annotate trunk/src/Lexer.d @ 94:0fe650a7a8d1

- Renamed Type enum to InfoType in module Information. - Changed 'TOK type' member of TypeDotIdExpression to 'Type type'. - Added module Types. - Implemented most of parseBasicType(). - Indented code of 'BasicType . Identifier' parser and creating Type instance.
author aziz
date Fri, 06 Jul 2007 15:23:04 +0000
parents fc645fb2fe72
children 511a1aa25896
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 {
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
27 Token token;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
28 string text;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
29 char* p;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
30 char* end;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
31
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
32 uint loc = 1; /// line of code
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
33
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
34 char[] fileName;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
35
69
24db7c5522d5 - Added module Information for compiler messages like warnings, info and errors to the user.
aziz
parents: 68
diff changeset
36 Information[] errors;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
37
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
38 Identifier[string] idtable;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
39
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
40 this(string text, string fileName)
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
41 {
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
42 this.fileName = fileName;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
43
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
44 this.text = text;
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
45 if (text[$-1] != 0)
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
46 {
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
47 this.text.length = this.text.length + 1;
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
48 this.text[$-1] = 0;
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
49 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
50
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
51 this.p = this.text.ptr;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
52 this.end = this.p + this.text.length;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
53
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
54 loadKeywords();
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
55 }
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
56
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
57 public void scan(out Token t)
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
58 {
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
59 assert(p < end);
0
8ba2570de175 Initial import.
aziz
parents:
diff changeset
60
10
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
61 uint c = *p;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
62
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
63 while (1)
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
64 {
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
65 t.start = p;
16
476e8e55c1d4 - Added Whitespace to the character properties table.
aziz
parents: 15
diff changeset
66
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
67 if (c == 0)
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
68 {
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
69 assert(*p == 0);
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
70 ++p;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
71 assert(p == end);
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
72 t.type = TOK.EOF;
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
73 t.end = p;
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
74 return;
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
75 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
76
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
77 if (c == '\n')
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
78 {
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
79 c = *++p;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
80 ++loc;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
81 continue;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
82 }
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
83 else if (c == '\r')
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
84 {
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
85 c = *++p;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
86 if (c != '\n')
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
87 ++loc;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
88 continue;
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
89 }
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
90 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
91 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
92 p += 3;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
93 c = *p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
94 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
95 }
17
9bd0bac79479 - Removed Whitespace from enum list.
aziz
parents: 16
diff changeset
96
13
e5211758b63c - Added isidbeg() function.
aziz
parents: 12
diff changeset
97 if (isidbeg(c))
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
98 {
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
99 if (c == 'r' && p[1] == '"' && ++p)
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
100 return scanRawStringLiteral(t);
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
101 if (c == 'x' && p[1] == '"')
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
102 return scanHexStringLiteral(t);
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
103 Lidentifier:
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
104 do
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
105 { c = *++p; }
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
106 while (isident(c) || c & 128 && isUniAlpha(decodeUTF8()))
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
107
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
108 t.end = p;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
109
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 64
diff changeset
110 string str = t.srcText;
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
111 Identifier* id = str in idtable;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
112
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
113 if (!id)
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
114 {
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
115 idtable[str] = Identifier.Identifier(TOK.Identifier, str);
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
116 id = str in idtable;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
117 }
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
118 assert(id);
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
119 t.type = id.type;
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
120 return;
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
121 }
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
122
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
123 if (isdigit(c))
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
124 return scanNumber(t);
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
125
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
126 if (c == '/')
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
127 {
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
128 c = *++p;
14
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
129 switch(c)
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
130 {
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
131 case '=':
14
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
132 ++p;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
133 t.type = TOK.DivAssign;
14
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
134 t.end = p;
cdf788d8bdaf - Parsing /= now.
aziz
parents: 13
diff changeset
135 return;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
136 case '+':
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
137 uint level = 1;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
138 while (1)
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
139 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
140 c = *++p;
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
141 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
142 switch (c)
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
143 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
144 case '\r':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
145 if (p[1] == '\n')
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
146 ++p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
147 case '\n':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
148 ++loc;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
149 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
150 case 0, _Z_:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
151 error(MID.UnterminatedNestedComment);
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
152 goto LreturnNC;
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
153 default:
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
154 }
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
155
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
156 c <<= 8;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
157 c |= *++p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
158 switch (c)
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
159 {
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
160 case 0x2F2B: // /+
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
161 ++level;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
162 continue;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
163 case 0x2B2F: // +/
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
164 if (--level == 0)
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
165 {
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
166 ++p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
167 LreturnNC:
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
168 t.type = TOK.Comment;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
169 t.end = p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
170 return;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
171 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
172 continue;
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
173 case 0xE280: // LS[0..1] || PS[0..1]
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
174 if (p[1] == LS[2] || p[1] == PS[2])
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
175 {
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
176 ++loc;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
177 ++p;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
178 }
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
179 continue;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
180 default:
42
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
181 c &= char.max;
c6a9974a6e3c - Optimized scanner of nested comments.
aziz
parents: 41
diff changeset
182 goto LswitchNC;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
183 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
184 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
185 case '*':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
186 while (1)
7
07e45c06a024 - Parsing nested comments correctly now.
aziz
parents: 5
diff changeset
187 {
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
188 c = *++p;
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
189 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
190 switch (c)
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
191 {
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
192 case '\r':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
193 if (p[1] == '\n')
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
194 ++p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
195 case '\n':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
196 ++loc;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
197 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
198 case 0, _Z_:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
199 error(MID.UnterminatedBlockComment);
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
200 goto LreturnBC;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
201 default:
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
202 }
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
203
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
204 c <<= 8;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
205 c |= *++p;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
206 switch (c)
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
207 {
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
208 case 0x2A2F: // */
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
209 ++p;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
210 LreturnBC:
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
211 t.type = TOK.Comment;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
212 t.end = p;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
213 return;
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
214 case 0xE280: // LS[0..1] || PS[0..1]
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
215 if (p[1] == LS[2] || p[1] == PS[2])
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
216 {
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
217 ++loc;
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
218 ++p;
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
219 }
43
1845c23dd056 - Matched some parts of the scanner of block comments to the scanner of nested comments.
