annotate trunk/src/Lexer.d @ 239:7911f6a92e6e

- Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK. - Tokens are stored in a doubly linked list by the Lexer now. - Added method scanNext() which can be used by peek() and nextToken(). - Saving 'token' in Parser.try_().
author aziz
date Mon, 30 Jul 2007 12:36:04 +0000
parents 043722eed46e
children 32d354584b28
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
8ba2570de175 Initial import.
aziz
parents:
diff changeset
3 License: GPL2
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;
2
81c6cc33f5c8 - Initializing ptable with a precomputed array literal.
aziz
parents: 1
diff changeset
11 import std.stdio;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
12 import std.utf;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
13 import std.uni;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
14 import std.c.stdlib;
66
8e84db78ad55 - Added support for variadic arguments in error messages.
aziz
parents: 65
diff changeset
15 import std.string;
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
16
11
dffcdaa7c47a - Added Unicode line and paragraph separators.
aziz
parents: 10
diff changeset
17 const char[3] LS = \u2028;
dffcdaa7c47a - Added Unicode line and paragraph separators.
aziz
parents: 10
diff changeset
18 const char[3] PS = \u2029;
dffcdaa7c47a - Added Unicode line and paragraph separators.
aziz
parents: 10
diff changeset
19
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
20 const dchar LSd = 0x2028;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
21 const dchar PSd = 0x2029;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
22
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
23 const uint _Z_ = 26; /// Control+Z
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
24
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
25 class Lexer
8ba2570de175 Initial import.
aziz
parents:
diff changeset
26 {
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
27 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
28 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
29 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
30 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
31 char* end; /// Points one character past the end of the source text.
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
32
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
33 uint loc = 1; /// line of code
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
34
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
35 char[] fileName;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
36
69
24db7c5522d5 - Added module Information for compiler messages like warnings, info and errors to the user.
aziz
parents: 68
diff changeset
37 Information[] errors;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
38
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
39 // 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
40
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
41 Identifier[string] idtable;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
42
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
43 this(string text, string fileName)
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
44 {
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
45 this.fileName = fileName;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
46
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
47 this.text = text;
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
48 if (text[$-1] != 0)
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
49 {
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
50 this.text.length = this.text.length + 1;
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
51 this.text[$-1] = 0;
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
52 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
53
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
54 this.p = this.text.ptr;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
55 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
56 // this.reportErrors = true;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
57 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
58
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
59 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
60 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
61 this.token = this.head;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
62 }
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
63
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
64 public void scan(out Token t)
207
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
65 in
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
66 {
207
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
67 assert(text.ptr <= p && p < end);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
68 }
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
69 out
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
70 {
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
71 assert(text.ptr <= t.start && t.start < end);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
72 assert(text.ptr < t.end && t.end <= end, std.string.format(t.type));
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
73 }
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
74 body
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
75 {
10
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
76 uint c = *p;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
77
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
78 while (1)
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
79 {
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
80 t.start = p;
16
476e8e55c1d4 - Added Whitespace to the character properties table.
aziz
parents: 15
diff changeset
81
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
82 if (c == 0 || c == _Z_)
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
83 {
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
84 assert(*p == 0 || *p == _Z_);
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
85 t.type = TOK.EOF;
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
86 t.end = p;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
87 assert(t.start == t.end);
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
88 return;
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
89 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
90
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
91 if (c == '\n')
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
92 {
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
93 c = *++p;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
94 ++loc;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
95 continue;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
96 }
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
97 else if (c == '\r')
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
98 {
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
99 c = *++p;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
100 if (c != '\n')
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
101 ++loc;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
102 continue;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
103 }
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
104 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
105 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
106 p += 3;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
107 c = *p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
108 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
109 }
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
110
13
e5211758b63c - Added isidbeg() function.
aziz
parents: 12
diff changeset
111 if (isidbeg(c))
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
112 {
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
113 if (c == 'r' && p[1] == '"' && ++p)
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
114 return scanRawStringLiteral(t);
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
115 if (c == 'x' && p[1] == '"')
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
116 return scanHexStringLiteral(t);
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
117 Lidentifier:
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
118 do
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
119 { c = *++p; }
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
120 while (isident(c) || c & 128 && isUniAlpha(decodeUTF8()))
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
121
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
122 t.end = p;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
123
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 64
diff changeset
124 string str = t.srcText;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
125 Identifier* id = str in idtable;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
126
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
127 if (!id)
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
128 {
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
129 idtable[str] = Identifier.Identifier(TOK.Identifier, str);
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
130 id = str in idtable;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
131 }
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
132 assert(id);
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
133 t.type = id.type;
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
134 return;
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
135 }
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
136
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
137 if (isdigit(c))
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
138 return scanNumber(t);
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
139
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
140 if (c == '/')
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
141 {
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
142 c = *++p;
14
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
143 switch(c)
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
144 {
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
145 case '=':
14
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
146 ++p;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
147 t.type = TOK.DivAssign;
14
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
148 t.end = p;
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
149 return;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
150 case '+':
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
151 uint level = 1;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
152 while (1)
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
153 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
154 c = *++p;
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
155 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
156 switch (c)
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
157 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
158 case '\r':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
159 if (p[1] == '\n')
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
160 ++p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
161 case '\n':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
162 ++loc;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
163 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
164 case 0, _Z_:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
165 error(MID.UnterminatedNestedComment);
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
166 goto LreturnNC;
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
167 default:
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
168 }
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
169
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
170 c <<= 8;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
171 c |= *++p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
172 switch (c)
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
173 {
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
174 case 0x2F2B: // /+
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
175 ++level;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
176 continue;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
177 case 0x2B2F: // +/
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
178 if (--level == 0)
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
179 {
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
180 ++p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
181 LreturnNC:
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
182 t.type = TOK.Comment;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
183 t.end = p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
184 return;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
185 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
186 continue;
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
187 case 0xE280: // LS[0..1] || PS[0..1]
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
188 if (p[1] == LS[2] || p[1] == PS[2])
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
189 {
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
190 ++loc;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
191 ++p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
192 }
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
193 continue;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
194 default:
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
195 c &= char.max;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
196 goto LswitchNC;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
197 }
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 case '*':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
200 while (1)
7
07e45c06a024 - Parsing nested comments correctly now.
