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 P yyextra->pool_
  53 #define A new(P)
  54 #define Y P.strmemdup(yytext, yyleng)
  55
  56 #define I(type, Type, value, highlight) do { \
  57     yylval->type ## _ = A CY ## Type; \
  58     F(value, highlight); \
  59 } while (false)
  60
  61 #define T yylval->newline_ = yyextra->state_ == CYNewLine; BEGIN(Div);
  62 #define C T yyextra->state_ = CYClear;
  63 #define R T yyextra->state_ = CYRestricted;
  64
  65 #define N \
  66     if (yyextra->state_ != CYNewLine) { \
  67         if (yyextra->state_ != CYRestricted) \
  68             yyextra->state_ = CYNewLine; \
  69         else { \
  70             yyextra->state_ = CYClear; \
  71             F(tk::NewLine, hi::Nothing); \
  72         } \
  73     }
  74
  75 #define V(more) { \
  76     if (const char *nl = reinterpret_cast<const char *>(memchr(yytext, '\n', yyleng))) { \
  77         unsigned lines(0); \
  78         size_t left; \
  79         do { \
  80             ++lines; \
  81             left = yyleng - (nl - yytext) - 1; \
  82             nl = reinterpret_cast<const char *>(memchr(nl + 1, '\n', left)); \
  83         } while (nl != NULL); \
  84         yylloc->step(); \
  85         yylloc->end.lines(lines); \
  86         yylloc->end.columns(left); \
  87         more \
  88     } else L \
  89 }
  90
  91 #define L { \
  92     yylloc->step(); \
  93     yylloc->end.columns(yyleng); \
  94 }
  95
  96 #define M { \
  97     if (yyextra->commented_) { \
  98         I(comment, Comment(Y), tk::Comment, hi::Comment); \
  99     } \
 100 }
 101
 102 #define E(message) { \
 103     CYDriver::Error error; \
 104     error.location_ = *yylloc; \
 105     error.message_ = "syntax error, " message; \
 106     yyextra->errors_.push_back(error); \
 107     yyterminate(); \
 108 }
 109
 110 int H(char c) {
 111     if (c >= '0' && c <= '9')
 112         return c - '0';
 113     if (c >= 'a' && c <= 'f')
 114         return c - 'a' + 10;
 115     if (c >= 'A' && c <= 'F')
 116         return c - 'A' + 10;
 117     return -1;
 118 }
 119
 120 static void U(char *&local, unsigned point) {
 121     if (false) {
 122     } else if (point < 0x000080) {
 123         *local++ = point;
 124     } else if (point < 0x000800) {
 125         *local++ = 0xc0 | point >> 0x06 & 0x1f;
 126         goto one;
 127     } else if (point < 0x010000) {
 128         *local++ = 0xe0 | point >> 0x0c & 0x0f;
 129         goto two;
 130     } else if (point < 0x110000) {
 131         *local++ = 0xf0 | point >> 0x12 & 0x07;
 132         *local++ = 0x80 | point >> 0x0c & 0x3f;
 133       two:
 134         *local++ = 0x80 | point >> 0x06 & 0x3f;
 135       one:
 136         *local++ = 0x80 | point >> 0x00 & 0x3f;
 137     } else _assert(false);
 138 }
 139
 140 static void U(char *&local, const char *text, yy_size_t &i) {
 141     unsigned point;
 142
 143     char next(text[++i]);
 144     if (next != '{') {
 145         point = H(text[i + 0]) << 12 | H(text[i + 1]) << 8 | H(text[i + 2]) << 4 | H(text[i + 3]);
 146         i += 3;
 147     } else {
 148         point = 0;
 149         for (;;) {
 150             next = text[++i];
 151             if (next == '}')
 152                 break;
 153             point = (point << 4) | H(next);
 154         }
 155     }
 156
 157     U(local, point);
 158 }
 159
 160 #define YY_INPUT(data, value, size) { \
 161     if (yyextra->data_.eof()) \
 162         value = YY_NULL; \
 163     else { \
 164         yyextra->data_.read(data, size); \
 165         size_t copy(yyextra->data_.gcount()); \
 166         value = copy == 0 ? YY_NULL : copy; \
 167     } \
 168 }
 169
 170 %}
 171
 172 %option prefix="cy"
