]> git.saurik.com Git - cycript.git/blob - Cycript.l.in
YYABORT does not actually lead to a parse failure.
[cycript.git] / Cycript.l.in
1 /* Cycript - Optimizing JavaScript Compiler/Runtime
2 * Copyright (C) 2009-2015 Jay Freeman (saurik)
3 */
4
5 /* GNU Affero General Public License, Version 3 {{{ */
6 /*
7 * This program is free software: you can redistribute it and/or modify
8 * it under the terms of the GNU Affero General Public License as published by
9 * the Free Software Foundation, either version 3 of the License, or
10 * (at your option) any later version.
11
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU Affero General Public License for more details.
16
17 * You should have received a copy of the GNU Affero General Public License
18 * along with this program. If not, see <http://www.gnu.org/licenses/>.
19 **/
20 /* }}} */
21
22 /* XXX: supposedly I will be screwed on very very long multi-line comments and need to replace these with a manual lexer. http://websrv.cs.fsu.edu/~engelen/courses/COP5621/Pr2.pdf */
23
24 %{
25
26 #if defined(__clang__)
27 #pragma clang diagnostic push
28 #pragma clang diagnostic ignored "-Wunknown-pragmas"
29 #pragma clang diagnostic ignored "-Wdeprecated-register"
30 #endif
31
32 #define YYLTYPE CYLocation
33 #include "Cycript.tab.hh"
34 typedef cy::parser::token tk;
35
36 #include "Highlight.hpp"
37
38 #define YY_EXTRA_TYPE CYDriver *
39
40 // do /not/ fold token to the return: this is a macro and the ordering is dependent
41 #define F(value, highlight) do { \
42 int token(value); \
43 @begin ObjectiveC
44 yyextra->no_.AtImplementation = false; \
45 @end
46 yyextra->no_.Function = false; \
47 yyextra->no_.OpenBrace = false; \
48 yylval->highlight_ = highlight; \
49 return token; \
50 } while (false)
51
52 #define A new($pool)
53 #define Y $pool.strmemdup(yytext, yyleng)
54
55 #define I(type, Type, value, highlight) do { \
56 yylval->type ## _ = A CY ## Type; \
57 F(value, highlight); \
58 } while (false)
59
60 #define T yylval->newline_ = yyextra->state_ == CYNewLine; BEGIN(Div);
61 #define C T yyextra->state_ = CYClear;
62 #define R T yyextra->state_ = CYRestricted;
63
64 #define N \
65 if (yyextra->state_ != CYNewLine) { \
66 if (yyextra->state_ != CYRestricted) \
67 yyextra->state_ = CYNewLine; \
68 else { \
69 yyextra->state_ = CYClear; \
70 F(tk::NewLine, hi::Nothing); \
71 } \
72 }
73
74 #define V(more) { \
75 if (const char *nl = reinterpret_cast<const char *>(memchr(yytext, '\n', yyleng))) { \
76 unsigned lines(0); \
77 size_t left; \
78 do { \
79 ++lines; \
80 left = yyleng - (nl - yytext) - 1; \
81 nl = reinterpret_cast<const char *>(memchr(nl + 1, '\n', left)); \
82 } while (nl != NULL); \
83 yylloc->step(); \
84 yylloc->end.lines(lines); \
85 yylloc->end.columns(left); \
86 more \
87 } else L \
88 }
89
90 #define L { \
91 yylloc->step(); \
92 yylloc->end.columns(yyleng); \
93 }
94
95 #define M { \
96 if (yyextra->commented_) { \
97 I(comment, Comment(Y), tk::Comment, hi::Comment); \
98 } \
99 }
100
101 int H(char c) {
102 if (c >= '0' && c <= '9')
103 return c - '0';
104 if (c >= 'a' && c <= 'f')
105 return c - 'a' + 10;
106 if (c >= 'A' && c <= 'F')
107 return c - 'A' + 10;
108 return -1;
109 }
110
111 #define YY_INPUT(data, value, size) { \
112 if (yyextra->data_.eof()) \
113 value = YY_NULL; \
114 else { \
115 yyextra->data_.read(data, size); \
116 size_t copy(yyextra->data_.gcount()); \
117 value = copy == 0 ? YY_NULL : copy; \
118 } \
119 }
120
121 %}
122
123 %option prefix="cy"
124 %option bison-bridge