aziz
parents: 5
diff changeset
201 {
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
202 c = *++p;
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
203 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
204 switch (c)
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
205 {
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
206 case '\r':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
207 if (p[1] == '\n')
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
208 ++p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
209 case '\n':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
210 ++loc;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
211 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
212 case 0, _Z_:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
213 error(MID.UnterminatedBlockComment);
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
214 goto LreturnBC;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
215 default:
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
216 }
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
217
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
218 c <<= 8;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
219 c |= *++p;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
220 switch (c)
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
221 {
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
222 case 0x2A2F: // */
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
223 ++p;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
224 LreturnBC:
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
225 t.type = TOK.Comment;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
226 t.end = p;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
227 return;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
228 case 0xE280: // LS[0..1] || PS[0..1]
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
229 if (p[1] == LS[2] || p[1] == PS[2])
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
230 {
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
231 ++loc;
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
232 ++p;
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
233 }
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
234 continue;
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
235 default:
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
236 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
237 goto LswitchBC;
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
238 }
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
239 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
240 assert(0);
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
241 case '/':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
242 while (1)
7
07e45c06a024 - Parsing nested comments correctly now.
aziz
parents: 5
diff changeset
243 {
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
244 c = *++p;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
245 switch (c)
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
246 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
247 case '\r':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
248 if (p[1] == '\n')
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
249 ++p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
250 case '\n':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
251 case 0, _Z_:
11
dffcdaa7c47a - Added Unicode line and paragraph separators.
aziz
parents: 10
diff changeset
252 break;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
253 case LS[0]:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
254 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
255 break;
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 default:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
258 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
259 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
260 t.type = TOK.Comment;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
261 t.end = p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
262 return;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
263 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
264 default:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
265 t.type = TOK.Div;
10
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
266 t.end = p;
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
267 return;
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
268 }
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
269 }
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
270
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
271 switch (c)
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
272 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
273 case '\'':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
274 return scanCharacterLiteral(t);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
275 case '`':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
276 return scanRawStringLiteral(t);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
277 case '"':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
278 return scanNormalStringLiteral(t);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
279 case '\\':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
280 char[] buffer;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
281 do
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
282 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
283 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
284 c = scanEscapeSequence();
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
285 if (c < 128)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
286 buffer ~= c;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
287 else
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
288 encodeUTF8(buffer, c);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
289 } while (*p == '\\')
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
290 buffer ~= 0;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
291 t.type = TOK.String;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
292 t.str = buffer;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
293 t.end = p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
294 return;
38
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
295 case '>': /* > >= >> >>= >>> >>>= */
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
296 c = *++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
297 switch (c)
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
298 {
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
299 case '=':
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
300 t.type = TOK.GreaterEqual;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
301 goto Lcommon;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
302 case '>':
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
303 if (p[1] == '>')
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
304 {
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
305 ++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
306 if (p[1] == '=')
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
307 { ++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
308 t.type = TOK.URShiftAssign;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
309 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
310 else
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
311 t.type = TOK.URShift;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
312 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
313 else if (p[1] == '=')
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
314 {
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
315 ++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
316 t.type = TOK.RShiftAssign;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
317 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
318 else
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
319 t.type = TOK.RShift;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
320 goto Lcommon;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
321 default:
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
322 t.type = TOK.Greater;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
323 goto Lcommon2;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
324 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
325 assert(0);
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
326 case '<': /* < <= <> <>= << <<= */
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
327 c = *++p;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
328 switch (c)
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
329 {
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
330 case '=':
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
331 t.type = TOK.LessEqual;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
332 goto Lcommon;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
333 case '<':
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
334 if (p[1] == '=') {
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
335 ++p;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
336 t.type = TOK.LShiftAssign;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
337 }
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
338 else
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
339 t.type = TOK.LShift;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
340 goto Lcommon;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
341 case '>':
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
342 if (p[1] == '=') {
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
343 ++p;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
344 t.type = TOK.LorEorG;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
345 }
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
346 else
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
347 t.type = TOK.LorG;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
348 goto Lcommon;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
349 default:
38
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
350 t.type = TOK.Less;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
351 goto Lcommon2;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
352 }
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
353 assert(0);
37
7f3bcb97d017 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 36
diff changeset
354 case '!': /* ! !< !> !<= !>= !<> !<>= */
35
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
355 c = *++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
356 switch (c)
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
357 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
358 case '<':
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
359 c = *++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
360 if (c == '>')
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
361 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
362 if (p[1] == '=') {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
363 ++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
364 t.type = TOK.Unordered;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
365 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
366 else
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
367 t.type = TOK.UorE;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
368 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
369 else if (c == '=')
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
370 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
371 t.type = TOK.UorG;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
372 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
373 else {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
374 t.type = TOK.UorGorE;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
375 goto Lcommon2;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
376 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
377 goto Lcommon;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
378 case '>':
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
379 if (p[1] == '=')
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
380 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
381 ++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
382 t.type = TOK.UorL;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
383 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
384 else
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
385 t.type = TOK.UorLorE;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
386 goto Lcommon;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
387 case '=':
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
388 t.type = TOK.NotEqual;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
389 goto Lcommon;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
390 default:
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
391 t.type = TOK.Not;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
392 goto Lcommon2;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
393 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
394 assert(0);
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
395 case '.': /* . .[0-9] .. ... */
22
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
396 if (p[1] == '.')
