]> git.saurik.com Git - cycript.git/blob - Scanner.lpp.in
Move LexSetRegEx to new "post-lookahead" paradigm.
[cycript.git] / Scanner.lpp.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 %top{
23 #if defined(__clang__)
24 #pragma clang diagnostic push
25 #pragma clang diagnostic ignored "-Wdeprecated-register"
26 #pragma clang diagnostic ignored "-Wunused-function"
27 #pragma clang diagnostic ignored "-Wunused-variable"
28 #else
29 #pragma GCC diagnostic push
30 #pragma GCC diagnostic ignored "-Wsign-compare"
31 #pragma GCC diagnostic ignored "-Wunused-function"
32 #pragma GCC diagnostic ignored "-Wunused-variable"
33 #endif
34 }
35
36 %{
37
38 #define YYLTYPE CYLocation
39 #include "Parser.hpp"
40 typedef cy::parser::token tk;
41
42 #include "Highlight.hpp"
43
44 #define YY_EXTRA_TYPE CYDriver *
45
46 #define F(value, highlight) do { \
47 yyextra->newline_ = yyextra->last_; \
48 yyextra->last_ = false; \
49 yyextra->next_ = false; \
50 BEGIN(yyextra->template_.top() ? DivOrTemplateTail : Div); \
51 yylval->highlight_ = highlight; \
52 return value; \
53 } while (false)
54
55 #define S(stack) do { \
56 if (yyextra->stack.size() == 1) \
57 E("invalid brace") \
58 yyextra->stack.pop(); \
59 } while (false)
60
61 #define P yyextra->pool_
62 #define A new(P)
63 #define Y P.strmemdup(yytext, yyleng)
64
65 #define I(type, Type, value, highlight) do { \
66 yylval->semantic_.type ## _ = A CY ## Type; \
67 F(value, highlight); \
68 } while (false)
69
70 #define N \
71 if (yyextra->last_ && yyextra->next_) { \
72 yyextra->last_ = false; \
73 F(tk::NewLine, hi::Nothing); \
74 }
75
76 #define V(more) { \
77 if (const char *nl = reinterpret_cast<const char *>(memchr(yytext, '\n', yyleng))) { \
78 unsigned lines(0); \
79 size_t left; \
80 do { \
81 ++lines; \
82 left = yyleng - (nl - yytext) - 1; \
83 nl = reinterpret_cast<const char *>(memchr(nl + 1, '\n', left)); \
84 } while (nl != NULL); \
85 yylloc->step(); \
86 yylloc->end.lines(lines); \
87 yylloc->end.columns(left); \
88 more \
89 } else L \
90 }
91
92 #define R yylloc->end.columns(yyleng);
93 #define L yylloc->step(); R
94
95 #define H(value, highlight) do { \
96 if (yyextra->highlight_) \
97 F(value, highlight); \
98 } while (false)
99
100 #define M \
101 H(tk::Comment, hi::Comment);
102
103 #define E(message) { \
104 CYDriver::Error error; \
105 error.location_ = *yylloc; \
106 error.message_ = "syntax error, " message; \
