annotate trunk/src/Lexer.d @ 323:6259fb93e3dd

- Rewrote scanSpecialToken(). - Added Filespec to enum TOK. - Added another nested struct to union in Token. - Deleted ExpectedNormalStringLiteral from enum MID and added ExpectedFilespec and UnterminatedFilespec. Adapted lang_de.d and lang_en.d accordingly. - Fixed some code in scanNumber(). Added some assert statements.
author aziz
date Tue, 21 Aug 2007 15:26:00 +0000
parents 29c33ce6c5bb
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
1 /++
8ba2570de175 Initial import.
aziz
parents:
diff changeset
2 Author: Aziz Köksal
249
32d354584b28 - Upgraded license notices to GPL3.
aziz
parents: 239
diff changeset
3 License: GPL3
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
4 +/
8ba2570de175 Initial import.
aziz
parents:
diff changeset
5 module Lexer;
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
6 import Token;
69
24db7c5522d5 - Added module Information for compiler messages like warnings, info and errors to the user.
aziz
parents: 68
diff changeset
7 import Information;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
8 import Keywords;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
9 import Identifier;
54
e55bd2270f94 - Relocated messages table to a separate module.
aziz
parents: 53
diff changeset
10 import Messages;
272
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
11 import HtmlEntities;
2
81c6cc33f5c8 - Initializing ptable with a precomputed array literal.
aziz
parents: 1
diff changeset
12 import std.stdio;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
13 import std.utf;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
14 import std.uni;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
15 import std.c.stdlib;
66
8e84db78ad55 - Added support for variadic arguments in error messages.
aziz
parents: 65
diff changeset
16 import std.string;
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
17
11
dffcdaa7c47a - Added Unicode line and paragraph separators.
aziz
parents: 10
diff changeset
18 const char[3] LS = \u2028;
dffcdaa7c47a - Added Unicode line and paragraph separators.
aziz
parents: 10
diff changeset
19 const char[3] PS = \u2029;
dffcdaa7c47a - Added Unicode line and paragraph separators.
aziz
parents: 10
diff changeset
20
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
21 const dchar LSd = 0x2028;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
22 const dchar PSd = 0x2029;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
23
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
24 const uint _Z_ = 26; /// Control+Z
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
25
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
26 class Lexer
8ba2570de175 Initial import.
aziz
parents:
diff changeset
27 {
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
28 Token* head; /// The head of the doubly linked token list.
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
29 Token* token; /// Points to the current token in the token list.
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
30 string text;
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
31 char* p; /// Points to the current character in the source text.
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
32 char* end; /// Points one character past the end of the source text.
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
33
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
34 uint loc = 1; /// line of code
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
35
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
36 char[] fileName;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
37
69
24db7c5522d5 - Added module Information for compiler messages like warnings, info and errors to the user.
aziz
parents: 68
diff changeset
38 Information[] errors;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
39
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
40 // bool reportErrors;
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
41
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
42 Identifier[string] idtable;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
43
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
44 this(string text, string fileName)
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
45 {
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
46 this.fileName = fileName;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
47
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
48 this.text = text;
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
49 if (text[$-1] != 0)
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
50 {
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
51 this.text.length = this.text.length + 1;
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
52 this.text[$-1] = 0;
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
53 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
54
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
55 this.p = this.text.ptr;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
56 this.end = this.p + this.text.length;
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
57 // this.reportErrors = true;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
58 loadKeywords();
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
59
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
60 this.head = new Token;
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
61 this.head.type = TOK.HEAD;
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
62 this.token = this.head;
315
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
63 scanShebang();
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
64 }
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
65
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
66 void scanShebang()
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
67 {
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
68 if (*p == '#' && p[1] == '!')
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
69 {
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
70 Token* t = new Token;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
71 t.start = p;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
72 t.type = TOK.Shebang;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
73 ++p;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
74 while (1)
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
75 {
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
76 switch (*++p)
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
77 {
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
78 case '\r':
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
79 if (p[1] == '\n')
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
80 ++p;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
81 case '\n':
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
82 ++loc;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
83 if (p[-1] == '\r')
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
84 t.end = p-1;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
85 else
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
86 t.end = p;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
87 break;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
88 case LS[0]:
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
89 t.end = p;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
90 if (p[1] == LS[1] && (p[2] == LS[2] || p[2] == PS[2]))
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
91 {
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
92 ++p; ++p;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
93 ++loc;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
94 }
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
95 break;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
96 case 0, _Z_:
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
97 t.end = p;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
98 break;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
99 default:
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
100 continue;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
101 }
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
102 break; // Exit loop.
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
103 }
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
104 this.head.next = t;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
105 t.prev = this.head;
29c33ce6c5bb - Added method scanShebang to class Lexer.
aziz
parents: 309
diff changeset
106 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
107 }
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
108
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
109 public void scan(out Token t)
207
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
110 in
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
111 {
207
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
112 assert(text.ptr <= p && p < end);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
113 }
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
114 out
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
115 {
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
116 assert(text.ptr <= t.start && t.start < end);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
117 assert(text.ptr < t.end && t.end <= end, std.string.format(t.type));
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
118 }
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
119 body
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
120 {
10
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
121 uint c = *p;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
122
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
123 while (1)
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
124 {
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
125 t.start = p;
16
476e8e55c1d4 - Added Whitespace to the character properties table.
aziz
parents: 15
diff changeset
126
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
127 if (c == 0 || c == _Z_)
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
128 {
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
129 assert(*p == 0 || *p == _Z_);
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
130 t.type = TOK.EOF;
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
131 t.end = p;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
132 assert(t.start == t.end);
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
133 return;
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
134 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
135
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
136 if (c == '\n')
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
137 {
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
138 c = *++p;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
139 ++loc;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
140 continue;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
141 }
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
142 else if (c == '\r')
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
143 {
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
144 c = *++p;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
145 if (c != '\n')
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
146 ++loc;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
147 continue;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
148 }
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
149 else if (c == LS[0] && p[1] == LS[1] && (p[2] == LS[2] || p[2] == PS[2]))
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
150 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
151 p += 3;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
152 c = *p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
153 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
154 }
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
155
13
e5211758b63c - Added isidbeg() function.
