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 %top{
  25 #if defined(__clang__)
  26 #pragma clang diagnostic push
  27 #pragma clang diagnostic ignored "-Wunused-variable"
  28 #pragma clang diagnostic ignored "-Wdeprecated-register"
  29 #else
  30 #pragma GCC diagnostic push
  31 #pragma GCC diagnostic ignored "-Wsign-compare"
  32 #pragma GCC diagnostic ignored "-Wunused-function"
  33 #pragma GCC diagnostic ignored "-Wunused-variable"
  34 #endif
  35 }
  36
  37 %{
  38
  39 #define YYLTYPE CYLocation
  40 #include "Cycript.tab.hh"
  41 typedef cy::parser::token tk;
  42
  43 #include "Highlight.hpp"
  44
  45 #define YY_EXTRA_TYPE CYDriver *
  46
  47 // do /not/ fold token to the return: this is a macro and the ordering is dependent
  48 #define F(value, highlight) do { \
  49     int token(value); \
  50 @begin ObjectiveC
  51     yyextra->no_.AtImplementation = false; \
  52 @end
  53     yyextra->no_.Function = false; \
  54     yyextra->no_.OpenBrace = false; \
  55     yylval->highlight_ = highlight; \
  56     return token; \
  57 } while (false)
  58
  59 #define P yyextra->pool_
  60 #define A new(P)
  61 #define Y P.strmemdup(yytext, yyleng)
  62
  63 #define I(type, Type, value, highlight) do { \
  64     yylval->type ## _ = A CY ## Type; \
  65     F(value, highlight); \
  66 } while (false)
  67
  68 #define T yylval->newline_ = yyextra->state_ == CYNewLine; BEGIN(Div);
  69 #define C T yyextra->state_ = CYClear;
  70 #define R T yyextra->state_ = CYRestricted;
  71
  72 #define N \
  73     if (yyextra->state_ != CYNewLine) { \
  74         if (yyextra->state_ != CYRestricted) \
  75             yyextra->state_ = CYNewLine; \
  76         else { \
  77             yyextra->state_ = CYClear; \
  78             F(tk::NewLine, hi::Nothing); \
  79         } \
  80     }
  81
  82 #define V(more) { \
  83     if (const char *nl = reinterpret_cast<const char *>(memchr(yytext, '\n', yyleng))) { \
  84         unsigned lines(0); \
  85         size_t left; \
  86         do { \
  87             ++lines; \
  88             left = yyleng - (nl - yytext) - 1; \
  89             nl = reinterpret_cast<const char *>(memchr(nl + 1, '\n', left)); \
  90         } while (nl != NULL); \
  91         yylloc->step(); \
  92         yylloc->end.lines(lines); \
  93         yylloc->end.columns(left); \
  94         more \
  95     } else L \
  96 }
  97
  98 #define L { \
  99     yylloc->step(); \
 100     yylloc->end.columns(yyleng); \
 101 }
 102
 103 #define M { \
 104     if (yyextra->commented_) { \
 105         I(comment, Comment(Y), tk::Comment, hi::Comment); \
 106     } \
 107 }
 108
 109 #define E(message) { \
 110     CYDriver::Error error; \
 111     error.location_ = *yylloc; \
 112     error.message_ = "syntax error, " message; \
 113     yyextra->errors_.push_back(error); \
 114     yyterminate(); \
 115 }
 116
 117 int H(char c) {
 118     if (c >= '0' && c <= '9')
 119         return c - '0';
 120     if (c >= 'a' && c <= 'f')
 121         return c - 'a' + 10;
 122     if (c >= 'A' && c <= 'F')
 123         return c - 'A' + 10;
 124     return -1;
 125 }
 126
 127 static void U(char *&local, unsigned point) {
 128     if (false) {
 129     } else if (point < 0x000080) {
 130         *local++ = point;
 131     } else if (point < 0x000800) {
 132         *local++ = 0xc0 | point >> 0x06 & 0x1f;
 133         goto one;
 134     } else if (point < 0x010000) {
 135         *local++ = 0xe0 | point >> 0x0c & 0x0f;
 136         goto two;
 137     } else if (point < 0x110000) {
 138         *local++ = 0xf0 | point >> 0x12 & 0x07;
 139         *local++ = 0x80 | point >> 0x0c & 0x3f;
 140       two:
 141         *local++ = 0x80 | point >> 0x06 & 0x3f;
 142       one:
 143         *local++ = 0x80 | point >> 0x00 & 0x3f;
 144     } else _assert(false);
 145 }
 146
 147 static void U(char *&local, const char *text, yy_size_t &i) {
 148     unsigned point;
 149
 150     char next(text[++i]);
 151     if (next != '{') {
 152         point = H(text[i + 0]) << 12 | H(text[i + 1]) << 8 | H(text[i + 2]) << 4 | H(text[i + 3]);
 153         i += 3;
 154     } else {
 155         point = 0;
 156         for (;;) {
 157             next = text[++i];
 158             if (next == '}')
 159                 break;
 160             point = (point << 4) | H(next);
 161         }
 162     }
 163
 164     U(local, point);
 165 }
 166
 167 #define YY_INPUT(data, value, size) { \
 168     if (yyextra->data_.eof()) \
 169         value = YY_NULL; \
 170     else { \
 171         yyextra->data_.read(data, size); \
 172         size_t copy(yyextra->data_.gcount()); \
 173         value = copy == 0 ? YY_NULL : copy; \
 174     } \
 175 }
 176
 177 %}
 178
 179 %option prefix="cy"