125 %option bison-locations
126 %option nodefault
127 %option noyywrap
128 %option yylineno
129 %option nounput
130 %option nounistd
131 %option 8bit
132 %option backup
133 %option batch
134 %option never-interactive
135 %option pointer
136 %option reentrant
137 %option stack
138
139 Exponent [eE][+-]?[0-9]+
140 Escape \\[\\'"bfnrtv]|\\0|\\x[0-9a-fA-F]{2}|\\u[0-9a-fA-F]{4}|\\\n
141
142 IdentifierStart [a-zA-Z$_]
143 IdentifierPart [a-zA-Z$_0-9]
144
145 NonTerminator [^\n]
146 BackslashSequence \\{NonTerminator}
147 RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence}
148 RegularExpressionChar [^\n\\/]|{BackslashSequence}
149 RegularExpressionFlags {IdentifierPart}*
150 RegularExpressionChars {RegularExpressionChar}*
151 RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}
152
153 @begin E4X
154 XMLNameStart [a-zA-Z_:]
155 XMLNamePart [a-zA-Z0-9.-_:]
156 XMLName {XMLNameStart}{XMLNamePart}*
157 @end
158
159 %s Div
160 %s RegExp
161
162 @begin E4X
163 %x XMLContent
164 %x XMLTag
165 @end
166
167 %%
168
169 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);
170
171 #![^\n]* L M
172
173 \/\/[^\n]* L M
174
175 /* http://ostermiller.org/findcomment.html */
176 /* XXX: unify these two rules using !? */
177 \/\*!([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V() C I(comment, Comment(Y), tk::Comment, hi::Comment);
178 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V(N) M
179
180 @begin E4X
181 <RegExp>"<>" L F(tk::LeftRight, hi::Structure);
182 <XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
183
184 <RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
185 <RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
186 <RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);
187
188 <XMLTag>"=" L F(tk::Equal, hi::Structure);
189 <XMLTag>">" L F(tk::Right, hi::Structure);
190 <XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
191 <XMLTag>"{" L F(tk::OpenBrace, hi::Structure);
192
193 <XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue, hi::Constant);
194 <XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
195 <XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);
196
197 <XMLContent>"{" L F(tk::OpenBrace, hi::Structure);
198 <XMLContent>"<" L F(tk::Left, hi::Structure);
199 <XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
200 @end
201
202 "..." L C F(tk::PeriodPeriodPeriod, hi::Meta);
203
204 @begin E4X
205 "::" L C F(tk::ColonColon, hi::Operator);
206 ".." L C F(tk::PeriodPeriod, hi::Operator);
207 @end
208
209 @begin E4X ObjectiveC
210 "@" L C F(tk::At, hi::Operator);
211 "#" L C F(tk::Pound, hi::Operator);
212 @end
213
214 "&" L C F(tk::Ampersand, hi::Operator);
215 "&&" L C F(tk::AmpersandAmpersand, hi::Operator);
216 "&=" L C F(tk::AmpersandEqual, hi::Operator);
217 "^" L C F(tk::Carrot, hi::Operator);
218 "^=" L C F(tk::CarrotEqual, hi::Operator);
219 "=" L C F(tk::Equal, hi::Operator);
220 "==" L C F(tk::EqualEqual, hi::Operator);
221 "===" L C F(tk::EqualEqualEqual, hi::Operator);
222 "=>" L C F(tk::EqualRight, hi::Operator);
223 "!" L C F(tk::Exclamation, hi::Operator);
224 "!=" L C F(tk::ExclamationEqual, hi::Operator);
225 "!==" L C F(tk::ExclamationEqualEqual, hi::Operator);
226 "-" L C F(tk::Hyphen, hi::Operator);
227 "-=" L C F(tk::HyphenEqual, hi::Operator);
228 "--" L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
229 "->" L C F(tk::HyphenRight, hi::Operator);
230 "<" L C F(tk::Left, hi::Operator);
231 "<=" L C F(tk::LeftEqual, hi::Operator);
232 "<<" L C F(tk::LeftLeft, hi::Operator);