aziz
parents: 42
diff changeset
220 continue;
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
221 default:
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
222 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
223 goto LswitchBC;
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
224 }
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
225 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
226 assert(0);
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
227 case '/':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
228 while (1)
7
07e45c06a024 - Parsing nested comments correctly now.
aziz
parents: 5
diff changeset
229 {
8
d4ba94a5a282 - Parsing /* */ comments now.
aziz
parents: 7
diff changeset
230 c = *++p;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
231 switch (c)
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
232 {
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
233 case '\r':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
234 if (p[1] == '\n')
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
235 ++p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
236 case '\n':
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
237 case 0, _Z_:
11
dffcdaa7c47a - Added Unicode line and paragraph separators.
aziz
parents: 10
diff changeset
238 break;
32
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
239 case LS[0]:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
240 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
241 break;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
242 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
243 default:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
244 continue;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
245 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
246 t.type = TOK.Comment;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
247 t.end = p;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
248 return;
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
249 }
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
250 default:
d7011daa4740 - Added missing commas after the items in the messages table.
aziz
parents: 31
diff changeset
251 t.type = TOK.Div;
10
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
252 t.end = p;
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
253 return;
3ee65d6e39c9 - Parsing // comments now.
aziz
parents: 9
diff changeset
254 }
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
255 }
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
256
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
257 switch (c)
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
258 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
259 case '\'':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
260 return scanCharacterLiteral(t);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
261 case '`':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
262 return scanRawStringLiteral(t);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
263 case '"':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
264 return scanNormalStringLiteral(t);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
265 case '\\':
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
266 char[] buffer;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
267 do
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
268 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
269 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
270 c = scanEscapeSequence();
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
271 if (c < 128)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
272 buffer ~= c;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
273 else
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
274 encodeUTF8(buffer, c);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
275 } while (*p == '\\')
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
276 buffer ~= 0;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
277 t.type = TOK.String;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
278 t.str = buffer;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
279 t.end = p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
280 return;
38
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
281 case '>': /* > >= >> >>= >>> >>>= */
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
282 c = *++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
283 switch (c)
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
284 {
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
285 case '=':
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
286 t.type = TOK.GreaterEqual;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
287 goto Lcommon;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
288 case '>':
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
289 if (p[1] == '>')
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
290 {
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
291 ++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
292 if (p[1] == '=')
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
293 { ++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
294 t.type = TOK.URShiftAssign;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
295 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
296 else
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
297 t.type = TOK.URShift;
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 else if (p[1] == '=')
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
300 {
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
301 ++p;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
302 t.type = TOK.RShiftAssign;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
303 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
304 else
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
305 t.type = TOK.RShift;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
306 goto Lcommon;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
307 default:
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
308 t.type = TOK.Greater;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
309 goto Lcommon2;
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
310 }
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
311 assert(0);
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
312 case '<': /* < <= <> <>= << <<= */
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
313 c = *++p;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
314 switch (c)
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
315 {
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
316 case '=':
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
317 t.type = TOK.LessEqual;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
318 goto Lcommon;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
319 case '<':
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
320 if (p[1] == '=') {
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
321 ++p;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
322 t.type = TOK.LShiftAssign;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
323 }
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
324 else
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
325 t.type = TOK.LShift;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
326 goto Lcommon;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
327 case '>':
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
328 if (p[1] == '=') {
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
329 ++p;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
330 t.type = TOK.LorEorG;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
331 }
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
332 else
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
333 t.type = TOK.LorG;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
334 goto Lcommon;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
335 default:
38
640c45aaaaee - Added code for parsing GreaterEqual, Greater, RShiftAssign, RShift, URShiftAssign and URShift tokens.
aziz
parents: 37
diff changeset
336 t.type = TOK.Less;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
337 goto Lcommon2;
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
338 }
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
339 assert(0);
37
7f3bcb97d017 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 36
diff changeset
340 case '!': /* ! !< !> !<= !>= !<> !<>= */
35
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
341 c = *++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
342 switch (c)
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
343 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
344 case '<':
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
345 c = *++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
346 if (c == '>')
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
347 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
348 if (p[1] == '=') {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
349 ++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
350 t.type = TOK.Unordered;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
351 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
352 else
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
353 t.type = TOK.UorE;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
354 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
355 else if (c == '=')
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
356 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
357 t.type = TOK.UorG;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
358 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
359 else {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
360 t.type = TOK.UorGorE;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
361 goto Lcommon2;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
362 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
363 goto Lcommon;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
364 case '>':
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
365 if (p[1] == '=')
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
366 {
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
367 ++p;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
368 t.type = TOK.UorL;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
369 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
370 else
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
371 t.type = TOK.UorLorE;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
372 goto Lcommon;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
373 case '=':
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
374 t.type = TOK.NotEqual;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
375 goto Lcommon;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
376 default:
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
377 t.type = TOK.Not;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
378 goto Lcommon2;
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
379 }
c470b9356e35 - Added code for parsing Unordered, UorE, UorG, UorGorE, UorL, UorLorE, NotEqual and Not tokens.
aziz
parents: 34
diff changeset
380 assert(0);
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
381 case '.': /* . .[0-9] .. ... */
22
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
382 if (p[1] == '.')
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
383 {
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
384 ++p;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
385 if (p[1] == '.') {
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
386 ++p;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
387 t.type = TOK.Ellipses;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
388 }
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
389 else
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
390 t.type = TOK.Slice;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
391 }
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
392 else if (isdigit(p[1]))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
393 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
394 return scanReal(t);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
395 }
22
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
396 else
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
397 t.type = TOK.Dot;
b05fff8e2ce4 - Added code for parsing Dot, Slice and Ellipses tokens.
aziz
parents: 21
diff changeset
398 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
399 case '|': /* | || |= */
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
400 c = *++p;
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
401 if (c == '=')
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
402 t.type = TOK.OrAssign;
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
403 else if (c == '|')
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
404 t.type = TOK.OrLogical;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
405 else {
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
406 t.type = TOK.OrBinary;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
407 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
408 }
23
1a7903701a3d - Added code for parsing OrAssign, OrLogical and OrBinary tokens.