aziz
parents: 12
diff changeset
156 if (isidbeg(c))
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
157 {
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
158 if (c == 'r' && p[1] == '"' && ++p)
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
159 return scanRawStringLiteral(t);
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
160 if (c == 'x' && p[1] == '"')
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
161 return scanHexStringLiteral(t);
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
162 Lidentifier:
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
163 do
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
164 { c = *++p; }
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
165 while (isident(c) || c & 128 && isUniAlpha(decodeUTF8()))
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
166
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
167 t.end = p;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
168
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 64
diff changeset
169 string str = t.srcText;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
170 Identifier* id = str in idtable;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
171
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
172 if (!id)
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
173 {
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
174 idtable[str] = Identifier.Identifier(TOK.Identifier, str);
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
175 id = str in idtable;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
176 }
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
177 assert(id);
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
178 t.type = id.type;
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
179 return;
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
180 }
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
181
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
182 if (isdigit(c))
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
183 return scanNumber(t);
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
184
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
185 if (c == '/')
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
186 {
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
187 c = *++p;
14
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
188 switch(c)
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
189 {
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
190 case '=':
14
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
191 ++p;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
192 t.type = TOK.DivAssign;
14
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
193 t.end = p;
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
194 return;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
195 case '+':
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
196 uint level = 1;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
197 while (1)
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
198 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
199 c = *++p;
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
200 LswitchNC: // only jumped to from default case of next switch(c)
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
201 switch (c)
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
202 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
203 case '\r':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
204 if (p[1] == '\n')
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
205 ++p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
206 case '\n':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
207 ++loc;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
208 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
209 case 0, _Z_:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
210 error(MID.UnterminatedNestedComment);
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
211 goto LreturnNC;
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
212 default:
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
213 }
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
214
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
215 c <<= 8;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
216 c |= *++p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
217 switch (c)
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
218 {
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
219 case 0x2F2B: // /+
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
220 ++level;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
221 continue;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
222 case 0x2B2F: // +/
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
223 if (--level == 0)
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
224 {
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
225 ++p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
226 LreturnNC:
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
227 t.type = TOK.Comment;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
228 t.end = p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
229 return;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
230 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
231 continue;
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
232 case 0xE280: // LS[0..1] || PS[0..1]
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
233 if (p[1] == LS[2] || p[1] == PS[2])
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
234 {
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
235 ++loc;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
236 ++p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
237 }
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
238 continue;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
239 default:
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
240 c &= char.max;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
241 goto LswitchNC;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
242 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
243 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
244 case '*':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
245 while (1)
7
07e45c06a024 - Parsing nested comments correctly now.
aziz
parents: 5
diff changeset
246 {
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
247 c = *++p;
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
248 LswitchBC: // only jumped to from default case of next switch(c)
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
249 switch (c)
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
250 {
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
251 case '\r':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
252 if (p[1] == '\n')
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
253 ++p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
254 case '\n':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
255 ++loc;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
256 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
257 case 0, _Z_:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
258 error(MID.UnterminatedBlockComment);
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
259 goto LreturnBC;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
260 default:
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
261 }
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
262
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
263 c <<= 8;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
264 c |= *++p;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
265 switch (c)
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
266 {
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
267 case 0x2A2F: // */
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
268 ++p;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
269 LreturnBC:
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
270 t.type = TOK.Comment;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
271 t.end = p;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
272 return;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
273 case 0xE280: // LS[0..1] || PS[0..1]
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
274 if (p[1] == LS[2] || p[1] == PS[2])
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
275 {
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
276 ++loc;
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
277 ++p;
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
278 }
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
279 continue;
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
280 default:
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
281 c &= char.max;
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
282 goto LswitchBC;
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
283 }
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
284 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
285 assert(0);
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
286 case '/':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
287 while (1)
7
07e45c06a024 - Parsing nested comments correctly now.
aziz
parents: 5
diff changeset
288 {
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
289 c = *++p;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
290 switch (c)
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
291 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
292 case '\r':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
293 if (p[1] == '\n')
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
294 ++p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
295 case '\n':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
296 case 0, _Z_:
11
dffcdaa7c47a - Added Unicode line and paragraph separators.
aziz
parents: 10
diff changeset
297 break;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
298 case LS[0]:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
299 if (p[1] == LS[1] && (p[2] == LS[2] || p[2] == PS[2]))
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
300 break;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
301 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
302 default:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
303 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
304 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
305 t.type = TOK.Comment;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
306 t.end = p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
307 return;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
308 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
309 default:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
310 t.type = TOK.Div;
10
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
311 t.end = p;
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
312 return;
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
313 }
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
314 }
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
315
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
316 switch (c)
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
317 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
318 case '\'':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
319 return scanCharacterLiteral(t);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
320 case '`':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
321 return scanRawStringLiteral(t);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
322 case '"':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
323 return scanNormalStringLiteral(t);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
324 case '\\':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
325 char[] buffer;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
326 do
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
327 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
328 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
329 c = scanEscapeSequence();
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
330 if (c < 128)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
331 buffer ~= c;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
332 else
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
333 encodeUTF8(buffer, c);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
334 } while (*p == '\\')
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
335 buffer ~= 0;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
336 t.type = TOK.String;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
337 t.str = buffer;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
338 t.end = p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
339 return;
38
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
340 case '>': /* > >= >> >>= >>> >>>= */
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
341 c = *++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
342 switch (c)
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
343 {
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
344 case '=':
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
345 t.type = TOK.GreaterEqual;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
346 goto Lcommon;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
347 case '>':
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
348 if (p[1] == '>')
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
349 {
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
350 ++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
351 if (p[1] == '=')
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
352 { ++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
353 t.type = TOK.URShiftAssign;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
354 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
355 else
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
356 t.type = TOK.URShift;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
357 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
358 else if (p[1] == '=')
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
359 {
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
360 ++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
361 t.type = TOK.RShiftAssign;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
362 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
363 else
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
364 t.type = TOK.RShift;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
365 goto Lcommon;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
366 default:
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
367 t.type = TOK.Greater;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
368 goto Lcommon2;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
369 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
370 assert(0);
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
371 case '<': /* < <= <> <>= << <<= */
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
372 c = *++p;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
373 switch (c)
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
374 {
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
375 case '=':
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
376 t.type = TOK.LessEqual;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
377 goto Lcommon;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
378 case '<':
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
379 if (p[1] == '=') {
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
380 ++p;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
381 t.type = TOK.LShiftAssign;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
382 }
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
383 else
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
384 t.type = TOK.LShift;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
385 goto Lcommon;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
386 case '>':
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
387 if (p[1] == '=') {
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
388 ++p;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
389 t.type = TOK.LorEorG;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
390 }
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
391 else
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
392 t.type = TOK.LorG;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
393 goto Lcommon;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
394 default:
38
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
395 t.type = TOK.Less;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
396 goto Lcommon2;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
397 }
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
398 assert(0);
37
7f3bcb97d017 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 36
diff changeset
399 case '!': /* ! !< !> !<= !>= !<> !<>= */
35
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
400 c = *++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
401 switch (c)
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
402 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
403 case '<':
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
404 c = *++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
405 if (c == '>')
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
406 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
407 if (p[1] == '=') {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
408 ++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
409 t.type = TOK.Unordered;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
410 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
411 else
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
412 t.type = TOK.UorE;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
413 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
414 else if (c == '=')
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
415 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
416 t.type = TOK.UorG;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
417 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
418 else {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
419 t.type = TOK.UorGorE;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
420 goto Lcommon2;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
421 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
422 goto Lcommon;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
423 case '>':
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
424 if (p[1] == '=')
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
425 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
426 ++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
427 t.type = TOK.UorL;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
428 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
429 else
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
430 t.type = TOK.UorLorE;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
431 goto Lcommon;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
432 case '=':
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
433 t.type = TOK.NotEqual;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
434 goto Lcommon;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
435 default:
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
436 t.type = TOK.Not;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
437 goto Lcommon2;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
438 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
439 assert(0);
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
440 case '.': /* . .[0-9] .. ... */
22
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
441 if (p[1] == '.')
