this.head.kind = TOK.HEAD; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
65 this.head.start = this.head.end = this.p; |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
223
diff
changeset
|
66 this.token = this.head; |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
67 // Initialize this.filePaths. |
755
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
68 newFilePath(this.srcText.filePath); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
69 // Add a newline as the first token after the head. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
70 auto newline = new Token; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
71 newline.kind = TOK.Newline; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
72 newline.setWhitespaceFlag(); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
73 newline.start = newline.end = this.p; |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
74 newline.newline.filePaths = this.filePaths; |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
75 newline.newline.oriLineNum = 1; |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
76 newline.newline.setLineNum = 0; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
77 // Link in. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
78 this.token.next = newline; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
79 newline.prev = this.token; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
80 this.token = newline; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
81 // this.newline = newline; |
315 | 82 scanShebang(); |
83 } | |
84 | |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
85 /// The destructor deletes the doubly-linked token list. |
344 | 86 ~this() |
87 { | |
88 auto token = head.next; | |
418
7354f15cd5e9
Applied some fixes to the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
416
diff
changeset
|
89 while (token !is null) |
344 | 90 { |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
91 assert(token.kind == TOK.EOF ? token == tail && token.next is null : 1); |
344 | 92 delete token.prev; |
93 token = token.next; | |
418
7354f15cd5e9
Applied some fixes to the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
416
diff
changeset
|
94 } |
344 | 95 delete tail; |
96 } | |
97 | |
755
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
98 char[] text() |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
99 { |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
100 return srcText.data; |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
101 } |
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
102 |
776
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
103 /// The "shebang" may optionally appear once at the beginning of a file. |
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
104 /// Regexp: #![^\EndOfLine]* |
315 | 105 void scanShebang() |
106 { | |
107 if (*p == '#' && p[1] == '!') | |
108 { | |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
109 auto t = new Token; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
110 t.kind = TOK.Shebang; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
111 t.setWhitespaceFlag(); |
315 | 112 t.start = p; |
113 ++p; | |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
114 while (!isEndOfLine(++p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
115 isascii(*p) || decodeUTF8(); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
116 t.end = p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
117 this.token.next = t; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
118 t.prev = this.token; |
315 | 119 } |
4 | 120 } |
121 | |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
122 /// Sets the value of the special token. |
343
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
123 void finalizeSpecialToken(ref Token t) |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
124 { |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
125 assert(t.srcText[0..2] == "__"); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
126 switch (t.kind) |
343
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
127 { |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
128 case TOK.FILE: |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
129 t.str = this.filePaths.setPath; |
343
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
130 break; |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
131 case TOK.LINE: |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
132 t.uint_ = this.errorLineNumber(this.lineNum); |
343
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
133 break; |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
134 case TOK.DATE, |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
135 TOK.TIME, |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
136 TOK.TIMESTAMP: |
776
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
137 auto time_str = Time.toString(); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
138 switch (t.kind) |
343
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
139 { |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
140 case TOK.DATE: |
776
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
141 time_str = Time.month_day(time_str) ~ ' ' ~ Time.year(time_str); break; |
343
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
142 case TOK.TIME: |
776
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
143 time_str = Time.time(time_str); break; |
343
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
144 case TOK.TIMESTAMP: |
776
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
145 break; // time_str is the timestamp. |
343
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
146 default: assert(0); |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
147 } |
776
580d4ca9f1ff
Added new module dil.Time.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
769
diff
changeset
|
148 time_str ~= '\0'; // Terminate with a zero. |
343
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
149 t.str = time_str; |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
150 break; |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
151 case TOK.VENDOR: |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
152 t.str = VENDOR; |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
153 break; |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
154 case TOK.VERSION: |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
155 t.uint_ = VERSION_MAJOR*1000 + VERSION_MINOR; |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
156 break; |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
157 default: |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
158 assert(0); |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
159 } |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
160 } |
95f1b6e43214
- Removed TOK.Special and added an own entry for each special token.
aziz
parents:
327
diff
changeset
|
161 |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
162 /// Sets a new file path. |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
163 void newFilePath(char[] newPath) |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
164 { |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
165 auto paths = new NewlineData.FilePaths; |
755
90668b83ae5e
Introduced new module dil.SourceText and class SourceText.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
742
diff
changeset
|
166 paths.oriPath = this.srcText.filePath; |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
167 paths.setPath = newPath; |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
168 this.filePaths = paths; |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
169 } |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
170 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
171 private void setLineBegin(char* p) |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
172 { |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
173 // Check that we can look behind one character. |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
174 assert((p-1) >= text.ptr && p < end); |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
175 // Check that previous character is a newline. |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
176 assert(isNewlineEnd(p - 1)); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
177 this.lineBegin = p; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
178 } |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
179 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
180 /// Scans the next token in the source text. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
181 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
182 /// Creates a new token if t.next is null and appends it to the list. |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
183 private void scanNext(ref Token* t) |
419
89e40d43065d
Added new member 'p_newl' to Lexer and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
418
diff
changeset
|
184 { |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
185 assert(t !is null); |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
186 if (t.next) |
434
cac9c8d637ad
Fix: members related to error info weren't updated correctly.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
433
diff
changeset
|
187 { |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
188 t = t.next; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
189 // if (t.kind == TOK.Newline) |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
190 // this.newline = t; |
434
cac9c8d637ad
Fix: members related to error info weren't updated correctly.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
433
diff
changeset
|
191 } |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
192 else if (t != this.tail) |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
193 { |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
194 Token* new_t = new Token; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
195 scan(*new_t); |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
196 new_t.prev = t; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
197 t.next = new_t; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
198 t = new_t; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
199 } |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
200 } |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
201 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
202 /// Advance t one token forward. |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
203 void peek(ref Token* t) |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
204 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
205 scanNext(t); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
206 } |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
207 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
208 /// Advance to the next token in the source text. |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
209 TOK nextToken() |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
210 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
211 scanNext(this.token); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
212 return this.token.kind; |
419
89e40d43065d
Added new member 'p_newl' to Lexer and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
418
diff
changeset
|
213 } |
89e40d43065d
Added new member 'p_newl' to Lexer and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
418
diff
changeset
|
214 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
215 /// Returns true if p points to the last character of a Newline. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
216 bool isNewlineEnd(char* p) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
217 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
218 if (*p == '\n' || *p == '\r') |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
219 return true; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
220 if (*p == LS[2] || *p == PS[2]) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
221 if ((p-2) >= text.ptr) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
222 if (p[-1] == LS[1] && p[-2] == LS[0]) |
434
cac9c8d637ad
Fix: members related to error info weren't updated correctly.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
433
diff
changeset
|
223 return true; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
224 return false; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
225 } |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
226 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
227 /// The main method which recognizes the characters that make up a token. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
228 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
229 /// Complicated tokens are scanned in separate methods. |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
230 public void scan(ref Token t) |
207 | 231 in |
3 | 232 { |
207 | 233 assert(text.ptr <= p && p < end); |
234 } | |
235 out | |
236 { | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
237 assert(text.ptr <= t.start && t.start < end, Token.toString(t.kind)); |
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
238 assert(text.ptr <= t.end && t.end <= end, Token.toString(t.kind)); |
207 | 239 } |
240 body | |
241 { | |
390
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
242 // Scan whitespace. |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
243 if (isspace(*p)) |
390
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
244 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
245 t.ws = p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
246 while (isspace(*++p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
247 {} |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
248 } |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
249 |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
250 // Scan a token. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
251 uint c = *p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
252 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
253 t.start = p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
254 // Newline. |
390
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
255 switch (*p) |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
256 { |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
257 case '\r': |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
258 if (p[1] == '\n') |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
259 ++p; |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
260 case '\n': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
261 assert(isNewlineEnd(p)); |
390
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
262 ++p; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
263 ++lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
264 setLineBegin(p); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
265 // this.newline = &t; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
266 t.kind = TOK.Newline; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
267 t.setWhitespaceFlag(); |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
268 t.newline.filePaths = this.filePaths; |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
269 t.newline.oriLineNum = lineNum; |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
270 t.newline.setLineNum = lineNum_hline; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
271 t.end = p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
272 return; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
273 default: |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
274 if (isUnicodeNewline(p)) |
390
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
275 { |
412
fb31af0fda73
Added struct Location, and token2LocTable to Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
411
diff
changeset
|
276 ++p; ++p; |
fb31af0fda73
Added struct Location, and token2LocTable to Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
411
diff
changeset
|
277 goto case '\n'; |
390
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
278 } |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
279 } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
280 // Identifier or string literal. |
13 | 281 if (isidbeg(c)) |
5 | 282 { |
33 | 283 if (c == 'r' && p[1] == '"' && ++p) |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
284 return scanRawStringLiteral(t); |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
285 if (c == 'x' && p[1] == '"') |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
286 return scanHexStringLiteral(t); |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
287 version(D2) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
288 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
289 if (c == 'q' && p[1] == '"') |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
290 return scanDelimitedStringLiteral(t); |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
291 if (c == 'q' && p[1] == '{') |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
292 return scanTokenStringLiteral(t); |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
293 } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
294 // Scan identifier. |
12
0989206cf73c
- Added code to decode Unicode characters in identifiers.
aziz
parents:
11
diff
changeset
|
295 Lidentifier: |
5 | 296 do |
297 { c = *++p; } | |
550
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
298 while (isident(c) || !isascii(c) && isUnicodeAlpha()) |
28 | 299 |
5 | 300 t.end = p; |
28 | 301 |
502
4e14cd1b24da
Refactored code and added modules related to tabulated Identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
499
diff
changeset
|
302 auto id = IdTable.lookup(t.srcText); |
769
5e3ef1b2011c
Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
764
diff
changeset
|
303 t.kind = id.kind; |
502
4e14cd1b24da
Refactored code and added modules related to tabulated Identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
499
diff
changeset
|
304 t.ident = id; |
28 | 305 |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
306 if (t.kind == TOK.Identifier || t.isKeyword) |
411 | 307 return; |
497
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
308 else if (t.isSpecialToken) |
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
309 finalizeSpecialToken(t); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
310 else if (t.kind == TOK.EOF) |
411 | 311 { |
312 tail = &t; | |
313 assert(t.srcText == "__EOF__"); | |
314 } | |
497
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
315 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
316 assert(0, "unexpected token type: " ~ Token.toString(t.kind)); |
5 | 317 return; |
318 } | |
319 | |
15 | 320 if (isdigit(c)) |
321 return scanNumber(t); | |
322 | |
8 | 323 if (c == '/') |
5 | 324 { |
8 | 325 c = *++p; |
14 | 326 switch(c) |
5 | 327 { |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
328 case '=': |
14 | 329 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
330 t.kind = TOK.DivAssign; |
14 | 331 t.end = p; |
332 return; | |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
333 case '+': |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
334 return scanNestedComment(t); |
32
d7011daa4740
- Added missing commas after the items in the messages table.
aziz
parents:
31
diff
changeset
|
335 case '*': |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
336 return scanBlockComment(t); |
32
d7011daa4740
- Added missing commas after the items in the messages table.
aziz
parents:
31
diff
changeset
|
337 case '/': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
338 while (!isEndOfLine(++p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
339 isascii(*p) || decodeUTF8(); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
340 t.kind = TOK.Comment; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
341 t.setWhitespaceFlag(); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
342 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
343 return; |
32
d7011daa4740
- Added missing commas after the items in the messages table.