aziz
parents: 22
diff changeset
409 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
410 case '&': /* & && &= */
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
411 c = *++p;
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
412 if (c == '=')
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
413 t.type = TOK.AndAssign;
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
414 else if (c == '&')
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
415 t.type = TOK.AndLogical;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
416 else {
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
417 t.type = TOK.AndBinary;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
418 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
419 }
24
903f91163f23 - Added code for parsing AndAssign, AndLogical and AndBinary tokens.
aziz
parents: 23
diff changeset
420 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
421 case '+': /* + ++ += */
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
422 c = *++p;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
423 if (c == '=')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
424 t.type = TOK.PlusAssign;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
425 else if (c == '+')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
426 t.type = TOK.PlusPlus;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
427 else {
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
428 t.type = TOK.Plus;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
429 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
430 }
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
431 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
432 case '-': /* - -- -= */
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
433 c = *++p;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
434 if (c == '=')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
435 t.type = TOK.MinusAssign;
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
436 else if (c == '-')
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
437 t.type = TOK.MinusMinus;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
438 else {
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
439 t.type = TOK.Minus;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
440 goto Lcommon2;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
441 }
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
442 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
443 case '=': /* = == */
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
444 if (p[1] == '=') {
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
445 ++p;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
446 t.type = TOK.Equal;
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
447 }
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
448 else
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
449 t.type = TOK.Assign;
25
9c866208b3f6 - Added code for parsing PlusAssign, PlusPlus, Plus, MinusAssign, MinusMinus, Minus tokens.
aziz
parents: 24
diff changeset
450 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
451 case '~': /* ~ ~= */
27
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
452 if (p[1] == '=') {
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
453 ++p;
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
454 t.type = TOK.CatAssign;
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
455 }
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
456 else
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
457 t.type = TOK.Tilde;
43b6bf56f0e9 - Added code for parsing CatAssign and Tilde tokens.
aziz
parents: 26
diff changeset
458 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
459 case '*': /* * *= */
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
460 if (p[1] == '=') {
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
461 ++p;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
462 t.type = TOK.MulAssign;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
463 }
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
464 else
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
465 t.type = TOK.Mul;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
466 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
467 case '^': /* ^ ^= */
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
468 if (p[1] == '=') {
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
469 ++p;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
470 t.type = TOK.XorAssign;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
471 }
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
472 else
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
473 t.type = TOK.Xor;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
474 goto Lcommon;
36
3c7210a722f7 - Added code for parsing LorEorG, LorG, LessEqual, LessThan, LShiftAssign and LShift tokens.
aziz
parents: 35
diff changeset
475 case '%': /* % %= */
29
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
476 if (p[1] == '=') {
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
477 ++p;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
478 t.type = TOK.ModAssign;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
479 }
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
480 else
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
481 t.type = TOK.Mod;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
482 goto Lcommon;
ef83eea26bbd - Added code for parsing MulAssign, Mul, ModAssign, Mod, XorAssign and Xor tokens.
aziz
parents: 28
diff changeset
483 // Single character tokens:
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
484 case '(':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
485 t.type = TOK.LParen;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
486 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
487 case ')':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
488 t.type = TOK.RParen;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
489 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
490 case '[':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
491 t.type = TOK.LBracket;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
492 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
493 case ']':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
494 t.type = TOK.RBracket;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
495 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
496 case '{':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
497 t.type = TOK.LBrace;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
498 goto Lcommon;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
499 case '}':
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
500 t.type = TOK.RBrace;
21
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
501 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
502 case ':':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
503 t.type = TOK.Colon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
504 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
505 case ';':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
506 t.type = TOK.Semicolon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
507 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
508 case '?':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
509 t.type = TOK.Question;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
510 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
511 case ',':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
512 t.type = TOK.Comma;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
513 goto Lcommon;
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
514 case '$':
c785c122e4e6 - Added code for parsing Colon, Semicolon, Question, Comma and Dollar.
aziz
parents: 20
diff changeset
515 t.type = TOK.Dollar;
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
516 Lcommon:
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
517 ++p;
26
c3d7373db241 - Added code for parsing Assign and Equal tokens.
aziz
parents: 25
diff changeset
518 Lcommon2:
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
519 t.end = p;
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
520 return;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
521 case '#':
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
522 scanSpecialToken();
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
523 c = *p;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
524 continue;
20
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
525 default:
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
526 }
d6adfbd7c513 - Added code for parsing braces.
aziz
parents: 19
diff changeset
527
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
528 if (c & 128 && isUniAlpha(decodeUTF8()))
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
529 goto Lidentifier;
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
530 c = *++p;
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
531 }
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
532 }
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
533
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
534 void peek(ref Token t)
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
535 {
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
536 char* tmp = p;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
537 uint len = errors.length;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
538 scan(t);
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
539 p = tmp;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
540 if (errors.length != len)
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
541 errors = errors[0..len];
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
542 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
543
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
544 void scanNormalStringLiteral(ref Token t)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
545 {
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
546 assert(*p == '"');
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
547 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
548 char[] buffer;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
549 t.type = TOK.String;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
550 while (1)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
551 {
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
552 switch (*p)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
553 {
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
554 case '"':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
555 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
556 Lreturn:
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
557 buffer ~= 0;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
558 t.str = buffer;
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
559 t.pf = scanPostfix();
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
560 t.end = p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
561 return;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
562 case '\\':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
563 ++p;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
564 dchar d = scanEscapeSequence();
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
565 if (d < 128)
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
566 buffer ~= d;
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
567 else
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
568 encodeUTF8(buffer, d);
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
569 continue;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
570 case '\r':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
571 if (p[1] == '\n')
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
572 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
573 case '\n':
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
574 ++p;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
575 ++loc;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
576 buffer ~= '\n';
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
577 continue;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
578 case 0, _Z_:
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
579 error(MID.UnterminatedString);
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
580 goto Lreturn;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
581 default:
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
582 if (*p & 128)
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
583 {
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
584 // char* begin = p;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
585 dchar d = decodeUTF8();
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
586
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
587 if (d == LSd || d == PSd)
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
588 goto case '\n';
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
589
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
590 // 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
591 // 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
592 // ++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
593 // buffer ~= begin[0 .. p - begin];
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 encodeUTF8(buffer, d);
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
595 continue;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
596 }
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
597 buffer ~= *p++;
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
598 }
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
599 }
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
600 assert(0);
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
601 }
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
602
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
603 void scanCharacterLiteral(ref Token t)
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
604 {
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
605 assert(*p == '\'');
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
606 MID id = MID.UnterminatedCharacterLiteral;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
607 ++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
608 TOK type = TOK.CharLiteral;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
609 switch (*p)
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
610 {
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
611 case '\\':
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
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 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
614 {
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
615 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
616 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
617 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
618 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
619 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
620 }
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
621 t.dchar_ = scanEscapeSequence();
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
622 break;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
623 case '\'':
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
624 ++p;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
625 id = MID.EmptyCharacterLiteral;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
626 case '\n', '\r', 0, _Z_:
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
627 goto Lerr;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
628 default:
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
629 uint c = *p;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
630 if (c & 128)
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
631 {
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