233 "<<=" L C F(tk::LeftLeftEqual, hi::Operator);
234 "%" L C F(tk::Percent, hi::Operator);
235 "%=" L C F(tk::PercentEqual, hi::Operator);
236 "." L C F(tk::Period, hi::Operator);
237 "|" L C F(tk::Pipe, hi::Operator);
238 "|=" L C F(tk::PipeEqual, hi::Operator);
239 "||" L C F(tk::PipePipe, hi::Operator);
240 "+" L C F(tk::Plus, hi::Operator);
241 "+=" L C F(tk::PlusEqual, hi::Operator);
242 "++" L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
243 ">" L C F(tk::Right, hi::Operator);
244 ">=" L C F(tk::RightEqual, hi::Operator);
245 ">>" L C F(tk::RightRight, hi::Operator);
246 ">>=" L C F(tk::RightRightEqual, hi::Operator);
247 ">>>" L C F(tk::RightRightRight, hi::Operator);
248 ">>>=" L C F(tk::RightRightRightEqual, hi::Operator);
249 "*" L C F(tk::Star, hi::Operator);
250 "*=" L C F(tk::StarEqual, hi::Operator);
251 "~" L C F(tk::Tilde, hi::Operator);
252
253 <Div>"/" L C F(tk::Slash, hi::Operator);
254 <Div>"/=" L C F(tk::SlashEqual, hi::Operator);
255
256 ":" L C F(tk::Colon, hi::Structure);
257 "," L C F(tk::Comma, hi::Structure);
258 "?" L C F(tk::Question, hi::Structure);
259 ";" L C F(tk::SemiColon, hi::Structure);
260
261 "(" L C F(tk::OpenParen, hi::Structure);
262 ")" L C F(tk::CloseParen, hi::Structure);
263
264 "{" L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
265 "}" L C F(tk::CloseBrace, hi::Structure);
266
267 "[" L C F(tk::OpenBracket, hi::Structure);
268 "]" L C F(tk::CloseBracket, hi::Structure);
269
270 "@error" L C F(tk::AtError, hi::Error);
271
272 @begin Java
273 "@class" L C F(tk::AtClass, hi::Meta);
274 @end
275
276 @begin C
277 "typedef" L C I(identifier, Identifier("typedef"), tk::Typedef, hi::Meta);
278 "unsigned" L C I(identifier, Identifier("unsigned"), tk::Unsigned, hi::Type);
279 "signed" L C I(identifier, Identifier("signed"), tk::Signed, hi::Type);
280 "extern" L C I(identifier, Identifier("extern"), tk::Extern, hi::Type);
281 @end
282
283 @begin C
284 "@encode" L C F(tk::AtEncode, hi::Meta);
285 @end
286
287 @begin ObjectiveC
288 "@end" L C F(tk::AtEnd, hi::Meta);
289 "@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta);
290 "@import" L C F(tk::AtImport, hi::Special);
291 "@selector" L C F(tk::AtSelector, hi::Meta);
292
293 "NULL" L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant);
294 "nil" L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant);
295 "YES" L C I(identifier, Identifier("YES"), tk::Yes, hi::Constant);
296 "NO" L C I(identifier, Identifier("NO"), tk::No, hi::Constant);
297
298 "bool" L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type);
299 "BOOL" L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type);
300 "id" L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type);
301 "SEL" L C I(identifier, Identifier("SEL"), tk::Identifier_, hi::Type);
302 @end
303
304 "undefined" L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator);
305
306 "false" L C I(false, False(), tk::False, hi::Constant);
307 "null" L C I(null, Null(), tk::Null, hi::Constant);
308 "true" L C I(true, True(), tk::True, hi::Constant);
309
310 "auto" L C I(word, Word("auto"), tk::Auto, hi::Meta);
311 "break" L R I(word, Word("break"), tk::Break, hi::Control);
312 "case" L C I(word, Word("case"), tk::Case, hi::Control);
313 "catch" L C I(word, Word("catch"), tk::Catch, hi::Control);
314 "continue" L R I(word, Word("continue"), tk::Continue, hi::Control);
315 "default" L C I(word, Word("default"), tk::Default, hi::Control);
316 "delete" L C I(word, Word("delete"), tk::Delete, hi::Operator);