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
442 {
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
443 ++p;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
444 if (p[1] == '.') {
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
445 ++p;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
446 t.type = TOK.Ellipses;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
447 }
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
448 else
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
449 t.type = TOK.Slice;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
450 }
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
451 else if (isdigit(p[1]))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
452 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
453 return scanReal(t);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
454 }
22
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
455 else
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
456 t.type = TOK.Dot;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
457 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
458 case '|': /* | || |= */
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
459 c = *++p;
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
460 if (c == '=')
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
461 t.type = TOK.OrAssign;
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
462 else if (c == '|')
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
463 t.type = TOK.OrLogical;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
464 else {
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
465 t.type = TOK.OrBinary;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
466 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
467 }
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
468 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
469 case '&': /* & && &= */
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
470 c = *++p;
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
471 if (c == '=')
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
472 t.type = TOK.AndAssign;
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
473 else if (c == '&')
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
474 t.type = TOK.AndLogical;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
475 else {
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
476 t.type = TOK.AndBinary;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
477 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
478 }
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
479 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
480 case '+': /* + ++ += */
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
481 c = *++p;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
482 if (c == '=')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
483 t.type = TOK.PlusAssign;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
484 else if (c == '+')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
485 t.type = TOK.PlusPlus;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
486 else {
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
487 t.type = TOK.Plus;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
488 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
489 }
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
490 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
491 case '-': /* - -- -= */
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
492 c = *++p;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
493 if (c == '=')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
494 t.type = TOK.MinusAssign;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
495 else if (c == '-')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
496 t.type = TOK.MinusMinus;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
497 else {
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
498 t.type = TOK.Minus;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
499 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
500 }
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
501 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
502 case '=': /* = == */
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
503 if (p[1] == '=') {
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
504 ++p;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
505 t.type = TOK.Equal;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
506 }
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
507 else
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
508 t.type = TOK.Assign;
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
509 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
510 case '~': /* ~ ~= */
27
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
511 if (p[1] == '=') {
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
512 ++p;
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
513 t.type = TOK.CatAssign;
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
514 }
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
515 else
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
516 t.type = TOK.Tilde;
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
517 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
518 case '*': /* * *= */
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
519 if (p[1] == '=') {
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
520 ++p;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
521 t.type = TOK.MulAssign;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
522 }
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
523 else
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
524 t.type = TOK.Mul;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
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 '^': /* ^ ^= */
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
527 if (p[1] == '=') {
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
528 ++p;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
529 t.type = TOK.XorAssign;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
530 }
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
531 else
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
532 t.type = TOK.Xor;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
533 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
534 case '%': /* % %= */
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
535 if (p[1] == '=') {
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
536 ++p;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
537 t.type = TOK.ModAssign;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
538 }
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
539 else
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
540 t.type = TOK.Mod;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
541 goto Lcommon;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
542 // Single character tokens:
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
543 case '(':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
544 t.type = TOK.LParen;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
545 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
546 case ')':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
547 t.type = TOK.RParen;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
548 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
549 case '[':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
550 t.type = TOK.LBracket;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
551 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
552 case ']':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
553 t.type = TOK.RBracket;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
554 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
555 case '{':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
556 t.type = TOK.LBrace;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
557 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
558 case '}':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
559 t.type = TOK.RBrace;
21
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
560 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
561 case ':':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
562 t.type = TOK.Colon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
563 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
564 case ';':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
565 t.type = TOK.Semicolon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
566 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
567 case '?':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
568 t.type = TOK.Question;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
569 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
570 case ',':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
571 t.type = TOK.Comma;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
572 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
573 case '$':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
574 t.type = TOK.Dollar;
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
575 Lcommon:
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
576 ++p;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
577 Lcommon2:
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
578 t.end = p;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
579 return;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
580 case '#':
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
581 return scanSpecialToken(t);
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
582 default:
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
583 }
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
584
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
585 if (c & 128 && isUniAlpha(decodeUTF8()))
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
586 goto Lidentifier;
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
587 c = *++p;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
588 }
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
589 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
590
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
591 void scanNormalStringLiteral(ref Token t)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
592 {
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
593 assert(*p == '"');
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
594 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
595 char[] buffer;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
596 t.type = TOK.String;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
597 while (1)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
598 {
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
599 switch (*p)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
600 {
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
601 case '"':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
602 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
603 Lreturn:
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
604 buffer ~= 0;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
605 t.str = buffer;
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
606 t.pf = scanPostfix();
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
607 t.end = p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
608 return;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
609 case '\\':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
610 ++p;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
611 dchar d = scanEscapeSequence();
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
612 if (d < 128)
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
613 buffer ~= d;
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
614 else
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
615 encodeUTF8(buffer, d);
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
616 continue;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
617 case '\r':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
618 if (p[1] == '\n')
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
619 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
620 case '\n':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
621 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
622 ++loc;
221
f26a2beb745e - Moved ++p below if-statement.
aziz
parents: 215
diff changeset
623 buffer ~= '\n'; // Convert EndOfLine to \n.
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
624 continue;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
625 case 0, _Z_:
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
626 error(MID.UnterminatedString);
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
627 goto Lreturn;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
628 default:
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
629 if (*p & 128)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
630 {
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
631 // char* begin = p;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
632 dchar d = decodeUTF8();
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
633 if (d == LSd || d == PSd)
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
634 goto case '\n';
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
635
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
636 // We don't copy per pointer because we might include
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
637 // invalid, skipped utf-8 sequences. See decodeUTF8().
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
638 // ++p;
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
639 // buffer ~= begin[0 .. p - begin];
221
f26a2beb745e - Moved ++p below if-statement.
aziz
parents: 215
diff changeset
640 ++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
641 encodeUTF8(buffer, d);
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
642 continue;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
643 }
223
043722eed46e - Fix: if-statements should be while-statements.
aziz
parents: 221
diff changeset
644 // Copy ASCII character.
