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 #define E(message) { \
 102     CYDriver::Error error; \
 103     error.location_ = *yylloc; \
 104     error.message_ = "syntax error, " message; \
 105     yyextra->errors_.push_back(error); \
 106     yyterminate(); \
 107 }
 108
 109 int H(char c) {
 110     if (c >= '0' && c <= '9')
 111         return c - '0';
 112     if (c >= 'a' && c <= 'f')
 113         return c - 'a' + 10;
 114     if (c >= 'A' && c <= 'F')
 115         return c - 'A' + 10;
 116     return -1;
 117 }
 118
 119 static void U(char *&local, unsigned point) {
 120     if (false) {
 121     } else if (point < 0x000080) {
 122         *local++ = point;
 123     } else if (point < 0x000800) {
 124         *local++ = 0xc0 | point >> 0x06 & 0x1f;
 125         goto one;
 126     } else if (point < 0x010000) {
 127         *local++ = 0xe0 | point >> 0x0c & 0x0f;
 128         goto two;
 129     } else if (point < 0x110000) {
 130         *local++ = 0xf0 | point >> 0x12 & 0x07;
 131         *local++ = 0x80 | point >> 0x0c & 0x3f;
 132       two:
 133         *local++ = 0x80 | point >> 0x06 & 0x3f;
 134       one:
 135         *local++ = 0x80 | point >> 0x00 & 0x3f;
 136     } else _assert(false);
 137 }
 138
 139 static void U(char *&local, const char *text, yy_size_t &i) {
 140     unsigned point;
 141
 142     char next(text[++i]);
 143     if (next != '{') {
 144         point = H(text[i + 0]) << 12 | H(text[i + 1]) << 8 | H(text[i + 2]) << 4 | H(text[i + 3]);
 145         i += 3;
 146     } else {
 147         point = 0;
 148         for (;;) {
 149             next = text[++i];
 150             if (next == '}')
 151                 break;
 152             point = (point << 4) | H(next);
 153         }
 154     }
 155
 156     U(local, point);
 157 }
 158
 159 #define YY_INPUT(data, value, size) { \
 160     if (yyextra->data_.eof()) \
 161         value = YY_NULL; \
 162     else { \
 163         yyextra->data_.read(data, size); \
 164         size_t copy(yyextra->data_.gcount()); \
 165         value = copy == 0 ? YY_NULL : copy; \
 166     } \
 167 }
 168
 169 %}
 170
 171 %option prefix="cy"