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
397 {
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
398 ++p;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
399 if (p[1] == '.') {
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
400 ++p;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
401 t.type = TOK.Ellipses;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
402 }
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
403 else
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
404 t.type = TOK.Slice;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
405 }
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
406 else if (isdigit(p[1]))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
407 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
408 return scanReal(t);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
409 }
22
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
410 else
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
411 t.type = TOK.Dot;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
412 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
413 case '|': /* | || |= */
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
414 c = *++p;
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
415 if (c == '=')
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
416 t.type = TOK.OrAssign;
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
417 else if (c == '|')
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
418 t.type = TOK.OrLogical;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
419 else {
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
420 t.type = TOK.OrBinary;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
421 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
422 }
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
423 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
424 case '&': /* & && &= */
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
425 c = *++p;
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
426 if (c == '=')
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
427 t.type = TOK.AndAssign;
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
428 else if (c == '&')
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
429 t.type = TOK.AndLogical;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
430 else {
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
431 t.type = TOK.AndBinary;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
432 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
433 }
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
434 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
435 case '+': /* + ++ += */
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
436 c = *++p;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
437 if (c == '=')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
438 t.type = TOK.PlusAssign;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
439 else if (c == '+')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
440 t.type = TOK.PlusPlus;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
441 else {
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
442 t.type = TOK.Plus;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
443 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
444 }
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
445 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
446 case '-': /* - -- -= */
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
447 c = *++p;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
448 if (c == '=')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
449 t.type = TOK.MinusAssign;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
450 else if (c == '-')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
451 t.type = TOK.MinusMinus;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
452 else {
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
453 t.type = TOK.Minus;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
454 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
455 }
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
456 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
457 case '=': /* = == */
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
458 if (p[1] == '=') {
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
459 ++p;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
460 t.type = TOK.Equal;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
461 }
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
462 else
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
463 t.type = TOK.Assign;
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
464 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
465 case '~': /* ~ ~= */
27
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
466 if (p[1] == '=') {
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
467 ++p;
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
468 t.type = TOK.CatAssign;
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
469 }
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
470 else
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
471 t.type = TOK.Tilde;
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
472 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
473 case '*': /* * *= */
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
474 if (p[1] == '=') {
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
475 ++p;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
476 t.type = TOK.MulAssign;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
477 }
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
478 else
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
479 t.type = TOK.Mul;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
480 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
481 case '^': /* ^ ^= */
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
482 if (p[1] == '=') {
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
483 ++p;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
484 t.type = TOK.XorAssign;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
485 }
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
486 else
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
487 t.type = TOK.Xor;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
488 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
489 case '%': /* % %= */
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
490 if (p[1] == '=') {
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
491 ++p;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
492 t.type = TOK.ModAssign;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
493 }
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
494 else
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
495 t.type = TOK.Mod;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
496 goto Lcommon;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
497 // Single character tokens:
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
498 case '(':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
499 t.type = TOK.LParen;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
500 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
501 case ')':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
502 t.type = TOK.RParen;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
503 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
504 case '[':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
505 t.type = TOK.LBracket;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
506 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
507 case ']':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
508 t.type = TOK.RBracket;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
509 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
510 case '{':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
511 t.type = TOK.LBrace;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
512 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
513 case '}':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
514 t.type = TOK.RBrace;
21
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
515 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
516 case ':':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
517 t.type = TOK.Colon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
518 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
519 case ';':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
520 t.type = TOK.Semicolon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
521 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
522 case '?':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
523 t.type = TOK.Question;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
524 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
525 case ',':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
526 t.type = TOK.Comma;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
527 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
528 case '$':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
529 t.type = TOK.Dollar;
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
530 Lcommon:
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
531 ++p;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
532 Lcommon2:
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
533 t.end = p;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
534 return;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
535 case '#':
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
536 scanSpecialToken();
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
537 c = *p;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
538 continue;
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
539 default:
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
540 }
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
541
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
542 if (c & 128 && isUniAlpha(decodeUTF8()))
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
543 goto Lidentifier;
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
544 c = *++p;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
545 }
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
546 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
547
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
548 void scanNormalStringLiteral(ref Token t)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
549 {
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
550 assert(*p == '"');
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
551 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
552 char[] buffer;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
553 t.type = TOK.String;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
554 while (1)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
555 {
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
556 switch (*p)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
557 {
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
558 case '"':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
559 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
560 Lreturn:
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
561 buffer ~= 0;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
562 t.str = buffer;
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
563 t.pf = scanPostfix();
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
564 t.end = p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
565 return;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
566 case '\\':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
567 ++p;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
568 dchar d = scanEscapeSequence();
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
569 if (d < 128)
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
570 buffer ~= d;
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
571 else
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
572 encodeUTF8(buffer, d);
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
573 continue;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
574 case '\r':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
575 if (p[1] == '\n')
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
576 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
577 case '\n':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
578 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
579 ++loc;
221
f26a2beb745e - Moved ++p below if-statement.
aziz
parents: 215
diff changeset
580 buffer ~= '\n'; // Convert EndOfLine to \n.
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
581 continue;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
582 case 0, _Z_:
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
583 error(MID.UnterminatedString);
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
584 goto Lreturn;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
585 default:
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
586 if (*p & 128)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
587 {
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
588 // char* begin = p;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
589 dchar d = decodeUTF8();
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
590 if (d == LSd || d == PSd)
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
591 goto case '\n';
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
592
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
593 // 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
594 // 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
595 // ++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
596 // buffer ~= begin[0 .. p - begin];
221
f26a2beb745e - Moved ++p below if-statement.
aziz
parents: 215
diff changeset
597 ++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
598 encodeUTF8(buffer, d);
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
599 continue;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
600 }
223
043722eed46e - Fix: if-statements should be while-statements.
aziz
parents: 221
diff changeset
601 // Copy ASCII character.