 180 %option bison-bridge
 181 %option bison-locations
 182 %option nodefault
 183 %option noyywrap
 184 %option noyylineno
 185 %option nounput
 186 %option nounistd
 187 %option 8bit
 188 %option backup
 189 %option batch
 190 %option never-interactive
 191 %option pointer
 192 %option reentrant
 193 %option stack
 194
 195 %option full
 196 %option ecs
 197
 198 U1 [\x00-\x7f]
 199 U0 [\x80-\xbf]
 200 U2 [\xc2-\xdf]
 201 U3 [\xe0-\xef]
 202 U4 [\xf0-\xf4]
 203
 204 HexDigit [0-9a-fA-F]
 205 LineTerminatorSequence \r?\n|\r|\xe2\x80[\xa8\xa9]
 206 WhiteSpace [\x09\x0b\x0c\x20]|\xc2\xa0|\xef\xbb\xbf
 207 UnicodeEscape \\u({HexDigit}{4}|\{{HexDigit}+\})
 208
 209 OctalEscape \\[1-7]|\\[4-7][0-7]|\\[0-3][0-7][0-7]?
 210 StringEscape \\['"\\bfnrtv]|\\0|{OctalEscape}|\\x{HexDigit}{2}|{UnicodeEscape}
 211 StringExtra {StringEscape}|\\{LineTerminatorSequence}
 212 SingleString ([^'\\\n]|{StringExtra})*
 213 DoubleString ([^"\\\n]|{StringExtra})*
 214 StringPrefix '{SingleString}|\"{DoubleString}
 215
 216 @include UnicodeIDStart.l
 217 @include UnicodeIDContinue.l
 218
 219 IdentifierMore [$_]
 220
 221 UnicodeStart {IdentifierMore}|{UnicodeIDStart}
 222 UnicodePart {IdentifierMore}|\xe2\x80[\x8c\x8d]|{UnicodeIDContinue}
 223 UnicodeFail {U2}|{U3}|{U3}{U0}|{U4}|{U4}{U0}|{U4}{U0}{U0}
 224 UnicodeScrap {UnicodePart}*{UnicodeFail}?
 225
 226 IdentifierStart {UnicodeStart}|{UnicodeEscape}
 227 IdentifierPart {UnicodePart}|{UnicodeEscape}
 228 IdentifierFail {UnicodeFail}|\\(u({HexDigit}{0,3}|\{{HexDigit}*))?
 229 IdentifierScrap {IdentifierPart}*{IdentifierFail}?
 230
 231 NonTerminator [^\n]
 232 BackslashSequence \\{NonTerminator}
 233 RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence}
 234 RegularExpressionChar [^\n\\/]|{BackslashSequence}
 235 RegularExpressionFlags {UnicodePart}*
 236 RegularExpressionChars {RegularExpressionChar}*
 237 RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}
 238
 239 @begin E4X
 240 XMLNameStart [a-zA-Z_:]
 241 XMLNamePart [a-zA-Z0-9.-_:]
 242 XMLName {XMLNameStart}{XMLNamePart}*
 243 @end
 244
 245 %s Div
 246 %s RegExp
 247
 248 @begin E4X
 249 %x XMLContent
 250 %x XMLTag
 251 @end
 252
 253 %%
 254
 255     /* RegEx {{{ */
 256 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);
 257 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags}{UnicodeFail} L E("invalid flags")
 258 <RegExp>\/{RegularExpressionBody}?\\? L E("unterminated regex")