 173 %option bison-bridge
 174 %option bison-locations
 175 %option nodefault
 176 %option noyywrap
 177 %option noyylineno
 178 %option nounput
 179 %option nounistd
 180 %option 8bit
 181 %option backup
 182 %option batch
 183 %option never-interactive
 184 %option pointer
 185 %option reentrant
 186 %option stack
 187
 188 %option full
 189 %option ecs
 190
 191 U1 [\x00-\x7f]
 192 U0 [\x80-\xbf]
 193 U2 [\xc2-\xdf]
 194 U3 [\xe0-\xef]
 195 U4 [\xf0-\xf4]
 196
 197 HexDigit [0-9a-fA-F]
 198 LineTerminatorSequence \r?\n|\r|\xe2\x80[\xa8\xa9]
 199 WhiteSpace [\x09\x0b\x0c\x20]|\xc2\xa0|\xef\xbb\xbf
 200 UnicodeEscape \\u({HexDigit}{4}|\{{HexDigit}+\})
 201
 202 OctalEscape \\[1-7]|\\[4-7][0-7]|\\[0-3][0-7][0-7]?
 203 StringEscape \\['"\\bfnrtv]|\\0|{OctalEscape}|\\x{HexDigit}{2}|{UnicodeEscape}
 204 StringExtra {StringEscape}|\\{LineTerminatorSequence}
 205 SingleString ([^'\\\n]|{StringExtra})*
 206 DoubleString ([^"\\\n]|{StringExtra})*
 207 StringPrefix '{SingleString}|\"{DoubleString}
 208
 209 @include UnicodeIDStart.l
 210 @include UnicodeIDContinue.l
 211
 212 IdentifierMore [$_]
 213
 214 UnicodeStart {IdentifierMore}|{UnicodeIDStart}
 215 UnicodePart {IdentifierMore}|\xe2\x80[\x8c\x8d]|{UnicodeIDContinue}
 216 UnicodeFail {U2}|{U3}|{U3}{U0}|{U4}|{U4}{U0}|{U4}{U0}{U0}
 217 UnicodeScrap {UnicodePart}*{UnicodeFail}?
 218
 219 IdentifierStart {UnicodeStart}|{UnicodeEscape}
 220 IdentifierPart {UnicodePart}|{UnicodeEscape}
 221 IdentifierFail {UnicodeFail}|\\(u({HexDigit}{0,3}|\{{HexDigit}*))?
 222 IdentifierScrap {IdentifierPart}*{IdentifierFail}?
 223
 224 NonTerminator [^\n]
 225 BackslashSequence \\{NonTerminator}
 226 RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence}
 227 RegularExpressionChar [^\n\\/]|{BackslashSequence}
 228 RegularExpressionFlags {UnicodePart}*
 229 RegularExpressionChars {RegularExpressionChar}*
 230 RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}
 231
 232 @begin E4X
 233 XMLNameStart [a-zA-Z_:]
 234 XMLNamePart [a-zA-Z0-9.-_:]
 235 XMLName {XMLNameStart}{XMLNamePart}*
 236 @end
 237
 238 %s Div
 239 %s RegExp
 240
 241 @begin E4X
 242 %x XMLContent
 243 %x XMLTag
 244 @end
 245
 246 %%
 247
 248 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);
 249 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags}{UnicodeFail} L E("invalid flags")
 250 <RegExp>\/{RegularExpressionBody}?\\? L E("unterminated regex")
 251
 252 #![^\n]* L M
 253
 254 \/\/[^\n]* L M
 255
 256         /* http://ostermiller.org/findcomment.html */
 257         /* XXX: unify these two rules using !? */
 258 \/\*!([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V() C I(comment, Comment(Y), tk::Comment, hi::Comment);
 259 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V(N) M
 260 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\** V() E("invalid comment")
 261
 262 @begin E4X
 263 <RegExp>"<>"      L F(tk::LeftRight, hi::Structure);
 264 <XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
 265
 266 <RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
 267 <RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
 268 <RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);
 269
 270 <XMLTag>"="  L F(tk::Equal, hi::Structure);
 271 <XMLTag>">"  L F(tk::Right, hi::Structure);
 272 <XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
 273 <XMLTag>"{"  L F(tk::OpenBrace, hi::Structure);
 274
 275 <XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue, hi::Constant);
 276 <XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
 277 <XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);
 278
 279 <XMLContent>"{"  L F(tk::OpenBrace, hi::Structure);
 280 <XMLContent>"<"  L F(tk::Left, hi::Structure);
 281 <XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
 282 @end
 283
 284 "..."  L C F(tk::PeriodPeriodPeriod, hi::Meta);
 285 ".."   L E("invalid operator")
 286
 287 @begin E4X
 288 "::"   L C F(tk::ColonColon, hi::Operator);