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
645 buffer ~= *p++;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
646 }
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
647 }
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
648 assert(0);
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
649 }
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
650
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
651 void scanCharacterLiteral(ref Token t)
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
652 {
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
653 assert(*p == '\'');
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
654 MID id = MID.UnterminatedCharacterLiteral;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
655 ++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
656 TOK type = TOK.CharLiteral;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
657 switch (*p)
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
658 {
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
659 case '\\':
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
660 ++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
661 switch (*p)
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
662 {
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
663 case 'u':
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
664 type = TOK.WCharLiteral; break;
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
665 case 'U':
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
666 type = TOK.DCharLiteral; break;
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
667 default:
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
668 }
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
669 t.dchar_ = scanEscapeSequence();
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
670 break;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
671 case '\'':
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
672 ++p;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
673 id = MID.EmptyCharacterLiteral;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
674 case '\n', '\r', 0, _Z_:
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
675 goto Lerr;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
676 default:
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
677 uint c = *p;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
678 if (c & 128)
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
679 {
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
680 c = decodeUTF8();
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
681 if (c == LSd || c == PSd)
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
682 goto Lerr;
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
683 if (c <= 0xFFFF)
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
684 type = TOK.WCharLiteral;
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
685 else
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
686 type = TOK.DCharLiteral;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
687 }
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
688 t.dchar_ = c;
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
689 ++p;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
690 }
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
691
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
692 if (*p == '\'')
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
693 ++p;
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
694 else
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
695 Lerr:
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
696 error(id);
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
697 t.type = type;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
698 t.end = p;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
699 }
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
700
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
701 char scanPostfix()
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
702 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
703 switch (*p)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
704 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
705 case 'c':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
706 case 'w':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
707 case 'd':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
708 return *p++;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
709 default:
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
710 return 0;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
711 }
112
004d98df65af - Implemented parseInterfaceDeclaration().
aziz
parents: 105
diff changeset
712 assert(0);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
713 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
714
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
715 void scanRawStringLiteral(ref Token t)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
716 {
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
717 uint delim = *p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
718 assert(delim == '`' || delim == '"' && p[-1] == 'r');
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
719 t.type = TOK.String;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
720 char[] buffer;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
721 uint c;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
722 while (1)
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
723 {
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
724 c = *++p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
725 switch (c)
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
726 {
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
727 case '\r':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
728 if (p[1] == '\n')
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
729 ++p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
730 c = '\n'; // Convert '\r' and '\r\n' to '\n'
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
731 case '\n':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
732 ++loc;
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
733 break;
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
734 case '`':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
735 case '"':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
736 if (c == delim)
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
737 {
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
738 ++p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
739 t.pf = scanPostfix();
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
740 Lreturn:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
741 t.str = buffer ~ '\0';
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
742 t.end = p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
743 return;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
744 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
745 break;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
746 case LS[0]:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
747 if (p[1] == LS[1] && (p[2] == LS[2] || p[2] == PS[2]))
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
748 {
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
749 c = '\n';
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
750 ++p; ++p;
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
751 ++loc;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
752 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
753 break;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
754 case 0, _Z_:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
755 if (delim == 'r')
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
756 error(MID.UnterminatedRawString);
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
757 else
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
758 error(MID.UnterminatedBackQuoteString);
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
759 goto Lreturn;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
760 default:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
761 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
762 buffer ~= c; // copy character to buffer
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
763 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
764 assert(0);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
765 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
766
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
767 void scanHexStringLiteral(ref Token t)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
768 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
769 assert(p[0] == 'x' && p[1] == '"');
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
770 t.type = TOK.String;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
771
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
772 uint c;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
773 ubyte[] buffer;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
774 ubyte h; // hex number
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
775 uint n; // number of hex digits
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
776
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
777 ++p;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
778 while (1)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
779 {
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
780 c = *++p;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
781 switch (c)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
782 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
783 case '"':
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
784 ++p;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
785 if (n & 1)
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
786 error(MID.OddNumberOfDigitsInHexString);
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
787 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
788 Lreturn:
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
789 buffer ~= 0;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
790 t.str = cast(string) buffer;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
791 t.end = p;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
792 return;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
793 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
794 if (p[1] == '\n')
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
795 ++p;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
796 case '\n':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
797 ++loc;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
798 continue;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
799 default:
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
800 if (ishexad(c))
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
801 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
802 if (c <= '9')
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
803 c -= '0';
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
804 else if (c <= 'F')
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
805 c -= 'A' - 10;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
806 else
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
807 c -= 'a' - 10;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
808
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
809 if (n & 1)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
810 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
811 h <<= 4;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
812 h |= c;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
813 buffer ~= h;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
814 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
815 else
112
004d98df65af - Implemented parseInterfaceDeclaration().
aziz
parents: 105
diff changeset
816 h = cast(ubyte)c;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
817 ++n;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
818 continue;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
819 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
820 else if (isspace(c))
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
821 continue;
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
822
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
823 if (c >= 128)
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
824 {
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
825 c = decodeUTF8();
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
826 if (c == LSd || c == PSd)
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
827 {
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
828 ++p; ++p;
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
829 ++loc;
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
830 continue;
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
831 }
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
832 }
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
833 else if (c == 0 || c == _Z_)
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
834 {
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
835 error(MID.UnterminatedHexString);
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
836 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
837 goto Lreturn;
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
838 }
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
839 error(MID.NonHexCharInHexString, cast(dchar)c);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
840 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
841 }
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
842 assert(0);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
843 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
844
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
845 dchar scanEscapeSequence()
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
846 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
847 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
848 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
849 {
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
850 ++p;
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
851 return c;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
852 }
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
853 uint digits = 2;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
854
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
855 switch (*p)
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
856 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
857 case 'x':
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
858 c = 0;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
859 while (1)
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
860 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
861 ++p;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
862 if (ishexad(*p))
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
863 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
864 c *= 16;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
865 if (*p <= '9')
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
866 c += *p - '0';
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
867 else if (*p <= 'F')
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
868 c += *p - 'A' + 10;
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
869 else
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
870 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
871
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
872 if (!--digits)
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
873 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
874 ++p;
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
875 break;
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
876 }
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
877 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
878 else
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
879 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
880 error(MID.InsufficientHexDigits);
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
881 break;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
882 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
883 }
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
884 if (!isValidDchar(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
885 error(MID.InvalidUnicodeCharacter);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
886 break;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
887 case 'u':
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
888 digits = 4;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
889 goto case 'x';
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
890 case 'U':
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
891 digits = 8;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
892 goto case 'x';
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
893 default:
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
894 if (isoctal(*p))
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
895 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
896 c = 0;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
897 c += *p - '0';
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
898 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
899 if (!isoctal(*p))
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
900 return c;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
901 c *= 8;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
902 c += *p - '0';
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
903 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
904 if (!isoctal(*p))
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
905 return c;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
906 c *= 8;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
907 c += *p - '0';
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
908 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
909 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
910 else if(*p == '&')
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
911 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
912 if (isalpha(*++p))
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
913 {
272
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
914 auto begin = p;
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
915 while (isalnum(*++p))
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
916 {}
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
917
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
918 if (*p == ';')
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
919 {
272
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
920 c = entity2Unicode(begin[0..p - begin]);
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
921 ++p;
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
922 if (c == 0xFFFF)
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
923 error(MID.UndefinedHTMLEntity, (begin-1)[0..p-(begin-1)]);
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
924 }
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
925 else
0bde32503976 - Added module HtmlEntities. It contains a table for converting HTML entities to Unicode characters.