107 yyextra->errors_.push_back(error); \
108 yyterminate(); \
109 }
110
111 int X(char c) {
112 if (c >= '0' && c <= '9')
113 return c - '0';
114 if (c >= 'a' && c <= 'f')
115 return c - 'a' + 10;
116 if (c >= 'A' && c <= 'F')
117 return c - 'A' + 10;
118 return -1;
119 }
120
121 template <typename Type_>
122 static void U(Type_ &local, unsigned point) {
123 if (false) {
124 } else if (point < 0x000080) {
125 *local++ = point;
126 } else if (point < 0x000800) {
127 *local++ = 0xc0 | point >> 0x06 & 0x1f;
128 goto one;
129 } else if (point < 0x010000) {
130 *local++ = 0xe0 | point >> 0x0c & 0x0f;
131 goto two;
132 } else if (point < 0x110000) {
133 *local++ = 0xf0 | point >> 0x12 & 0x07;
134 *local++ = 0x80 | point >> 0x0c & 0x3f;
135 two:
136 *local++ = 0x80 | point >> 0x06 & 0x3f;
137 one:
138 *local++ = 0x80 | point >> 0x00 & 0x3f;
139 } else _assert(false);
140 }
141
142 static void U(char *&local, const char *text, yy_size_t &i) {
143 unsigned point;
144
145 char next(text[++i]);
146 if (next != '{') {
147 point = X(text[i + 0]) << 12 | X(text[i + 1]) << 8 | X(text[i + 2]) << 4 | X(text[i + 3]);
148 i += 3;
149 } else {
150 point = 0;
151 for (;;) {
152 next = text[++i];
153 if (next == '}')
154 break;
155 point = (point << 4) | X(next);
156 }
157 }
158
159 U(local, point);
160 }
161
162 #define CYLexBufferPoint(point) do { \
163 std::back_insert_iterator<std::vector<char> > inserter(yyextra->buffer_); \
164 U(inserter, point); \
165 } while (false)
166
167 #define CYLexBufferUnit(value) do { \
168 yyextra->buffer_.push_back(value); \
169 } while (false)
170
171 #define CYLexBufferUnits(data, size) do { \
172 yyextra->buffer_.insert(yyextra->buffer_.end(), data, data + size); \
173 } while (false)
174
175 #define CYLexBufferStart(condition) do { \
176 yyextra->buffer_.clear(); \
177 yy_push_state(condition, yyscanner); \
178 } while (false)
179
180 #define CYLexBufferEnd(type, Type, value, highlight) do { \
181 yy_pop_state(yyscanner); \
182 I(type, Type(P.strmemdup(yyextra->buffer_.data(), yyextra->buffer_.size()), yyextra->buffer_.size()), value, highlight); \
183 } while (false)
184
185 #define YY_INPUT(data, value, size) { \
186 if (yyextra->data_.eof()) \
187 value = YY_NULL; \
188 else { \
189 yyextra->data_.read(data, size); \
190 size_t copy(yyextra->data_.gcount()); \
191 value = copy == 0 ? YY_NULL : copy; \
192 } \
193 }
194
195 %}
196
197 %option prefix="cy"