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
602 buffer ~= *p++;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
603 }
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
604 }
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
605 assert(0);
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
606 }
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
607
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
608 void scanCharacterLiteral(ref Token t)
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
609 {
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
610 assert(*p == '\'');
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
611 MID id = MID.UnterminatedCharacterLiteral;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
612 ++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
613 TOK type = TOK.CharLiteral;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
614 switch (*p)
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
615 {
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
616 case '\\':
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
617 ++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
618 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
619 {
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
620 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
621 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
622 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
623 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
624 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
625 }
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
626 t.dchar_ = scanEscapeSequence();
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
627 break;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
628 case '\'':
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
629 ++p;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
630 id = MID.EmptyCharacterLiteral;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
631 case '\n', '\r', 0, _Z_:
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
632 goto Lerr;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
633 default:
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
634 uint c = *p;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
635 if (c & 128)
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
636 {
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
637 c = decodeUTF8();
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
638 if (c == LSd || c == PSd)
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
639 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
640 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
641 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
642 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
643 type = TOK.DCharLiteral;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
644 }
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
645 t.dchar_ = c;
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
646 ++p;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
647 }
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
648
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
649 if (*p == '\'')
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
650 ++p;
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
651 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
652 Lerr:
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
653 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
654 t.type = type;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
655 t.end = p;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
656 }
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
657
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
658 char scanPostfix()
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
659 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
660 switch (*p)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
661 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
662 case 'c':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
663 case 'w':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
664 case 'd':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
665 return *p++;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
666 default:
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
667 return 0;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
668 }
112
004d98df65af - Implemented parseInterfaceDeclaration().
aziz
parents: 105
diff changeset
669 assert(0);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
670 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
671
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
672 void scanRawStringLiteral(ref Token t)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
673 {
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
674 uint delim = *p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
675 assert(delim == '`' || delim == '"' && p[-1] == 'r');
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
676 t.type = TOK.String;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
677 char[] buffer;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
678 uint c;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
679 while (1)
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
680 {
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
681 c = *++p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
682 switch (c)
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
683 {
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
684 case '\r':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
685 if (p[1] == '\n')
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
686 ++p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
687 c = '\n'; // Convert '\r' and '\r\n' to '\n'
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
688 case '\n':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
689 ++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
690 break;
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
691 case '`':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
692 case '"':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
693 if (c == delim)
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
694 {
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
695 ++p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
696 t.pf = scanPostfix();
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
697 Lreturn:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
698 t.str = buffer ~ '\0';
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
699 t.end = p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
700 return;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
701 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
702 break;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
703 case LS[0]:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
704 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
705 {
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
706 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
707 ++p; ++p;
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
708 ++loc;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
709 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
710 break;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
711 case 0, _Z_:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
712 if (delim == 'r')
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
713 error(MID.UnterminatedRawString);
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
714 else
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
715 error(MID.UnterminatedBackQuoteString);
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
716 goto Lreturn;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
717 default:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
718 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
719 buffer ~= c; // copy character to buffer
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
720 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
721 assert(0);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
722 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
723
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
724 void scanHexStringLiteral(ref Token t)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
725 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
726 assert(p[0] == 'x' && p[1] == '"');
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
727 t.type = TOK.String;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
728
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
729 uint c;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
730 ubyte[] buffer;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
731 ubyte h; // hex number
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
732 uint n; // number of hex digits
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
733
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
734 ++p;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
735 while (1)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
736 {
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
737 c = *++p;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
738 switch (c)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
739 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
740 case '"':
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
741 ++p;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
742 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
743 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
744 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
745 Lreturn:
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
746 buffer ~= 0;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
747 t.str = cast(string) buffer;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
748 t.end = p;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
749 return;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
750 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
751 if (p[1] == '\n')
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
752 ++p;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
753 case '\n':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
754 ++loc;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
755 continue;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
756 default:
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
757 if (ishexad(c))
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
758 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
759 if (c <= '9')
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
760 c -= '0';
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
761 else if (c <= 'F')
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
762 c -= 'A' - 10;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
763 else
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
764 c -= 'a' - 10;
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 if (n & 1)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
767 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
768 h <<= 4;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
769 h |= c;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
770 buffer ~= h;
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 else
112
004d98df65af - Implemented parseInterfaceDeclaration().
aziz
parents: 105
diff changeset
773 h = cast(ubyte)c;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
774 ++n;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
775 continue;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
776 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
777 else if (isspace(c))
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
778 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
779
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
780 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
781 {
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
782 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
783 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
784 {
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
785 ++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
786 ++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
787 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
788 }
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
789 }
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
790 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
791 {
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
792 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
793 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
794 goto Lreturn;
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
795 }
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
796 error(MID.NonHexCharInHexString, cast(dchar)c);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
797 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
798 }
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
799 assert(0);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
800 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
801
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
802 dchar scanEscapeSequence()
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
803 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
804 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
805 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
806 {
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
807 ++p;
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
808 return c;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
809 }
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
810 uint digits = 2;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
811
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
812 switch (*p)
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
813 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
814 case 'x':
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
815 c = 0;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
816 while (1)
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
817 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
818 ++p;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
819 if (ishexad(*p))
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
820 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
821 c *= 16;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
822 if (*p <= '9')
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
823 c += *p - '0';
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
824 else if (*p <= 'F')
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
825 c += *p - 'A' + 10;
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
826 else
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
827 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
828
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
829 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
830 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
831 ++p;
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
832 break;
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
833 }
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
834 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
835 else
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
836 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
837 error(MID.InsufficientHexDigits);
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
838 break;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
839 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
840 }
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
841 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
842 error(MID.InvalidUnicodeCharacter);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
843 break;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
844 case 'u':
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
845 digits = 4;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
846 goto case 'x';
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
847 case 'U':
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
848 digits = 8;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
849 goto case 'x';
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
850 default:
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
851 if (isoctal(*p))
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
852 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
853 c = 0;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
854 c += *p - '0';
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
855 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
856 if (!isoctal(*p))
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
857 return c;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
858 c *= 8;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
859 c += *p - '0';
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
860 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
861 if (!isoctal(*p))
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
862 return c;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
863 c *= 8;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
864 c += *p - '0';
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
865 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
866 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
867 else if(*p == '&')
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
868 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
869 if (isalpha(*++p))
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
870 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
871 while (1)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
872 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
873 if (isalnum(*++p))
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
874 continue;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
875 if (*p == ';') {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
876 // TODO: convert entity to unicode codepoint.