aziz
parents:
31
diff
changeset
|
344 default: |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
345 t.kind = TOK.Div; |
10 | 346 t.end = p; |
347 return; | |
348 } | |
5 | 349 } |
350 | |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
351 switch (c) |
20 | 352 { |
48 | 353 case '\'': |
354 return scanCharacterLiteral(t); | |
355 case '`': | |
356 return scanRawStringLiteral(t); | |
357 case '"': | |
358 return scanNormalStringLiteral(t); | |
359 case '\\': | |
360 char[] buffer; | |
361 do | |
362 { | |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
363 bool isBinary; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
364 c = scanEscapeSequence(isBinary); |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
365 if (isascii(c) || isBinary) |
48 | 366 buffer ~= c; |
367 else | |
368 encodeUTF8(buffer, c); | |
369 } while (*p == '\\') | |
370 buffer ~= 0; | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
371 t.kind = TOK.String; |
48 | 372 t.str = buffer; |
373 t.end = p; | |
374 return; | |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
375 case '>': /* > >= >> >>= >>> >>>= */ |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
376 c = *++p; |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
377 switch (c) |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
378 { |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
379 case '=': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
380 t.kind = TOK.GreaterEqual; |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
381 goto Lcommon; |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
382 case '>': |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
383 if (p[1] == '>') |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
384 { |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
385 ++p; |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
386 if (p[1] == '=') |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
387 { ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
388 t.kind = TOK.URShiftAssign; |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
389 } |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
390 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
391 t.kind = TOK.URShift; |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
392 } |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
393 else if (p[1] == '=') |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
394 { |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
395 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
396 t.kind = TOK.RShiftAssign; |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
397 } |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
398 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
399 t.kind = TOK.RShift; |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
400 goto Lcommon; |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
401 default: |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
402 t.kind = TOK.Greater; |
38
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
403 goto Lcommon2; |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
404 } |
640c45aaaaee
- Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents:
37
diff
changeset
|
405 assert(0); |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
406 case '<': /* < <= <> <>= << <<= */ |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
407 c = *++p; |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
408 switch (c) |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
409 { |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
410 case '=': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
411 t.kind = TOK.LessEqual; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
412 goto Lcommon; |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
413 case '<': |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
414 if (p[1] == '=') { |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
415 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
416 t.kind = TOK.LShiftAssign; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
417 } |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
418 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
419 t.kind = TOK.LShift; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
420 goto Lcommon; |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
421 case '>': |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
422 if (p[1] == '=') { |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
423 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
424 t.kind = TOK.LorEorG; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
425 } |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
426 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
427 t.kind = TOK.LorG; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
428 goto Lcommon; |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
429 default: |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
430 t.kind = TOK.Less; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
431 goto Lcommon2; |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
432 } |
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
433 assert(0); |
37
7f3bcb97d017
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
36
diff
changeset
|
434 case '!': /* ! !< !> !<= !>= !<> !<>= */ |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
435 c = *++p; |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
436 switch (c) |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
437 { |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
438 case '<': |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
439 c = *++p; |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
440 if (c == '>') |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
441 { |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
442 if (p[1] == '=') { |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
443 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
444 t.kind = TOK.Unordered; |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
445 } |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
446 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
447 t.kind = TOK.UorE; |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
448 } |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
449 else if (c == '=') |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
450 { |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
451 t.kind = TOK.UorG; |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
452 } |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
453 else { |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
454 t.kind = TOK.UorGorE; |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
455 goto Lcommon2; |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
456 } |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
457 goto Lcommon; |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
458 case '>': |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
459 if (p[1] == '=') |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
460 { |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
461 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
462 t.kind = TOK.UorL; |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
463 } |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
464 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
465 t.kind = TOK.UorLorE; |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
466 goto Lcommon; |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
467 case '=': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
468 t.kind = TOK.NotEqual; |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
469 goto Lcommon; |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
470 default: |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
471 t.kind = TOK.Not; |
35
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
472 goto Lcommon2; |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
473 } |
c470b9356e35
- Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents:
34
diff
changeset
|
474 assert(0); |
63 | 475 case '.': /* . .[0-9] .. ... */ |
22
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
476 if (p[1] == '.') |
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
477 { |
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
478 ++p; |
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
479 if (p[1] == '.') { |
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
480 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
481 t.kind = TOK.Ellipses; |
22
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
482 } |
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
483 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
484 t.kind = TOK.Slice; |
22
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
485 } |
63 | 486 else if (isdigit(p[1])) |
487 { | |
488 return scanReal(t); | |
489 } | |
22
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
490 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
491 t.kind = TOK.Dot; |
22
b05fff8e2ce4
- Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents:
21
diff
changeset
|
492 goto Lcommon; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
493 case '|': /* | || |= */ |
23
1a7903701a3d
- Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents:
22
diff
changeset
|
494 c = *++p; |
1a7903701a3d
- Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents:
22
diff
changeset
|
495 if (c == '=') |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
496 t.kind = TOK.OrAssign; |
23
1a7903701a3d
- Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents:
22
diff
changeset
|
497 else if (c == '|') |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
498 t.kind = TOK.OrLogical; |
26 | 499 else { |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
500 t.kind = TOK.OrBinary; |
26 | 501 goto Lcommon2; |
502 } | |
23
1a7903701a3d
- Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents:
22
diff
changeset
|
503 goto Lcommon; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
504 case '&': /* & && &= */ |
24
903f91163f23
- Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents:
23
diff
changeset
|
505 c = *++p; |
903f91163f23
- Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents:
23
diff
changeset
|
506 if (c == '=') |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
507 t.kind = TOK.AndAssign; |
24
903f91163f23
- Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents:
23
diff
changeset
|
508 else if (c == '&') |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
509 t.kind = TOK.AndLogical; |
26 | 510 else { |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
511 t.kind = TOK.AndBinary; |
26 | 512 goto Lcommon2; |
513 } | |
24
903f91163f23
- Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents:
23
diff
changeset
|
514 goto Lcommon; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
515 case '+': /* + ++ += */ |
25
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
516 c = *++p; |
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
517 if (c == '=') |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
518 t.kind = TOK.PlusAssign; |
25
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
519 else if (c == '+') |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
520 t.kind = TOK.PlusPlus; |
26 | 521 else { |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
522 t.kind = TOK.Plus; |
26 | 523 goto Lcommon2; |
524 } | |
25
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
525 goto Lcommon; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
526 case '-': /* - -- -= */ |
25
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
527 c = *++p; |
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
528 if (c == '=') |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
529 t.kind = TOK.MinusAssign; |
25
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
530 else if (c == '-') |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
531 t.kind = TOK.MinusMinus; |
26 | 532 else { |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
533 t.kind = TOK.Minus; |
26 | 534 goto Lcommon2; |
535 } | |
536 goto Lcommon; | |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
537 case '=': /* = == */ |
26 | 538 if (p[1] == '=') { |
539 ++p; | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
540 t.kind = TOK.Equal; |
26 | 541 } |
25
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
542 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
543 t.kind = TOK.Assign; |
25
9c866208b3f6
- Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents:
24
diff
changeset
|
544 goto Lcommon; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
545 case '~': /* ~ ~= */ |
27 | 546 if (p[1] == '=') { |
547 ++p; | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
548 t.kind = TOK.CatAssign; |
27 | 549 } |
550 else | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
551 t.kind = TOK.Tilde; |
27 | 552 goto Lcommon; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
553 case '*': /* * *= */ |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
554 if (p[1] == '=') { |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
555 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
556 t.kind = TOK.MulAssign; |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
557 } |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
558 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
559 t.kind = TOK.Mul; |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
560 goto Lcommon; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
561 case '^': /* ^ ^= */ |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
562 if (p[1] == '=') { |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
563 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
564 t.kind = TOK.XorAssign; |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
565 } |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
566 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
567 t.kind = TOK.Xor; |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
568 goto Lcommon; |
36
3c7210a722f7
- Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents:
35
diff
changeset
|
569 case '%': /* % %= */ |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
570 if (p[1] == '=') { |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
571 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
572 t.kind = TOK.ModAssign; |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
573 } |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
574 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
575 t.kind = TOK.Mod; |
29
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
576 goto Lcommon; |
ef83eea26bbd
- Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents:
28
diff
changeset
|
577 // Single character tokens: |
20 | 578 case '(': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
579 t.kind = TOK.LParen; |
20 | 580 goto Lcommon; |
581 case ')': | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
582 t.kind = TOK.RParen; |
20 | 583 goto Lcommon; |
584 case '[': | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
585 t.kind = TOK.LBracket; |
20 | 586 goto Lcommon; |
587 case ']': | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
588 t.kind = TOK.RBracket; |
20 | 589 goto Lcommon; |
590 case '{': | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
591 t.kind = TOK.LBrace; |
20 | 592 goto Lcommon; |
593 case '}': | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
594 t.kind = TOK.RBrace; |
21
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
595 goto Lcommon; |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
596 case ':': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
597 t.kind = TOK.Colon; |
21
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
598 goto Lcommon; |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
599 case ';': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
600 t.kind = TOK.Semicolon; |
21
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
601 goto Lcommon; |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
602 case '?': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
603 t.kind = TOK.Question; |
21
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
604 goto Lcommon; |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
605 case ',': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
606 t.kind = TOK.Comma; |
21
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
607 goto Lcommon; |
c785c122e4e6
- Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents:
20
diff
changeset
|
608 case '$': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
609 t.kind = TOK.Dollar; |
20 | 610 Lcommon: |
611 ++p; | |
26 | 612 Lcommon2: |
20 | 613 t.end = p; |
614 return; | |
30 | 615 case '#': |
360
b6a3755eba94
- Renamed scanSpecialToken() to scanSpecialTokenSequence().
aziz
parents:
350
diff
changeset
|
616 return scanSpecialTokenSequence(t); |
20 | 617 default: |
618 } | |
619 | |
411 | 620 // Check for EOF |
497
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
621 if (isEOF(c)) |
411 | 622 { |
497
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
623 assert(isEOF(*p), ""~*p); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
624 t.kind = TOK.EOF; |
411 | 625 t.end = p; |
626 tail = &t; | |
627 assert(t.start == t.end); | |
628 return; | |
629 } | |
630 | |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
631 if (!isascii(c)) |
390
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
632 { |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
633 c = decodeUTF8(); |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
634 if (isUniAlpha(c)) |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
635 goto Lidentifier; |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
636 } |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
637 |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
638 error(t.start, MID.IllegalCharacter, cast(dchar)c); |
390
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
639 |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
640 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
641 t.kind = TOK.Illegal; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
642 t.setWhitespaceFlag(); |
390
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
643 t.dchar_ = c; |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
644 t.end = p; |
4d36eea1bbc9
Refactored Lexer.scan().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
389
diff
changeset
|
645 return; |
4 | 646 } |
3 | 647 } |
4 | 648 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
649 /// Converts a string literal to an integer. |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
650 template toUint(char[] T) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
651 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
652 static assert(0 < T.length && T.length <= 4); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
653 static if (T.length == 1) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
654 const uint toUint = T[0]; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
655 else |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
656 const uint toUint = (T[0] << ((T.length-1)*8)) | toUint!(T[1..$]); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
657 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
658 static assert(toUint!("\xAA\xBB\xCC\xDD") == 0xAABBCCDD); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
659 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
660 /// Constructs case statements. E.g.: |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
661 /// --- |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
662 //// // case_!("<", "Less", "Lcommon") -> |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
663 /// case 60u: |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
664 /// t.kind = TOK.Less; |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
665 /// goto Lcommon; |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
666 /// --- |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
667 /// Note:Can't use this yet due to a $(DMDBUG 1534, bug) in DMD. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
668 template case_(char[] str, char[] kind, char[] label) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
669 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
670 const char[] case_ = |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
671 `case `~toUint!(str).stringof~`:` |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
672 `t.kind = TOK.`~kind~`;` |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
673 `goto `~label~`;`; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
674 } |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
675 //pragma(msg, case_!("<", "Less", "Lcommon")); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
676 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
677 template case_L4(char[] str, TOK kind) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
678 { |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
679 const char[] case_L4 = case_!(str, kind, "Lcommon_4"); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
680 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
681 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
682 template case_L3(char[] str, TOK kind) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
683 { |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
684 const char[] case_L3 = case_!(str, kind, "Lcommon_3"); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
685 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
686 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
687 template case_L2(char[] str, TOK kind) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
688 { |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
689 const char[] case_L2 = case_!(str, kind, "Lcommon_2"); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
690 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
691 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
692 template case_L1(char[] str, TOK kind) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
693 { |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
694 const char[] case_L3 = case_!(str, kind, "Lcommon"); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
695 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
696 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
697 /// An alternative scan method. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
698 /// Profiling shows it's a bit slower. |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
699 public void scan_(ref Token t) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
700 in |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
701 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
702 assert(text.ptr <= p && p < end); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
703 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
704 out |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
705 { |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
706 assert(text.ptr <= t.start && t.start < end, Token.toString(t.kind)); |
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
707 assert(text.ptr <= t.end && t.end <= end, Token.toString(t.kind)); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
708 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
709 body |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
710 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
711 // Scan whitespace. |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
712 if (isspace(*p)) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
713 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
714 t.ws = p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
715 while (isspace(*++p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
716 {} |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
717 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
718 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
719 // Scan a token. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
720 t.start = p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
721 // Newline. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
722 switch (*p) |
418
7354f15cd5e9
Applied some fixes to the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
416
diff
changeset
|
723 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
724 case '\r': |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
725 if (p[1] == '\n') |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
726 ++p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
727 case '\n': |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
728 assert(isNewlineEnd(p)); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
729 ++p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
730 ++lineNum; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
731 setLineBegin(p); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
732 // this.newline = &t; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
733 t.kind = TOK.Newline; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
734 t.setWhitespaceFlag(); |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
735 t.newline.filePaths = this.filePaths; |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
736 t.newline.oriLineNum = lineNum; |
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
737 t.newline.setLineNum = lineNum_hline; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
738 t.end = p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
739 return; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
740 default: |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
741 if (isUnicodeNewline(p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
742 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
743 ++p; ++p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
744 goto case '\n'; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
745 } |
418
7354f15cd5e9
Applied some fixes to the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
416
diff
changeset
|
746 } |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
747 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
748 uint c = *p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
749 assert(end - p != 0); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
750 switch (end - p) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
751 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
752 case 1: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
753 goto L1character; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
754 case 2: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
755 c <<= 8; c |= p[1]; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
756 goto L2characters; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
757 case 3: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
758 c <<= 8; c |= p[1]; c <<= 8; c |= p[2]; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
759 goto L3characters; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
760 default: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
761 version(BigEndian) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
762 c = *cast(uint*)p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
763 else |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
764 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
765 c <<= 8; c |= p[1]; c <<= 8; c |= p[2]; c <<= 8; c |= p[3]; |
414
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
766 /+ |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
767 c = *cast(uint*)p; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
768 asm |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
769 { |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
770 mov EDX, c; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
771 bswap EDX; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
772 mov c, EDX; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
773 } |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
774 +/ |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
775 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
776 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
777 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
778 // 4 character tokens. |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
779 switch (c) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
780 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
781 case toUint!(">>>="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
782 t.kind = TOK.RShiftAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
783 goto Lcommon_4; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
784 case toUint!("!<>="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
785 t.kind = TOK.Unordered; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
786 Lcommon_4: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
787 p += 4; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
788 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
789 return; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
790 default: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
791 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
792 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
793 c >>>= 8; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
794 L3characters: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
795 assert(p == t.start); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
796 // 3 character tokens. |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
797 switch (c) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
798 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
799 case toUint!(">>="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
800 t.kind = TOK.RShiftAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
801 goto Lcommon_3; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
802 case toUint!(">>>"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
803 t.kind = TOK.URShift; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
804 goto Lcommon_3; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
805 case toUint!("<>="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
806 t.kind = TOK.LorEorG; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
807 goto Lcommon_3; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
808 case toUint!("<<="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
809 t.kind = TOK.LShiftAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
810 goto Lcommon_3; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
811 case toUint!("!<="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
812 t.kind = TOK.UorG; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
813 goto Lcommon_3; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
814 case toUint!("!>="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
815 t.kind = TOK.UorL; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
816 goto Lcommon_3; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
817 case toUint!("!<>"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
818 t.kind = TOK.UorE; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
819 goto Lcommon_3; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
820 case toUint!("..."): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
821 t.kind = TOK.Ellipses; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
822 Lcommon_3: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
823 p += 3; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
824 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
825 return; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
826 default: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
827 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
828 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
829 c >>>= 8; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
830 L2characters: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
831 assert(p == t.start); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
832 // 2 character tokens. |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
833 switch (c) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
834 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
835 case toUint!("/+"): |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
836 ++p; // Skip / |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
837 return scanNestedComment(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
838 case toUint!("/*"): |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
839 ++p; // Skip / |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
840 return scanBlockComment(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
841 case toUint!("//"): |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
842 ++p; // Skip / |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