198 %option bison-bridge
199 %option bison-locations
200 %option nodefault
201 %option noyywrap
202 %option noyylineno
203 %option nounput
204 %option nounistd
205 %option 8bit
206 %option backup
207 %option batch
208 %option never-interactive
209 %option pointer
210 %option reentrant
211 %option stack
212
213 U1 [\x00-\x7f]
214 U0 [\x80-\xbf]
215 U2 [\xc2-\xdf]
216 U3 [\xe0-\xef]
217 U4 [\xf0-\xf4]
218 UN [\xc0-\xc1\xf5-\xff]
219
220 HexDigit [0-9a-fA-F]
221 LineTerminatorSequence \r?\n|\r|\xe2\x80[\xa8\xa9]
222 WhiteSpace [\x09\x0b\x0c\x20]|\xc2\xa0|\xef\xbb\xbf
223 UnicodeEscape \\u({HexDigit}{4}|\{{HexDigit}+\})
224
225 @include NotLineTerminator.l
226 NoneTerminatorCharacter [^\r\n\x80-\xff]|{NotLineTerminator}
227 RegExCharacter [^/[\\]{-}[\r\n\x80-\xff]|{NotLineTerminator}
228 RegClsCharacter [^]\\]{-}[\r\n\x80-\xff]|{NotLineTerminator}
229 CommentCharacter [^*/]{-}[\r\n\x80-\xff]|{NotLineTerminator}
230 SingleCharacter [^'\\]{-}[\r\n\x80-\xff]|{NotLineTerminator}
231 DoubleCharacter [^"\\]{-}[\r\n\x80-\xff]|{NotLineTerminator}
232 PlateCharacter [^$`\\]{-}[\r\n\x80-\xff]|{NotLineTerminator}
233
234 @include UnicodeIDStart.l
235 @include UnicodeIDContinue.l
236 IdentifierMore [$_]
237
238 UnicodeStart {IdentifierMore}|{UnicodeIDStart}
239 UnicodePart {IdentifierMore}|\xe2\x80[\x8c\x8d]|{UnicodeIDContinue}
240 UnicodeFail {U2}|{U3}|{U3}{U0}|{U4}|{U4}{U0}|{U4}{U0}{U0}|{UN}|{U0}
241 UnicodeScrap {UnicodePart}*{UnicodeFail}?
242
243 IdentifierStart {UnicodeStart}|{UnicodeEscape}
244 IdentifierPart {UnicodePart}|{UnicodeEscape}
245 IdentifierFail {UnicodeFail}|\\(u({HexDigit}{0,3}|\{{HexDigit}*))?
246 IdentifierScrap {IdentifierPart}*{IdentifierFail}?
247
248 RegularExpressionBackslashSequence \\{NoneTerminatorCharacter}
249 RegularExpressionClassChars ({RegClsCharacter}|{RegularExpressionBackslashSequence})*
250
251 @begin E4X
252 XMLNameStart [a-zA-Z_:]
253 XMLNamePart [a-zA-Z0-9.-_:]
254 XMLName {XMLNameStart}{XMLNamePart}*
255 @end
256
257 %x RegularExpression
258 %x MultiLine
259
260 %x LegacySingleString
261 %x LegacyDoubleString
262
263 %x StrictSingleString
264 %x StrictDoubleString
265 %x StrictAccentString
266
267 %s Div
268 %s DivOrTemplateTail
269
270 @begin E4X
271 %x XMLContent
272 %x XMLTag
273 @end
274
275 %%
276
277 /* RegEx {{{ */
278 <RegularExpression>{
279 \/{UnicodePart}* R CYLexBufferUnits(yytext, yyleng); CYLexBufferEnd(literal, RegEx, tk::RegularExpressionLiteral, hi::Constant);
280 \/{UnicodePart}*{UnicodeFail} R E("invalid flags")
281
282 {RegExCharacter}+ R CYLexBufferUnits(yytext, yyleng);
283 {RegExCharacter}*{UnicodeFail} R E("invalid character");
284
285 {RegularExpressionBackslashSequence} R CYLexBufferUnits(yytext, yyleng);
286 \\{UnicodeFail}? R E("invalid escape")
287
288 "["{RegularExpressionClassChars}"]" R CYLexBufferUnits(yytext, yyleng);
289 "["{RegularExpressionClassChars}\\? R E("invalid class");
290 "["{RegularExpressionClassChars}\\?{UnicodeFail} R E("invalid character");
291 "["{RegularExpressionClassChars}\\?{LineTerminatorSequence} R E("invalid newline");
292