632 c = decodeUTF8();
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
633 if (c == LSd || c == PSd)
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
634 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
635 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
636 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
637 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
638 type = TOK.DCharLiteral;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
639 }
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
640 t.dchar_ = c;
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
641 ++p;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
642 }
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
643
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
644 if (*p == '\'')
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
645 ++p;
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
646 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
647 Lerr:
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
648 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
649 t.type = type;
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
650 t.end = p;
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
651 }
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
652
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
653 char scanPostfix()
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
654 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
655 switch (*p)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
656 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
657 case 'c':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
658 case 'w':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
659 case 'd':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
660 return *p++;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
661 default:
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
662 return 0;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
663 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
664 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
665
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
666 void scanRawStringLiteral(ref Token t)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
667 {
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
668 uint delim = *p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
669 assert(delim == '`' || delim == '"' && p[-1] == 'r');
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
670 t.type = TOK.String;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
671 char[] buffer;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
672 uint c;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
673 while (1)
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
674 {
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
675 c = *++p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
676 switch (c)
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
677 {
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
678 case '\r':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
679 if (p[1] == '\n')
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
680 ++p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
681 c = '\n'; // Convert '\r' and '\r\n' to '\n'
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
682 case '\n':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
683 ++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
684 break;
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
685 case '`':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
686 case '"':
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
687 if (c == delim)
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
688 {
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
689 ++p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
690 t.pf = scanPostfix();
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
691 Lreturn:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
692 t.str = buffer ~ '\0';
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
693 t.end = p;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
694 return;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
695 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
696 break;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
697 case LS[0]:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
698 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
699 {
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
700 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
701 ++p; ++p;
33
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
702 ++loc;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
703 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
704 break;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
705 case 0, _Z_:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
706 if (delim == 'r')
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
707 error(MID.UnterminatedRawString);
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
708 else
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
709 error(MID.UnterminatedBackQuoteString);
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
710 goto Lreturn;
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
711 default:
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
712 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
713 buffer ~= c; // copy character to buffer
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
714 }
cf3047cf3cd2 - Added code for parsing back quote and raw strings.
aziz
parents: 32
diff changeset
715 assert(0);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
716 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
717
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
718 void scanHexStringLiteral(ref Token t)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
719 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
720 assert(p[0] == 'x' && p[1] == '"');
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
721 t.type = TOK.String;
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 uint c;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
724 ubyte[] buffer;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
725 ubyte h; // hex number
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
726 uint n; // number of hex digits
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
727
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
728 ++p;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
729 while (1)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
730 {
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
731 c = *++p;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
732 switch (c)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
733 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
734 case '"':
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
735 ++p;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
736 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
737 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
738 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
739 Lreturn:
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
740 buffer ~= 0;
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
741 t.str = cast(string) buffer;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
742 t.end = p;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
743 return;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
744 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
745 if (p[1] == '\n')
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
746 ++p;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
747 case '\n':
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
748 ++loc;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
749 continue;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
750 default:
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
751 if (ishexad(c))
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
752 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
753 if (c <= '9')
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
754 c -= '0';
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
755 else if (c <= 'F')
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
756 c -= 'A' - 10;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
757 else
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
758 c -= 'a' - 10;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
759
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
760 if (n & 1)
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
761 {
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
762 h <<= 4;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
763 h |= c;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
764 buffer ~= h;
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 else
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
767 h = c;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
768 ++n;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
769 continue;
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
770 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
771 else if (isspace(c))
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
772 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
773
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
774 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
775 {
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
776 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
777 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
778 {
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 ++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
780 ++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
781 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
782 }
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
783 }
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
784 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
785 {
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
786 error(MID.UnterminatedHexString);
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
787 t.pf = 0;
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
788 goto Lreturn;
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
789 }
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
790 error(MID.NonHexCharInHexString, cast(dchar)c);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
791 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
792 }
53
1786c2825491 - Fixed scanner for hex string literals. Terminating string with 0. Relocated some code to the bottom.
aziz
parents: 52
diff changeset
793 assert(0);
31
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
794 }
94f09f4e988e - Added struct for strings to Token with 'pf' = postfix.
aziz
parents: 30
diff changeset
795
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
796 dchar scanEscapeSequence()
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
797 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
798 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
799 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
800 {
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
801 ++p;
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
802 return c;
46
e0dbd4722f74 - Scanning character literals correctly now.
aziz
parents: 45
diff changeset
803 }
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
804 uint digits = 2;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
805
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
806 switch (*p)
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
807 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
808 case 'x':
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
809 c = 0;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
810 while (1)
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
811 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
812 ++p;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
813 if (ishexad(*p))
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
814 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
815 c *= 16;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
816 if (*p <= '9')
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
817 c += *p - '0';
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
818 else if (*p <= 'F')
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
819 c += *p - 'A' + 10;
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
820 else
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
821 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
822
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
823 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
824 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
825 ++p;
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
826 break;
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
827 }
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
828 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
829 else
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
830 {
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
831 error(MID.InsufficientHexDigits);
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
832 break;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
833 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
834 }
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
835 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
836 error(MID.InvalidUnicodeCharacter);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
837 break;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
838 case 'u':
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
839 digits = 4;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
840 goto case 'x';
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
841 case 'U':
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
842 digits = 8;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
843 goto case 'x';
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
844 default:
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
845 if (isoctal(*p))
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
846 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
847 c = 0;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
848 c += *p - '0';
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
849 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
850 if (!isoctal(*p))
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
851 return c;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
852 c *= 8;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
853 c += *p - '0';
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
854 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
855 if (!isoctal(*p))
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
856 return c;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
857 c *= 8;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
858 c += *p - '0';
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
859 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
860 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
861 else if(*p == '&')
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
862 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
863 if (isalpha(*++p))
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
864 {
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
865 while (1)
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 if (isalnum(*++p))
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
868 continue;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
869 if (*p == ';') {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
870 // TODO: convert entity to unicode codepoint.