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
877 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
878 break;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
879 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
880 else {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
881 error(MID.UnterminatedHTMLEntity);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
882 break;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
883 }
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
884 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
885 }
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
886 else
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
887 error(MID.InvalidBeginHTMLEntity);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
888 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
889 else
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
890 error(MID.UndefinedEscapeSequence);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
891 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
892
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
893 return c;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
894 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
895
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
896 /*
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
897 IntegerLiteral:= (Dec|Hex|Bin|Oct)Suffix?
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
898 Dec:= (0|[1-9][0-9_]*)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
899 Hex:= 0[xX] HexDigits
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
900 Bin:= 0[bB][01_]+
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
901 Oct:= 0[0-7_]+
68
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
902 Suffix:= (L[uU]?|[uU]L?)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
903 HexDigits:= [0-9a-zA-Z_]+
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
904
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
905 Invalid: "0b_", "0x_", "._"
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
906 */
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
907 void scanNumber(ref Token t)
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
908 {
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
909 ulong ulong_;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
910 bool overflow;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
911 bool isDecimal;
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
912 size_t digits;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
913
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
914 if (*p != '0')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
915 goto LscanInteger;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
916 ++p; // skip zero
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
917 // check for xX bB ...
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
918 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
919 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
920 case 'x','X':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
921 goto LscanHex;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
922 case 'b','B':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
923 goto LscanBin;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
924 case 'L':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
925 if (p[1] == 'i')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
926 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
927 case '.':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
928 if (p[1] == '.')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
929 break;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
930 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
931 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
932 case '_':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
933 ++p;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
934 goto LscanOct;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
935 default:
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
936 if (isoctal(*p))
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
937 goto LscanOct;
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
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
940 ulong_ = p[-1];
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
941 isDecimal = true;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
942 goto Lfinalize;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
943
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
944 LscanInteger:
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
945 isDecimal = true;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
946 while (1)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
947 {
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
948 if (*++p == '_')
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
949 continue;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
950 if (!isdigit(*p))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
951 break;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
952 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
953 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
954 ulong_ *= 10;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
955 ulong_ += *p - '0';
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
956 continue;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
957 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
958 // Overflow: skip following digits.
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
959 overflow = true;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
960 while (isdigit(*++p)) {}
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
961 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
962 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
963
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
964 // 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
965 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
966 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
967 case '.':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
968 if (p[1] != '.')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
969 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
970 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
971 case 'L':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
972 if (p[1] != 'i')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
973 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
974 case 'i', 'f', 'F', 'e', 'E':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
975 goto LscanReal;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
976 default:
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
977 }
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
978
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
979 if (overflow)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
980 error(MID.OverflowDecimalNumber);
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
981
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
982 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
983 goto Lfinalize;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
984
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
985 LscanHex:
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
986 assert(digits == 0);
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
987 while (1)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
988 {
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
989 if (*++p == '_')
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
990 continue;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
991 if (!ishexad(*p))
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
992 break;
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
993 ++digits;
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
994 ulong_ *= 16;
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
995 if (*p <= '9')
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
996 ulong_ += *p - '0';
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
997 else if (*p <= 'F')
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
998 ulong_ += *p - 'A' + 10;
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
999 else
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1000 ulong_ += *p - 'a' + 10;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1001 }
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1002
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1003 switch (*p)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1004 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1005 case '.':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1006 if (p[1] != '.')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1007 goto LscanHexReal;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1008 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1009 case 'L':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1010 if (p[1] != 'i')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1011 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1012 case 'i', 'p', 'P':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1013 goto LscanHexReal;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1014 default:
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1015 }
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1016 if (digits == 0)
59
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1017 error(MID.NoDigitsInHexNumber);
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1018 else if (digits > 16)
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1019 {
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1020 // Overflow: skip following digits.
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1021 error(MID.OverflowHexNumber);
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1022 while (ishexad(*++p)) {}
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1023 }
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1024 goto Lfinalize;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1025 LscanHexReal:
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1026 return scanHexReal(t);
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1027
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1028 LscanBin:
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1029 assert(digits == 0);
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1030 while (1)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1031 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1032 if (*++p == '0')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1033 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1034 ++digits;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1035 ulong_ *= 2;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1036 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1037 if (*p == '1')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1038 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1039 ++digits;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1040 ulong_ *= 2;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1041 ulong_ += *p - '0';
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1042 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1043 if (*p == '_')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1044 continue;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1045 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1046 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1047
59
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1048 if (digits == 0)
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1049 error(MID.NoDigitsInBinNumber);
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1050
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1051 if (digits > 64)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1052 error(MID.OverflowBinaryNumber);
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1053 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
1054 goto Lfinalize;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1055
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1056 LscanOct:
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1057 while (1)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1058 {
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1059 if (*++p == '_')
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1060 continue;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1061 if (!isoctal(*p))
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1062 break;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1063 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
1064 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1065 ulong_ *= 8;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1066 ulong_ += *p - '0';
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1067 ++p;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1068 continue;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1069 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1070 // Overflow: skip following digits.