293 (\\|{RegExCharacter}+)?{LineTerminatorSequence} R E("invalid newline");
294 <<EOF>> R E("unterminated regex")
295 }
296 /* }}} */
297 /* Comment {{{ */
298 #![^\n]* L M
299 \/\/[^\n]* L M
300
301 \/\* L yy_push_state(MultiLine, yyscanner);
302
303 <MultiLine>{
304 \**\*\/ R yy_pop_state(yyscanner); M N
305 \**{LineTerminatorSequence} yylloc->end.lines(); yyextra->last_ = true;
306 \**{CommentCharacter}|\/ R
307 \**({UnicodeFail}|\*) R E("invalid comment");
308 <<EOF>> R E("invalid comment")
309 }
310 /* }}} */
311 /* Element {{{ */
312 @begin E4X
313 <RegExp>"<>" L F(tk::LeftRight, hi::Structure);
314 <XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
315
316 <RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
317 <RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
318 <RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);
319
320 <XMLTag>"=" L F(tk::Equal, hi::Structure);
321 <XMLTag>">" L F(tk::Right, hi::Structure);
322 <XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
323 <XMLTag>"{" L F(tk::OpenBrace, hi::Structure);
324
325 <XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue, hi::Constant);
326 <XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
327 <XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);
328
329 <XMLContent>"{" L F(tk::OpenBrace, hi::Structure);
330 <XMLContent>"<" L F(tk::Left, hi::Structure);
331 <XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
332 @end
333 /* }}} */
334 /* Operator {{{ */
335 "..." L F(tk::PeriodPeriodPeriod, hi::Meta);
336 ".." L E("invalid operator")
337
338 @begin E4X
339 "::" L F(tk::ColonColon, hi::Operator);
340 ".." L F(tk::PeriodPeriod, hi::Operator);
341 @end
342
343 @begin E4X ObjectiveC
344 "@" L F(tk::At, hi::Operator);
345 "#" L F(tk::Pound, hi::Operator);
346 @end
347
348 "&" L F(tk::Ampersand, hi::Operator);
349 "&&" L F(tk::AmpersandAmpersand, hi::Operator);
350 "&=" L F(tk::AmpersandEqual, hi::Operator);
351 "^" L F(tk::Carrot, hi::Operator);
352 "^=" L F(tk::CarrotEqual, hi::Operator);
353 "=" L F(tk::Equal, hi::Operator);
354 "==" L F(tk::EqualEqual, hi::Operator);
355 "===" L F(tk::EqualEqualEqual, hi::Operator);
356 "=>" L F(yyextra->newline_ ? tk::EqualRight_ : tk::EqualRight, hi::Operator);
357 "!" L F(tk::Exclamation, hi::Operator);
358 "!=" L F(tk::ExclamationEqual, hi::Operator);
359 "!==" L F(tk::ExclamationEqualEqual, hi::Operator);
360 "-" L F(tk::Hyphen, hi::Operator);
361 "-=" L F(tk::HyphenEqual, hi::Operator);
362 "--" L F(yyextra->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
363 "->" L F(tk::HyphenRight, hi::Operator);
364 "<" L F(tk::Left, hi::Operator);
365 "<=" L F(tk::LeftEqual, hi::Operator);
366 "<<" L F(tk::LeftLeft, hi::Operator);
367 "<<=" L F(tk::LeftLeftEqual, hi::Operator);
368 "%" L F(tk::Percent, hi::Operator);
369 "%=" L F(tk::PercentEqual, hi::Operator);
370 "." L F(tk::Period, hi::Operator);
371 "|" L F(tk::Pipe, hi::Operator);
372 "|=" L F(tk::PipeEqual, hi::Operator);
373 "||" L F(tk::PipePipe, hi::Operator);
374 "+" L F(tk::Plus, hi::Operator);
375 "+=" L F(tk::PlusEqual, hi::Operator);