aziz
parents: 249
diff changeset
926 error(MID.UnterminatedHTMLEntity);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
927 }
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
928 else
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
929 error(MID.InvalidBeginHTMLEntity);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
930 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
931 else
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
932 error(MID.UndefinedEscapeSequence);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
933 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
934
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
935 return c;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
936 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
937
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
938 /*
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
939 IntegerLiteral:= (Dec|Hex|Bin|Oct)Suffix?
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
940 Dec:= (0|[1-9][0-9_]*)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
941 Hex:= 0[xX] HexDigits
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
942 Bin:= 0[bB][01_]+
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
943 Oct:= 0[0-7_]+
68
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
944 Suffix:= (L[uU]?|[uU]L?)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
945 HexDigits:= [0-9a-zA-Z_]+
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
946
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
947 Invalid: "0b_", "0x_", "._"
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
948 */
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
949 void scanNumber(ref Token t)
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
950 {
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
951 ulong ulong_;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
952 bool overflow;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
953 bool isDecimal;
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
954 size_t digits;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
955
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
956 if (*p != '0')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
957 goto LscanInteger;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
958 ++p; // skip zero
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
959 // check for xX bB ...
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
960 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
961 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
962 case 'x','X':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
963 goto LscanHex;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
964 case 'b','B':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
965 goto LscanBin;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
966 case 'L':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
967 if (p[1] == 'i')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
968 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
969 case '.':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
970 if (p[1] == '.')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
971 break;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
972 case 'i','f','F', 'e', 'E': // Imaginary and float literal suffix
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
973 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
974 default:
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
975 if (*p == '_' || isoctal(*p))
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
976 goto LscanOct;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
977 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
978
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
979 // Number 0
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
980 assert(p[-1] == '0');
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
981 assert(ulong_ == 0);
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
982 isDecimal = true;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
983 goto Lfinalize;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
984
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
985 LscanInteger:
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
986 assert(*p != 0 && isdigit(*p));
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
987 isDecimal = true;
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
988 goto Lenter_loop_int;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
989 while (1)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
990 {
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
991 if (*++p == '_')
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
992 continue;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
993 if (!isdigit(*p))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
994 break;
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
995 Lenter_loop_int:
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
996 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
997 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
998 ulong_ *= 10;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
999 ulong_ += *p - '0';
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1000 continue;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1001 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1002 // Overflow: skip following digits.
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1003 overflow = true;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1004 while (isdigit(*++p)) {}
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1005 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1006 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1007
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1008 // 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
1009 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1010 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1011 case '.':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1012 if (p[1] != '.')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1013 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1014 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1015 case 'L':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1016 if (p[1] != 'i')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1017 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1018 case 'i', 'f', 'F', 'e', 'E':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1019 goto LscanReal;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1020 default:
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1021 }
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1022
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1023 if (overflow)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1024 error(MID.OverflowDecimalNumber);
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1025
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1026 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
1027 goto Lfinalize;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1028
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1029 LscanHex:
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1030 assert(digits == 0);
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1031 assert(*p == 'x');
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1032 while (1)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1033 {
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1034 if (*++p == '_')
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1035 continue;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1036 if (!ishexad(*p))
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1037 break;
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1038 ++digits;
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1039 ulong_ *= 16;
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1040 if (*p <= '9')
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1041 ulong_ += *p - '0';
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1042 else if (*p <= 'F')
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1043 ulong_ += *p - 'A' + 10;
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1044 else
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1045 ulong_ += *p - 'a' + 10;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1046 }
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1047
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1048 switch (*p)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1049 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1050 case '.':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1051 if (p[1] != '.')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1052 goto LscanHexReal;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1053 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1054 case 'L':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1055 if (p[1] != 'i')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1056 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1057 case 'i', 'p', 'P':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1058 goto LscanHexReal;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1059 default:
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1060 }
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1061 if (digits == 0)
59
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1062 error(MID.NoDigitsInHexNumber);
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1063 else if (digits > 16)
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1064 {
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1065 // Overflow: skip following digits.
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1066 error(MID.OverflowHexNumber);
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1067 while (ishexad(*++p)) {}
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1068 }
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1069 goto Lfinalize;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1070 LscanHexReal:
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1071 return scanHexReal(t);
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1072
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1073 LscanBin:
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1074 assert(digits == 0);
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1075 assert(*p == 'b');
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1076 while (1)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1077 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1078 if (*++p == '0')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1079 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1080 ++digits;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1081 ulong_ *= 2;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1082 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1083 if (*p == '1')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1084 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1085 ++digits;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1086 ulong_ *= 2;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1087 ulong_ += *p - '0';
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1088 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1089 if (*p == '_')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1090 continue;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1091 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1092 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1093
59
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1094 if (digits == 0)
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1095 error(MID.NoDigitsInBinNumber);
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1096
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1097 if (digits > 64)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1098 error(MID.OverflowBinaryNumber);
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1099 assert((p[-1] == '0' || p[-1] == '1' || p[-1] == '_') && !(*p == '0' || *p == '1' || *p == '_'));
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1100 goto Lfinalize;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1101
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1102 LscanOct:
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1103 assert(*p == '_' || isoctal(*p));
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1104 if (*p != '_')
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1105 goto Lenter_loop_oct;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1106 while (1)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1107 {
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1108 if (*++p == '_')
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1109 continue;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1110 if (!isoctal(*p))
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1111 break;
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1112 Lenter_loop_oct:
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1113 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
1114 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1115 ulong_ *= 8;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1116 ulong_ += *p - '0';
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1117 ++p;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1118 continue;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1119 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1120 // Overflow: skip following digits.