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
871 ++p;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
872 break;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
873 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
874 else {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
875 error(MID.UnterminatedHTMLEntity);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
876 break;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
877 }
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
878 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
879 }
48
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.InvalidBeginHTMLEntity);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
882 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
883 else
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
884 error(MID.UndefinedEscapeSequence);
45
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
885 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
886
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
887 return c;
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
888 }
cc84b9ca9d0a - Implemented escape sequence scanner.
aziz
parents: 44
diff changeset
889
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
890 /*
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
891 IntegerLiteral:= (Dec|Hex|Bin|Oct)Suffix?
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
892 Dec:= (0|[1-9][0-9_]*)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
893 Hex:= 0[xX] HexDigits
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
894 Bin:= 0[bB][01_]+
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
895 Oct:= 0[0-7_]+
68
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
896 Suffix:= (L[uU]?|[uU]L?)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
897 HexDigits:= [0-9a-zA-Z_]+
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
898
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
899 Invalid: "0b_", "0x_", "._"
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
900 */
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
901 void scanNumber(ref Token t)
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
902 {
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
903 ulong ulong_;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
904 bool overflow;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
905 bool isDecimal;
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
906 size_t digits;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
907
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
908 if (*p != '0')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
909 goto LscanInteger;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
910 ++p; // skip zero
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
911 // check for xX bB ...
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
912 switch (*p)
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 case 'x','X':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
915 goto LscanHex;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
916 case 'b','B':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
917 goto LscanBin;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
918 case 'L':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
919 if (p[1] == 'i')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
920 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
921 case '.':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
922 if (p[1] == '.')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
923 break;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
924 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
925 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
926 case '_':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
927 ++p;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
928 goto LscanOct;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
929 default:
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
930 if (isoctal(*p))
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
931 goto LscanOct;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
932 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
933
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
934 ulong_ = p[-1];
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
935 isDecimal = true;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
936 goto Lfinalize;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
937
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
938 LscanInteger:
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
939 isDecimal = true;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
940 while (1)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
941 {
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
942 if (*++p == '_')
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
943 continue;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
944 if (!isdigit(*p))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
945 break;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
946 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
947 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
948 ulong_ *= 10;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
949 ulong_ += *p - '0';
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
950 continue;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
951 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
952 // Overflow: skip following digits.
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
953 overflow = true;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
954 while (isdigit(*++p)) {}
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
955 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
956 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
957
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
958 // 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
959 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
960 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
961 case '.':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
962 if (p[1] != '.')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
963 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
964 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
965 case 'L':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
966 if (p[1] != 'i')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
967 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
968 case 'i', 'f', 'F', 'e', 'E':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
969 goto LscanReal;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
970 default:
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
971 }
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
972
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
973 if (overflow)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
974 error(MID.OverflowDecimalNumber);
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
975
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
976 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
977 goto Lfinalize;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
978
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
979 LscanHex:
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
980 assert(digits == 0);
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
981 while (1)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
982 {
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
983 if (*++p == '_')
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
984 continue;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
985 if (!ishexad(*p))
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
986 break;
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
987 ++digits;
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
988 ulong_ *= 16;
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
989 if (*p <= '9')
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
990 ulong_ += *p - '0';
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
991 else if (*p <= 'F')
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
992 ulong_ += *p - 'A' + 10;
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
993 else
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
994 ulong_ += *p - 'a' + 10;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
995 }
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
996
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
997 switch (*p)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
998 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
999 case '.':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1000 if (p[1] != '.')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1001 goto LscanHexReal;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1002 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1003 case 'L':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1004 if (p[1] != 'i')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1005 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1006 case 'i', 'p', 'P':
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 default:
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1009 }
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1010 if (digits == 0)
59
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1011 error(MID.NoDigitsInHexNumber);
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1012 else if (digits > 16)
61
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1013 {
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1014 // Overflow: skip following digits.
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1015 error(MID.OverflowHexNumber);
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1016 while (ishexad(*++p)) {}
512cd2248dfc - Fix: issueing error on hexadecimal number overflow.
aziz
parents: 60
diff changeset
1017 }
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1018 goto Lfinalize;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1019 LscanHexReal:
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1020 return scanHexReal(t);
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1021
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1022 LscanBin:
57
c0f1c8be3a47 - Added code for converting hex characters to binary numbers.
aziz
parents: 56
diff changeset
1023 assert(digits == 0);
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1024 while (1)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1025 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1026 if (*++p == '0')
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 ++digits;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1029 ulong_ *= 2;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1030 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1031 if (*p == '1')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1032 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1033 ++digits;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1034 ulong_ *= 2;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1035 ulong_ += *p - '0';
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 == '_')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1038 continue;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1039 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1040 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1041
59
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1042 if (digits == 0)
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1043 error(MID.NoDigitsInBinNumber);
3e594725899a - Issuing error when no digits were found in hex and binary numbers.
aziz
parents: 58
diff changeset
1044
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1045 if (digits > 64)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1046 error(MID.OverflowBinaryNumber);
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1047 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
1048 goto Lfinalize;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1049
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1050 LscanOct:
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1051 while (1)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1052 {
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1053 if (*++p == '_')
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1054 continue;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1055 if (!isoctal(*p))
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1056 break;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1057 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
1058 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1059 ulong_ *= 8;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1060 ulong_ += *p - '0';
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1061 ++p;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1062 continue;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1063 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1064 // Overflow: skip following digits.