843 assert(*p == '/'); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
844 while (!isEndOfLine(++p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
845 isascii(*p) || decodeUTF8(); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
846 t.kind = TOK.Comment; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
847 t.setWhitespaceFlag(); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
848 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
849 return; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
850 case toUint!(">="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
851 t.kind = TOK.GreaterEqual; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
852 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
853 case toUint!(">>"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
854 t.kind = TOK.RShift; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
855 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
856 case toUint!("<<"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
857 t.kind = TOK.LShift; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
858 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
859 case toUint!("<="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
860 t.kind = TOK.LessEqual; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
861 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
862 case toUint!("<>"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
863 t.kind = TOK.LorG; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
864 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
865 case toUint!("!<"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
866 t.kind = TOK.UorGorE; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
867 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
868 case toUint!("!>"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
869 t.kind = TOK.UorLorE; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
870 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
871 case toUint!("!="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
872 t.kind = TOK.NotEqual; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
873 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
874 case toUint!(".."): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
875 t.kind = TOK.Slice; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
876 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
877 case toUint!("&&"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
878 t.kind = TOK.AndLogical; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
879 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
880 case toUint!("&="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
881 t.kind = TOK.AndAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
882 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
883 case toUint!("||"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
884 t.kind = TOK.OrLogical; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
885 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
886 case toUint!("|="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
887 t.kind = TOK.OrAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
888 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
889 case toUint!("++"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
890 t.kind = TOK.PlusPlus; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
891 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
892 case toUint!("+="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
893 t.kind = TOK.PlusAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
894 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
895 case toUint!("--"): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
896 t.kind = TOK.MinusMinus; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
897 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
898 case toUint!("-="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
899 t.kind = TOK.MinusAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
900 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
901 case toUint!("=="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
902 t.kind = TOK.Equal; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
903 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
904 case toUint!("~="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
905 t.kind = TOK.CatAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
906 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
907 case toUint!("*="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
908 t.kind = TOK.MulAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
909 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
910 case toUint!("/="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
911 t.kind = TOK.DivAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
912 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
913 case toUint!("^="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
914 t.kind = TOK.XorAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
915 goto Lcommon_2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
916 case toUint!("%="): |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
917 t.kind = TOK.ModAssign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
918 Lcommon_2: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
919 p += 2; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
920 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
921 return; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
922 default: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
923 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
924 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
925 c >>>= 8; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
926 L1character: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
927 assert(p == t.start); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
928 assert(*p == c, Format("p={0},c={1}", *p, cast(dchar)c)); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
929 // 1 character tokens. |
414
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
930 // TODO: consider storing the token type in ptable. |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
931 switch (c) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
932 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
933 case '\'': |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
934 return scanCharacterLiteral(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
935 case '`': |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
936 return scanRawStringLiteral(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
937 case '"': |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
938 return scanNormalStringLiteral(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
939 case '\\': |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
940 char[] buffer; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
941 do |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
942 { |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
943 bool isBinary; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
944 c = scanEscapeSequence(isBinary); |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
945 if (isascii(c) || isBinary) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
946 buffer ~= c; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
947 else |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
948 encodeUTF8(buffer, c); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
949 } while (*p == '\\') |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
950 buffer ~= 0; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
951 t.kind = TOK.String; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
952 t.str = buffer; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
953 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
954 return; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
955 case '<': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
956 t.kind = TOK.Greater; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
957 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
958 case '>': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
959 t.kind = TOK.Less; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
960 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
961 case '^': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
962 t.kind = TOK.Xor; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
963 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
964 case '!': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
965 t.kind = TOK.Not; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
966 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
967 case '.': |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
968 if (isdigit(p[1])) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
969 return scanReal(t); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
970 t.kind = TOK.Dot; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
971 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
972 case '&': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
973 t.kind = TOK.AndBinary; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
974 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
975 case '|': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
976 t.kind = TOK.OrBinary; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
977 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
978 case '+': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
979 t.kind = TOK.Plus; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
980 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
981 case '-': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
982 t.kind = TOK.Minus; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
983 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
984 case '=': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
985 t.kind = TOK.Assign; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
986 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
987 case '~': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
988 t.kind = TOK.Tilde; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
989 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
990 case '*': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
991 t.kind = TOK.Mul; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
992 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
993 case '/': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
994 t.kind = TOK.Div; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
995 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
996 case '%': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
997 t.kind = TOK.Mod; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
998 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
999 case '(': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1000 t.kind = TOK.LParen; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1001 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1002 case ')': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1003 t.kind = TOK.RParen; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1004 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1005 case '[': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1006 t.kind = TOK.LBracket; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1007 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1008 case ']': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1009 t.kind = TOK.RBracket; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1010 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1011 case '{': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1012 t.kind = TOK.LBrace; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1013 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1014 case '}': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1015 t.kind = TOK.RBrace; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1016 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1017 case ':': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1018 t.kind = TOK.Colon; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1019 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1020 case ';': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1021 t.kind = TOK.Semicolon; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1022 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1023 case '?': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1024 t.kind = TOK.Question; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1025 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1026 case ',': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1027 t.kind = TOK.Comma; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1028 goto Lcommon; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1029 case '$': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1030 t.kind = TOK.Dollar; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1031 Lcommon: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1032 ++p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1033 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1034 return; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1035 case '#': |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1036 return scanSpecialTokenSequence(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1037 default: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1038 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1039 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1040 assert(p == t.start); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1041 assert(*p == c); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1042 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1043 // TODO: consider moving isidbeg() and isdigit() up. |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1044 if (isidbeg(c)) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1045 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1046 if (c == 'r' && p[1] == '"' && ++p) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1047 return scanRawStringLiteral(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1048 if (c == 'x' && p[1] == '"') |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1049 return scanHexStringLiteral(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1050 version(D2) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1051 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1052 if (c == 'q' && p[1] == '"') |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1053 return scanDelimitedStringLiteral(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1054 if (c == 'q' && p[1] == '{') |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1055 return scanTokenStringLiteral(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1056 } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1057 // Scan identifier. |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1058 Lidentifier: |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1059 do |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1060 { c = *++p; } |
550
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
1061 while (isident(c) || !isascii(c) && isUnicodeAlpha()) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1062 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1063 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1064 |
502
4e14cd1b24da
Refactored code and added modules related to tabulated Identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
499
diff
changeset
|
1065 auto id = IdTable.lookup(t.srcText); |
769
5e3ef1b2011c
Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
764
diff
changeset
|
1066 t.kind = id.kind; |
502
4e14cd1b24da
Refactored code and added modules related to tabulated Identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
499
diff
changeset
|
1067 t.ident = id; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1068 |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1069 if (t.kind == TOK.Identifier || t.isKeyword) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1070 return; |
497
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
1071 else if (t.isSpecialToken) |
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
1072 finalizeSpecialToken(t); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1073 else if (t.kind == TOK.EOF) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1074 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1075 tail = &t; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1076 assert(t.srcText == "__EOF__"); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1077 } |
497
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
1078 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1079 assert(0, "unexpected token type: " ~ Token.toString(t.kind)); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1080 return; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1081 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1082 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1083 if (isdigit(c)) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1084 return scanNumber(t); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1085 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1086 // Check for EOF |
497
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
1087 if (isEOF(c)) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1088 { |
497
0ffcc4ff82f3
Refactored a few things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
496
diff
changeset
|
1089 assert(isEOF(*p), *p~""); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1090 t.kind = TOK.EOF; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1091 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1092 tail = &t; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1093 assert(t.start == t.end); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1094 return; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1095 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1096 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1097 if (!isascii(c)) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1098 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1099 c = decodeUTF8(); |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1100 if (isUniAlpha(c)) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1101 goto Lidentifier; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1102 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1103 |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1104 error(t.start, MID.IllegalCharacter, cast(dchar)c); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1105 |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1106 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1107 t.kind = TOK.Illegal; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
1108 t.setWhitespaceFlag(); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1109 t.dchar_ = c; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1110 t.end = p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1111 return; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1112 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1113 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1114 /// Scans a block comment. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1115 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1116 /// BlockComment := "/*" AnyChar* "*/" |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1117 void scanBlockComment(ref Token t) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1118 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1119 assert(p[-1] == '/' && *p == '*'); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1120 auto tokenLineNum = lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1121 auto tokenLineBegin = lineBegin; |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1122 Loop: |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1123 while (1) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1124 { |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1125 switch (*++p) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1126 { |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1127 case '*': |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1128 if (p[1] != '/') |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1129 continue; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1130 p += 2; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1131 break Loop; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1132 case '\r': |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1133 if (p[1] == '\n') |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1134 ++p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1135 case '\n': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1136 assert(isNewlineEnd(p)); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1137 ++lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1138 setLineBegin(p+1); |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1139 break; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1140 default: |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1141 if (!isascii(*p)) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1142 { |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1143 if (isUnicodeNewlineChar(decodeUTF8())) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1144 goto case '\n'; |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1145 } |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1146 else if (isEOF(*p)) |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1147 { |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1148 error(tokenLineNum, tokenLineBegin, t.start, MID.UnterminatedBlockComment); |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1149 break Loop; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1150 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1151 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1152 } |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1153 t.kind = TOK.Comment; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
1154 t.setWhitespaceFlag(); |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1155 t.end = p; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1156 return; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1157 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1158 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1159 /// Scans a nested comment. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1160 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1161 /// NestedComment := "/+" (AnyChar* | NestedComment) "+/" |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1162 void scanNestedComment(ref Token t) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1163 { |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1164 assert(p[-1] == '/' && *p == '+'); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1165 auto tokenLineNum = lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1166 auto tokenLineBegin = lineBegin; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1167 uint level = 1; |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1168 Loop: |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1169 while (1) |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1170 { |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1171 switch (*++p) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1172 { |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1173 case '/': |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1174 if (p[1] == '+') |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1175 ++p, ++level; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1176 continue; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1177 case '+': |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1178 if (p[1] != '/') |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1179 continue; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1180 ++p; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1181 if (--level != 0) |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1182 continue; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1183 ++p; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1184 break Loop; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1185 case '\r': |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1186 if (p[1] == '\n') |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1187 ++p; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1188 case '\n': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1189 assert(isNewlineEnd(p)); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1190 ++lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1191 setLineBegin(p+1); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1192 continue; |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1193 default: |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1194 if (!isascii(*p)) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1195 { |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1196 if (isUnicodeNewlineChar(decodeUTF8())) |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1197 goto case '\n'; |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1198 } |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1199 else if (isEOF(*p)) |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1200 { |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1201 error(tokenLineNum, tokenLineBegin, t.start, MID.UnterminatedNestedComment); |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1202 break Loop; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1203 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1204 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1205 } |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1206 t.kind = TOK.Comment; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
1207 t.setWhitespaceFlag(); |
498
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1208 t.end = p; |
49c201b5c465
Refactored scanners for block and nested comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
497
diff
changeset
|
1209 return; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1210 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
1211 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1212 /// Scans the postfix character of a string literal. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1213 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1214 /// PostfixChar := "c" | "w" | "d" |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1215 char scanPostfix() |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1216 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1217 assert(p[-1] == '"' || p[-1] == '`' || |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1218 { version(D2) return p[-1] == '}'; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1219 else return 0; }() |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1220 ); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1221 switch (*p) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1222 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1223 case 'c': |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1224 case 'w': |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1225 case 'd': |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1226 return *p++; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1227 default: |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1228 return 0; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1229 } |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1230 assert(0); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1231 } |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1232 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1233 /// Scans a normal string literal. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1234 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1235 /// NormalStringLiteral := "\"" Char* "\"" |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1236 void scanNormalStringLiteral(ref Token t) |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1237 { |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1238 assert(*p == '"'); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1239 auto tokenLineNum = lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1240 auto tokenLineBegin = lineBegin; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1241 t.kind = TOK.String; |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1242 char[] buffer; |
384
f7ce725e79c3
Refactored scanNormalStringLiteral().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
383
diff
changeset
|
1243 uint c; |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1244 while (1) |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1245 { |
384
f7ce725e79c3
Refactored scanNormalStringLiteral().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
383
diff
changeset
|
1246 c = *++p; |
f7ce725e79c3
Refactored scanNormalStringLiteral().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
383
diff
changeset
|
1247 switch (c) |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1248 { |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1249 case '"': |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1250 ++p; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1251 t.pf = scanPostfix(); |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1252 Lreturn: |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1253 t.str = buffer ~ '\0'; |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1254 t.end = p; |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1255 return; |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1256 case '\\': |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1257 bool isBinary; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1258 c = scanEscapeSequence(isBinary); |
384
f7ce725e79c3
Refactored scanNormalStringLiteral().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
383
diff
changeset
|
1259 --p; |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1260 if (isascii(c) || isBinary) |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1261 buffer ~= c; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1262 else |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1263 encodeUTF8(buffer, c); |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1264 continue; |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1265 case '\r': |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1266 if (p[1] == '\n') |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1267 ++p; |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1268 case '\n': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1269 assert(isNewlineEnd(p)); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1270 c = '\n'; // Convert Newline to \n. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1271 ++lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1272 setLineBegin(p+1); |
384
f7ce725e79c3
Refactored scanNormalStringLiteral().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
383
diff
changeset
|
1273 break; |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1274 case 0, _Z_: |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1275 error(tokenLineNum, tokenLineBegin, t.start, MID.UnterminatedString); |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1276 goto Lreturn; |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1277 default: |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1278 if (!isascii(c)) |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1279 { |
384
f7ce725e79c3
Refactored scanNormalStringLiteral().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
383
diff
changeset
|
1280 c = decodeUTF8(); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1281 if (isUnicodeNewlineChar(c)) |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1282 goto case '\n'; |
384
f7ce725e79c3
Refactored scanNormalStringLiteral().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
383
diff
changeset
|
1283 encodeUTF8(buffer, c); |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1284 continue; |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1285 } |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1286 } |
424
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
1287 assert(isascii(c)); |
384
f7ce725e79c3
Refactored scanNormalStringLiteral().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
383
diff
changeset
|
1288 buffer ~= c; |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1289 } |
67
996065105910
- Fix: variadic arguments are local to the variadic function. Parameters are converted to an array of strings first and then passed to the constructor of class Problem.