 289 ".."   L C F(tk::PeriodPeriod, hi::Operator);
 290 @end
 291
 292 @begin E4X ObjectiveC
 293 "@"    L C F(tk::At, hi::Operator);
 294 "#"    L C F(tk::Pound, hi::Operator);
 295 @end
 296
 297 "&"    L C F(tk::Ampersand, hi::Operator);
 298 "&&"   L C F(tk::AmpersandAmpersand, hi::Operator);
 299 "&="   L C F(tk::AmpersandEqual, hi::Operator);
 300 "^"    L C F(tk::Carrot, hi::Operator);
 301 "^="   L C F(tk::CarrotEqual, hi::Operator);
 302 "="    L C F(tk::Equal, hi::Operator);
 303 "=="   L C F(tk::EqualEqual, hi::Operator);
 304 "==="  L C F(tk::EqualEqualEqual, hi::Operator);
 305 "=>"   L C F(yylval->newline_ ? tk::EqualRight_ : tk::EqualRight, hi::Operator);
 306 "!"    L C F(tk::Exclamation, hi::Operator);
 307 "!="   L C F(tk::ExclamationEqual, hi::Operator);
 308 "!=="  L C F(tk::ExclamationEqualEqual, hi::Operator);
 309 "-"    L C F(tk::Hyphen, hi::Operator);
 310 "-="   L C F(tk::HyphenEqual, hi::Operator);
 311 "--"   L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
 312 "->"   L C F(tk::HyphenRight, hi::Operator);
 313 "<"    L C F(tk::Left, hi::Operator);
 314 "<="   L C F(tk::LeftEqual, hi::Operator);
 315 "<<"   L C F(tk::LeftLeft, hi::Operator);
 316 "<<="  L C F(tk::LeftLeftEqual, hi::Operator);
 317 "%"    L C F(tk::Percent, hi::Operator);
 318 "%="   L C F(tk::PercentEqual, hi::Operator);
 319 "."    L C F(tk::Period, hi::Operator);
 320 "|"    L C F(tk::Pipe, hi::Operator);
 321 "|="   L C F(tk::PipeEqual, hi::Operator);
 322 "||"   L C F(tk::PipePipe, hi::Operator);
 323 "+"    L C F(tk::Plus, hi::Operator);
 324 "+="   L C F(tk::PlusEqual, hi::Operator);
 325 "++"   L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
 326 ">"    L C F(tk::Right, hi::Operator);
 327 ">="   L C F(tk::RightEqual, hi::Operator);
 328 ">>"   L C F(tk::RightRight, hi::Operator);
 329 ">>="  L C F(tk::RightRightEqual, hi::Operator);
 330 ">>>"  L C F(tk::RightRightRight, hi::Operator);
 331 ">>>=" L C F(tk::RightRightRightEqual, hi::Operator);
 332 "*"    L C F(tk::Star, hi::Operator);
 333 "*="   L C F(tk::StarEqual, hi::Operator);
 334 "~"    L C F(tk::Tilde, hi::Operator);
 335
 336 <Div>"/"  L C F(tk::Slash, hi::Operator);
 337 <Div>"/=" L C F(tk::SlashEqual, hi::Operator);
 338
 339 ":"    L C F(tk::Colon, hi::Structure);
 340 ","    L C F(tk::Comma, hi::Structure);
 341 "?"    L C F(tk::Question, hi::Structure);
 342 ";"    L C F(tk::SemiColon, hi::Structure);
 343
 344 "("    L C F(tk::OpenParen, hi::Structure);
 345 ")"    L C F(tk::CloseParen, hi::Structure);
 346
 347 "{"    L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
 348 "}"    L C F(tk::CloseBrace, hi::Structure);
 349
 350 "["    L C F(tk::OpenBracket, hi::Structure);
 351 "]"    L C F(tk::CloseBracket, hi::Structure);
 352
 353 "@error"          L C F(tk::AtError, hi::Error);
 354
 355 @begin Java
 356 "@class"          L C F(tk::AtClass, hi::Meta);
 357 @end
 358
 359 @begin C
 360 "typedef"         L C I(identifier, Identifier("typedef"), tk::Typedef, hi::Meta);
 361 "unsigned"        L C I(identifier, Identifier("unsigned"), tk::Unsigned, hi::Type);
 362 "signed"          L C I(identifier, Identifier("signed"), tk::Signed, hi::Type);
 363 "extern"          L C I(identifier, Identifier("extern"), tk::Extern, hi::Type);
 364 @end
 365
 366 @begin C
 367 "@encode"         L C F(tk::AtEncode, hi::Meta);
 368 @end
 369
 370 @begin ObjectiveC
 371 "@end"            L C F(tk::AtEnd, hi::Meta);
 372 "@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta);
 373 "@import"         L C F(tk::AtImport, hi::Special);
 374 "@selector"       L C F(tk::AtSelector, hi::Meta);
 375
 376 "@null"           L C F(tk::AtNull, hi::Constant);
 377 "@YES"            L C F(tk::AtYes, hi::Constant);
 378 "@NO"             L C F(tk::AtNo, hi::Constant);
 379 "@true"           L C F(tk::AtTrue, hi::Constant);
 380 "@false"          L C F(tk::AtFalse, hi::Constant);
 381
 382 "NULL"         L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant);