376 "++" L F(yyextra->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
377 ">" L F(tk::Right, hi::Operator);
378 ">=" L F(tk::RightEqual, hi::Operator);
379 ">>" L F(tk::RightRight, hi::Operator);
380 ">>=" L F(tk::RightRightEqual, hi::Operator);
381 ">>>" L F(tk::RightRightRight, hi::Operator);
382 ">>>=" L F(tk::RightRightRightEqual, hi::Operator);
383 "*" L F(tk::Star, hi::Operator);
384 "*=" L F(tk::StarEqual, hi::Operator);
385 "~" L F(tk::Tilde, hi::Operator);
386
387 "/" L F(tk::Slash, hi::Operator);
388 "/=" L F(tk::SlashEqual, hi::Operator);
389
390 ":" L F(tk::Colon, hi::Structure);
391 "," L F(tk::Comma, hi::Structure);
392 "?" L F(tk::Question, hi::Structure);
393 ";" L F(tk::SemiColon, hi::Structure);
394
395 "(" L F(tk::OpenParen, hi::Structure);
396 ")" L F(tk::CloseParen, hi::Structure);
397
398 "{" L yyextra->template_.push(false); F(yyextra->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
399 <Div>"}" L S(template_); F(tk::CloseBrace, hi::Structure);
400
401 "[" L F(tk::OpenBracket, hi::Structure);
402 "]" L F(tk::CloseBracket, hi::Structure);
403 /* }}} */
404 /* Keyword {{{ */
405 "@error" L F(tk::At_error_, hi::Error);
406
407 @begin Java
408 "@class" L F(tk::At_class_, hi::Meta);
409 @end
410
411 @begin C
412 "@encode" L F(tk::At_encode_, hi::Meta);
413 @end
414
415 @begin ObjectiveC
416 "@end" L F(tk::At_end_, hi::Meta);
417 "@false" L F(tk::At_false_, hi::Constant);
418 "@implementation" L F(tk::At_implementation_, hi::Meta);
419 "@import" L F(tk::At_import_, hi::Special);
420 "@NO" L F(tk::At_NO_, hi::Constant);
421 "@null" L F(tk::At_null_, hi::Constant);
422 "@selector" L F(tk::At_selector_, hi::Meta);
423 "@true" L F(tk::At_true_, hi::Constant);
424 "@YES" L F(tk::At_YES_, hi::Constant);
425 @end
426
427 @({UnicodeStart}{UnicodeScrap}|{UnicodeFail}) L E("invalid keyword")
428 /* }}} */
429 /* Highlight {{{ */
430 "undefined" L F(tk::_undefined_, hi::Operator);
431
432 @begin ObjectiveC
433 "bool" L F(tk::_bool_, hi::Type);
434 "BOOL" L F(tk::_BOOL_, hi::Type);
435 "id" L F(tk::_id_, hi::Type);
436 "nil" L F(tk::_nil_, hi::Constant);
437 "NULL" L F(tk::_NULL_, hi::Constant);
438 "SEL" L F(tk::_SEL_, hi::Type);
439 @end
440 /* }}} */
441 /* Reserved {{{ */
442 "abstract" L /*FII*/ F(tk::_abstract_, hi::Meta);
443 "await" L /*II?*/ F(tk::_await_, hi::Meta);
444 "boolean" L /*FII*/ F(tk::_boolean_, hi::Type);
445 "break" L /*KKK*/ F(tk::_break_, hi::Control);
446 "byte" L /*FII*/ F(tk::_byte_, hi::Type);
447 "case" L /*KKK*/ F(tk::_case_, hi::Control);
448 "catch" L /*KKK*/ F(tk::_catch_, hi::Control);
449 "char" L /*FII*/ F(tk::_char_, hi::Type);
450 "class" L /*FFK*/ F(tk::_class_, hi::Meta);
451 "const" L /*FFK*/ F(tk::_const_, hi::Meta);
452 "constructor" L /*III*/ F(tk::_constructor_, hi::Special);
453 "continue" L /*KKK*/ F(tk::_continue_, hi::Control);
454 "debugger" L /*FKK*/ F(tk::_debugger_, hi::Meta);
455 "default" L /*KKK*/ F(tk::_default_, hi::Control);
456 "delete" L /*KKK*/ F(tk::_delete_, hi::Operator);
457 "do" L /*KKK*/ F(tk::_do_, hi::Control);
458 "double" L /*FII*/ F(tk::_double_, hi::Type);