aziz
parents:
66
diff
changeset
|
1290 assert(0); |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1291 } |
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
1292 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1293 /// Scans a character literal. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1294 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1295 /// CharLiteral := "'" Char "'" |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1296 void scanCharacterLiteral(ref Token t) |
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1297 { |
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1298 assert(*p == '\''); |
46 | 1299 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1300 t.kind = TOK.CharLiteral; |
46 | 1301 switch (*p) |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1302 { |
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1303 case '\\': |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1304 bool notused; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1305 t.dchar_ = scanEscapeSequence(notused); |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1306 break; |
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1307 case '\'': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1308 error(t.start, MID.EmptyCharacterLiteral); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1309 break; |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1310 default: |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1311 if (isEndOfLine(p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1312 break; |
46 | 1313 uint c = *p; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1314 if (!isascii(c)) |
49 | 1315 c = decodeUTF8(); |
46 | 1316 t.dchar_ = c; |
1317 ++p; | |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1318 } |
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1319 |
46 | 1320 if (*p == '\'') |
1321 ++p; | |
1322 else | |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1323 error(t.start, MID.UnterminatedCharacterLiteral); |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1324 t.end = p; |
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1325 } |
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
1326 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1327 /// Scans a raw string literal. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1328 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1329 /// RawStringLiteral := "r\"" AnyChar* "\"" | "`" AnyChar* "`" |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1330 void scanRawStringLiteral(ref Token t) |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1331 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1332 assert(*p == '`' || *p == '"' && p[-1] == 'r'); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1333 auto tokenLineNum = lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1334 auto tokenLineBegin = lineBegin; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1335 t.kind = TOK.String; |
33 | 1336 uint delim = *p; |
1337 char[] buffer; | |
1338 uint c; | |
1339 while (1) | |
1340 { | |
1341 c = *++p; | |
1342 switch (c) | |
1343 { | |
1344 case '\r': | |
1345 if (p[1] == '\n') | |
1346 ++p; | |
1347 case '\n': | |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1348 assert(isNewlineEnd(p)); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1349 c = '\n'; // Convert Newline to '\n'. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1350 ++lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1351 setLineBegin(p+1); |
52
f65a83c27638
- Fixed the raw string literal scanner. Newlines weren't copied to the buffer. Converting LS and PS to '\n' as well.
aziz
parents:
51
diff
changeset
|
1352 break; |
33 | 1353 case '`': |
1354 case '"': | |
1355 if (c == delim) | |
1356 { | |
1357 ++p; | |
1358 t.pf = scanPostfix(); | |
1359 Lreturn: | |
1360 t.str = buffer ~ '\0'; | |
1361 t.end = p; | |
1362 return; | |
1363 } | |
1364 break; | |
1365 case 0, _Z_: | |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1366 error(tokenLineNum, tokenLineBegin, t.start, |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1367 delim == 'r' ? MID.UnterminatedRawString : MID.UnterminatedBackQuoteString); |
33 | 1368 goto Lreturn; |
1369 default: | |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1370 if (!isascii(c)) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1371 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1372 c = decodeUTF8(); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1373 if (isUnicodeNewlineChar(c)) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1374 goto case '\n'; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1375 encodeUTF8(buffer, c); |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1376 continue; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1377 } |
33 | 1378 } |
424
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
1379 assert(isascii(c)); |
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
1380 buffer ~= c; |
33 | 1381 } |
1382 assert(0); | |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1383 } |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1384 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1385 /// Scans a hexadecimal string literal. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1386 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1387 /// HexStringLiteral := "x\"" (HexChar HexChar)* "\"" |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1388 void scanHexStringLiteral(ref Token t) |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1389 { |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1390 assert(p[0] == 'x' && p[1] == '"'); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1391 t.kind = TOK.String; |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1392 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1393 auto tokenLineNum = lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1394 auto tokenLineBegin = lineBegin; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1395 |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1396 uint c; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1397 ubyte[] buffer; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1398 ubyte h; // hex number |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1399 uint n; // number of hex digits |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1400 |
53
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1401 ++p; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1402 assert(*p == '"'); |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1403 while (1) |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1404 { |
53
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1405 c = *++p; |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1406 switch (c) |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1407 { |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1408 case '"': |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1409 if (n & 1) |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1410 error(tokenLineNum, tokenLineBegin, t.start, MID.OddNumberOfDigitsInHexString); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1411 ++p; |
53
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1412 t.pf = scanPostfix(); |
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1413 Lreturn: |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1414 t.str = cast(string) (buffer ~= 0); |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1415 t.end = p; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1416 return; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1417 case '\r': |
53
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1418 if (p[1] == '\n') |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1419 ++p; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1420 case '\n': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1421 assert(isNewlineEnd(p)); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1422 ++lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1423 setLineBegin(p+1); |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1424 continue; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1425 default: |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1426 if (ishexad(c)) |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1427 { |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1428 if (c <= '9') |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1429 c -= '0'; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1430 else if (c <= 'F') |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1431 c -= 'A' - 10; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1432 else |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1433 c -= 'a' - 10; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1434 |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1435 if (n & 1) |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1436 { |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1437 h <<= 4; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1438 h |= c; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1439 buffer ~= h; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1440 } |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1441 else |
112 | 1442 h = cast(ubyte)c; |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1443 ++n; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1444 continue; |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1445 } |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1446 else if (isspace(c)) |
419
89e40d43065d
Added new member 'p_newl' to Lexer and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
418
diff
changeset
|
1447 continue; // Skip spaces. |
496
5a607597dc22
Improved error recovery in the Parser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
494
diff
changeset
|
1448 else if (isEOF(c)) |
53
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1449 { |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1450 error(tokenLineNum, tokenLineBegin, t.start, MID.UnterminatedHexString); |
53
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1451 t.pf = 0; |
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1452 goto Lreturn; |
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1453 } |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1454 else |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1455 { |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1456 auto errorAt = p; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1457 if (!isascii(c)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1458 { |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1459 c = decodeUTF8(); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1460 if (isUnicodeNewlineChar(c)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1461 goto case '\n'; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1462 } |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1463 error(errorAt, MID.NonHexCharInHexString, cast(dchar)c); |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1464 } |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1465 } |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1466 } |
53
1786c2825491
- Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents:
52
diff
changeset
|
1467 assert(0); |
31
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1468 } |
94f09f4e988e
- Added struct for strings to Token with 'pf' = postfix.
aziz
parents:
30
diff
changeset
|
1469 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1470 version(DDoc) |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1471 { |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1472 /// Scans a delimited string literal. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1473 void scanDelimitedStringLiteral(ref Token t); |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1474 /// Scans a token string literal. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1475 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1476 /// TokenStringLiteral := "q{" Token* "}" |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1477 void scanTokenStringLiteral(ref Token t); |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1478 } |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1479 else |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1480 version(D2) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1481 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1482 void scanDelimitedStringLiteral(ref Token t) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1483 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1484 assert(p[0] == 'q' && p[1] == '"'); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1485 t.kind = TOK.String; |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1486 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1487 auto tokenLineNum = lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1488 auto tokenLineBegin = lineBegin; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1489 |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1490 char[] buffer; |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1491 dchar opening_delim = 0, // 0 if no nested delimiter or '[', '(', '<', '{' |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1492 closing_delim; // Will be ']', ')', '>', '}, |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1493 // the first character of an identifier or |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1494 // any other Unicode/ASCII character. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1495 char[] str_delim; // Identifier delimiter. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1496 uint level = 1; // Counter for nestable delimiters. |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1497 |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1498 ++p; ++p; // Skip q" |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1499 uint c = *p; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1500 switch (c) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1501 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1502 case '(': |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1503 opening_delim = c; |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1504 closing_delim = ')'; // c + 1 |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1505 break; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1506 case '[', '<', '{': |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1507 opening_delim = c; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1508 closing_delim = c + 2; // Get to closing counterpart. Feature of ASCII table. |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1509 break; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1510 default: |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1511 dchar scanNewline() |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1512 { |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1513 switch (*p) |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1514 { |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1515 case '\r': |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1516 if (p[1] == '\n') |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1517 ++p; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1518 case '\n': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1519 assert(isNewlineEnd(p)); |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1520 ++p; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1521 ++lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1522 setLineBegin(p); |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1523 return '\n'; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1524 default: |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1525 if (isUnicodeNewline(p)) |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1526 { |
419
89e40d43065d
Added new member 'p_newl' to Lexer and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
418
diff
changeset
|
1527 ++p; ++p; |
89e40d43065d
Added new member 'p_newl' to Lexer and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
418
diff
changeset
|
1528 goto case '\n'; |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1529 } |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1530 } |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1531 return 0; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1532 } |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1533 // Skip leading newlines: |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1534 while (scanNewline() != 0) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1535 {} |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1536 assert(!isNewline(p)); |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1537 |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1538 char* begin = p; |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1539 c = *p; |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1540 closing_delim = c; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1541 // TODO: Check for non-printable characters? |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1542 if (!isascii(c)) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1543 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1544 closing_delim = decodeUTF8(); |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1545 if (!isUniAlpha(closing_delim)) |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1546 break; // Not an identifier. |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1547 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1548 else if (!isidbeg(c)) |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1549 break; // Not an identifier. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1550 |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1551 // Parse Identifier + EndOfLine |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1552 do |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1553 { c = *++p; } |
550
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
1554 while (isident(c) || !isascii(c) && isUnicodeAlpha()) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1555 // Store identifier |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1556 str_delim = begin[0..p-begin]; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1557 // Scan newline |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1558 if (scanNewline() == '\n') |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1559 --p; // Go back one because of "c = *++p;" in main loop. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1560 else |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1561 { |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1562 // TODO: error(p, MID.ExpectedNewlineAfterIdentDelim); |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1563 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1564 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1565 |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1566 bool checkStringDelim(char* p) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1567 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1568 assert(str_delim.length != 0); |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1569 if (buffer[$-1] == '\n' && // Last character copied to buffer must be '\n'. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1570 end-p >= str_delim.length && // Check remaining length. |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1571 p[0..str_delim.length] == str_delim) // Compare. |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1572 return true; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1573 return false; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1574 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1575 |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1576 while (1) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1577 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1578 c = *++p; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1579 switch (c) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1580 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1581 case '\r': |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1582 if (p[1] == '\n') |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1583 ++p; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1584 case '\n': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1585 assert(isNewlineEnd(p)); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1586 c = '\n'; // Convert Newline to '\n'. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1587 ++lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1588 setLineBegin(p+1); |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1589 break; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1590 case 0, _Z_: |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1591 // TODO: error(tokenLineNum, tokenLineBegin, t.start, MID.UnterminatedDelimitedString); |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1592 goto Lreturn3; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1593 default: |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1594 if (!isascii(c)) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1595 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1596 auto begin = p; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1597 c = decodeUTF8(); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1598 if (isUnicodeNewlineChar(c)) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1599 goto case '\n'; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1600 if (c == closing_delim) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1601 { |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1602 if (str_delim.length) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1603 { |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1604 if (checkStringDelim(begin)) |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1605 { |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1606 p = begin + str_delim.length; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1607 goto Lreturn2; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1608 } |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1609 } |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1610 else |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1611 { |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1612 assert(level == 1); |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1613 --level; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1614 goto Lreturn; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1615 } |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1616 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1617 encodeUTF8(buffer, c); |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1618 continue; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1619 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1620 else |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1621 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1622 if (c == opening_delim) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1623 ++level; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1624 else if (c == closing_delim) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1625 { |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1626 if (str_delim.length) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1627 { |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1628 if (checkStringDelim(p)) |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1629 { |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1630 p += str_delim.length; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1631 goto Lreturn2; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1632 } |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1633 } |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1634 else if (--level == 0) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1635 goto Lreturn; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1636 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1637 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1638 } |
424
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
1639 assert(isascii(c)); |
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
1640 buffer ~= c; |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1641 } |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1642 Lreturn: // Character delimiter. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1643 assert(c == closing_delim); |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1644 assert(level == 0); |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1645 ++p; // Skip closing delimiter. |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1646 Lreturn2: // String delimiter. |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1647 if (*p == '"') |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1648 ++p; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1649 else |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1650 { |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1651 // TODO: error(p, MID.ExpectedDblQuoteAfterDelim, str_delim.length ? str_delim : closing_delim~""); |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1652 } |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1653 |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1654 t.pf = scanPostfix(); |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1655 Lreturn3: // Error. |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1656 t.str = buffer ~ '\0'; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1657 t.end = p; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1658 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1659 |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1660 void scanTokenStringLiteral(ref Token t) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1661 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1662 assert(p[0] == 'q' && p[1] == '{'); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1663 t.kind = TOK.String; |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1664 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1665 auto tokenLineNum = lineNum; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1666 auto tokenLineBegin = lineBegin; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1667 |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1668 // A guard against changes to particular members: |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1669 // this.lineNum_hline and this.errorPath |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1670 ++inTokenString; |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1671 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1672 uint lineNum = this.lineNum; |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1673 uint level = 1; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1674 |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1675 ++p; ++p; // Skip q{ |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1676 |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1677 auto prev_t = &t; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1678 Token* token; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1679 while (1) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1680 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1681 token = new Token; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1682 scan(*token); |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1683 // Save the tokens in a doubly linked list. |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1684 // Could be useful for various tools. |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1685 token.prev = prev_t; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1686 prev_t.next = token; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1687 prev_t = token; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1688 switch (token.kind) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1689 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1690 case TOK.LBrace: |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1691 ++level; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1692 continue; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1693 case TOK.RBrace: |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1694 if (--level == 0) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1695 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1696 t.tok_str = t.next; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1697 t.next = null; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1698 break; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1699 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1700 continue; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1701 case TOK.EOF: |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1702 // TODO: error(tokenLineNum, tokenLineBegin, t.start, MID.UnterminatedTokenString); |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1703 t.tok_str = t.next; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1704 t.next = token; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1705 break; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1706 default: |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1707 continue; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1708 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1709 break; // Exit loop. |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1710 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1711 |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1712 assert(token.kind == TOK.RBrace || token.kind == TOK.EOF); |
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1713 assert(token.kind == TOK.RBrace && t.next is null || |
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1714 token.kind == TOK.EOF && t.next !is null); |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1715 |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1716 char[] buffer; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1717 // token points to } or EOF |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
1718 if (token.kind == TOK.EOF) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1719 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1720 t.end = token.start; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1721 buffer = t.srcText[2..$].dup ~ '\0'; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1722 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1723 else |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1724 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1725 // Assign to buffer before scanPostfix(). |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1726 t.end = p; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1727 buffer = t.srcText[2..$-1].dup ~ '\0'; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1728 t.pf = scanPostfix(); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1729 t.end = p; // Assign again because of postfix. |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1730 } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1731 // Convert newlines to '\n'. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1732 if (lineNum != this.lineNum) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1733 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1734 assert(buffer[$-1] == '\0'); |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1735 uint i, j; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1736 for (; i < buffer.length; ++i) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1737 switch (buffer[i]) |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1738 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1739 case '\r': |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1740 if (buffer[i+1] == '\n') |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1741 ++i; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1742 case '\n': |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1743 assert(isNewlineEnd(buffer.ptr + i)); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1744 buffer[j++] = '\n'; // Convert Newline to '\n'. |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1745 break; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1746 default: |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1747 if (isUnicodeNewline(buffer.ptr + i)) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1748 { |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1749 ++i; ++i; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1750 goto case '\n'; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1751 } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1752 buffer[j++] = buffer[i]; // Copy. |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1753 } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1754 buffer.length = j; // Adjust length. |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1755 } |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1756 assert(buffer[$-1] == '\0'); |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1757 t.str = buffer; |
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1758 |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1759 --inTokenString; |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1760 } |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1761 } // version(D2) |
383
6a5fc22cae34
Implemented scanner for new string literals and applied some fixes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
382
diff
changeset
|
1762 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1763 /// Scans an escape sequence. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1764 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1765 /// EscapeSequence := "\" (Octal{1,3} | ("x" Hex{2}) | |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1766 /// ("u" Hex{4}) | ("U" Hex{8}) | |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1767 /// "'" | "\"" | "\\" | "?" | "a" | |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1768 /// "b" | "f" | "n" | "r" | "t" | "v") |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1769 /// Params: |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1770 /// isBinary = set to true for octal and hexadecimal escapes. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1771 /// Returns: the escape value. |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1772 dchar scanEscapeSequence(ref bool isBinary) |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1773 out(result) |
510
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
1774 { assert(isValidChar(result)); } |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1775 body |
45 | 1776 { |
386
392a0068fc61
Refactored code related to scanning escape sequences.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
384
diff
changeset
|
1777 assert(*p == '\\'); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1778 |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1779 auto sequenceStart = p; // Used for error reporting. |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1780 |
386
392a0068fc61
Refactored code related to scanning escape sequences.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
384
diff
changeset
|
1781 ++p; |
45 | 1782 uint c = char2ev(*p); |
82
fc645fb2fe72
- scanEscapeSequences() doesn't return 0xFFFF as an error value anymore, because it is a valid codepoint usable by the user.