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1121 overflow = true;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1122 while (isdigit(*++p)) {}
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1123 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1124 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1125
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1126 bool hasDecimalDigits;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1127 if (isdigit(*p))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1128 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1129 hasDecimalDigits = true;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1130 while (isdigit(*++p)) {}
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1131 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1132
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1133 // 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
1134 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1135 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1136 case '.':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1137 if (p[1] != '.')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1138 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1139 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1140 case 'L':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1141 if (p[1] != 'i')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1142 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1143 case 'i', 'f', 'F', 'e', 'E':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1144 goto LscanReal;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1145 default:
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1146 }
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1147
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1148 if (hasDecimalDigits)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1149 error(MID.OctalNumberHasDecimals);
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1150 if (overflow)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1151 error(MID.OverflowOctalNumber);
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1152 // goto Lfinalize;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1153
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1154 Lfinalize:
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1155 enum Suffix
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1156 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1157 None = 0,
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1158 Unsigned = 1,
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1159 Long = 2
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1160 }
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1161
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1162 Suffix suffix;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1163 while (1)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1164 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1165 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1166 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1167 case 'L':
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1168 if (suffix & Suffix.Long)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1169 break;
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1170 suffix |= Suffix.Long;
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1171 ++p;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1172 continue;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1173 case 'u', 'U':
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1174 if (suffix & Suffix.Unsigned)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1175 break;
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1176 suffix |= Suffix.Unsigned;
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1177 ++p;
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1178 continue;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1179 default:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1180 break;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1181 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1182 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1183 }
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1184
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1185 switch (suffix)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1186 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1187 case Suffix.None:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1188 if (ulong_ & 0x8000000000000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1189 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1190 if (isDecimal)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1191 error(MID.OverflowDecimalSign);
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1192 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1193 }
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1194 else if (ulong_ & 0xFFFFFFFF00000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1195 t.type = TOK.Int64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1196 else if (ulong_ & 0x80000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1197 t.type = isDecimal ? TOK.Int64 : TOK.Uint32;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1198 else
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1199 t.type = TOK.Int32;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1200 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1201 case Suffix.Unsigned:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1202 if (ulong_ & 0xFFFFFFFF00000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1203 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1204 else
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1205 t.type = TOK.Uint32;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1206 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1207 case Suffix.Long:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1208 if (ulong_ & 0x8000000000000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1209 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1210 if (isDecimal)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1211 error(MID.OverflowDecimalSign);
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1212 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1213 }
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1214 else
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1215 t.type = TOK.Int64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1216 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1217 case Suffix.Unsigned | Suffix.Long:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1218 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1219 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1220 default:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1221 assert(0);
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1222 }
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1223 t.ulong_ = ulong_;
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
1224 t.end = p;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1225 return;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1226 LscanReal:
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1227 scanReal(t);
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1228 return;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1229 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1230
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1231 /*
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1232 FloatLiteral:= Float[fFL]?i?
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1233 Float:= DecFloat | HexFloat
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
1234 DecFloat:= ([0-9][0-9_]*[.][0-9_]*DecExponent?) | [.][0-9][0-9_]*DecExponent? | [0-9][0-9_]*DecExponent
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1235 DecExponent:= [eE][+-]?[0-9][0-9_]*
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1236 HexFloat:= 0[xX](HexDigits[.]HexDigits | [.][0-9a-zA-Z]HexDigits? | HexDigits)HexExponent
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1237 HexExponent:= [pP][+-]?[0-9][0-9_]*
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1238 */
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1239 void scanReal(ref Token t)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1240 {
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1241 if (*p == '.')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1242 // This function was called by scan() or scanNumber().
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1243 while (isdigit(*++p) || *p == '_') {}
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1244 else
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1245 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1246 // This function was called by scanNumber().
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1247 debug switch (*p)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1248 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1249 case 'L':
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1250 if (p[1] != 'i')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1251 assert(0);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1252 case 'i', 'f', 'F', 'e', 'E': break;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1253 default: assert(0);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1254 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1255 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1256
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1257 // Scan exponent.
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1258 if (*p == 'e' || *p == 'E')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1259 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1260 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1261 if (*p == '-' || *p == '+')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1262 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1263 if (!isdigit(*p))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1264 error(MID.FloatExponentDigitExpected);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1265 else
64
dd4c5a0e47dd - Improved while loop.
aziz
parents: 63
diff changeset
1266 while (isdigit(*++p) || *p == '_') {}
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1267 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1268
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1269 // Copy string to buffer ignoring underscores.
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1270 char[] buffer;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1271 char* end = p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1272 p = t.start;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1273 do
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1274 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1275 if (*p == '_')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1276 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1277 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1278 continue;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1279 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1280 buffer ~= *p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1281 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1282 } while (p != end)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1283 buffer ~= 0;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1284 finalizeFloat(t, buffer);
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1285 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1286
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1287 void scanHexReal(ref Token t)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1288 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1289 assert(*p == '.' || *p == 'i' || *p == 'p' || *p == 'P' || (*p == 'L' && p[1] == 'i'));
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1290 MID mid;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1291 if (*p == '.')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1292 while (ishexad(*++p) || *p == '_') {}
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1293 if (*p != 'p' && *p != 'P')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1294 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1295 mid = MID.HexFloatExponentRequired;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1296 goto Lerr;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1297 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1298 // Copy mantissa to a buffer ignoring underscores.
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1299 char* end = p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1300 p = t.start;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1301 char[] buffer;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1302 do
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1303 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1304 if (*p == '_')
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1305 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1306 ++p;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1307 continue;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1308 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1309 buffer ~= *p;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1310 ++p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1311 } while (p != end)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1312
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1313 assert(p == end && (*p == 'p' || *p == 'P'));
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1314 // Scan and copy the exponent.
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1315 buffer ~= 'p';
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1316 size_t bufflen = buffer.length;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1317 while (1)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1318 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1319 if (*++p == '_')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1320 continue;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1321 if (isdigit(*p))
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1322 buffer ~= *p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1323 else
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1324 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1325 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1326 // When the buffer length hasn't changed, no digits were copied.
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1327 if (bufflen == buffer.length) {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1328 mid = MID.HexFloatMissingExpDigits;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1329 goto Lerr;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1330 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1331 buffer ~= 0; // Terminate for C functions.
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1332 finalizeFloat(t, buffer);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1333 return;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1334 Lerr:
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1335 t.type = TOK.Float32;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1336 t.end = p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1337 error(mid);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1338 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1339
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1340 void finalizeFloat(ref Token t, string buffer)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1341 {
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1342 // 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
1343 switch (*p)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1344 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1345 case 'f', 'F':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1346 t.type = TOK.Float32;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1347 t.float_ = strtof(buffer.ptr, null);
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1348 ++p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1349 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1350 case 'L':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1351 t.type = TOK.Float80;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1352 t.real_ = strtold(buffer.ptr, null);
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1353 ++p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1354 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1355 default:
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1356 t.type = TOK.Float64;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1357 t.double_ = strtod(buffer.ptr, null);
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1358 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1359 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1360 if (*p == 'i')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1361 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1362 ++p;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1363 t.type += 3; // Switch to imaginary counterpart.