 172 %option bison-bridge
 173 %option bison-locations
 174 %option nodefault
 175 %option noyywrap
 176 %option noyylineno
 177 %option nounput
 178 %option nounistd
 179 %option 8bit
 180 %option backup
 181 %option batch
 182 %option never-interactive
 183 %option pointer
 184 %option reentrant
 185 %option stack
 186
 187 %option full
 188 %option ecs
 189
 190 U1 [\x00-\x7f]
 191 U0 [\x80-\xbf]
 192 U2 [\xc2-\xdf]
 193 U3 [\xe0-\xef]
 194 U4 [\xf0-\xf4]
 195
 196 HexDigit [0-9a-fA-F]
 197 LineTerminatorSequence \r?\n|\r|\xe2\x80[\xa8\xa9]
 198 WhiteSpace [\x09\x0b\x0c\x20]|\xc2\xa0|\xef\xbb\xbf
 199 UnicodeEscape \\u({HexDigit}{4}|\{{HexDigit}+\})
 200
 201 OctalEscape \\[1-7]|\\[4-7][0-7]|\\[0-3][0-7][0-7]?
 202 StringEscape \\['"\\bfnrtv]|\\0|{OctalEscape}|\\x{HexDigit}{2}|{UnicodeEscape}
 203 StringExtra {StringEscape}|\\{LineTerminatorSequence}
 204 SingleString ([^'\\\n]|{StringExtra})*
 205 DoubleString ([^"\\\n]|{StringExtra})*
 206 StringPrefix '{SingleString}|\"{DoubleString}
 207
 208 @include UnicodeIDStart.l
 209 @include UnicodeIDContinue.l
 210
 211 IdentifierMore [$_]
 212
 213 UnicodeStart {IdentifierMore}|{UnicodeIDStart}
 214 UnicodePart {IdentifierMore}|\xe2\x80[\x8c\x8d]|{UnicodeIDContinue}
 215 UnicodeFail {U2}|{U3}|{U3}{U0}|{U4}|{U4}{U0}|{U4}{U0}{U0}
 216 UnicodeScrap {UnicodePart}*{UnicodeFail}?
 217
 218 IdentifierStart {UnicodeStart}|{UnicodeEscape}
 219 IdentifierPart {UnicodePart}|{UnicodeEscape}
 220 IdentifierFail {UnicodeFail}|\\(u({HexDigit}{0,3}|\{{HexDigit}*))?
 221 IdentifierScrap {IdentifierPart}*{IdentifierFail}?
 222
 223 NonTerminator [^\n]
 224 BackslashSequence \\{NonTerminator}
 225 RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence}
 226 RegularExpressionChar [^\n\\/]|{BackslashSequence}
 227 RegularExpressionFlags {UnicodePart}*
 228 RegularExpressionChars {RegularExpressionChar}*
 229 RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}
 230
 231 @begin E4X
 232 XMLNameStart [a-zA-Z_:]
 233 XMLNamePart [a-zA-Z0-9.-_:]
 234 XMLName {XMLNameStart}{XMLNamePart}*
 235 @end
 236
 237 %s Div
 238 %s RegExp
 239
 240 @begin E4X
 241 %x XMLContent
 242 %x XMLTag
 243 @end
 244
 245 %%
 246
 247 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);
 248 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags}{UnicodeFail} L E("invalid flags")
 249 <RegExp>\/{RegularExpressionBody}?\\? L E("unterminated regex")
 250
 251 #![^\n]* L M
 252
 253 \/\/[^\n]* L M
 254
 255         /* http://ostermiller.org/findcomment.html */
 256         /* XXX: unify these two rules using !? */
 257 \/\*!([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V() C I(comment, Comment(Y), tk::Comment, hi::Comment);
 258 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V(N) M
 259 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\** V() E("invalid comment")
 260
 261 @begin E4X
 262 <RegExp>"<>"      L F(tk::LeftRight, hi::Structure);
 263 <XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
 264
 265 <RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
 266 <RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
 267 <RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);
 268
 269 <XMLTag>"="  L F(tk::Equal, hi::Structure);
 270 <XMLTag>">"  L F(tk::Right, hi::Structure);
 271 <XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
 272 <XMLTag>"{"  L F(tk::OpenBrace, hi::Structure);
 273
 274 <XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue, hi::Constant);
 275 <XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
 276 <XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);
 277
 278 <XMLContent>"{"  L F(tk::OpenBrace, hi::Structure);
 279 <XMLContent>"<"  L F(tk::Left, hi::Structure);
 280 <XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
 281 @end
 282
 283 "..."  L C F(tk::PeriodPeriodPeriod, hi::Meta);
 284 ".."   L E("invalid operator")
 285
 286 @begin E4X
 287 "::"   L C F(tk::ColonColon, hi::Operator);