aziz
parents:
71
diff
changeset
|
1783 if (c) |
fc645fb2fe72
- scanEscapeSequences() doesn't return 0xFFFF as an error value anymore, because it is a valid codepoint usable by the user.
aziz
parents:
71
diff
changeset
|
1784 { |
46 | 1785 ++p; |
45 | 1786 return c; |
46 | 1787 } |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1788 |
45 | 1789 uint digits = 2; |
1790 | |
1791 switch (*p) | |
1792 { | |
1793 case 'x': | |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1794 isBinary = true; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1795 case_Unicode: |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1796 assert(c == 0); |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1797 assert(digits == 2 || digits == 4 || digits == 8); |
45 | 1798 while (1) |
1799 { | |
1800 ++p; | |
1801 if (ishexad(*p)) | |
1802 { | |
1803 c *= 16; | |
1804 if (*p <= '9') | |
48 | 1805 c += *p - '0'; |
45 | 1806 else if (*p <= 'F') |
48 | 1807 c += *p - 'A' + 10; |
45 | 1808 else |
48 | 1809 c += *p - 'a' + 10; |
82
fc645fb2fe72
- scanEscapeSequences() doesn't return 0xFFFF as an error value anymore, because it is a valid codepoint usable by the user.
aziz
parents:
71
diff
changeset
|
1810 |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1811 if (--digits == 0) |
82
fc645fb2fe72
- scanEscapeSequences() doesn't return 0xFFFF as an error value anymore, because it is a valid codepoint usable by the user.
aziz
parents:
71
diff
changeset
|
1812 { |
48 | 1813 ++p; |
510
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
1814 if (isValidChar(c)) |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1815 return c; // Return valid escape value. |
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1816 |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1817 error(sequenceStart, MID.InvalidUnicodeEscapeSequence, |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1818 sequenceStart[0..p-sequenceStart]); |
45 | 1819 break; |
48 | 1820 } |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1821 continue; |
45 | 1822 } |
432
3ead178e0662
Added message MID.InvalidUnicodeEscapeSequence.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
430
diff
changeset
|
1823 |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1824 error(sequenceStart, MID.InsufficientHexDigits, |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1825 sequenceStart[0..p-sequenceStart]); |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1826 break; |
424
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
1827 } |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1828 break; |
45 | 1829 case 'u': |
1830 digits = 4; | |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1831 goto case_Unicode; |
45 | 1832 case 'U': |
1833 digits = 8; | |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1834 goto case_Unicode; |
45 | 1835 default: |
48 | 1836 if (isoctal(*p)) |
45 | 1837 { |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1838 isBinary = true; |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1839 assert(c == 0); |
48 | 1840 c += *p - '0'; |
1841 ++p; | |
1842 if (!isoctal(*p)) | |
1843 return c; | |
1844 c *= 8; | |
1845 c += *p - '0'; | |
1846 ++p; | |
1847 if (!isoctal(*p)) | |
1848 return c; | |
1849 c *= 8; | |
1850 c += *p - '0'; | |
1851 ++p; | |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
1852 return c & 0xFF; // Return valid escape value. |
48 | 1853 } |
1854 else if(*p == '&') | |
1855 { | |
1856 if (isalpha(*++p)) | |
45 | 1857 { |
272
0bde32503976
- Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents:
249
diff
changeset
|
1858 auto begin = p; |
0bde32503976
- Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents:
249
diff
changeset
|
1859 while (isalnum(*++p)) |
0bde32503976
- Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents:
249
diff
changeset
|
1860 {} |
0bde32503976
- Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents:
249
diff
changeset
|
1861 |
0bde32503976
- Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents:
249
diff
changeset
|
1862 if (*p == ';') |
48 | 1863 { |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1864 // Pass entity excluding '&' and ';'. |
272
0bde32503976
- Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents:
249
diff
changeset
|
1865 c = entity2Unicode(begin[0..p - begin]); |
386
392a0068fc61
Refactored code related to scanning escape sequences.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
384
diff
changeset
|
1866 ++p; // Skip ; |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1867 if (c != 0xFFFF) |
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1868 return c; // Return valid escape value. |
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1869 else |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1870 error(sequenceStart, MID.UndefinedHTMLEntity, sequenceStart[0 .. p - sequenceStart]); |
272
0bde32503976
- Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents:
249
diff
changeset
|
1871 } |
0bde32503976
- Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents:
249
diff
changeset
|
1872 else |
432
3ead178e0662
Added message MID.InvalidUnicodeEscapeSequence.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
430
diff
changeset
|
1873 error(sequenceStart, MID.UnterminatedHTMLEntity, sequenceStart[0 .. p - sequenceStart]); |
45 | 1874 } |
48 | 1875 else |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1876 error(sequenceStart, MID.InvalidBeginHTMLEntity); |
45 | 1877 } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1878 else if (isEndOfLine(p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1879 error(sequenceStart, MID.UndefinedEscapeSequence, |
496
5a607597dc22
Improved error recovery in the Parser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
494
diff
changeset
|
1880 isEOF(*p) ? `\EOF` : `\NewLine`); |
45 | 1881 else |
386
392a0068fc61
Refactored code related to scanning escape sequences.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
384
diff
changeset
|
1882 { |
392a0068fc61
Refactored code related to scanning escape sequences.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
384
diff
changeset
|
1883 char[] str = `\`; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1884 if (isascii(c)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1885 str ~= *p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
1886 else |
387
ad0cbd1c8881
Undefined escape sequences are passed to error() now.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
386
diff
changeset
|
1887 encodeUTF8(str, decodeUTF8()); |
ad0cbd1c8881
Undefined escape sequences are passed to error() now.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
386
diff
changeset
|
1888 ++p; |
ad0cbd1c8881
Undefined escape sequences are passed to error() now.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
386
diff
changeset
|
1889 // TODO: check for unprintable character? |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1890 error(sequenceStart, MID.UndefinedEscapeSequence, str); |
386
392a0068fc61
Refactored code related to scanning escape sequences.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
384
diff
changeset
|
1891 } |
45 | 1892 } |
433
063cd016c913
Fixed some issues in scanEscapeSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
432
diff
changeset
|
1893 return REPLACEMENT_CHAR; // Error: return replacement character. |
45 | 1894 } |
1895 | |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1896 /// Scans a number literal. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1897 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1898 /// $(PRE |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1899 /// IntegerLiteral := (Dec|Hex|Bin|Oct)Suffix? |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1900 /// Dec := (0|[1-9][0-9_]*) |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1901 /// Hex := 0[xX][_]*[0-9a-zA-Z][0-9a-zA-Z_]* |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1902 /// Bin := 0[bB][_]*[01][01_]* |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1903 /// Oct := 0[0-7_]* |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1904 /// Suffix := (L[uU]?|[uU]L?) |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1905 /// ) |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
1906 /// Invalid: "0b_", "0x_", "._" etc. |
15 | 1907 void scanNumber(ref Token t) |
1908 { | |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1909 ulong ulong_; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1910 bool overflow; |
58 | 1911 bool isDecimal; |
57
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
1912 size_t digits; |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1913 |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1914 if (*p != '0') |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1915 goto LscanInteger; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1916 ++p; // skip zero |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1917 // check for xX bB ... |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1918 switch (*p) |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1919 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1920 case 'x','X': |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1921 goto LscanHex; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1922 case 'b','B': |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1923 goto LscanBinary; |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1924 case 'L': |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1925 if (p[1] == 'i') |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1926 goto LscanReal; // 0Li |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1927 break; // 0L |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1928 case '.': |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1929 if (p[1] == '.') |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1930 break; // 0.. |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1931 // 0. |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1932 case 'i','f','F', // Imaginary and float literal suffixes. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
1933 'e', 'E': // Float exponent. |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1934 goto LscanReal; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1935 default: |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1936 if (*p == '_') |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1937 goto LscanOctal; // 0_ |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1938 else if (isdigit(*p)) |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1939 { |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1940 if (*p == '8' || *p == '9') |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1941 goto Loctal_hasDecimalDigits; // 08 or 09 |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1942 else |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1943 goto Loctal_enter_loop; // 0[0-7] |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1944 } |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1945 } |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1946 |
323 | 1947 // Number 0 |
1948 assert(p[-1] == '0'); | |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
1949 assert(*p != '_' && !isdigit(*p)); |
323 | 1950 assert(ulong_ == 0); |
58 | 1951 isDecimal = true; |
1952 goto Lfinalize; | |
1953 | |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1954 LscanInteger: |
323 | 1955 assert(*p != 0 && isdigit(*p)); |
58 | 1956 isDecimal = true; |
323 | 1957 goto Lenter_loop_int; |
63 | 1958 while (1) |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1959 { |
63 | 1960 if (*++p == '_') |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1961 continue; |
63 | 1962 if (!isdigit(*p)) |
1963 break; | |
323 | 1964 Lenter_loop_int: |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1965 if (ulong_ < ulong.max/10 || (ulong_ == ulong.max/10 && *p <= '5')) |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1966 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1967 ulong_ *= 10; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1968 ulong_ += *p - '0'; |
58 | 1969 continue; |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1970 } |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1971 // Overflow: skip following digits. |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1972 overflow = true; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1973 while (isdigit(*++p)) {} |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1974 break; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1975 } |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1976 |
61 | 1977 // The number could be a float, so check overflow below. |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1978 switch (*p) |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1979 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1980 case '.': |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1981 if (p[1] != '.') |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1982 goto LscanReal; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1983 break; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1984 case 'L': |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1985 if (p[1] != 'i') |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1986 break; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1987 case 'i', 'f', 'F', 'e', 'E': |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1988 goto LscanReal; |
58 | 1989 default: |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1990 } |
58 | 1991 |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1992 if (overflow) |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
1993 error(t.start, MID.OverflowDecimalNumber); |
58 | 1994 |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1995 assert((isdigit(p[-1]) || p[-1] == '_') && !isdigit(*p) && *p != '_'); |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1996 goto Lfinalize; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1997 |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
1998 LscanHex: |
61 | 1999 assert(digits == 0); |
410
4d9ee8e60712
Added destructors for two particular tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
403
diff
changeset
|
2000 assert(*p == 'x' || *p == 'X'); |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2001 while (1) |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2002 { |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2003 if (*++p == '_') |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2004 continue; |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2005 if (!ishexad(*p)) |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2006 break; |
61 | 2007 ++digits; |
57
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
2008 ulong_ *= 16; |
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
2009 if (*p <= '9') |
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
2010 ulong_ += *p - '0'; |
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
2011 else if (*p <= 'F') |
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
2012 ulong_ += *p - 'A' + 10; |
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
2013 else |
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
2014 ulong_ += *p - 'a' + 10; |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2015 } |
57
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
2016 |
410
4d9ee8e60712
Added destructors for two particular tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
403
diff
changeset
|
2017 assert(ishexad(p[-1]) || p[-1] == '_' || p[-1] == 'x' || p[-1] == 'X'); |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2018 assert(!ishexad(*p) && *p != '_'); |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2019 |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2020 switch (*p) |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2021 { |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2022 case '.': |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2023 if (p[1] == '.') |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2024 break; |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2025 case 'p', 'P': |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2026 return scanHexReal(t); |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2027 default: |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2028 } |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2029 |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2030 if (digits == 0 || digits > 16) |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2031 error(t.start, digits == 0 ? MID.NoDigitsInHexNumber : MID.OverflowHexNumber); |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2032 |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2033 goto Lfinalize; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2034 |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
2035 LscanBinary: |
57
c0f1c8be3a47
- Added code for converting hex characters to binary numbers.