 383 "nil"          L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant);
 384 "YES"          L C I(identifier, Identifier("YES"), tk::Yes, hi::Constant);
 385 "NO"           L C I(identifier, Identifier("NO"), tk::No, hi::Constant);
 386
 387 "bool"         L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type);
 388 "BOOL"         L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type);
 389 "id"           L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type);
 390 "SEL"          L C I(identifier, Identifier("SEL"), tk::Identifier_, hi::Type);
 391 @end
 392
 393 "undefined"    L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator);
 394
 395 "false"        L C I(false, False(), tk::False, hi::Constant);
 396 "null"         L C I(null, Null(), tk::Null, hi::Constant);
 397 "true"         L C I(true, True(), tk::True, hi::Constant);
 398
 399 "auto"         L C I(word, Word("auto"), tk::Auto, hi::Meta);
 400 "break"        L R I(word, Word("break"), tk::Break, hi::Control);
 401 "case"         L C I(word, Word("case"), tk::Case, hi::Control);
 402 "catch"        L C I(word, Word("catch"), tk::Catch, hi::Control);
 403 "continue"     L R I(word, Word("continue"), tk::Continue, hi::Control);
 404 "default"      L C I(word, Word("default"), tk::Default, hi::Control);
 405 "delete"       L C I(word, Word("delete"), tk::Delete, hi::Operator);
 406 "do"           L C I(word, Word("do"), tk::Do, hi::Control);
 407 "else"         L C I(word, Word("else"), tk::Else, hi::Control);
 408 "finally"      L C I(word, Word("finally"), tk::Finally, hi::Control);
 409 "for"          L C I(word, Word("for"), tk::For, hi::Control);
 410 "function"     L C I(word, Word("function"), yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta);
 411 "if"           L C I(word, Word("if"), tk::If, hi::Control);
 412 "in"           L C I(word, Word("in"), yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator);
 413 "instanceof"   L C I(word, Word("instanceof"), tk::InstanceOf, hi::Operator);
 414 "new"          L C I(word, Word("new"), tk::New, hi::Operator);
 415 "return"       L R I(word, Word("return"), tk::Return, hi::Control);
 416 "switch"       L C I(word, Word("switch"), tk::Switch, hi::Control);
 417 "this"         L C I(this, This(), tk::This, hi::Constant);
 418 "throw"        L R I(word, Word("throw"), tk::Throw, hi::Control);
 419 "try"          L C I(word, Word("try"), tk::Try, hi::Control);
 420 "typeof"       L C I(word, Word("typeof"), tk::TypeOf, hi::Operator);
 421 "var"          L C I(word, Word("var"), tk::Var, hi::Meta);
 422 "void"         L C I(word, Word("void"), tk::Void, hi::Operator);
 423 "while"        L C I(word, Word("while"), tk::While, hi::Control);
 424 "with"         L C I(word, Word("with"), tk::With, hi::Control);
 425
 426 "debugger"     L C I(word, Word("debugger"), tk::Debugger, hi::Meta);
 427
 428 "const"        L C I(word, Word("const"), tk::Const, hi::Meta);
 429
 430 "class"        L C I(word, Word("class"), tk::Class, hi::Meta);
 431 "enum"         L C I(word, Word("enum"), tk::Enum, hi::Meta);
 432 "export"       L C I(word, Word("export"), tk::Export, hi::Meta);
 433 "extends"      L C I(word, Word("extends"), tk::Extends, hi::Meta);
 434 "import"       L C I(word, Word("import"), tk::Import, hi::Meta);
 435 "super"        L C I(word, Word("super"), tk::Super, hi::Constant);
 436
 437 "implements"   L C I(identifier, Identifier("implements"), tk::Implements, hi::Meta);
 438 "interface"    L C I(identifier, Identifier("interface"), tk::Interface, hi::Meta);
 439 "package"      L C I(identifier, Identifier("package"), tk::Package, hi::Meta);
 440 "private"      L C I(identifier, Identifier("private"), tk::Private, hi::Meta);
 441 "protected"    L C I(identifier, Identifier("protected"), tk::Protected, hi::Meta);
 442 "public"       L C I(identifier, Identifier("public"), tk::Public, hi::Meta);