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1071 overflow = true;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1072 while (isdigit(*++p)) {}
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1073 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1074 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1075
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1076 bool hasDecimalDigits;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1077 if (isdigit(*p))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1078 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1079 hasDecimalDigits = true;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1080 while (isdigit(*++p)) {}
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1081 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1082
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1083 // 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
1084 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1085 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1086 case '.':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1087 if (p[1] != '.')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1088 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1089 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1090 case 'L':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1091 if (p[1] != 'i')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1092 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1093 case 'i', 'f', 'F', 'e', 'E':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1094 goto LscanReal;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1095 default:
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1096 }
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1097
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1098 if (hasDecimalDigits)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1099 error(MID.OctalNumberHasDecimals);
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1100 if (overflow)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1101 error(MID.OverflowOctalNumber);
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1102 // goto Lfinalize;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1103
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1104 Lfinalize:
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1105 enum Suffix
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1106 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1107 None = 0,
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1108 Unsigned = 1,
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1109 Long = 2
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1110 }
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1111
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1112 Suffix suffix;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1113 while (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 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1116 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1117 case 'L':
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1118 if (suffix & Suffix.Long)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1119 break;
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1120 suffix |= Suffix.Long;
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1121 ++p;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1122 continue;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1123 case 'u', 'U':
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1124 if (suffix & Suffix.Unsigned)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1125 break;
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1126 suffix |= Suffix.Unsigned;
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1127 ++p;
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1128 continue;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1129 default:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1130 break;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1131 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1132 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1133 }
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1134
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1135 switch (suffix)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1136 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1137 case Suffix.None:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1138 if (ulong_ & 0x8000000000000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1139 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1140 if (isDecimal)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1141 error(MID.OverflowDecimalSign);
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1142 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1143 }
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1144 else if (ulong_ & 0xFFFFFFFF00000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1145 t.type = TOK.Int64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1146 else if (ulong_ & 0x80000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1147 t.type = isDecimal ? TOK.Int64 : TOK.Uint32;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1148 else
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1149 t.type = TOK.Int32;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1150 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1151 case Suffix.Unsigned:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1152 if (ulong_ & 0xFFFFFFFF00000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1153 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1154 else
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1155 t.type = TOK.Uint32;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1156 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1157 case Suffix.Long:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1158 if (ulong_ & 0x8000000000000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1159 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1160 if (isDecimal)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1161 error(MID.OverflowDecimalSign);
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1162 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1163 }
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1164 else
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1165 t.type = TOK.Int64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1166 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1167 case Suffix.Unsigned | Suffix.Long:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1168 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1169 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1170 default:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1171 assert(0);
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1172 }
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1173 t.ulong_ = ulong_;
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
1174 t.end = p;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1175 return;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1176 LscanReal:
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1177 scanReal(t);
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1178 return;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1179 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1180
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1181 /*
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1182 FloatLiteral:= Float[fFL]?i?
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1183 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
1184 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
1185 DecExponent:= [eE][+-]?[0-9][0-9_]*
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1186 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
1187 HexExponent:= [pP][+-]?[0-9][0-9_]*
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1188 */
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1189 void scanReal(ref Token t)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1190 {
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1191 if (*p == '.')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1192 // This function was called by scan() or scanNumber().
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1193 while (isdigit(*++p) || *p == '_') {}
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1194 else
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1195 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1196 // This function was called by scanNumber().
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1197 debug switch (*p)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1198 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1199 case 'L':
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1200 if (p[1] != 'i')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1201 assert(0);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1202 case 'i', 'f', 'F', 'e', 'E': break;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1203 default: assert(0);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1204 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1205 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1206
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1207 // Scan exponent.
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1208 if (*p == 'e' || *p == 'E')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1209 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1210 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1211 if (*p == '-' || *p == '+')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1212 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1213 if (!isdigit(*p))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1214 error(MID.FloatExponentDigitExpected);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1215 else
64
dd4c5a0e47dd - Improved while loop.
aziz
parents: 63
diff changeset
1216 while (isdigit(*++p) || *p == '_') {}
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1217 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1218
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1219 // Copy string to buffer ignoring underscores.
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1220 char[] buffer;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1221 char* end = p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1222 p = t.start;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1223 do
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1224 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1225 if (*p == '_')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1226 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1227 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1228 continue;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1229 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1230 buffer ~= *p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1231 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1232 } while (p != end)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1233 buffer ~= 0;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1234 finalizeFloat(t, buffer);
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1235 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1236
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1237 void scanHexReal(ref Token t)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1238 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1239 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
1240 MID mid;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1241 if (*p == '.')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1242 while (ishexad(*++p) || *p == '_') {}
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1243 if (*p != 'p' && *p != 'P')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1244 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1245 mid = MID.HexFloatExponentRequired;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1246 goto Lerr;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1247 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1248 // 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
1249 char* end = p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1250 p = t.start;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1251 char[] buffer;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1252 do
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1253 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1254 if (*p == '_')
63
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 ++p;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1257 continue;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1258 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1259 buffer ~= *p;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1260 ++p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1261 } while (p != end)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1262
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1263 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
1264 // Scan and copy the exponent.
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1265 buffer ~= 'p';
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1266 size_t bufflen = buffer.length;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1267 while (1)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1268 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1269 if (*++p == '_')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1270 continue;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1271 if (isdigit(*p))
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1272 buffer ~= *p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1273 else
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1274 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1275 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1276 // 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
1277 if (bufflen == buffer.length) {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1278 mid = MID.HexFloatMissingExpDigits;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1279 goto Lerr;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1280 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1281 buffer ~= 0; // Terminate for C functions.
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1282 finalizeFloat(t, buffer);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1283 return;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1284 Lerr:
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1285 t.type = TOK.Float32;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1286 t.end = p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1287 error(mid);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1288 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1289
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1290 void finalizeFloat(ref Token t, string buffer)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1291 {
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1292 // 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
1293 switch (*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 case 'f', 'F':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1296 t.type = TOK.Float32;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1297 t.float_ = strtof(buffer.ptr, null);
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1298 ++p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1299 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1300 case 'L':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1301 t.type = TOK.Float80;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1302 t.real_ = strtold(buffer.ptr, null);
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1303 ++p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1304 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1305 default:
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1306 t.type = TOK.Float64;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1307 t.double_ = strtod(buffer.ptr, null);
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1308 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1309 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1310 if (*p == 'i')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1311 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1312 ++p;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1313 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
1314 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1315 if (getErrno == ERANGE)
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1316 error(MID.OverflowFloatNumber);
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1317 t.end = p;
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
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1320 /// Scan special token: #line Integer [Filespec] EndOfLine
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1321 // TODO: Handle case like: #line 0 #line 2
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1322 void scanSpecialToken()
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1323 {
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
1324 assert(*p == '#');
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1325
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
1326 ++p;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1327 MID mid;
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1328 Token* t;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1329 uint oldloc = this.loc, newloc;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1330
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1331 peek(t);
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 64
diff changeset
1332 if (!(this.loc == oldloc && p == t.start && t.type == TOK.Identifier && t.srcText == "line"))
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1333 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1334 this.loc = oldloc; // reset this.loc because we took a peek at the next token
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1335 mid = MID.ExpectedIdentifierSTLine;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1336 goto Lerr;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1337 }
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1338 p = t.end; // consume token
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1339
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1340 peek(t);
131
ce636f3981cc - Removed TOK.Number.