 288 ".."   L C F(tk::PeriodPeriod, hi::Operator);
 289 @end
 290
 291 @begin E4X ObjectiveC
 292 "@"    L C F(tk::At, hi::Operator);
 293 "#"    L C F(tk::Pound, hi::Operator);
 294 @end
 295
 296 "&"    L C F(tk::Ampersand, hi::Operator);
 297 "&&"   L C F(tk::AmpersandAmpersand, hi::Operator);
 298 "&="   L C F(tk::AmpersandEqual, hi::Operator);
 299 "^"    L C F(tk::Carrot, hi::Operator);
 300 "^="   L C F(tk::CarrotEqual, hi::Operator);
 301 "="    L C F(tk::Equal, hi::Operator);
 302 "=="   L C F(tk::EqualEqual, hi::Operator);
 303 "==="  L C F(tk::EqualEqualEqual, hi::Operator);
 304 "=>"   L C F(yylval->newline_ ? tk::EqualRight_ : tk::EqualRight, hi::Operator);
 305 "!"    L C F(tk::Exclamation, hi::Operator);
 306 "!="   L C F(tk::ExclamationEqual, hi::Operator);
 307 "!=="  L C F(tk::ExclamationEqualEqual, hi::Operator);
 308 "-"    L C F(tk::Hyphen, hi::Operator);
 309 "-="   L C F(tk::HyphenEqual, hi::Operator);
 310 "--"   L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
 311 "->"   L C F(tk::HyphenRight, hi::Operator);
 312 "<"    L C F(tk::Left, hi::Operator);
 313 "<="   L C F(tk::LeftEqual, hi::Operator);
 314 "<<"   L C F(tk::LeftLeft, hi::Operator);
 315 "<<="  L C F(tk::LeftLeftEqual, hi::Operator);
 316 "%"    L C F(tk::Percent, hi::Operator);
 317 "%="   L C F(tk::PercentEqual, hi::Operator);
 318 "."    L C F(tk::Period, hi::Operator);
 319 "|"    L C F(tk::Pipe, hi::Operator);
 320 "|="   L C F(tk::PipeEqual, hi::Operator);
 321 "||"   L C F(tk::PipePipe, hi::Operator);
 322 "+"    L C F(tk::Plus, hi::Operator);
 323 "+="   L C F(tk::PlusEqual, hi::Operator);
 324 "++"   L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
 325 ">"    L C F(tk::Right, hi::Operator);
 326 ">="   L C F(tk::RightEqual, hi::Operator);
 327 ">>"   L C F(tk::RightRight, hi::Operator);
 328 ">>="  L C F(tk::RightRightEqual, hi::Operator);
 329 ">>>"  L C F(tk::RightRightRight, hi::Operator);
 330 ">>>=" L C F(tk::RightRightRightEqual, hi::Operator);
 331 "*"    L C F(tk::Star, hi::Operator);
 332 "*="   L C F(tk::StarEqual, hi::Operator);
 333 "~"    L C F(tk::Tilde, hi::Operator);
 334
 335 <Div>"/"  L C F(tk::Slash, hi::Operator);
 336 <Div>"/=" L C F(tk::SlashEqual, hi::Operator);
 337
 338 ":"    L C F(tk::Colon, hi::Structure);
 339 ","    L C F(tk::Comma, hi::Structure);
 340 "?"    L C F(tk::Question, hi::Structure);
 341 ";"    L C F(tk::SemiColon, hi::Structure);
 342
 343 "("    L C F(tk::OpenParen, hi::Structure);
 344 ")"    L C F(tk::CloseParen, hi::Structure);
 345
 346 "{"    L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
 347 "}"    L C F(tk::CloseBrace, hi::Structure);
 348
 349 "["    L C F(tk::OpenBracket, hi::Structure);
 350 "]"    L C F(tk::CloseBracket, hi::Structure);
 351
 352 "@error"          L C F(tk::AtError, hi::Error);
 353
 354 @begin Java
 355 "@class"          L C F(tk::AtClass, hi::Meta);
 356 @end
 357
 358 @begin C
 359 "typedef"         L C I(identifier, Identifier("typedef"), tk::Typedef, hi::Meta);
 360 "unsigned"        L C I(identifier, Identifier("unsigned"), tk::Unsigned, hi::Type);
 361 "signed"          L C I(identifier, Identifier("signed"), tk::Signed, hi::Type);
 362 "extern"          L C I(identifier, Identifier("extern"), tk::Extern, hi::Type);
 363 @end
 364
 365 @begin C
 366 "@encode"         L C F(tk::AtEncode, hi::Meta);
 367 @end
 368
 369 @begin ObjectiveC
 370 "@end"            L C F(tk::AtEnd, hi::Meta);
 371 "@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta);
 372 "@import"         L C F(tk::AtImport, hi::Special);
 373 "@selector"       L C F(tk::AtSelector, hi::Meta);
 374
 375 "@null"           L C F(tk::AtNull, hi::Constant);
 376 "@YES"            L C F(tk::AtYes, hi::Constant);
 377 "@NO"             L C F(tk::AtNo, hi::Constant);
 378 "@true"           L C F(tk::AtTrue, hi::Constant);
 379 "@false"          L C F(tk::AtFalse, hi::Constant);
 380
 381 "NULL"         L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant);