aziz
parents:
56
diff
changeset
|
2036 assert(digits == 0); |
410
4d9ee8e60712
Added destructors for two particular tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
403
diff
changeset
|
2037 assert(*p == 'b' || *p == 'B'); |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2038 while (1) |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2039 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2040 if (*++p == '0') |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2041 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2042 ++digits; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2043 ulong_ *= 2; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2044 } |
367
dda55fae37de
- ImportGraph.execute() can parse all modules depending on the imports of the root module.
aziz
parents:
365
diff
changeset
|
2045 else if (*p == '1') |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2046 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2047 ++digits; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2048 ulong_ *= 2; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2049 ulong_ += *p - '0'; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2050 } |
367
dda55fae37de
- ImportGraph.execute() can parse all modules depending on the imports of the root module.
aziz
parents:
365
diff
changeset
|
2051 else if (*p == '_') |
dda55fae37de
- ImportGraph.execute() can parse all modules depending on the imports of the root module.
aziz
parents:
365
diff
changeset
|
2052 continue; |
dda55fae37de
- ImportGraph.execute() can parse all modules depending on the imports of the root module.
aziz
parents:
365
diff
changeset
|
2053 else |
dda55fae37de
- ImportGraph.execute() can parse all modules depending on the imports of the root module.
aziz
parents:
365
diff
changeset
|
2054 break; |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2055 } |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2056 |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2057 if (digits == 0 || digits > 64) |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2058 error(t.start, digits == 0 ? MID.NoDigitsInBinNumber : MID.OverflowBinaryNumber); |
59
3e594725899a
- Issuing error when no digits were found in hex and binary numbers.
aziz
parents:
58
diff
changeset
|
2059 |
410
4d9ee8e60712
Added destructors for two particular tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
403
diff
changeset
|
2060 assert(p[-1] == '0' || p[-1] == '1' || p[-1] == '_' || p[-1] == 'b' || p[-1] == 'B', p[-1] ~ ""); |
367
dda55fae37de
- ImportGraph.execute() can parse all modules depending on the imports of the root module.
aziz
parents:
365
diff
changeset
|
2061 assert( !(*p == '0' || *p == '1' || *p == '_') ); |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2062 goto Lfinalize; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2063 |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
2064 LscanOctal: |
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
2065 assert(*p == '_'); |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2066 while (1) |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2067 { |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2068 if (*++p == '_') |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2069 continue; |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2070 if (!isoctal(*p)) |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2071 break; |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
2072 Loctal_enter_loop: |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2073 if (ulong_ < ulong.max/2 || (ulong_ == ulong.max/2 && *p <= '1')) |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2074 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2075 ulong_ *= 8; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2076 ulong_ += *p - '0'; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2077 continue; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2078 } |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2079 // Overflow: skip following digits. |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2080 overflow = true; |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
2081 while (isoctal(*++p)) {} |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2082 break; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2083 } |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2084 |
63 | 2085 bool hasDecimalDigits; |
2086 if (isdigit(*p)) | |
2087 { | |
396
0a4619735ce9
Applied fixes to Lexer, Parser and other classes.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
393
diff
changeset
|
2088 Loctal_hasDecimalDigits: |
63 | 2089 hasDecimalDigits = true; |
2090 while (isdigit(*++p)) {} | |
2091 } | |
2092 | |
2093 // The number could be a float, so check errors below. | |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2094 switch (*p) |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2095 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2096 case '.': |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2097 if (p[1] != '.') |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2098 goto LscanReal; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2099 break; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2100 case 'L': |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2101 if (p[1] != 'i') |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2102 break; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2103 case 'i', 'f', 'F', 'e', 'E': |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2104 goto LscanReal; |
63 | 2105 default: |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2106 } |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2107 |
63 | 2108 if (hasDecimalDigits) |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2109 error(t.start, MID.OctalNumberHasDecimals); |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2110 |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2111 if (overflow) |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2112 error(t.start, MID.OverflowOctalNumber); |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2113 // goto Lfinalize; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2114 |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2115 Lfinalize: |
58 | 2116 enum Suffix |
2117 { | |
2118 None = 0, | |
2119 Unsigned = 1, | |
60 | 2120 Long = 2 |
58 | 2121 } |
60 | 2122 |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2123 // Scan optional suffix: L, Lu, LU, u, uL, U or UL. |
58 | 2124 Suffix suffix; |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2125 while (1) |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2126 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2127 switch (*p) |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2128 { |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2129 case 'L': |
58 | 2130 if (suffix & Suffix.Long) |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2131 break; |
60 | 2132 suffix |= Suffix.Long; |
2133 ++p; | |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2134 continue; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2135 case 'u', 'U': |
58 | 2136 if (suffix & Suffix.Unsigned) |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2137 break; |
60 | 2138 suffix |= Suffix.Unsigned; |
2139 ++p; | |
2140 continue; | |
58 | 2141 default: |
2142 break; | |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2143 } |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2144 break; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2145 } |
58 | 2146 |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2147 // Determine type of Integer. |
58 | 2148 switch (suffix) |
2149 { | |
2150 case Suffix.None: | |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2151 if (ulong_ & 0x8000_0000_0000_0000) |
58 | 2152 { |
2153 if (isDecimal) | |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2154 error(t.start, MID.OverflowDecimalSign); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2155 t.kind = TOK.Uint64; |
58 | 2156 } |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2157 else if (ulong_ & 0xFFFF_FFFF_0000_0000) |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2158 t.kind = TOK.Int64; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2159 else if (ulong_ & 0x8000_0000) |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2160 t.kind = isDecimal ? TOK.Int64 : TOK.Uint32; |
58 | 2161 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2162 t.kind = TOK.Int32; |
58 | 2163 break; |
2164 case Suffix.Unsigned: | |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2165 if (ulong_ & 0xFFFF_FFFF_0000_0000) |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2166 t.kind = TOK.Uint64; |
58 | 2167 else |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2168 t.kind = TOK.Uint32; |
58 | 2169 break; |
2170 case Suffix.Long: | |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2171 if (ulong_ & 0x8000_0000_0000_0000) |
58 | 2172 { |
2173 if (isDecimal) | |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2174 error(t.start, MID.OverflowDecimalSign); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2175 t.kind = TOK.Uint64; |
58 | 2176 } |
2177 else | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2178 t.kind = TOK.Int64; |
58 | 2179 break; |
2180 case Suffix.Unsigned | Suffix.Long: | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2181 t.kind = TOK.Uint64; |
58 | 2182 break; |
2183 default: | |
2184 assert(0); | |
2185 } | |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2186 t.ulong_ = ulong_; |
15 | 2187 t.end = p; |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2188 return; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2189 LscanReal: |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2190 scanReal(t); |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2191 return; |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2192 } |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2193 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2194 /// Scans a floating point number literal. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2195 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2196 /// $(PRE |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2197 /// FloatLiteral := Float[fFL]?i? |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2198 /// Float := DecFloat | HexFloat |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2199 /// DecFloat := ([0-9][0-9_]*[.][0-9_]*DecExponent?) | |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2200 /// [.][0-9][0-9_]*DecExponent? | [0-9][0-9_]*DecExponent |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2201 /// DecExponent := [eE][+-]?[0-9][0-9_]* |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2202 /// HexFloat := 0[xX](HexDigits[.]HexDigits | |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2203 /// [.][0-9a-zA-Z]HexDigits? | |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2204 /// HexDigits)HexExponent |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2205 /// HexExponent := [pP][+-]?[0-9][0-9_]* |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2206 /// ) |
56
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2207 void scanReal(ref Token t) |
63af7ddf52e1
- Started properly implementing number scanner. Added stub for real numbers.
aziz
parents:
55
diff
changeset
|
2208 { |
63 | 2209 if (*p == '.') |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2210 { |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2211 assert(p[1] != '.'); |
63 | 2212 // This function was called by scan() or scanNumber(). |
2213 while (isdigit(*++p) || *p == '_') {} | |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2214 } |
63 | 2215 else |
2216 // This function was called by scanNumber(). | |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2217 assert(delegate () |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2218 { |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2219 switch (*p) |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2220 { |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2221 case 'L': |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2222 if (p[1] != 'i') |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2223 return false; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2224 case 'i', 'f', 'F', 'e', 'E': |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2225 return true; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2226 default: |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2227 } |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2228 return false; |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2229 }() |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2230 ); |
63 | 2231 |
2232 // Scan exponent. | |
2233 if (*p == 'e' || *p == 'E') | |
2234 { | |
2235 ++p; | |
2236 if (*p == '-' || *p == '+') | |
2237 ++p; | |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2238 if (isdigit(*p)) |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2239 while (isdigit(*++p) || *p == '_') {} |
63 | 2240 else |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2241 error(t.start, MID.FloatExpMustStartWithDigit); |
63 | 2242 } |
2243 | |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2244 // Copy whole number and remove underscores from buffer. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2245 char[] buffer = t.start[0..p-t.start].dup; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2246 uint j; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2247 foreach (c; buffer) |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2248 if (c != '_') |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2249 buffer[j++] = c; |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2250 buffer.length = j; // Adjust length. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2251 buffer ~= 0; // Terminate for C functions. |
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2252 |
63 | 2253 finalizeFloat(t, buffer); |
30 | 2254 } |
2255 | |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2256 /// Scans a hexadecimal floating point number literal. |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2257 void scanHexReal(ref Token t) |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2258 { |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2259 assert(*p == '.' || *p == 'p' || *p == 'P'); |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2260 MID mid; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2261 if (*p == '.') |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2262 while (ishexad(*++p) || *p == '_') |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2263 {} |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2264 // Decimal exponent is required. |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2265 if (*p != 'p' && *p != 'P') |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2266 { |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2267 mid = MID.HexFloatExponentRequired; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2268 goto Lerr; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2269 } |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2270 // Scan exponent |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2271 assert(*p == 'p' || *p == 'P'); |
393
fce1e6133dac
Applied fix to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
392
diff
changeset
|
2272 ++p; |
fce1e6133dac
Applied fix to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
392
diff
changeset
|
2273 if (*p == '+' || *p == '-') |
fce1e6133dac
Applied fix to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
392
diff
changeset
|
2274 ++p; |
fce1e6133dac
Applied fix to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
392
diff
changeset
|
2275 if (!isdigit(*p)) |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2276 { |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2277 mid = MID.HexFloatExpMustStartWithDigit; |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2278 goto Lerr; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2279 } |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2280 while (isdigit(*++p) || *p == '_') |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2281 {} |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2282 // Copy whole number and remove underscores from buffer. |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2283 char[] buffer = t.start[0..p-t.start].dup; |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2284 uint j; |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2285 foreach (c; buffer) |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2286 if (c != '_') |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2287 buffer[j++] = c; |
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2288 buffer.length = j; // Adjust length. |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2289 buffer ~= 0; // Terminate for C functions. |
63 | 2290 finalizeFloat(t, buffer); |
2291 return; | |
2292 Lerr: | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2293 t.kind = TOK.Float32; |
63 | 2294 t.end = p; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2295 error(t.start, mid); |
63 | 2296 } |
2297 | |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2298 /// Sets the value of the token. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2299 /// Params: |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2300 /// t = receives the value. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2301 /// buffer = the well-formed float number. |
63 | 2302 void finalizeFloat(ref Token t, string buffer) |
2303 { | |
389
c4bfceab7246
Applied fixes and improvements to hex float scanner.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
388
diff
changeset
|
2304 assert(buffer[$-1] == 0); |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2305 // Float number is well-formed. Check suffixes and do conversion. |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2306 switch (*p) |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2307 { |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2308 case 'f', 'F': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2309 t.kind = TOK.Float32; |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2310 t.float_ = strtof(buffer.ptr, null); |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2311 ++p; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2312 break; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2313 case 'L': |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2314 t.kind = TOK.Float80; |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2315 t.real_ = strtold(buffer.ptr, null); |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2316 ++p; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2317 break; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2318 default: |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2319 t.kind = TOK.Float64; |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2320 t.double_ = strtod(buffer.ptr, null); |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2321 } |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2322 if (*p == 'i') |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2323 { |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2324 ++p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2325 t.kind += 3; // Switch to imaginary counterpart. |
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2326 assert(t.kind == TOK.Imaginary32 || |
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2327 t.kind == TOK.Imaginary64 || |
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2328 t.kind == TOK.Imaginary80); |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2329 } |
391
33b566df6af4
Migrated project to Tango.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
390
diff
changeset
|
2330 if (errno() == ERANGE) |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2331 error(t.start, MID.OverflowFloatNumber); |
62
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2332 t.end = p; |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2333 } |
96af5653acef
- Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents:
61
diff
changeset
|
2334 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2335 /// Scans a special token sequence. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2336 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2337 /// SpecialTokenSequence := "#line" Integer Filespec? EndOfLine |
360
b6a3755eba94
- Renamed scanSpecialToken() to scanSpecialTokenSequence().
aziz
parents:
350
diff
changeset
|
2338 void scanSpecialTokenSequence(ref Token t) |
30 | 2339 { |
47
8aa37a78937b
- Properly implemented scanner for normal string literals.