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1364 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1365 if (getErrno == ERANGE)
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1366 error(MID.OverflowFloatNumber);
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1367 t.end = p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1368 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1369
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1370 /// Scan special token: #line Integer [Filespec] EndOfLine
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1371 void scanSpecialToken(ref Token t)
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1372 {
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
1373 assert(*p == '#');
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1374
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1375 t.type = TOK.HashLine;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1376
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1377 MID mid;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1378
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1379 ++p;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1380 if (p[0] != 'l' || p[1] != 'i' || p[2] != 'n' || p[3] != 'e')
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1381 {
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1382 mid = MID.ExpectedNumberAfterSTLine;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1383 goto Lerr;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1384 }
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1385 p += 3;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1386
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1387 enum State
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1388 { Number, Filespec, End }
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1389
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1390 State state;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1391
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1392 Loop:
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1393 while (1)
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1394 {
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1395 switch (*++p)
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1396 {
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1397 case '\r':
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1398 if (p[1] == '\n')
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1399 ++p;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1400 case '\n', 0, _Z_:
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1401 break Loop;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1402 case LS[0]:
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1403 if (p[1] == LS[1] && (p[2] == LS[2] || p[2] == PS[2]))
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1404 {
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1405 ++p; ++p;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1406 break Loop;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1407 }
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1408 goto default;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1409 default:
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1410 if (isspace(*p))
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1411 continue;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1412 if (state == State.Number)
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1413 {
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1414 if (!isdigit(*p))
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1415 {
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1416 mid = MID.ExpectedNumberAfterSTLine;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1417 goto Lerr;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1418 }
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1419 t.line_num = new Token;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1420 scan(*t.line_num);
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1421 --p;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1422 state = State.Filespec;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1423 }
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1424 else if (state == State.Filespec)
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1425 {
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1426 if (*p != '"')
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1427 {
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1428 mid = MID.ExpectedFilespec;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1429 goto Lerr;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1430 }
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1431 t.line_filespec = new Token;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1432 t.line_filespec.start = p;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1433 t.line_filespec.type = TOK.Filespec;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1434 while (1)
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1435 {
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1436 switch (*++p)
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1437 {
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1438 case '"':
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1439 break;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1440 case LS[0]:
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1441 if (!(p[1] == LS[1] && (p[2] == LS[2] || p[2] == PS[2])))
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1442 goto default;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1443 case '\r', '\n', 0, _Z_:
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1444 mid = MID.UnterminatedFilespec;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1445 t.line_filespec.end = p;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1446 goto Lerr;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1447 default:
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1448 if (*p & 128)
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1449 decodeUTF8();
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1450 continue;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1451 }
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1452 break; // Exit loop.
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1453 }
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1454 auto start = t.line_filespec.start +1; // +1 skips '"'
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1455 t.line_filespec.str = start[0 .. p - start];
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1456 t.line_filespec.end = p + 1;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1457 state = State.End;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1458 }
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1459 else/+ if (state == State.End)+/
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1460 {
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1461 mid = MID.UnterminatedSpecialToken;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1462 goto Lerr;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1463 }
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1464 }
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1465 }
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1466
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1467 if (state == State.Number)
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1468 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1469 mid = MID.ExpectedNumberAfterSTLine;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1470 goto Lerr;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1471 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1472
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1473 this.loc = t.line_num.uint_ - 1;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1474 if (t.line_filespec)
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1475 this.fileName = t.line_filespec.str;
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1476 t.end = p;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1477
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1478 return;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1479 Lerr:
323
6259fb93e3dd - Rewrote scanSpecialToken().
aziz
parents: 315
diff changeset
1480 t.end = p;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1481 error(mid);
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
1482 }
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
1483
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
1484 dchar decodeUTF8()
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1485 {
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1486 assert(*p & 128, "check for ASCII char before calling decodeUTF8().");
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1487 size_t idx;
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
1488 dchar d;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1489 try
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1490 {
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1491 d = std.utf.decode(p[0 .. end-p], idx);
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1492 p += idx -1;
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1493 }
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1494 catch (UtfException e)
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1495 {
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1496 error(MID.InvalidUTF8Sequence);
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1497 // Skip to next valid utf-8 sequence
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1498 while (UTF8stride[*++p] != 0xFF) {}
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
1499 --p;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1500 }
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1501 return d;
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1502 }
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1503
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1504 void loadKeywords()
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1505 {
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1506 foreach(k; keywords)
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1507 idtable[k.str] = k;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1508 }
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1509 /+
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1510 struct State
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1511 {
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1512 Lexer lexer;
209
921d15b1beeb - Added members token and loc to State.
aziz
parents: 207
diff changeset
1513 Token token;
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1514 char* scanPointer;
209
921d15b1beeb - Added members token and loc to State.
aziz
parents: 207
diff changeset
1515 int loc;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1516 string fileName;
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1517 size_t errorLen;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1518 static State opCall(Lexer lx)
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1519 {
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1520 State s;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1521 s.lexer = lx;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1522 s.token = lx.token;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1523 s.scanPointer = lx.p;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1524 s.loc = lx.loc;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1525 s.fileName = lx.fileName;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1526 s.errorLen = lx.errors.length;
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1527 return s;
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1528 }
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1529 void restore()
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1530 {
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1531 lexer.p = scanPointer;
209
921d15b1beeb - Added members token and loc to State.
aziz
parents: 207
diff changeset
1532 lexer.token = token;
921d15b1beeb - Added members token and loc to State.
aziz
parents: 207
diff changeset
1533 lexer.loc = loc;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1534 lexer.fileName = fileName;
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1535 lexer.errors = lexer.errors[0..errorLen];
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1536 }
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1537 }
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1538
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1539 State getState()
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1540 {
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1541 return State(this);
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1542 }
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1543 +/
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1544
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1545 private void scanNext(ref Token* t)
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
1546 {
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1547 assert(t !is null);
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1548 if (t.next)
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1549 t = t.next;
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1550 else if (t.type != TOK.EOF)
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
1551 {
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1552 Token* new_t = new Token;
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1553 scan(*new_t);
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1554 new_t.prev = t;
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1555 t.next = new_t;
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1556 t = new_t;
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
1557 }
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1558 }
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1559
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1560 void peek(ref Token* t)
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1561 {
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1562 scanNext(t);
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1563 }
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1564
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1565 TOK nextToken()
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1566 {
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1567 scanNext(this.token);
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1568 return this.token.type;
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
1569 }
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
1570
66
8e84db78ad55 - Added support for variadic arguments in error messages.
aziz
parents: 65
diff changeset
1571 void error(MID id, ...)
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
1572 {
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1573 // if (reportErrors)
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1574 errors ~= new Information(InfoType.Lexer, id, loc, arguments(_arguments, _argptr));
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
1575 }
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
1576
207
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1577 unittest
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1578 {
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1579 string sourceText = "unittest { }";
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1580 auto lx = new Lexer(sourceText, null);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1581
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1582 Token next;
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1583 lx.peek(next);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1584 assert(next == TOK.Unittest);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1585 lx.peek(next);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1586 assert(next == TOK.LBrace);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1587 lx.peek(next);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1588 assert(next == TOK.RBrace);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1589 lx.peek(next);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1590 assert(next == TOK.EOF);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1591 writefln("end of peek() unittest");
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1592 }
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1593
306
9c866aadcb5b - Moved code out of main() to separate functions.
aziz
parents: 272
diff changeset
1594 Token* getTokens()
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1595 {
306
9c866aadcb5b - Moved code out of main() to separate functions.
aziz
parents: 272
diff changeset
1596 while (nextToken() != TOK.EOF)
9c866aadcb5b - Moved code out of main() to separate functions.
aziz
parents: 272
diff changeset
1597 {}
9c866aadcb5b - Moved code out of main() to separate functions.