 382 "nil"          L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant);
 383 "YES"          L C I(identifier, Identifier("YES"), tk::Yes, hi::Constant);
 384 "NO"           L C I(identifier, Identifier("NO"), tk::No, hi::Constant);
 385
 386 "bool"         L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type);
 387 "BOOL"         L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type);
 388 "id"           L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type);
 389 "SEL"          L C I(identifier, Identifier("SEL"), tk::Identifier_, hi::Type);
 390 @end
 391
 392 "undefined"    L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator);
 393
 394 "false"        L C I(false, False(), tk::False, hi::Constant);
 395 "null"         L C I(null, Null(), tk::Null, hi::Constant);
 396 "true"         L C I(true, True(), tk::True, hi::Constant);
 397
 398 "auto"         L C I(word, Word("auto"), tk::Auto, hi::Meta);
 399 "break"        L R I(word, Word("break"), tk::Break, hi::Control);
 400 "case"         L C I(word, Word("case"), tk::Case, hi::Control);
 401 "catch"        L C I(word, Word("catch"), tk::Catch, hi::Control);
 402 "continue"     L R I(word, Word("continue"), tk::Continue, hi::Control);
 403 "default"      L C I(word, Word("default"), tk::Default, hi::Control);
 404 "delete"       L C I(word, Word("delete"), tk::Delete, hi::Operator);
 405 "do"           L C I(word, Word("do"), tk::Do, hi::Control);
 406 "else"         L C I(word, Word("else"), tk::Else, hi::Control);
 407 "finally"      L C I(word, Word("finally"), tk::Finally, hi::Control);
 408 "for"          L C I(word, Word("for"), tk::For, hi::Control);
 409 "function"     L C I(word, Word("function"), yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta);
 410 "if"           L C I(word, Word("if"), tk::If, hi::Control);
 411 "in"           L C I(word, Word("in"), yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator);
 412 "instanceof"   L C I(word, Word("instanceof"), tk::InstanceOf, hi::Operator);
 413 "new"          L C I(word, Word("new"), tk::New, hi::Operator);
 414 "return"       L R I(word, Word("return"), tk::Return, hi::Control);
 415 "switch"       L C I(word, Word("switch"), tk::Switch, hi::Control);
 416 "this"         L C I(this, This(), tk::This, hi::Constant);
 417 "throw"        L R I(word, Word("throw"), tk::Throw, hi::Control);
 418 "try"          L C I(word, Word("try"), tk::Try, hi::Control);
 419 "typeof"       L C I(word, Word("typeof"), tk::TypeOf, hi::Operator);
 420 "var"          L C I(word, Word("var"), tk::Var, hi::Meta);
 421 "void"         L C I(word, Word("void"), tk::Void, hi::Operator);
 422 "while"        L C I(word, Word("while"), tk::While, hi::Control);
 423 "with"         L C I(word, Word("with"), tk::With, hi::Control);
 424
 425 "debugger"     L C I(word, Word("debugger"), tk::Debugger, hi::Meta);
 426
 427 "const"        L C I(word, Word("const"), tk::Const, hi::Meta);
 428
 429 "class"        L C I(word, Word("class"), tk::Class, hi::Meta);
 430 "enum"         L C I(word, Word("enum"), tk::Enum, hi::Meta);
 431 "export"       L C I(word, Word("export"), tk::Export, hi::Meta);
 432 "extends"      L C I(word, Word("extends"), tk::Extends, hi::Meta);
 433 "import"       L C I(word, Word("import"), tk::Import, hi::Meta);
 434 "super"        L C I(word, Word("super"), tk::Super, hi::Constant);
 435
 436 "implements"   L C I(identifier, Identifier("implements"), tk::Implements, hi::Meta);
 437 "interface"    L C I(identifier, Identifier("interface"), tk::Interface, hi::Meta);
 438 "package"      L C I(identifier, Identifier("package"), tk::Package, hi::Meta);
 439 "private"      L C I(identifier, Identifier("private"), tk::Private, hi::Meta);
 440 "protected"    L C I(identifier, Identifier("protected"), tk::Protected, hi::Meta);
 441 "public"       L C I(identifier, Identifier("public"), tk::Public, hi::Meta);