aziz
parents:
46
diff
changeset
|
2340 assert(*p == '#'); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2341 t.kind = TOK.HashLine; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
2342 t.setWhitespaceFlag(); |
323 | 2343 |
30 | 2344 MID mid; |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2345 char* errorAtColumn = p; |
763
f26f13b5a3a3
Fixed code in scanSpecialTokenSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
762
diff
changeset
|
2346 char* tokenEnd = ++p; |
30 | 2347 |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2348 if (!(p[0] == 'l' && p[1] == 'i' && p[2] == 'n' && p[3] == 'e')) |
51 | 2349 { |
370 | 2350 mid = MID.ExpectedIdentifierSTLine; |
51 | 2351 goto Lerr; |
2352 } | |
323 | 2353 p += 3; |
763
f26f13b5a3a3
Fixed code in scanSpecialTokenSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
762
diff
changeset
|
2354 tokenEnd = p + 1; |
323 | 2355 |
388
ae154eceba65
Applied some fixes to scanning and printing #line tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
387
diff
changeset
|
2356 // TODO: #line58"path/file" is legal. Require spaces? |
ae154eceba65
Applied some fixes to scanning and printing #line tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
387
diff
changeset
|
2357 // State.Space could be used for that purpose. |
323 | 2358 enum State |
388
ae154eceba65
Applied some fixes to scanning and printing #line tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
387
diff
changeset
|
2359 { /+Space,+/ Integer, Filespec, End } |
51 | 2360 |
388
ae154eceba65
Applied some fixes to scanning and printing #line tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
387
diff
changeset
|
2361 State state = State.Integer; |
763
f26f13b5a3a3
Fixed code in scanSpecialTokenSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
762
diff
changeset
|
2362 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2363 while (!isEndOfLine(++p)) |
51 | 2364 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2365 if (isspace(*p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2366 continue; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2367 if (state == State.Integer) |
323 | 2368 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2369 if (!isdigit(*p)) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2370 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2371 errorAtColumn = p; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2372 mid = MID.ExpectedIntegerAfterSTLine; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2373 goto Lerr; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2374 } |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2375 t.tokLineNum = new Token; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2376 scan(*t.tokLineNum); |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2377 tokenEnd = p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2378 if (t.tokLineNum.kind != TOK.Int32 && t.tokLineNum.kind != TOK.Uint32) |
323 | 2379 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2380 errorAtColumn = t.tokLineNum.start; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2381 mid = MID.ExpectedIntegerAfterSTLine; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2382 goto Lerr; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2383 } |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2384 --p; // Go one back because scan() advanced p past the integer. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2385 state = State.Filespec; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2386 } |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2387 else if (state == State.Filespec && *p == '"') |
763
f26f13b5a3a3
Fixed code in scanSpecialTokenSequence().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
762
diff
changeset
|
2388 { // MID.ExpectedFilespec is deprecated. |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2389 // if (*p != '"') |
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2390 // { |
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2391 // errorAtColumn = p; |
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2392 // mid = MID.ExpectedFilespec; |
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2393 // goto Lerr; |
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2394 // } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2395 t.tokLineFilespec = new Token; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2396 t.tokLineFilespec.start = p; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2397 t.tokLineFilespec.kind = TOK.Filespec; |
552
3bc7801c207e
Refactored the way how tokens are flagged as whitespace.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
550
diff
changeset
|
2398 t.tokLineFilespec.setWhitespaceFlag(); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2399 while (*++p != '"') |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2400 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2401 if (isEndOfLine(p)) |
323 | 2402 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2403 errorAtColumn = t.tokLineFilespec.start; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2404 mid = MID.UnterminatedFilespec; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2405 t.tokLineFilespec.end = p; |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2406 tokenEnd = p; |
323 | 2407 goto Lerr; |
2408 } | |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2409 isascii(*p) || decodeUTF8(); |
323 | 2410 } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2411 auto start = t.tokLineFilespec.start +1; // +1 skips '"' |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2412 t.tokLineFilespec.str = start[0 .. p - start]; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2413 t.tokLineFilespec.end = p + 1; |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2414 tokenEnd = p + 1; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2415 state = State.End; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2416 } |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2417 else/+ if (state == State.End)+/ |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2418 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2419 mid = MID.UnterminatedSpecialToken; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2420 goto Lerr; |
323 | 2421 } |
51 | 2422 } |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2423 assert(isEndOfLine(p)); |
323 | 2424 |
388
ae154eceba65
Applied some fixes to scanning and printing #line tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
387
diff
changeset
|
2425 if (state == State.Integer) |
51 | 2426 { |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2427 errorAtColumn = p; |
388
ae154eceba65
Applied some fixes to scanning and printing #line tokens.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
387
diff
changeset
|
2428 mid = MID.ExpectedIntegerAfterSTLine; |
30 | 2429 goto Lerr; |
2430 } | |
2431 | |
392
bb935c6f9b7a
Applied fixes and improvements to the Lexer class.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
391
diff
changeset
|
2432 // Evaluate #line only when not in token string. |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2433 if (!inTokenString && t.tokLineNum) |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2434 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2435 this.lineNum_hline = this.lineNum - t.tokLineNum.uint_ + 1; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2436 if (t.tokLineFilespec) |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
2437 newFilePath(t.tokLineFilespec.str); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2438 } |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2439 p = tokenEnd; |
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2440 t.end = tokenEnd; |
30 | 2441 |
2442 return; | |
2443 Lerr: | |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2444 p = tokenEnd; |
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2445 t.end = tokenEnd; |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2446 error(errorAtColumn, mid); |
371 | 2447 } |
2448 | |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2449 /// Inserts an empty dummy token (TOK.Empty) before t. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2450 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2451 /// Useful in the parsing phase for representing a node in the AST |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2452 /// that doesn't consume an actual token from the source text. |
414
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2453 Token* insertEmptyTokenBefore(Token* t) |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2454 { |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2455 assert(t !is null && t.prev !is null); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2456 assert(text.ptr <= t.start && t.start < end, Token.toString(t.kind)); |
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2457 assert(text.ptr <= t.end && t.end <= end, Token.toString(t.kind)); |
414
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2458 |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2459 auto prev_t = t.prev; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2460 auto new_t = new Token; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2461 new_t.kind = TOK.Empty; |
414
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2462 new_t.start = new_t.end = prev_t.end; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2463 // Link in new token. |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2464 prev_t.next = new_t; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2465 new_t.prev = prev_t; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2466 new_t.next = t; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2467 t.prev = new_t; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2468 return new_t; |
9c69615a4876
Added method insertEmptyTokenBefore() to Lexer. Made some fixes, too.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
413
diff
changeset
|
2469 } |
105 | 2470 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2471 /// Returns the error line number. |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2472 uint errorLineNumber(uint lineNum) |
239
7911f6a92e6e
- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents:
223
diff
changeset
|
2473 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2474 return lineNum - this.lineNum_hline; |
18
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
2475 } |
c48d2125f1e2
- Moved code for scanning character literals to separate function.
aziz
parents:
17
diff
changeset
|
2476 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2477 /// Forwards error parameters. |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2478 void error(char* columnPos, MID mid, ...) |
207 | 2479 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2480 error_(this.lineNum, this.lineBegin, columnPos, mid, _arguments, _argptr); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2481 } |
207 | 2482 |
798
c24be8d4f6ab
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
797
diff
changeset
|
2483 /// ditto |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2484 void error(uint lineNum, char* lineBegin, char* columnPos, MID mid, ...) |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2485 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2486 error_(lineNum, lineBegin, columnPos, mid, _arguments, _argptr); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2487 } |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2488 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2489 /// Creates an error report and appends it to a list. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2490 /// Params: |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2491 /// lineNum = the line number. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2492 /// lineBegin = points to the first character of the current line. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2493 /// columnPos = points to the character where the error is located. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2494 /// mid = the message ID. |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2495 void error_(uint lineNum, char* lineBegin, char* columnPos, MID mid, |
762
c909a3d3fa52
Fixed vararg issue with gdc.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
755
diff
changeset
|
2496 TypeInfo[] _arguments, Arg _argptr) |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2497 { |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2498 lineNum = this.errorLineNumber(lineNum); |
607
2ed1e6d638cd
Making use of struct NewlineData.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
601
diff
changeset
|
2499 auto errorPath = this.filePaths.setPath; |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2500 auto location = new Location(errorPath, lineNum, lineBegin, columnPos); |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2501 auto msg = Format(_arguments, _argptr, GetMsg(mid)); |
520
f203c5248d0b
Added 'compile' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
519
diff
changeset
|
2502 auto error = new LexerError(location, msg); |
f203c5248d0b
Added 'compile' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
519
diff
changeset
|
2503 errors ~= error; |
f203c5248d0b
Added 'compile' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
519
diff
changeset
|
2504 if (infoMan !is null) |
f203c5248d0b
Added 'compile' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
519
diff
changeset
|
2505 infoMan ~= error; |
207 | 2506 } |
2507 | |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2508 /// Scans the whole source text until EOF is encountered. |
487
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2509 void scanAll() |
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2510 { |
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2511 while (nextToken() != TOK.EOF) |
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2512 {} |
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2513 } |
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2514 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2515 /// Returns the first token of the source text. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2516 /// This can be the EOF token. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2517 /// Structure: HEAD -> Newline -> First Token |
487
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2518 Token* firstToken() |
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2519 { |
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2520 return this.head.next.next; |
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2521 } |
bccca748d745
Added 'tokenize' command.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
485
diff
changeset
|
2522 |
510
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2523 /// Returns true if str is a valid D identifier. |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2524 static bool isIdentifierString(char[] str) |
365 | 2525 { |
510
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2526 if (str.length == 0 || isdigit(str[0])) |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2527 return false; |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2528 size_t idx; |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2529 do |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2530 { |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2531 auto c = dil.Unicode.decode(str, idx); |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2532 if (c == ERROR_CHAR || !(isident(c) || !isascii(c) && isUniAlpha(c))) |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2533 return false; |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2534 } while (idx < str.length) |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2535 return true; |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2536 } |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2537 |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2538 /// Returns true if str is a keyword or a special token (__FILE__, __LINE__ etc.) |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2539 static bool isReservedIdentifier(char[] str) |
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2540 { |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2541 if (!isIdentifierString(str)) |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2542 return false; // str is not a valid identifier. |
382
66477017cb95
Applied fixes to isNonReservedIdentifier().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
371
diff
changeset
|
2543 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2544 auto id = IdTable.inStatic(str); |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2545 if (id is null || id.kind == TOK.Identifier) |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2546 return false; // str is not in the table or a normal identifier. |
365 | 2547 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2548 return true; |
424
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
2549 } |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2550 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2551 /// Returns true if the current character to be decoded is |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2552 /// a Unicode alpha character. |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2553 /// |
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2554 /// The current pointer 'p' is not advanced if false is returned. |
550
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2555 bool isUnicodeAlpha() |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2556 { |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2557 assert(!isascii(*p), "check for ASCII char before calling decodeUTF8()."); |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2558 char* p = this.p; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2559 dchar d = *p; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2560 ++p; // Move to second byte. |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2561 // Error if second byte is not a trail byte. |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2562 if (!isTrailByte(*p)) |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2563 return false; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2564 // Check for overlong sequences. |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2565 switch (d) |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2566 { |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2567 case 0xE0, 0xF0, 0xF8, 0xFC: |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2568 if ((*p & d) == 0x80) |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2569 return false; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2570 default: |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2571 if ((d & 0xFE) == 0xC0) // 1100000x |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2572 return false; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2573 } |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2574 const char[] checkNextByte = "if (!isTrailByte(*++p))" |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2575 " return false;"; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2576 const char[] appendSixBits = "d = (d << 6) | *p & 0b0011_1111;"; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2577 // Decode |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2578 if ((d & 0b1110_0000) == 0b1100_0000) |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2579 { |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2580 d &= 0b0001_1111; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2581 mixin(appendSixBits); |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2582 } |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2583 else if ((d & 0b1111_0000) == 0b1110_0000) |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2584 { |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2585 d &= 0b0000_1111; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2586 mixin(appendSixBits ~ |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2587 checkNextByte ~ appendSixBits); |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2588 } |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2589 else if ((d & 0b1111_1000) == 0b1111_0000) |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2590 { |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2591 d &= 0b0000_0111; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2592 mixin(appendSixBits ~ |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2593 checkNextByte ~ appendSixBits ~ |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2594 checkNextByte ~ appendSixBits); |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2595 } |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2596 else |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2597 return false; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2598 |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2599 assert(isTrailByte(*p)); |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2600 if (!isValidChar(d) || !isUniAlpha(d)) |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2601 return false; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2602 // Only advance pointer if this is a Unicode alpha character. |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2603 this.p = p; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2604 return true; |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2605 } |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2606 |
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2607 /// Decodes the next UTF-8 sequence. |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2608 dchar decodeUTF8() |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2609 { |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2610 assert(!isascii(*p), "check for ASCII char before calling decodeUTF8()."); |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2611 char* p = this.p; |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2612 dchar d = *p; |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2613 |
426
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2614 ++p; // Move to second byte. |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2615 // Error if second byte is not a trail byte. |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2616 if (!isTrailByte(*p)) |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2617 goto Lerr2; |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2618 |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2619 // Check for overlong sequences. |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2620 switch (d) |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2621 { |
426
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2622 case 0xE0, // 11100000 100xxxxx |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2623 0xF0, // 11110000 1000xxxx |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2624 0xF8, // 11111000 10000xxx |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2625 0xFC: // 11111100 100000xx |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2626 if ((*p & d) == 0x80) |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2627 goto Lerr; |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2628 default: |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2629 if ((d & 0xFE) == 0xC0) // 1100000x |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2630 goto Lerr; |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2631 } |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2632 |
426
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2633 const char[] checkNextByte = "if (!isTrailByte(*++p))" |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2634 " goto Lerr2;"; |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2635 const char[] appendSixBits = "d = (d << 6) | *p & 0b0011_1111;"; |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2636 |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2637 // Decode |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2638 if ((d & 0b1110_0000) == 0b1100_0000) |
550
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2639 { // 110xxxxx 10xxxxxx |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2640 d &= 0b0001_1111; |
426
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2641 mixin(appendSixBits); |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2642 } |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2643 else if ((d & 0b1111_0000) == 0b1110_0000) |
550
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2644 { // 1110xxxx 10xxxxxx 10xxxxxx |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2645 d &= 0b0000_1111; |
426
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2646 mixin(appendSixBits ~ |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2647 checkNextByte ~ appendSixBits); |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2648 } |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2649 else if ((d & 0b1111_1000) == 0b1111_0000) |
550
082126d78f90
Fixed lexing of identifiers.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
538
diff
changeset
|
2650 { // 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2651 d &= 0b0000_0111; |
426
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2652 mixin(appendSixBits ~ |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2653 checkNextByte ~ appendSixBits ~ |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2654 checkNextByte ~ appendSixBits); |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2655 } |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2656 else |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2657 // 5 and 6 byte UTF-8 sequences are not allowed yet. |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2658 // 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2659 // 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2660 goto Lerr; |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2661 |
426
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2662 assert(isTrailByte(*p)); |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2663 |
510
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2664 if (!isValidChar(d)) |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2665 { |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2666 Lerr: |
426
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2667 // Three cases: |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2668 // *) the UTF-8 sequence was successfully decoded but the resulting |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2669 // character is invalid. |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2670 // p points to last trail byte in the sequence. |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2671 // *) the UTF-8 sequence is overlong. |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2672 // p points to second byte in the sequence. |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2673 // *) the UTF-8 sequence has more than 4 bytes or starts with |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2674 // a trail byte. |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2675 // p points to second byte in the sequence. |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2676 assert(isTrailByte(*p)); |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2677 // Move to next ASCII character or lead byte of a UTF-8 sequence. |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2678 while (p < (end-1) && isTrailByte(*p)) |
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2679 ++p; |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2680 --p; |
426
3f7790d3f9d6
Improved decodeUTF8().