459 "else" L /*KKK*/ F(tk::_else_, hi::Control);
460 "enum" L /*FFF*/ F(tk::_enum_, hi::Meta);
461 "export" L /*FFK*/ F(tk::_export_, hi::Meta);
462 "extends" L /*FFK*/ F(tk::_extends_, hi::Meta);
463 "false" L /*LLL*/ F(tk::_false_, hi::Constant);
464 "final" L /*FII*/ F(tk::_final_, hi::Meta);
465 "finally" L /*KKK*/ F(tk::_finally_, hi::Control);
466 "float" L /*FII*/ F(tk::_float_, hi::Type);
467 "for" L /*KKK*/ F(tk::_for_, hi::Control);
468 "from" L /*III*/ F(tk::_from_, hi::Meta);
469 "function" L /*KKK*/ F(tk::_function_, hi::Meta);
470 "goto" L /*FII*/ F(tk::_goto_, hi::Control);
471 "get" L /*III*/ F(tk::_get_, hi::Meta);
472 "if" L /*KKK*/ F(tk::_if_, hi::Control);
473 "implements" L /*FSS*/ F(tk::_implements_, hi::Meta);
474 "import" L /*FFK*/ F(tk::_import_, hi::Meta);
475 "in" L /*KKK*/ F(yyextra->in_.top() ? tk::_in__ : tk::_in_, hi::Operator);
476 "instanceof" L /*KKK*/ F(tk::_instanceof_, hi::Operator);
477 "int" L /*FII*/ F(tk::_int_, hi::Type);
478 "interface" L /*FSS*/ F(tk::_interface_, hi::Meta);
479 "let" L /*IS?*/ F(tk::_let_, hi::Meta);
480 "long" L /*FII*/ F(tk::_long_, hi::Type);
481 "native" L /*FII*/ F(tk::_native_, hi::Meta);
482 "new" L /*KKK*/ F(tk::_new_, hi::Operator);
483 "null" L /*LLL*/ F(tk::_null_, hi::Constant);
484 "package" L /*FSS*/ F(tk::_package_, hi::Meta);
485 "private" L /*FSS*/ F(tk::_private_, hi::Meta);
486 "protected" L /*FSS*/ F(tk::_protected_, hi::Meta);
487 "prototype" L /*III*/ F(tk::_prototype_, hi::Special);
488 "public" L /*FSS*/ F(tk::_public_, hi::Meta);
489 "return" L /*KKK*/ F(yyextra->return_.top() ? tk::_return__ : tk::_return_, hi::Control);
490 "set" L /*III*/ F(tk::_set_, hi::Meta);
491 "short" L /*FII*/ F(tk::_short_, hi::Type);
492 "static" L /*FS?*/ F(tk::_static_, hi::Meta);
493 "super" L /*FFK*/ F(yyextra->super_.top() ? tk::_super__ : tk::_super_, hi::Constant);
494 "switch" L /*KKK*/ F(tk::_switch_, hi::Control);
495 "synchronized" L /*FII*/ F(tk::_synchronized_, hi::Meta);
496 "this" L /*KKK*/ F(tk::_this_, hi::Constant);
497 "throw" L /*KKK*/ F(tk::_throw_, hi::Control);
498 "throws" L /*FII*/ F(tk::_throws_, hi::Meta);
499 "transient" L /*FII*/ F(tk::_transient_, hi::Meta);
500 "true" L /*LLL*/ F(tk::_true_, hi::Constant);
501 "try" L /*KKK*/ F(tk::_try_, hi::Control);
502 "typeof" L /*KKK*/ F(tk::_typeof_, hi::Operator);
503 "var" L /*KKK*/ F(tk::_var_, hi::Meta);
504 "void" L /*KKK*/ F(tk::_void_, hi::Operator);
505 "volatile" L /*FII*/ F(tk::_volatile_, hi::Meta);
506 "while" L /*KKK*/ F(tk::_while_, hi::Control);
507 "with" L /*KKK*/ F(tk::_with_, hi::Control);
508 "yield" L /*IS?*/ F(yyextra->yield_.top() ? tk::_yield__ : tk::_yield_, hi::Control);
509
510 "auto" L F(tk::_auto_, hi::Meta);
511 "each" L F(tk::_each_, hi::Control);
512 "of" L F(tk::_of_, hi::Operator);
513
514 @begin C
515 "extern" L F(tk::_extern_, hi::Type);
516 "signed" L F(tk::_signed_, hi::Type);
517 "typedef" L F(tk::_typedef_, hi::Meta);
518 "unsigned" L F(tk::_unsigned_, hi::Type);
519 @end
520
521 @begin ObjectiveC
522 "NO" L F(tk::_NO_, hi::Constant);