317 "do" L C I(word, Word("do"), tk::Do, hi::Control);
318 "else" L C I(word, Word("else"), tk::Else, hi::Control);
319 "finally" L C I(word, Word("finally"), tk::Finally, hi::Control);
320 "for" L C I(word, Word("for"), tk::For, hi::Control);
321 "function" L C I(word, Word("function"), yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta);
322 "if" L C I(word, Word("if"), tk::If, hi::Control);
323 "in" L C I(word, Word("in"), yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator);
324 "instanceof" L C I(word, Word("instanceof"), tk::InstanceOf, hi::Operator);
325 "new" L C I(word, Word("new"), tk::New, hi::Operator);
326 "return" L R I(word, Word("return"), tk::Return, hi::Control);
327 "switch" L C I(word, Word("switch"), tk::Switch, hi::Control);
328 "this" L C I(this, This(), tk::This, hi::Constant);
329 "throw" L R I(word, Word("throw"), tk::Throw, hi::Control);
330 "try" L C I(word, Word("try"), tk::Try, hi::Control);
331 "typeof" L C I(word, Word("typeof"), tk::TypeOf, hi::Operator);
332 "var" L C I(word, Word("var"), tk::Var, hi::Meta);
333 "void" L C I(word, Word("void"), tk::Void, hi::Operator);
334 "while" L C I(word, Word("while"), tk::While, hi::Control);
335 "with" L C I(word, Word("with"), tk::With, hi::Control);
336
337 "debugger" L C I(word, Word("debugger"), tk::Debugger, hi::Meta);
338
339 "const" L C I(word, Word("const"), tk::Const, hi::Meta);
340
341 "class" L C I(word, Word("class"), tk::Class, hi::Meta);
342 "enum" L C I(word, Word("enum"), tk::Enum, hi::Meta);
343 "export" L C I(word, Word("export"), tk::Export, hi::Meta);
344 "extends" L C I(word, Word("extends"), tk::Extends, hi::Meta);
345 "import" L C I(word, Word("import"), tk::Import, hi::Meta);
346 "super" L C I(word, Word("super"), tk::Super, hi::Constant);
347
348 "implements" L C I(identifier, Identifier("implements"), tk::Implements, hi::Meta);
349 "interface" L C I(identifier, Identifier("interface"), tk::Interface, hi::Meta);
350 "package" L C I(identifier, Identifier("package"), tk::Package, hi::Meta);
351 "private" L C I(identifier, Identifier("private"), tk::Private, hi::Meta);
352 "protected" L C I(identifier, Identifier("protected"), tk::Protected, hi::Meta);
353 "public" L C I(identifier, Identifier("public"), tk::Public, hi::Meta);
354 "static" L C I(identifier, Identifier("static"), tk::Static, hi::Meta);
355
356 "abstract" L C I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta);
357 "boolean" L C I(identifier, Identifier("boolean"), tk::Boolean, hi::Type);
358 "byte" L C I(identifier, Identifier("byte"), tk::Byte, hi::Type);
359 "char" L C I(identifier, Identifier("char"), tk::Char, hi::Type);
360 "double" L C I(identifier, Identifier("double"), tk::Double, hi::Type);
361 "final" L C I(identifier, Identifier("final"), tk::Final, hi::Meta);
362 "float" L C I(identifier, Identifier("float"), tk::Float, hi::Type);
363 "goto" L C I(identifier, Identifier("goto"), tk::Goto, hi::Control);
364 "int" L C I(identifier, Identifier("int"), tk::Int, hi::Type);
365 "long" L C I(identifier, Identifier("long"), tk::Long, hi::Type);
366 "native" L C I(identifier, Identifier("native"), tk::Native, hi::Meta);
367 "short" L C I(identifier, Identifier("short"), tk::Short, hi::Type);
368 "synchronized" L C I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta);
369 "throws" L C I(identifier, Identifier("throws"), tk::Throws, hi::Meta);
370 "transient" L C I(identifier, Identifier("transient"), tk::Transient, hi::Meta);
371 "volatile" L C I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta);