 443 "static"       L C I(identifier, Identifier("static"), tk::Static, hi::Meta);
 444
 445 "abstract"     L C I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta);
 446 "boolean"      L C I(identifier, Identifier("boolean"), tk::Boolean, hi::Type);
 447 "byte"         L C I(identifier, Identifier("byte"), tk::Byte, hi::Type);
 448 "char"         L C I(identifier, Identifier("char"), tk::Char, hi::Type);
 449 "double"       L C I(identifier, Identifier("double"), tk::Double, hi::Type);
 450 "final"        L C I(identifier, Identifier("final"), tk::Final, hi::Meta);
 451 "float"        L C I(identifier, Identifier("float"), tk::Float, hi::Type);
 452 "goto"         L C I(identifier, Identifier("goto"), tk::Goto, hi::Control);
 453 "int"          L C I(identifier, Identifier("int"), tk::Int, hi::Type);
 454 "long"         L C I(identifier, Identifier("long"), tk::Long, hi::Type);
 455 "native"       L C I(identifier, Identifier("native"), tk::Native, hi::Meta);
 456 "short"        L C I(identifier, Identifier("short"), tk::Short, hi::Type);
 457 "synchronized" L C I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta);
 458 "throws"       L C I(identifier, Identifier("throws"), tk::Throws, hi::Meta);
 459 "transient"    L C I(identifier, Identifier("transient"), tk::Transient, hi::Meta);
 460 "volatile"     L C I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta);
 461
 462 "let"          L C I(identifier, Identifier("let"), tk::Let, hi::Meta);
 463 "yield"        L R I(identifier, Identifier("yield"), tk::Yield, hi::Control);
 464
 465 "each"         L C I(identifier, Identifier("each"), tk::Each, hi::Control);
 466 "of"           L C I(identifier, Identifier("of"), tk::Of, hi::Operator);
 467
 468 @begin E4X
 469 "namespace"    L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta);
 470 "xml"          L C I(identifier, Identifier("xml"), tk::XML, hi::Meta);
 471 @end
 472
 473 {UnicodeStart}{UnicodePart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);
 474
 475 {IdentifierStart}{IdentifierPart}* L C {
 476     char *value(A char[yyleng + 1]);
 477     char *local(value);
 478
 479     for (yy_size_t i(0), e(yyleng); i != e; ++i) {
 480         char next(yytext[i]);
 481         if (next != '\\')
 482             *local++ = next;
 483         else
 484             U(local, yytext, ++i);
 485     }
 486
 487     *local = '\0';
 488     I(identifier, Identifier(value), tk::Identifier_, hi::Identifier);
 489 }
 490
 491 ({IdentifierStart}{IdentifierPart}*)?{IdentifierFail} L E("invalid identifier")
 492
 493 0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant);
 494 0[0-9]+ L C I(number, Number(strtoull(yytext + 1, NULL, 10)), tk::NumericLiteral, hi::Constant);
 495
 496 0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant);
 497 0[oO][0-7]+ L C I(number, Number(strtoull(yytext + 2, NULL, 8)), tk::NumericLiteral, hi::Constant);
 498 0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant);
 499
 500 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)([eE][+-]?[0-9]+)? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
 501 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)[eE][+-]?{IdentifierScrap} L E("invalid exponent")
 502 (\.?[0-9]|(0|[1-9][0-9]*)\.){IdentifierScrap} L E("invalid number")
 503
 504 '{SingleString}'|\"{DoubleString}\" L C {
 505     char *value(A char[yyleng]);
 506     char *local(value);
 507
 508     for (yy_size_t i(1), e(yyleng - 1); i != e; ++i) {
 509         char next(yytext[i]);
 510
 511         if (yytext[i] == '\\')
 512             // XXX: support more line continuation characters
 513             if (false) line: {
 514                 yylloc->end.lines(1);
 515                 yylloc->end.columns(yyleng - i);
 516             } else switch (next = yytext[++i]) {
 517                 case '\n': goto line;
 518
 519                 case '\\': next = '\\'; break;
 520                 case '\'': next = '\''; break;