523 "YES" L F(tk::_YES_, hi::Constant);
524 @end
525
526 @begin E4X
527 "namespace" L F(tk::_namespace_, hi::Meta);
528 "xml" L F(tk::_xml_, hi::Meta);
529 @end
530 /* }}} */
531 /* Identifier {{{ */
532 {UnicodeStart}{UnicodePart}* L I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);
533
534 {IdentifierStart}{IdentifierPart}* L {
535 char *value(A char[yyleng + 1]);
536 char *local(value);
537
538 for (yy_size_t i(0), e(yyleng); i != e; ++i) {
539 char next(yytext[i]);
540 if (next != '\\')
541 *local++ = next;
542 else
543 U(local, yytext, ++i);
544 }
545
546 *local = '\0';
547 I(identifier, Identifier(value), tk::Identifier_, hi::Identifier);
548 }
549
550 ({IdentifierStart}{IdentifierPart}*)?{IdentifierFail} L E("invalid identifier")
551 /* }}} */
552 /* Number {{{ */
553 0[0-7]+ L I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant);
554 0[0-9]+ L I(number, Number(strtoull(yytext + 1, NULL, 10)), tk::NumericLiteral, hi::Constant);
555
556 0[xX][0-9a-fA-F]+ L I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant);
557 0[oO][0-7]+ L I(number, Number(strtoull(yytext + 2, NULL, 8)), tk::NumericLiteral, hi::Constant);
558 0[bB][0-1]+ L I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant);
559
560 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)([eE][+-]?[0-9]+)? L I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
561 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)[eE][+-]?{IdentifierScrap} L E("invalid exponent")
562 (\.?[0-9]|(0|[1-9][0-9]*)\.){IdentifierScrap} L E("invalid number")
563 /* }}} */
564 /* String {{{ */
565 \' L CYLexBufferStart(LegacySingleString);
566 <LegacySingleString,StrictSingleString>{
567 \' R CYLexBufferEnd(string, String, tk::StringLiteral, hi::Constant);
568 {SingleCharacter}+ R CYLexBufferUnits(yytext, yyleng);
569 {SingleCharacter}*{UnicodeFail} R E("invalid character");
570 {LineTerminatorSequence} R E("invalid newline");
571 }
572
573 \" L CYLexBufferStart(LegacyDoubleString);
574 <LegacyDoubleString,StrictDoubleString>{
575 \" R CYLexBufferEnd(string, String, tk::StringLiteral, hi::Constant);
576 {DoubleCharacter}+ R CYLexBufferUnits(yytext, yyleng);
577 {DoubleCharacter}*{UnicodeFail} R E("invalid character");
578 {LineTerminatorSequence} R E("invalid newline");
579 }
580 /* }}} */
581 /* Template {{{ */
582 "`" L yyextra->tail_ = false; CYLexBufferStart(StrictAccentString);
583 <DivOrTemplateTail>"}" L yyextra->tail_ = true; S(template_); CYLexBufferStart(StrictAccentString);
584
585 <StrictAccentString>{
586 "`" R CYLexBufferEnd(string, String, yyextra->tail_ ? tk::TemplateTail : tk::NoSubstitutionTemplate, hi::Constant);
587 "${" R yyextra->template_.push(true); CYLexBufferEnd(string, String, yyextra->tail_ ? tk::TemplateMiddle : tk::TemplateHead, hi::Constant);
588
589 "$" R CYLexBufferUnit('$');
590
591 {PlateCharacter}+ R CYLexBufferUnits(yytext, yyleng);
592 {PlateCharacter}*{UnicodeFail} R E("invalid character");
593 {LineTerminatorSequence} R E("invalid newline");
594 }
595 /* }}} */
596 /* Escapes {{{ */
597 <LegacySingleString,LegacyDoubleString>{