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1065 overflow = true;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1066 while (isdigit(*++p)) {}
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1067 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1068 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1069
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1070 bool hasDecimalDigits;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1071 if (isdigit(*p))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1072 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1073 hasDecimalDigits = true;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1074 while (isdigit(*++p)) {}
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1075 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1076
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1077 // 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
1078 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1079 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1080 case '.':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1081 if (p[1] != '.')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1082 goto LscanReal;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1083 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1084 case 'L':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1085 if (p[1] != 'i')
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1086 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1087 case 'i', 'f', 'F', 'e', 'E':
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1088 goto LscanReal;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1089 default:
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1090 }
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1091
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1092 if (hasDecimalDigits)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1093 error(MID.OctalNumberHasDecimals);
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1094 if (overflow)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1095 error(MID.OverflowOctalNumber);
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1096 // goto Lfinalize;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1097
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1098 Lfinalize:
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1099 enum Suffix
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1100 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1101 None = 0,
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1102 Unsigned = 1,
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1103 Long = 2
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1104 }
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1105
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1106 Suffix suffix;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1107 while (1)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1108 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1109 switch (*p)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1110 {
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1111 case 'L':
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1112 if (suffix & Suffix.Long)
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1113 break;
60
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1114 suffix |= Suffix.Long;
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1115 ++p;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1116 continue;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1117 case 'u', 'U':
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1118 if (suffix & Suffix.Unsigned)
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.Unsigned;
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1121 ++p;
32cc23bd217b - Fixed number suffix scanning.
aziz
parents: 59
diff changeset
1122 continue;
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1123 default:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1124 break;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1125 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1126 break;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1127 }
58
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1128
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1129 switch (suffix)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1130 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1131 case Suffix.None:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1132 if (ulong_ & 0x8000000000000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1133 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1134 if (isDecimal)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1135 error(MID.OverflowDecimalSign);
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1136 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1137 }
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1138 else if (ulong_ & 0xFFFFFFFF00000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1139 t.type = TOK.Int64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1140 else if (ulong_ & 0x80000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1141 t.type = isDecimal ? TOK.Int64 : TOK.Uint32;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1142 else
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1143 t.type = TOK.Int32;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1144 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1145 case Suffix.Unsigned:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1146 if (ulong_ & 0xFFFFFFFF00000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1147 t.type = TOK.Uint64;
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.Uint32;
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.Long:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1152 if (ulong_ & 0x8000000000000000)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1153 {
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1154 if (isDecimal)
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1155 error(MID.OverflowDecimalSign);
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1156 t.type = TOK.Uint64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1157 }
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1158 else
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1159 t.type = TOK.Int64;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1160 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1161 case Suffix.Unsigned | Suffix.Long:
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 break;
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1164 default:
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1165 assert(0);
50bb7fc9db44 - The types of integers are recognized now.
aziz
parents: 57
diff changeset
1166 }
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1167 t.ulong_ = ulong_;
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
1168 t.end = p;
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1169 return;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1170 LscanReal:
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1171 scanReal(t);
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1172 return;
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1173 }
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1174
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1175 /*
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1176 FloatLiteral:= Float[fFL]?i?
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1177 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
1178 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
1179 DecExponent:= [eE][+-]?[0-9][0-9_]*
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1180 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
1181 HexExponent:= [pP][+-]?[0-9][0-9_]*
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1182 */
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1183 void scanReal(ref Token t)
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1184 {
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1185 if (*p == '.')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1186 // This function was called by scan() or scanNumber().
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1187 while (isdigit(*++p) || *p == '_') {}
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1188 else
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1189 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1190 // This function was called by scanNumber().
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1191 debug switch (*p)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1192 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1193 case 'L':
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1194 if (p[1] != 'i')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1195 assert(0);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1196 case 'i', 'f', 'F', 'e', 'E': break;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1197 default: assert(0);
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 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1200
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1201 // Scan exponent.
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1202 if (*p == 'e' || *p == 'E')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1203 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1204 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1205 if (*p == '-' || *p == '+')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1206 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1207 if (!isdigit(*p))
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1208 error(MID.FloatExponentDigitExpected);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1209 else
64
dd4c5a0e47dd - Improved while loop.
aziz
parents: 63
diff changeset
1210 while (isdigit(*++p) || *p == '_') {}
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1211 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1212
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1213 // Copy string to buffer ignoring underscores.
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1214 char[] buffer;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1215 char* end = p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1216 p = t.start;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1217 do
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 if (*p == '_')
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1220 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1221 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1222 continue;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1223 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1224 buffer ~= *p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1225 ++p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1226 } while (p != end)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1227 buffer ~= 0;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1228 finalizeFloat(t, buffer);
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1229 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1230
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1231 void scanHexReal(ref Token t)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1232 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1233 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
1234 MID mid;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1235 if (*p == '.')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1236 while (ishexad(*++p) || *p == '_') {}
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1237 if (*p != 'p' && *p != 'P')
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 mid = MID.HexFloatExponentRequired;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1240 goto Lerr;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1241 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1242 // 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
1243 char* end = p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1244 p = t.start;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1245 char[] buffer;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1246 do
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 if (*p == '_')
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1249 {
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1250 ++p;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1251 continue;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1252 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1253 buffer ~= *p;
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1254 ++p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1255 } while (p != end)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1256
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1257 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
1258 // Scan and copy the exponent.
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1259 buffer ~= 'p';
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1260 size_t bufflen = buffer.length;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1261 while (1)
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 if (*++p == '_')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1264 continue;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1265 if (isdigit(*p))
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1266 buffer ~= *p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1267 else
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1268 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1269 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1270 // 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
1271 if (bufflen == buffer.length) {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1272 mid = MID.HexFloatMissingExpDigits;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1273 goto Lerr;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1274 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1275 buffer ~= 0; // Terminate for C functions.