 521                 case '"': next = '"'; break;
 522                 case 'b': next = '\b'; break;
 523                 case 'f': next = '\f'; break;
 524                 case 'n': next = '\n'; break;
 525                 case 'r': next = '\r'; break;
 526                 case 't': next = '\t'; break;
 527                 case 'v': next = '\v'; break;
 528
 529                 case '0': case '1': case '2': case '3':
 530                     if (yytext[i + 1] < '0' || yytext[i + 1] > '7')
 531                         next = H(yytext[i]), i += 0;
 532                     else if (yytext[i + 2] < '0' || yytext[i + 2] > '7')
 533                         next = H(yytext[i]) << 3 | H(yytext[i + 1]), i += 1;
 534                     else
 535                         next = H(yytext[i]) << 6 | H(yytext[i + 1]) << 3 | H(yytext[i + 2]), i += 2;
 536                 break;
 537
 538                 case '4': case '5': case '6': case '7':
 539                     if (yytext[i + 1] < '0' || yytext[i + 1] > '7')
 540                         next = H(yytext[i]), i += 0;
 541                     else
 542                         next = H(yytext[i]) << 3 | H(yytext[i + 1]), i += 1;
 543                 break;
 544
 545                 case 'x':
 546                     U(local, H(yytext[i + 1]) << 4 | H(yytext[i + 2]));
 547                     i += 2;
 548                 continue;
 549
 550                 case 'u':
 551                     U(local, yytext, i);
 552                 continue;
 553             }
 554
 555         *local++ = next;
 556     }
 557
 558     *local = '\0';
 559     I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
 560 }
 561
 562 {StringPrefix}\\(x.{0,2}|u([^{].{0,3}|\{[^}]*)?|{UnicodeFail})? L E("invalid escape")
 563 {StringPrefix} L E("invalid string")
 564
 565 {LineTerminatorSequence} yylloc->step(); yylloc->end.lines(); N
 566
 567 {WhiteSpace} L
 568
 569 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
 570
 571 @({UnicodeStart}{UnicodeScrap}|{UnicodeFail}) L E("invalid keyword")
 572
 573 . L E("invalid character")
 574
 575 %%
 576
 577 void CYDriver::ScannerInit() {
 578     cylex_init(&scanner_);
 579     cyset_extra(this, scanner_);
 580 }
 581
 582 void CYDriver::ScannerDestroy() {
 583     cylex_destroy(scanner_);
 584 }
 585
 586 CYDriver::Condition CYDriver::GetCondition() {
 587     switch (yy_top_state(scanner_)) {
 588         case RegExp:
 589             return RegExpCondition;
 590 @begin E4X
 591         case XMLContent:
 592             return XMLContentCondition;
 593         case XMLTag:
 594             return XMLTagCondition;
 595 @end
 596         default:
 597             _assert(false);
 598     }
 599 }
 600
 601 void CYDriver::SetCondition(Condition condition) {
 602     struct yyguts_t *yyg(reinterpret_cast<struct yyguts_t *>(scanner_));
 603
 604     switch (condition) {
 605         case RegExpCondition:
 606             BEGIN(RegExp);
 607             break;
 608 @begin E4X
 609         case XMLContentCondition:
 610             BEGIN(XMLContent);
 611             break;
 612         case XMLTagCondition:
 613             BEGIN(XMLTag);
 614             break;
 615 @end
 616         default:
 617             _assert(false);
 618     }
 619 }
 620
 621 void CYDriver::PushCondition(Condition condition) {
 622     switch (condition) {
 623         case RegExpCondition:
 624             yy_push_state(RegExp, scanner_);
 625             break;
 626 @begin E4X
 627         case XMLContentCondition:
 628             yy_push_state(XMLContent, scanner_);
 629             break;
 630         case XMLTagCondition:
 631             yy_push_state(XMLTag, scanner_);
 632             break;
 633 @end
 634         default:
 635             _assert(false);
 636     }
 637 }
 638
 639 void CYDriver::PopCondition() {
 640     yy_pop_state(scanner_);
 641 }
 642
 643 #if defined(__clang__)
 644 #pragma clang diagnostic pop
 645 #endif