372
373 "let" L C I(identifier, Identifier("let"), tk::Let, hi::Meta);
374 "yield" L C I(identifier, Identifier("yield"), tk::Yield, hi::Control);
375
376 "each" L C I(identifier, Identifier("each"), tk::Each, hi::Control);
377 "of" L C I(identifier, Identifier("of"), tk::Of, hi::Operator);
378
379 @begin E4X
380 "namespace" L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta);
381 "xml" L C I(identifier, Identifier("xml"), tk::XML, hi::Meta);
382 @end
383
384 {IdentifierStart}{IdentifierPart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);
385
386 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?){Exponent}? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
387
388 0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant);
389 0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant);
390 0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant);
391
392 \"([^"\\\n]|{Escape})*\"|'([^'\\\n]|{Escape})*' L C {
393 char *value(A char[yyleng]);
394 char *local(value);
395
396 for (yy_size_t i(1), e(yyleng - 1); i != e; ++i) {
397 char next(yytext[i]);
398
399 if (yytext[i] == '\\')
400 switch (next = yytext[++i]) {
401 case '\n': continue;
402 case '\\': next = '\\'; break;
403 case '\'': next = '\''; break;
404 case '"': next = '"'; break;
405 case 'b': next = '\b'; break;
406 case 'f': next = '\f'; break;
407 case 'n': next = '\n'; break;
408 case 'r': next = '\r'; break;
409 case 't': next = '\t'; break;
410 case 'v': next = '\v'; break;
411 case '0': next = '\0'; break;
412
413 case 'x':
414 next = H(yytext[i + 1]) << 4 | H(yytext[i + 2]);
415 i += 2;
416 break;
417 }
418
419 *local++ = next;
420 }
421
422 *local = '\0';
423 I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
424 }
425
426 \r?\n|\r|\xe2\x80[\xa8\xa9] yylloc->step(); yylloc->end.lines(); N
427
428 [ \t] L
429
430 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
431
432 . L {
433 CYDriver::Error error;
434 error.location_ = *yylloc;
435 error.message_ = "syntax error, unknown token";
436 yyextra->errors_.push_back(error);
437 yyterminate();
438 }
439
440 %%
441
442 void CYDriver::ScannerInit() {
443 cylex_init(&scanner_);
444 cyset_extra(this, scanner_);
445 }
446
447 void CYDriver::ScannerDestroy() {
448 cylex_destroy(scanner_);
449 }
450
451 CYDriver::Condition CYDriver::GetCondition() {
452 switch (yy_top_state(scanner_)) {
453 case RegExp:
454 return RegExpCondition;
455 @begin E4X
456 case XMLContent:
457 return XMLContentCondition;
458 case XMLTag:
459 return XMLTagCondition;
460 @end
461 default:
462 _assert(false);
463 }
464 }
465
466 void CYDriver::SetCondition(Condition condition) {
467 struct yyguts_t *yyg(reinterpret_cast<struct yyguts_t *>(scanner_));
468
469 switch (condition) {
470 case RegExpCondition:
471 BEGIN(RegExp);
472 break;
473 @begin E4X
474 case XMLContentCondition:
475 BEGIN(XMLContent);
476 break;
477 case XMLTagCondition:
478 BEGIN(XMLTag);
479 break;
480 @end
481 default:
482 _assert(false);
483 }
484 }
485
486 void CYDriver::PushCondition(Condition condition) {
487 switch (condition) {
488 case RegExpCondition:
489 yy_push_state(RegExp, scanner_);
490 break;
491 @begin E4X
492 case XMLContentCondition:
493 yy_push_state(XMLContent, scanner_);
494 break;
495 case XMLTagCondition:
496 yy_push_state(XMLTag, scanner_);
497 break;
498 @end
499 default:
500 _assert(false);
501 }
502 }
503
504 void CYDriver::PopCondition() {
505 yy_pop_state(scanner_);
506 }
507
508 #if defined(__clang__)
509 #pragma clang diagnostic pop
510 #endif