aziz
parents: 272
diff changeset
1598 return head;
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
1599 }
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1600
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1601 private void encodeUTF8(inout char[] str, dchar d)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1602 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1603 char[6] b;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1604 assert(d > 0x7F, "check for ASCII char before calling encodeUTF8().");
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1605 if (d < 0x800)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1606 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1607 b[0] = 0xC0 | (d >> 6);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1608 b[1] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1609 str ~= b[0..2];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1610 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1611 else if (d < 0x10000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1612 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1613 b[0] = 0xE0 | (d >> 12);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1614 b[1] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1615 b[2] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1616 str ~= b[0..3];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1617 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1618 else if (d < 0x200000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1619 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1620 b[0] = 0xF0 | (d >> 18);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1621 b[1] = 0x80 | ((d >> 12) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1622 b[2] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1623 b[3] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1624 str ~= b[0..4];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1625 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1626 else if (d < 0x4000000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1627 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1628 b[0] = 0xF8 | (d >> 24);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1629 b[1] = 0x80 | ((d >> 18) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1630 b[2] = 0x80 | ((d >> 12) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1631 b[3] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1632 b[4] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1633 str ~= b[0..5];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1634 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1635 else if (d < 0x80000000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1636 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1637 b[0] = 0xFC | (d >> 30);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1638 b[1] = 0x80 | ((d >> 24) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1639 b[2] = 0x80 | ((d >> 18) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1640 b[3] = 0x80 | ((d >> 12) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1641 b[4] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1642 b[5] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1643 str ~= b[0..6];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1644 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1645 else
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1646 error(MID.InvalidUnicodeCharacter);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1647 }
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
1648 }
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1649
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1650 unittest
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1651 {
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1652 string[] toks = [
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1653 ">", ">=", ">>", ">>=", ">>>", ">>>=", "<", "<=", "<>",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1654 "<>=", "<<", "<<=", "!", "!<", "!>", "!<=", "!>=", "!<>",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1655 "!<>=", ".", "..", "...", "&", "&&", "&=", "+", "++",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1656 "+=", "-", "--", "-=", "=", "==", "~", "~=", "*",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1657 "*=", "/", "/=", "^", "^=", "%", "%=", "(", ")",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1658 "[", "]", "{", "}", ":", ";", "?", ",", "$"
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1659 ];
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1660
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1661 char[] src;
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1662
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1663 foreach (op; toks)
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1664 src ~= op ~ " ";
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1665
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1666 auto lx = new Lexer(src, "");
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1667 auto tokens = lx.getTokens();
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1668
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1669 tokens = tokens[0..$-1]; // exclude TOK.EOF
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1670
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1671 assert(tokens.length == toks.length );
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1672
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1673 foreach (i, t; tokens)
207
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1674 assert(t.srcText == toks[i], std.string.format("Lexed '%s' but expected '%s'", t.srcText, toks[i]));
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
1675 }
55
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1676
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
1677 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
1678 {
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
1679 // Numbers unittest
68
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1680 // 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
1681 // 0u 0U 0uL 0UL 0L 0LU 0Lu
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1682 // 0Li 0f 0F 0fi 0Fi 0i
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1683 // 0b_1_LU 0b1000u
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1684 // 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
1685 }
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
1686
55
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1687 /// ASCII character properties table.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1688 static const int ptable[256] = [
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1689 0, 0, 0, 0, 0, 0, 0, 0, 0,32, 0,32,32, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1690 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1691 32, 0, 0x2200, 0, 0, 0, 0, 0x2700, 0, 0, 0, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1692 7, 7, 7, 7, 7, 7, 7, 7, 6, 6, 0, 0, 0, 0, 0, 0x3f00,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1693 0,12,12,12,12,12,12, 8, 8, 8, 8, 8, 8, 8, 8, 8,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1694 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 0, 0x5c00, 0, 0,16,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1695 0, 0x70c, 0x80c,12,12,12, 0xc0c, 8, 8, 8, 8, 8, 8, 8, 0xa08, 8,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1696 8, 8, 0xd08, 8, 0x908, 8, 0xb08, 8, 8, 8, 8, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1697 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1698 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1699 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1700 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1701 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1702 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1703 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1704 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1705 ];
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1706
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1707 enum CProperty
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1708 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1709 Octal = 1,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1710 Digit = 1<<1,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1711 Hex = 1<<2,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1712 Alpha = 1<<3,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1713 Underscore = 1<<4,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1714 Whitespace = 1<<5
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1715 }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1716
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1717 const uint EVMask = 0xFF00; // Bit mask for escape value
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1718
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1719 private alias CProperty CP;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1720 int isoctal(char c) { return ptable[c] & CP.Octal; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1721 int isdigit(char c) { return ptable[c] & CP.Digit; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1722 int ishexad(char c) { return ptable[c] & CP.Hex; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1723 int isalpha(char c) { return ptable[c] & CP.Alpha; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1724 int isalnum(char c) { return ptable[c] & (CP.Alpha | CP.Digit); }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1725 int isidbeg(char c) { return ptable[c] & (CP.Alpha | CP.Underscore); }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1726 int isident(char c) { return ptable[c] & (CP.Alpha | CP.Underscore | CP.Digit); }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1727 int isspace(char c) { return ptable[c] & CP.Whitespace; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1728 int char2ev(char c) { return ptable[c] >> 8; /*(ptable[c] & EVMask) >> 8;*/ }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1729
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1730 version(gen_ptable)
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1731 static this()
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1732 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1733 alias ptable p;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1734 // Initialize character properties table.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1735 for (int i; i < p.length; ++i)
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1736 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1737 p[i] = 0;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1738 if ('0' <= i && i <= '7')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1739 p[i] |= CP.Octal;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1740 if ('0' <= i && i <= '9')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1741 p[i] |= CP.Digit;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1742 if (isdigit(i) || 'a' <= i && i <= 'f' || 'A' <= i && i <= 'F')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1743 p[i] |= CP.Hex;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1744 if ('a' <= i && i <= 'z' || 'A' <= i && i <= 'Z')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1745 p[i] |= CP.Alpha;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1746 if (i == '_')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1747 p[i] |= CP.Underscore;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1748 if (i == ' ' || i == '\t' || i == '\v' || i == '\f')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1749 p[i] |= CP.Whitespace;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1750 }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1751 // Store escape sequence values in second byte.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1752 assert(CProperty.max <= ubyte.max, "character property flags and escape value byte overlap.");
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1753 p['\''] |= 39 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1754 p['"'] |= 34 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1755 p['?'] |= 63 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1756 p['\\'] |= 92 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1757 p['a'] |= 7 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1758 p['b'] |= 8 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1759 p['f'] |= 12 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1760 p['n'] |= 10 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1761 p['r'] |= 13 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1762 p['t'] |= 9 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1763 p['v'] |= 11 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1764 // Print a formatted array literal.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1765 char[] array = "[\n";
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1766 for (int i; i < p.length; ++i)
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1767 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1768 int c = p[i];
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1769 array ~= std.string.format(c>255?" 0x%x,":"%2d,", c, ((i+1) % 16) ? "":"\n");
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1770 }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1771 array[$-2..$] = "\n]";
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1772 writefln(array);
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1773 }