598 \\[0-3][0-7][0-7] R CYLexBufferPoint(X(yytext[1]) << 6 | X(yytext[2]) << 3 | X(yytext[3]));
599 \\[0-7][0-7] R CYLexBufferUnit(X(yytext[1]) << 3 | X(yytext[2]));
600 \\[0-7] R CYLexBufferUnit(X(yytext[1]));
601 }
602
603 <StrictSingleString,StrictDoubleString,StrictAccentString>{
604 \\0[0-7] R E("legacy escape");
605 \\0 R CYLexBufferUnit('\0');
606 }
607
608 <LegacySingleString,LegacyDoubleString,StrictSingleString,StrictDoubleString,StrictAccentString>{
609 \\b R CYLexBufferUnit('\b');
610 \\f R CYLexBufferUnit('\f');
611 \\n R CYLexBufferUnit('\n');
612 \\r R CYLexBufferUnit('\r');
613 \\t R CYLexBufferUnit('\t');
614 \\v R CYLexBufferUnit('\v');
615
616 \\x{HexDigit}{2} R CYLexBufferPoint(X(yytext[2]) << 4 | X(yytext[3]));
617
618 \\u{HexDigit}{4} R CYLexBufferPoint(X(yytext[2]) << 12 | X(yytext[3]) << 8 | X(yytext[4]) << 4 | X(yytext[5]));
619
620 \\u\{{HexDigit}+\} R {
621 unsigned point(0);
622 for (yy_size_t i(3); i != yyleng - 1; ++i)
623 point = point << 4 | X(yytext[i]);
624 CYLexBufferPoint(point);
625 }
626
627 \\{LineTerminatorSequence} yylloc->end.lines();
628 \\(.|{NotLineTerminator}) R CYLexBufferUnits(yytext + 1, yyleng - 1);
629
630 \\(x{HexDigit}{0,1}|u({HexDigit}{0,3}|\{{HexDigit}*)|{UnicodeFail})? R E("invalid escape");
631 <<EOF>> R E("invalid string");
632 }
633 /* }}} */
634
635 {LineTerminatorSequence} yylloc->step(); yylloc->end.lines(); yyextra->last_ = true; N
636 {WhiteSpace} L
637
638 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
639
640 . L E("invalid character")
641
642 %%
643
644 #undef yyextra
645 #define yyextra this
646 #define yyscanner scanner_
647
648 void CYDriver::ScannerInit() {
649 cylex_init(&scanner_);
650 cyset_extra(this, scanner_);
651 }
652
653 void CYDriver::ScannerDestroy() {
654 cylex_destroy(scanner_);
655 }
656
657 void CYDriver::SetRegEx(bool equal) {
658 CYLexBufferStart(RegularExpression);
659 CYLexBufferUnit('/');
660 if (equal)
661 CYLexBufferUnit('=');
662 }
663
664 void CYDriver::SetCondition(Condition condition) {
665 struct yyguts_t *yyg(reinterpret_cast<struct yyguts_t *>(scanner_));
666
667 switch (condition) {
668 @begin E4X
669 case XMLContentCondition:
670 BEGIN(XMLContent);
671 break;
672 case XMLTagCondition:
673 BEGIN(XMLTag);
674 break;
675 @end
676 default:
677 _assert(false);
678 }
679 }
680
681 void CYDriver::PushCondition(Condition condition) {
682 switch (condition) {
683 @begin E4X
684 case XMLContentCondition:
685 yy_push_state(XMLContent, scanner_);
686 break;
687 case XMLTagCondition:
688 yy_push_state(XMLTag, scanner_);
689 break;
690 @end
691 default:
692 _assert(false);
693 }
694 }
695
696 void CYDriver::PopCondition() {
697 yy_pop_state(scanner_);
698 }
699
700 bool CYLexerHighlight(hi::Value &highlight, CYLocation &location, void *scanner) {
701 YYSTYPE value;
702 if (cylex(&value, &location, scanner) == 0)
703 return false;
704 highlight = value.highlight_;
705 return true;
706 }
707
708 #if defined(__clang__)
709 #pragma clang diagnostic pop
710 #else
711 // must not pop -Wunused-function
712 //#pragma GCC diagnostic pop
713 #endif