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1276 finalizeFloat(t, buffer);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1277 return;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1278 Lerr:
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1279 t.type = TOK.Float32;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1280 t.end = p;
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1281 error(mid);
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1282 }
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1283
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1284 void finalizeFloat(ref Token t, string buffer)
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1285 {
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1286 // 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
1287 switch (*p)
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1288 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1289 case 'f', 'F':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1290 t.type = TOK.Float32;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1291 t.float_ = strtof(buffer.ptr, null);
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1292 ++p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1293 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1294 case 'L':
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1295 t.type = TOK.Float80;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1296 t.real_ = strtold(buffer.ptr, null);
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1297 ++p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1298 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1299 default:
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1300 t.type = TOK.Float64;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1301 t.double_ = strtod(buffer.ptr, null);
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1302 break;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1303 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1304 if (*p == 'i')
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1305 {
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1306 ++p;
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1307 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
1308 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1309 if (getErrno == ERANGE)
63
c29229fbf2f7 - Recognizing floats that start with a dot.
aziz
parents: 62
diff changeset
1310 error(MID.OverflowFloatNumber);
62
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1311 t.end = p;
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1312 }
96af5653acef - Fixed loop of hex number scanner. Moved checks under the switch block.
aziz
parents: 61
diff changeset
1313
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1314 /// Scan special token: #line Integer [Filespec] EndOfLine
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1315 // TODO: Handle case like: #line 0 #line 2
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1316 void scanSpecialToken()
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1317 {
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
1318 assert(*p == '#');
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1319
47
8aa37a78937b - Properly implemented scanner for normal string literals.
aziz
parents: 46
diff changeset
1320 ++p;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1321 MID mid;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1322 Token t;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1323 uint oldloc = this.loc, newloc;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1324
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1325 peek(t);
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 64
diff changeset
1326 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
1327 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1328 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
1329 mid = MID.ExpectedIdentifierSTLine;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1330 goto Lerr;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1331 }
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1332 p = t.end; // consume token
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1333
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1334 peek(t);
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1335 if (this.loc == oldloc && t.type == TOK.Number)
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1336 {
56
63af7ddf52e1 - Started properly implementing number scanner. Added stub for real numbers.
aziz
parents: 55
diff changeset
1337 newloc = t.uint_ - 1;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1338 p = t.end;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1339 }
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1340 else
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1341 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1342 this.loc = oldloc;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1343 mid = MID.ExpectedNumberAfterSTLine;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1344 goto Lerr;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1345 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1346
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1347 peek(t);
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1348 if (this.loc != oldloc)
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1349 {
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1350 this.loc = oldloc;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1351 mid = MID.NewlineInSpecialToken;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1352 goto Lerr;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1353 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1354 if (t.type == TOK.String)
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 if (*t.start != '"')
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1357 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1358 mid = MID.ExpectedNormalStringLiteral;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1359 goto Lerr;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1360 }
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 64
diff changeset
1361 fileName = t.srcText[1..$-1]; // contents of "..."
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1362 p = t.end;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1363 }
65
6c21ae79fbb3 - Renamed function Token.span to Token.srcText.
aziz
parents: 64
diff changeset
1364 else if (t.type == TOK.Identifier && t.srcText == "__FILE__")
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1365 {
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1366 p = t.end;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1367 }
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1368 /+
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1369 peek(t);
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1370 if (this.loc == oldloc && t.type != TOK.EOF)
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1371 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1372 mid = MID.UnterminatedSpecialToken;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1373 goto Lerr;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1374 }
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1375 +/
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1376 while (1)
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1377 {
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1378 switch (*p)
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1379 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1380 case '\r':
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1381 if (p[1] == '\n')
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1382 ++p;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1383 case '\n':
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1384 ++p;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1385 break;
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1386 case LS[0]:
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1387 if (p[1] == LS[1] && (p[2] == LS[2] || p[2] == PS[2]))
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1388 {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1389 p += 2;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1390 break;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1391 }
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1392 case 0, _Z_:
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1393 break;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1394 default:
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1395 if (isspace(*p)) {
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1396 ++p;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1397 continue;
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1398 }
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1399 mid = MID.UnterminatedSpecialToken;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1400 goto Lerr;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1401 }
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1402 break;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1403 }
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1404
51
cadd2bfe686c - Displaying error messages in XML.
aziz
parents: 50
diff changeset
1405 this.loc = newloc;
30
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1406 return;
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1407 Lerr:
426767b94635 - Added code for parsing the '#line' special token.
aziz
parents: 29
diff changeset
1408 error(mid);
15
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
1409 }
c70c028e47dd - Started implementation of lexing numbers.
aziz
parents: 14
diff changeset
1410
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
1411 dchar decodeUTF8()
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1412 {
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1413 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
1414 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
1415 dchar d;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1416 try
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1417 {
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1418 d = std.utf.decode(p[0 .. end-p], idx);
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1419 p += idx -1;
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1420 }
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1421 catch (UtfException e)
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1422 {
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1423 error(MID.InvalidUTF8Sequence);
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1424 // Skip to next valid utf-8 sequence
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1425 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
1426 --p;
49
7f0fa15dcffc - Renamed function.
aziz
parents: 48
diff changeset
1427 }
12
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1428 return d;
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1429 }
0989206cf73c - Added code to decode Unicode characters in identifiers.
aziz
parents: 11
diff changeset
1430
28
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1431 void loadKeywords()
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1432 {
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1433 foreach(k; keywords)
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1434 idtable[k.str] = k;
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1435 }
3a9daccf7d96 - Added table for identifiers to Lexer.
aziz
parents: 27
diff changeset
1436
66
8e84db78ad55 - Added support for variadic arguments in error messages.
aziz
parents: 65
diff changeset
1437 void error(MID id, ...)
18
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
1438 {
94
0fe650a7a8d1 - Renamed Type enum to InfoType in module Information.
aziz
parents: 82
diff changeset
1439 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
1440 }
c48d2125f1e2 - Moved code for scanning character literals to separate function.
aziz
parents: 17
diff changeset
1441
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
1442 public TOK nextToken()
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
1443 {
4
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1444 scan(this.token);
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1445 return this.token.type;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1446 }
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1447
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1448 Token[] getTokens()
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1449 {
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1450 Token[] tokens;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1451 while (nextToken() != TOK.EOF)
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1452 tokens ~= this.token;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1453 tokens ~= this.token;
92df59b1ec4a - Started implementation of scan().
aziz
parents: 3
diff changeset
1454 return tokens;
3
4bbce78bfb1e - Added TOK enum.