 442 "static"       L C I(identifier, Identifier("static"), tk::Static, hi::Meta);
 443
 444 "abstract"     L C I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta);
 445 "boolean"      L C I(identifier, Identifier("boolean"), tk::Boolean, hi::Type);
 446 "byte"         L C I(identifier, Identifier("byte"), tk::Byte, hi::Type);
 447 "char"         L C I(identifier, Identifier("char"), tk::Char, hi::Type);
 448 "double"       L C I(identifier, Identifier("double"), tk::Double, hi::Type);
 449 "final"        L C I(identifier, Identifier("final"), tk::Final, hi::Meta);
 450 "float"        L C I(identifier, Identifier("float"), tk::Float, hi::Type);
 451 "goto"         L C I(identifier, Identifier("goto"), tk::Goto, hi::Control);
 452 "int"          L C I(identifier, Identifier("int"), tk::Int, hi::Type);
 453 "long"         L C I(identifier, Identifier("long"), tk::Long, hi::Type);
 454 "native"       L C I(identifier, Identifier("native"), tk::Native, hi::Meta);
 455 "short"        L C I(identifier, Identifier("short"), tk::Short, hi::Type);
 456 "synchronized" L C I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta);
 457 "throws"       L C I(identifier, Identifier("throws"), tk::Throws, hi::Meta);
 458 "transient"    L C I(identifier, Identifier("transient"), tk::Transient, hi::Meta);
 459 "volatile"     L C I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta);
 460
 461 "let"          L C I(identifier, Identifier("let"), tk::Let, hi::Meta);
 462 "yield"        L R I(identifier, Identifier("yield"), tk::Yield, hi::Control);
 463
 464 "each"         L C I(identifier, Identifier("each"), tk::Each, hi::Control);
 465 "of"           L C I(identifier, Identifier("of"), tk::Of, hi::Operator);
 466
 467 @begin E4X
 468 "namespace"    L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta);
 469 "xml"          L C I(identifier, Identifier("xml"), tk::XML, hi::Meta);
 470 @end
 471
 472 {UnicodeStart}{UnicodePart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);
 473
 474 {IdentifierStart}{IdentifierPart}* L C {
 475     char *value(A char[yyleng + 1]);
 476     char *local(value);
 477
 478     for (yy_size_t i(0), e(yyleng); i != e; ++i) {
 479         char next(yytext[i]);
 480         if (next != '\\')
 481             *local++ = next;
 482         else
 483             U(local, yytext, ++i);
 484     }
 485
 486     *local = '\0';
 487     I(identifier, Identifier(value), tk::Identifier_, hi::Identifier);
 488 }
 489
 490 ({IdentifierStart}{IdentifierPart}*)?{IdentifierFail} L E("invalid identifier")
 491
 492 0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant);
 493 0[0-9]+ L C I(number, Number(strtoull(yytext + 1, NULL, 10)), tk::NumericLiteral, hi::Constant);
 494
 495 0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant);
 496 0[oO][0-7]+ L C I(number, Number(strtoull(yytext + 2, NULL, 8)), tk::NumericLiteral, hi::Constant);
 497 0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant);
 498
 499 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)([eE][+-]?[0-9]+)? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
 500 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)[eE][+-]?{IdentifierScrap} L E("invalid exponent")
 501 (\.?[0-9]|(0|[1-9][0-9]*)\.){IdentifierScrap} L E("invalid number")
 502
 503 '{SingleString}'|\"{DoubleString}\" L C {
 504     char *value(A char[yyleng]);
 505     char *local(value);
 506
 507     for (yy_size_t i(1), e(yyleng - 1); i != e; ++i) {
 508         char next(yytext[i]);
 509
 510         if (yytext[i] == '\\')
 511             // XXX: support more line continuation characters
 512             if (false) line: {
 513                 yylloc->end.lines(1);
 514                 yylloc->end.columns(yyleng - i);
 515             } else switch (next = yytext[++i]) {
 516                 case '\n': goto line;
 517
 518                 case '\\': next = '\\'; break;
 519                 case '\'': next = '\''; break;