aziz
parents: 112
diff changeset
1341 if (this.loc == oldloc && t.type == TOK.Int32)
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1342 {
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1343 newloc = t.uint_ - 1;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1344 p = t.end;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1345 }
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1346 else
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1347 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1348 this.loc = oldloc;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1349 mid = MID.ExpectedNumberAfterSTLine;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1350 goto Lerr;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1351 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1352
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1353 peek(t);
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1354 if (this.loc != oldloc)
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1355 {
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1356 this.loc = oldloc;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1357 mid = MID.NewlineInSpecialToken;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1358 goto Lerr;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1359 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1360 if (t.type == TOK.String)
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1361 {
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1362 if (*t.start != '"')
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1363 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1364 mid = MID.ExpectedNormalStringLiteral;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1365 goto Lerr;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1366 }
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 64
diff changeset
1367 fileName = t.srcText[1..$-1]; // contents of "..."
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1368 p = t.end;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1369 }
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 64
diff changeset
1370 else if (t.type == TOK.Identifier && t.srcText == "__FILE__")
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1371 {
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1372 p = t.end;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1373 }
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1374 /+
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1375 peek(t);
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1376 if (this.loc == oldloc && t.type != TOK.EOF)
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1377 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1378 mid = MID.UnterminatedSpecialToken;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1379 goto Lerr;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1380 }
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1381 +/
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1382 while (1)
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1383 {
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1384 switch (*p)
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1385 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1386 case '\r':
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1387 if (p[1] == '\n')
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1388 ++p;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1389 case '\n':
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1390 ++p;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1391 break;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1392 case LS[0]:
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1393 if (p[1] == LS[1] && (p[2] == LS[2] || p[2] == PS[2]))
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1394 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1395 p += 2;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1396 break;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1397 }
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1398 case 0, _Z_:
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1399 break;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1400 default:
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1401 if (isspace(*p)) {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1402 ++p;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1403 continue;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1404 }
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1405 mid = MID.UnterminatedSpecialToken;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1406 goto Lerr;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1407 }
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1408 break;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1409 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1410
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1411 this.loc = newloc;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1412 return;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1413 Lerr:
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1414 error(mid);
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
1415 }
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
1416
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
1417 dchar decodeUTF8()
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1418 {
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1419 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
1420 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
1421 dchar d;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1422 try
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1423 {
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1424 d = std.utf.decode(p[0 .. end-p], idx);
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1425 p += idx -1;
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1426 }
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1427 catch (UtfException e)
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1428 {
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1429 error(MID.InvalidUTF8Sequence);
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1430 // Skip to next valid utf-8 sequence
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1431 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
1432 --p;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1433 }
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1434 return d;
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1435 }
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1436
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1437 void loadKeywords()
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1438 {
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1439 foreach(k; keywords)
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1440 idtable[k.str] = k;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1441 }
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1442 /+
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1443 struct State
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1444 {
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1445 Lexer lexer;
209
921d15b1beeb - Added members token and loc to State.
aziz
parents: 207
diff changeset
1446 Token token;
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1447 char* scanPointer;
209
921d15b1beeb - Added members token and loc to State.
aziz
parents: 207
diff changeset
1448 int loc;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1449 string fileName;
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1450 size_t errorLen;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1451 static State opCall(Lexer lx)
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1452 {
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1453 State s;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1454 s.lexer = lx;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1455 s.token = lx.token;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1456 s.scanPointer = lx.p;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1457 s.loc = lx.loc;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1458 s.fileName = lx.fileName;
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1459 s.errorLen = lx.errors.length;
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1460 return s;
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1461 }
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1462 void restore()
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1463 {
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1464 lexer.p = scanPointer;
209
921d15b1beeb - Added members token and loc to State.
aziz
parents: 207
diff changeset
1465 lexer.token = token;
921d15b1beeb - Added members token and loc to State.
aziz
parents: 207
diff changeset
1466 lexer.loc = loc;
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1467 lexer.fileName = fileName;
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1468 lexer.errors = lexer.errors[0..errorLen];
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1469 }
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1470 }
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1471
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1472 State getState()
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1473 {
212
c9b9c979a620 - Fix: checking for _Z_.
aziz
parents: 209
diff changeset
1474 return State(this);
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1475 }
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1476 +/
105
df34ec47fb81 - Added getState() method and State struct to Lexer.
aziz
parents: 103
diff changeset
1477
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1478 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
1479 {
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1480 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
1481 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
1482 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
1483 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
1484 {
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1485 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
1486 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
1487 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
1488 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
1489 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
1490 }
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1491 }
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1492
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1493 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
1494 {
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1495 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
1496 }
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1497
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1498 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
1499 {
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1500 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
1501 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
1502 }
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
1503
66
8e84db78ad55 - Added support for variadic arguments in error messages.
aziz
parents: 65
diff changeset
1504 void error(MID id, ...)