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
425
diff
changeset
|
2681 assert(!isTrailByte(p[1])); |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2682 Lerr2: |
430
e6c759e151cd
Fixed a few things regarding encoding/decoding UTF-8 sequences.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
426
diff
changeset
|
2683 d = REPLACEMENT_CHAR; |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2684 error(this.p, MID.InvalidUTF8Sequence, formatBytes(this.p, p)); |
425
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2685 } |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2686 |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2687 this.p = p; |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2688 return d; |
6bf936bf3356
Added own UTF-8 decoding function.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
424
diff
changeset
|
2689 } |
424
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
2690 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2691 /// Encodes the character d and appends it to str. |
519 | 2692 static void encodeUTF8(ref char[] str, dchar d) |
48 | 2693 { |
430
e6c759e151cd
Fixed a few things regarding encoding/decoding UTF-8 sequences.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
426
diff
changeset
|
2694 assert(!isascii(d), "check for ASCII char before calling encodeUTF8()."); |
510
dd3ce87b3569
Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
507
diff
changeset
|
2695 assert(isValidChar(d), "check if character is valid before calling encodeUTF8()."); |
430
e6c759e151cd
Fixed a few things regarding encoding/decoding UTF-8 sequences.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
426
diff
changeset
|
2696 |
519 | 2697 char[6] b = void; |
48 | 2698 if (d < 0x800) |
2699 { | |
2700 b[0] = 0xC0 | (d >> 6); | |
2701 b[1] = 0x80 | (d & 0x3F); | |
2702 str ~= b[0..2]; | |
2703 } | |
2704 else if (d < 0x10000) | |
2705 { | |
2706 b[0] = 0xE0 | (d >> 12); | |
2707 b[1] = 0x80 | ((d >> 6) & 0x3F); | |
2708 b[2] = 0x80 | (d & 0x3F); | |
2709 str ~= b[0..3]; | |
2710 } | |
2711 else if (d < 0x200000) | |
2712 { | |
2713 b[0] = 0xF0 | (d >> 18); | |
2714 b[1] = 0x80 | ((d >> 12) & 0x3F); | |
2715 b[2] = 0x80 | ((d >> 6) & 0x3F); | |
2716 b[3] = 0x80 | (d & 0x3F); | |
2717 str ~= b[0..4]; | |
2718 } | |
424
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
2719 /+ // There are no 5 and 6 byte UTF-8 sequences yet. |
48 | 2720 else if (d < 0x4000000) |
2721 { | |
2722 b[0] = 0xF8 | (d >> 24); | |
2723 b[1] = 0x80 | ((d >> 18) & 0x3F); | |
2724 b[2] = 0x80 | ((d >> 12) & 0x3F); | |
2725 b[3] = 0x80 | ((d >> 6) & 0x3F); | |
2726 b[4] = 0x80 | (d & 0x3F); | |
2727 str ~= b[0..5]; | |
2728 } | |
2729 else if (d < 0x80000000) | |
2730 { | |
2731 b[0] = 0xFC | (d >> 30); | |
2732 b[1] = 0x80 | ((d >> 24) & 0x3F); | |
2733 b[2] = 0x80 | ((d >> 18) & 0x3F); | |
2734 b[3] = 0x80 | ((d >> 12) & 0x3F); | |
2735 b[4] = 0x80 | ((d >> 6) & 0x3F); | |
2736 b[5] = 0x80 | (d & 0x3F); | |
2737 str ~= b[0..6]; | |
2738 } | |
424
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
2739 +/ |
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
2740 else |
bb3cb00feeb2
Applied some fixes to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
422
diff
changeset
|
2741 assert(0); |
48 | 2742 } |
789
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2743 |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2744 /// Formats the bytes between start and end. |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2745 /// Returns: e.g.: abc -> \x61\x62\x63 |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2746 static char[] formatBytes(char* start, char* end) |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2747 { |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2748 auto strLen = end-start; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2749 const formatLen = `\xXX`.length; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2750 char[] result = new char[strLen*formatLen]; // Reserve space. |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2751 result.length = 0; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2752 foreach (c; cast(ubyte[])start[0..strLen]) |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2753 result ~= Format("\\x{:X}", c); |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2754 return result; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2755 } |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2756 |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2757 /// Searches for an invalid UTF-8 sequence in str. |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2758 /// Returns: a formatted string of the invalid sequence (e.g. \xC0\x80). |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2759 static string findInvalidUTF8Sequence(string str) |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2760 { |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2761 char* p = str.ptr, end = p + str.length; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2762 while (p < end) |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2763 { |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2764 if (decode(p, end) == ERROR_CHAR) |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2765 { |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2766 auto begin = p; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2767 // Skip trail-bytes. |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2768 while (++p < end && isTrailByte(*p)) |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2769 {} |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2770 return Lexer.formatBytes(begin, p); |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2771 } |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2772 } |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2773 assert(p == end); |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2774 return ""; |
c1d5cfd7aa44
Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
776
diff
changeset
|
2775 } |
5 | 2776 } |
39
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
2777 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2778 /// Tests the lexer with a list of tokens. |
39
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
2779 unittest |
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
2780 { |
391
33b566df6af4
Migrated project to Tango.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
390
diff
changeset
|
2781 Stdout("Testing Lexer.\n"); |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2782 struct Pair |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2783 { |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2784 char[] tokenText; |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2785 TOK kind; |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2786 } |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2787 static Pair[] pairs = [ |
494
9a7ca8c56e59
Refactored a few things in the Parser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
490
diff
changeset
|
2788 {"#!äöüß", TOK.Shebang}, {"\n", TOK.Newline}, |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2789 {"//çay", TOK.Comment}, {"\n", TOK.Newline}, |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2790 {"&", TOK.AndBinary}, |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2791 {"/*çağ*/", TOK.Comment}, {"&&", TOK.AndLogical}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2792 {"/+çak+/", TOK.Comment}, {"&=", TOK.AndAssign}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2793 {">", TOK.Greater}, {"+", TOK.Plus}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2794 {">=", TOK.GreaterEqual}, {"++", TOK.PlusPlus}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2795 {">>", TOK.RShift}, {"+=", TOK.PlusAssign}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2796 {">>=", TOK.RShiftAssign}, {"-", TOK.Minus}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2797 {">>>", TOK.URShift}, {"--", TOK.MinusMinus}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2798 {">>>=", TOK.URShiftAssign}, {"-=", TOK.MinusAssign}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2799 {"<", TOK.Less}, {"=", TOK.Assign}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2800 {"<=", TOK.LessEqual}, {"==", TOK.Equal}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2801 {"<>", TOK.LorG}, {"~", TOK.Tilde}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2802 {"<>=", TOK.LorEorG}, {"~=", TOK.CatAssign}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2803 {"<<", TOK.LShift}, {"*", TOK.Mul}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2804 {"<<=", TOK.LShiftAssign}, {"*=", TOK.MulAssign}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2805 {"!", TOK.Not}, {"/", TOK.Div}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2806 {"!=", TOK.NotEqual}, {"/=", TOK.DivAssign}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2807 {"!<", TOK.UorGorE}, {"^", TOK.Xor}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2808 {"!>", TOK.UorLorE}, {"^=", TOK.XorAssign}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2809 {"!<=", TOK.UorG}, {"%", TOK.Mod}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2810 {"!>=", TOK.UorL}, {"%=", TOK.ModAssign}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2811 {"!<>", TOK.UorE}, {"(", TOK.LParen}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2812 {"!<>=", TOK.Unordered}, {")", TOK.RParen}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2813 {".", TOK.Dot}, {"[", TOK.LBracket}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2814 {"..", TOK.Slice}, {"]", TOK.RBracket}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2815 {"...", TOK.Ellipses}, {"{", TOK.LBrace}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2816 {"|", TOK.OrBinary}, {"}", TOK.RBrace}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2817 {"||", TOK.OrLogical}, {":", TOK.Colon}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2818 {"|=", TOK.OrAssign}, {";", TOK.Semicolon}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2819 {"?", TOK.Question}, {",", TOK.Comma}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2820 {"$", TOK.Dollar}, {"cam", TOK.Identifier}, |
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2821 {"çay", TOK.Identifier}, {".0", TOK.Float64}, |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2822 {"0", TOK.Int32}, {"\n", TOK.Newline}, |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2823 {"\r", TOK.Newline}, {"\r\n", TOK.Newline}, |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2824 {"\u2028", TOK.Newline}, {"\u2029", TOK.Newline} |
40 | 2825 ]; |
39
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
2826 |
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
2827 char[] src; |
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
2828 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2829 // Join all token texts into a single string. |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2830 foreach (i, pair; pairs) |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2831 if (pair.kind == TOK.Comment && pair.tokenText[1] == '/' || // Line comment. |
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2832 pair.kind == TOK.Shebang) |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2833 { |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2834 assert(pairs[i+1].kind == TOK.Newline); // Must be followed by a newline. |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2835 src ~= pair.tokenText; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2836 } |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2837 else |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2838 src ~= pair.tokenText ~ " "; |
39
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
2839 |
764
4579e8505d5e
Fixed unittests and removed dil.File.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
763
diff
changeset
|
2840 auto lx = new Lexer(new SourceText("", src)); |
350 | 2841 auto token = lx.getTokens(); |
39
69b940398d7b
- Added unittest to test correct parsing of operator tokens.
aziz
parents:
38
diff
changeset
|
2842 |
350 | 2843 uint i; |
2844 assert(token == lx.head); | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2845 assert(token.next.kind == TOK.Newline); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2846 token = token.next.next; |
350 | 2847 do |
2848 { | |
413
0fd78fdcb982
Added an alternative scan() method to class Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
412
diff
changeset
|
2849 assert(i < pairs.length); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2850 assert(token.srcText == pairs[i].tokenText, Format("Scanned '{0}' but expected '{1}'", token.srcText, pairs[i].tokenText)); |
350 | 2851 ++i; |
2852 token = token.next; | |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2853 } while (token.kind != TOK.EOF) |
41 | 2854 } |
55 | 2855 |
797
cf2ad5df025c
Added documentation comments.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
789
diff
changeset
|
2856 /// Tests the Lexer's peek() method. |
67
996065105910
- Fix: variadic arguments are local to the variadic function. Parameters are converted to an array of strings first and then passed to the constructor of class Problem.
aziz
parents:
66
diff
changeset
|
2857 unittest |
996065105910
- Fix: variadic arguments are local to the variadic function. Parameters are converted to an array of strings first and then passed to the constructor of class Problem.
aziz
parents:
66
diff
changeset
|
2858 { |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2859 Stdout("Testing method Lexer.peek()\n"); |
764
4579e8505d5e
Fixed unittests and removed dil.File.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
763
diff
changeset
|
2860 auto sourceText = new SourceText("", "unittest { }"); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2861 auto lx = new Lexer(sourceText, null); |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2862 |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2863 auto next = lx.head; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2864 lx.peek(next); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2865 assert(next.kind == TOK.Newline); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2866 lx.peek(next); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2867 assert(next.kind == TOK.Unittest); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2868 lx.peek(next); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2869 assert(next.kind == TOK.LBrace); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2870 lx.peek(next); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2871 assert(next.kind == TOK.RBrace); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2872 lx.peek(next); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2873 assert(next.kind == TOK.EOF); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2874 |
764
4579e8505d5e
Fixed unittests and removed dil.File.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
763
diff
changeset
|
2875 lx = new Lexer(new SourceText("", "")); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2876 next = lx.head; |
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2877 lx.peek(next); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2878 assert(next.kind == TOK.Newline); |
485
ea8c7459f1c4
Changed a lot of things in the Lexer.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
464
diff
changeset
|
2879 lx.peek(next); |
679
ff6971637f88
Renamed Token member type to kind.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
607
diff
changeset
|
2880 assert(next.kind == TOK.EOF); |
422
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2881 } |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2882 |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2883 unittest |
ad7977fe315a
Added support for column numbers in error messages.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
420
diff
changeset
|
2884 { |
67
996065105910
- Fix: variadic arguments are local to the variadic function. Parameters are converted to an array of strings first and then passed to the constructor of class Problem.
aziz
parents:
66
diff
changeset
|
2885 // Numbers unittest |
68
7eb83dd38901
- Simplified suffix rule and added a few more numbers to unittest.
aziz
parents:
67
diff
changeset
|
2886 // 0L 0ULi 0_L 0_UL 0x0U 0x0p2 0_Fi 0_e2 0_F 0_i |
7eb83dd38901
- Simplified suffix rule and added a few more numbers to unittest.
aziz
parents:
67
diff
changeset
|
2887 // 0u 0U 0uL 0UL 0L 0LU 0Lu |
7eb83dd38901
- Simplified suffix rule and added a few more numbers to unittest.
aziz
parents:
67
diff
changeset
|
2888 // 0Li 0f 0F 0fi 0Fi 0i |
7eb83dd38901
- Simplified suffix rule and added a few more numbers to unittest.
aziz
parents:
67
diff
changeset
|
2889 // 0b_1_LU 0b1000u |
7eb83dd38901
- Simplified suffix rule and added a few more numbers to unittest.
aziz
parents:
67
diff
changeset
|
2890 // 0x232Lu |
67
996065105910
- Fix: variadic arguments are local to the variadic function. Parameters are converted to an array of strings first and then passed to the constructor of class Problem.
aziz
parents:
66
diff
changeset
|
2891 } |