 520                 case '"': next = '"'; break;
 521                 case 'b': next = '\b'; break;
 522                 case 'f': next = '\f'; break;
 523                 case 'n': next = '\n'; break;
 524                 case 'r': next = '\r'; break;
 525                 case 't': next = '\t'; break;
 526                 case 'v': next = '\v'; break;
 527
 528                 case '0': case '1': case '2': case '3':
 529                     if (yytext[i + 1] < '0' || yytext[i + 1] > '7')
 530                         next = H(yytext[i]), i += 0;
 531                     else if (yytext[i + 2] < '0' || yytext[i + 2] > '7')
 532                         next = H(yytext[i]) << 3 | H(yytext[i + 1]), i += 1;
 533                     else
 534                         next = H(yytext[i]) << 6 | H(yytext[i + 1]) << 3 | H(yytext[i + 2]), i += 2;
 535                 break;
 536
 537                 case '4': case '5': case '6': case '7':
 538                     if (yytext[i + 1] < '0' || yytext[i + 1] > '7')
 539                         next = H(yytext[i]), i += 0;
 540                     else
 541                         next = H(yytext[i]) << 3 | H(yytext[i + 1]), i += 1;
 542                 break;
 543
 544                 case 'x':
 545                     U(local, H(yytext[i + 1]) << 4 | H(yytext[i + 2]));
 546                     i += 2;
 547                 continue;
 548
 549                 case 'u':
 550                     U(local, yytext, i);
 551                 continue;
 552             }
 553
 554         *local++ = next;
 555     }
 556
 557     *local = '\0';
 558     I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
 559 }
 560
 561 {StringPrefix}\\(x.{0,2}|u([^{].{0,3}|\{[^}]*)?|{UnicodeFail})? L E("invalid escape")
 562 {StringPrefix} L E("invalid string")
 563
 564 {LineTerminatorSequence} yylloc->step(); yylloc->end.lines(); N
 565
 566 {WhiteSpace} L
 567
 568 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
 569
 570 @({UnicodeStart}{UnicodeScrap}|{UnicodeFail}) L E("invalid keyword")
 571
 572 . L E("invalid character")
 573
 574 %%
 575
 576 void CYDriver::ScannerInit() {
 577     cylex_init(&scanner_);
 578     cyset_extra(this, scanner_);
 579 }
 580
 581 void CYDriver::ScannerDestroy() {
 582     cylex_destroy(scanner_);
 583 }
 584
 585 CYDriver::Condition CYDriver::GetCondition() {
 586     switch (yy_top_state(scanner_)) {
 587         case RegExp:
 588             return RegExpCondition;
 589 @begin E4X
 590         case XMLContent:
 591             return XMLContentCondition;
 592         case XMLTag:
 593             return XMLTagCondition;
 594 @end
 595         default:
 596             _assert(false);
 597     }
 598 }
 599
 600 void CYDriver::SetCondition(Condition condition) {
 601     struct yyguts_t *yyg(reinterpret_cast<struct yyguts_t *>(scanner_));
 602
 603     switch (condition) {
 604         case RegExpCondition:
 605             BEGIN(RegExp);
 606             break;
 607 @begin E4X
 608         case XMLContentCondition:
 609             BEGIN(XMLContent);
 610             break;
 611         case XMLTagCondition:
 612             BEGIN(XMLTag);
 613             break;
 614 @end
 615         default:
 616             _assert(false);
 617     }
 618 }
 619
 620 void CYDriver::PushCondition(Condition condition) {
 621     switch (condition) {
 622         case RegExpCondition:
 623             yy_push_state(RegExp, scanner_);
 624             break;
 625 @begin E4X
 626         case XMLContentCondition:
 627             yy_push_state(XMLContent, scanner_);
 628             break;
 629         case XMLTagCondition:
 630             yy_push_state(XMLTag, scanner_);
 631             break;
 632 @end
 633         default:
 634             _assert(false);
 635     }
 636 }
 637
 638 void CYDriver::PopCondition() {
 639     yy_pop_state(scanner_);
 640 }
 641
 642 #if defined(__clang__)
 643 #pragma clang diagnostic pop
 644 #endif