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
1505 {
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1506 // 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
1507 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
1508 }
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
1509
207
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1510 unittest
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1511 {
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1512 string sourceText = "unittest { }";
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1513 auto lx = new Lexer(sourceText, null);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1514
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1515 Token next;
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1516 lx.peek(next);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1517 assert(next == TOK.Unittest);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1518 lx.peek(next);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1519 assert(next == TOK.LBrace);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1520 lx.peek(next);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1521 assert(next == TOK.RBrace);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1522 lx.peek(next);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1523 assert(next == TOK.EOF);
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1524 writefln("end of peek() unittest");
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1525 }
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1526
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1527 Token[] getTokens()
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1528 {
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1529 Token[] tokens;
239
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1530 // while (nextToken() != TOK.EOF)
7911f6a92e6e - Added 'new' and 'delete' declarations to Token and uncommented next and prev members. Added HEAD to TOK.
aziz
parents: 223
diff changeset
1531 // tokens ~= 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
1532 // tokens ~= this.token;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1533 return tokens;
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
1534 }
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1535
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1536 private void encodeUTF8(inout char[] str, dchar d)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1537 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1538 char[6] b;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1539 assert(d > 0x7F, "check for ASCII char before calling encodeUTF8().");
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1540 if (d < 0x800)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1541 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1542 b[0] = 0xC0 | (d >> 6);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1543 b[1] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1544 str ~= b[0..2];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1545 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1546 else if (d < 0x10000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1547 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1548 b[0] = 0xE0 | (d >> 12);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1549 b[1] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1550 b[2] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1551 str ~= b[0..3];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1552 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1553 else if (d < 0x200000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1554 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1555 b[0] = 0xF0 | (d >> 18);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1556 b[1] = 0x80 | ((d >> 12) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1557 b[2] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1558 b[3] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1559 str ~= b[0..4];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1560 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1561 else if (d < 0x4000000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1562 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1563 b[0] = 0xF8 | (d >> 24);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1564 b[1] = 0x80 | ((d >> 18) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1565 b[2] = 0x80 | ((d >> 12) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1566 b[3] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1567 b[4] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1568 str ~= b[0..5];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1569 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1570 else if (d < 0x80000000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1571 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1572 b[0] = 0xFC | (d >> 30);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1573 b[1] = 0x80 | ((d >> 24) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1574 b[2] = 0x80 | ((d >> 18) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1575 b[3] = 0x80 | ((d >> 12) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1576 b[4] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1577 b[5] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1578 str ~= b[0..6];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1579 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1580 else
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1581 error(MID.InvalidUnicodeCharacter);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1582 }
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
1583 }
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1584
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1585 unittest
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1586 {
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1587 string[] toks = [
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1588 ">", ">=", ">>", ">>=", ">>>", ">>>=", "<", "<=", "<>",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1589 "<>=", "<<", "<<=", "!", "!<", "!>", "!<=", "!>=", "!<>",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1590 "!<>=", ".", "..", "...", "&", "&&", "&=", "+", "++",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1591 "+=", "-", "--", "-=", "=", "==", "~", "~=", "*",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1592 "*=", "/", "/=", "^", "^=", "%", "%=", "(", ")",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1593 "[", "]", "{", "}", ":", ";", "?", ",", "$"
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1594 ];
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1595
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1596 char[] src;
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1597
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1598 foreach (op; toks)
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1599 src ~= op ~ " ";
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1600
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1601 auto lx = new Lexer(src, "");
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1602 auto tokens = lx.getTokens();
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1603
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1604 tokens = tokens[0..$-1]; // exclude TOK.EOF
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1605
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1606 assert(tokens.length == toks.length );
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1607
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1608 foreach (i, t; tokens)
207
481ed2b63a49 - Added contracts to method scan().
aziz
parents: 131
diff changeset
1609 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
1610 }
55
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1611
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
1612 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
1613 {
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
1614 // Numbers unittest
68
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1615 // 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
1616 // 0u 0U 0uL 0UL 0L 0LU 0Lu
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1617 // 0Li 0f 0F 0fi 0Fi 0i
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1618 // 0b_1_LU 0b1000u
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1619 // 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
1620 }
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
1621
55
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1622 /// ASCII character properties table.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1623 static const int ptable[256] = [
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1624 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
1625 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
1626 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
1627 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
1628 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
1629 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
1630 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
1631 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
1632 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
1633 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
1634 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
1635 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
1636 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
1637 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
1638 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
1639 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
1640 ];
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1641
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1642 enum CProperty
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1643 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1644 Octal = 1,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1645 Digit = 1<<1,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1646 Hex = 1<<2,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1647 Alpha = 1<<3,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1648 Underscore = 1<<4,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1649 Whitespace = 1<<5
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1650 }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1651
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1652 const uint EVMask = 0xFF00; // Bit mask for escape value
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1653
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1654 private alias CProperty CP;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1655 int isoctal(char c) { return ptable[c] & CP.Octal; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1656 int isdigit(char c) { return ptable[c] & CP.Digit; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1657 int ishexad(char c) { return ptable[c] & CP.Hex; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1658 int isalpha(char c) { return ptable[c] & CP.Alpha; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1659 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
1660 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
1661 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
1662 int isspace(char c) { return ptable[c] & CP.Whitespace; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1663 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
1664
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1665 version(gen_ptable)
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1666 static this()
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1667 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1668 alias ptable p;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1669 // Initialize character properties table.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1670 for (int i; i < p.length; ++i)
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1671 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1672 p[i] = 0;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1673 if ('0' <= i && i <= '7')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1674 p[i] |= CP.Octal;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1675 if ('0' <= i && i <= '9')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1676 p[i] |= CP.Digit;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1677 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
1678 p[i] |= CP.Hex;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1679 if ('a' <= i && i <= 'z' || 'A' <= i && i <= 'Z')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1680 p[i] |= CP.Alpha;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1681 if (i == '_')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1682 p[i] |= CP.Underscore;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1683 if (i == ' ' || i == '\t' || i == '\v' || i == '\f')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1684 p[i] |= CP.Whitespace;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1685 }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1686 // Store escape sequence values in second byte.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1687 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
1688 p['\''] |= 39 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1689 p['"'] |= 34 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1690 p['?'] |= 63 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1691 p['\\'] |= 92 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1692 p['a'] |= 7 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1693 p['b'] |= 8 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1694 p['f'] |= 12 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1695 p['n'] |= 10 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1696 p['r'] |= 13 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1697 p['t'] |= 9 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1698 p['v'] |= 11 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1699 // Print a formatted array literal.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1700 char[] array = "[\n";
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1701 for (int i; i < p.length; ++i)
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1702 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1703 int c = p[i];
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1704 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
1705 }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1706 array[$-2..$] = "\n]";
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1707 writefln(array);
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1708 }