aziz
parents: 2
diff changeset
1455 }
48
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1456
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1457 private void encodeUTF8(inout char[] str, dchar d)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1458 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1459 char[6] b;
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1460 assert(d > 0x7F, "check for ASCII char before calling encodeUTF8().");
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1461 if (d < 0x800)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1462 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1463 b[0] = 0xC0 | (d >> 6);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1464 b[1] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1465 str ~= b[0..2];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1466 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1467 else if (d < 0x10000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1468 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1469 b[0] = 0xE0 | (d >> 12);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1470 b[1] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1471 b[2] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1472 str ~= b[0..3];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1473 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1474 else if (d < 0x200000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1475 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1476 b[0] = 0xF0 | (d >> 18);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1477 b[1] = 0x80 | ((d >> 12) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1478 b[2] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1479 b[3] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1480 str ~= b[0..4];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1481 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1482 else if (d < 0x4000000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1483 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1484 b[0] = 0xF8 | (d >> 24);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1485 b[1] = 0x80 | ((d >> 18) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1486 b[2] = 0x80 | ((d >> 12) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1487 b[3] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1488 b[4] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1489 str ~= b[0..5];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1490 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1491 else if (d < 0x80000000)
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1492 {
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1493 b[0] = 0xFC | (d >> 30);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1494 b[1] = 0x80 | ((d >> 24) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1495 b[2] = 0x80 | ((d >> 18) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1496 b[3] = 0x80 | ((d >> 12) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1497 b[4] = 0x80 | ((d >> 6) & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1498 b[5] = 0x80 | (d & 0x3F);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1499 str ~= b[0..6];
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1500 }
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1501 else
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1502 error(MID.InvalidUnicodeCharacter);
c2e0e0269c28 - Added code for scanning escape string literals.
aziz
parents: 47
diff changeset
1503 }
5
79b4e8848794 - Started writing XML generator.
aziz
parents: 4
diff changeset
1504 }
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1505
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1506 unittest
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1507 {
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1508 string[] toks = [
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1509 ">", ">=", ">>", ">>=", ">>>", ">>>=", "<", "<=", "<>",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1510 "<>=", "<<", "<<=", "!", "!<", "!>", "!<=", "!>=", "!<>",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1511 "!<>=", ".", "..", "...", "&", "&&", "&=", "+", "++",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1512 "+=", "-", "--", "-=", "=", "==", "~", "~=", "*",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1513 "*=", "/", "/=", "^", "^=", "%", "%=", "(", ")",
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1514 "[", "]", "{", "}", ":", ";", "?", ",", "$"
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1515 ];
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1516
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1517 char[] src;
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1518
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1519 foreach (op; toks)
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1520 src ~= op ~ " ";
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1521
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1522 auto lx = new Lexer(src, "");
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1523 auto tokens = lx.getTokens();
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1524
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1525 tokens = tokens[0..$-1]; // exclude TOK.EOF
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1526
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1527 assert(tokens.length == toks.length );
39
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1528
69b940398d7b - Added unittest to test correct parsing of operator tokens.
aziz
parents: 38
diff changeset
1529 foreach (i, t; tokens)
40
9d5ceb0f8be9 - Added more tokens for testing.
aziz
parents: 39
diff changeset
1530 assert(t.span == toks[i], std.string.format("Lexed '%s' but expected '%s'", t.span, toks[i]));
41
2b7be1d67d4d - Optimized scanner of block comments.
aziz
parents: 40
diff changeset
1531 }
55
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1532
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
1533 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
1534 {
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
1535 // Numbers unittest
68
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1536 // 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
1537 // 0u 0U 0uL 0UL 0L 0LU 0Lu
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1538 // 0Li 0f 0F 0fi 0Fi 0i
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1539 // 0b_1_LU 0b1000u
7eb83dd38901 - Simplified suffix rule and added a few more numbers to unittest.
aziz
parents: 67
diff changeset
1540 // 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
1541 }
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
1542
55
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1543 /// ASCII character properties table.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1544 static const int ptable[256] = [
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1545 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
1546 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
1547 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
1548 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
1549 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
1550 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
1551 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
1552 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
1553 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
1554 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
1555 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
1556 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
1557 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
1558 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
1559 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
1560 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
1561 ];
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1562
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1563 enum CProperty
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1564 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1565 Octal = 1,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1566 Digit = 1<<1,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1567 Hex = 1<<2,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1568 Alpha = 1<<3,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1569 Underscore = 1<<4,
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1570 Whitespace = 1<<5
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1571 }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1572
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1573 const uint EVMask = 0xFF00; // Bit mask for escape value
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1574
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1575 private alias CProperty CP;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1576 int isoctal(char c) { return ptable[c] & CP.Octal; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1577 int isdigit(char c) { return ptable[c] & CP.Digit; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1578 int ishexad(char c) { return ptable[c] & CP.Hex; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1579 int isalpha(char c) { return ptable[c] & CP.Alpha; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1580 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
1581 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
1582 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
1583 int isspace(char c) { return ptable[c] & CP.Whitespace; }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1584 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
1585
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1586 version(gen_ptable)
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1587 static this()
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1588 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1589 alias ptable p;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1590 // Initialize character properties table.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1591 for (int i; i < p.length; ++i)
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1592 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1593 p[i] = 0;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1594 if ('0' <= i && i <= '7')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1595 p[i] |= CP.Octal;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1596 if ('0' <= i && i <= '9')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1597 p[i] |= CP.Digit;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1598 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
1599 p[i] |= CP.Hex;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1600 if ('a' <= i && i <= 'z' || 'A' <= i && i <= 'Z')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1601 p[i] |= CP.Alpha;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1602 if (i == '_')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1603 p[i] |= CP.Underscore;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1604 if (i == ' ' || i == '\t' || i == '\v' || i == '\f')
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1605 p[i] |= CP.Whitespace;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1606 }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1607 // Store escape sequence values in second byte.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1608 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
1609 p['\''] |= 39 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1610 p['"'] |= 34 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1611 p['?'] |= 63 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1612 p['\\'] |= 92 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1613 p['a'] |= 7 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1614 p['b'] |= 8 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1615 p['f'] |= 12 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1616 p['n'] |= 10 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1617 p['r'] |= 13 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1618 p['t'] |= 9 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1619 p['v'] |= 11 << 8;
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1620 // Print a formatted array literal.
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1621 char[] array = "[\n";
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1622 for (int i; i < p.length; ++i)
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1623 {
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1624 int c = p[i];
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1625 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
1626 }
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1627 array[$-2..$] = "\n]";
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1628 writefln(array);
5887751f8e04 - Relocated ptable to the bottom of the source file.
aziz
parents: 54
diff changeset
1629 }