 259     /* }}} */
 260     /* Comment {{{ */
 261 #![^\n]* L M
 262 \/\/[^\n]* L M
 263
 264     /* http://ostermiller.org/findcomment.html */
 265     /* XXX: unify these two rules using !? */
 266 \/\*!([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V() C I(comment, Comment(Y), tk::Comment, hi::Comment);
 267 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V(N) M
 268 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\** V() E("invalid comment")
 269     /* }}} */
 270     /* Element {{{ */
 271 @begin E4X
 272 <RegExp>"<>"      L F(tk::LeftRight, hi::Structure);
 273 <XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
 274
 275 <RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
 276 <RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
 277 <RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);
 278
 279 <XMLTag>"="  L F(tk::Equal, hi::Structure);
 280 <XMLTag>">"  L F(tk::Right, hi::Structure);
 281 <XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
 282 <XMLTag>"{"  L F(tk::OpenBrace, hi::Structure);
 283
 284 <XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue, hi::Constant);
 285 <XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
 286 <XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);
 287
 288 <XMLContent>"{"  L F(tk::OpenBrace, hi::Structure);
 289 <XMLContent>"<"  L F(tk::Left, hi::Structure);
 290 <XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
 291 @end
 292     /* }}} */
 293     /* Operator {{{ */
 294 "..."  L C F(tk::PeriodPeriodPeriod, hi::Meta);
 295 ".."   L E("invalid operator")
 296
 297 @begin E4X
 298 "::"   L C F(tk::ColonColon, hi::Operator);
 299 ".."   L C F(tk::PeriodPeriod, hi::Operator);
 300 @end
 301
 302 @begin E4X ObjectiveC
 303 "@"    L C F(tk::At, hi::Operator);
 304 "#"    L C F(tk::Pound, hi::Operator);
 305 @end
 306
 307 "&"    L C F(tk::Ampersand, hi::Operator);
 308 "&&"   L C F(tk::AmpersandAmpersand, hi::Operator);
 309 "&="   L C F(tk::AmpersandEqual, hi::Operator);
 310 "^"    L C F(tk::Carrot, hi::Operator);
 311 "^="   L C F(tk::CarrotEqual, hi::Operator);
 312 "="    L C F(tk::Equal, hi::Operator);
 313 "=="   L C F(tk::EqualEqual, hi::Operator);
 314 "==="  L C F(tk::EqualEqualEqual, hi::Operator);
 315 "=>"   L C F(yylval->newline_ ? tk::EqualRight_ : tk::EqualRight, hi::Operator);
 316 "!"    L C F(tk::Exclamation, hi::Operator);
 317 "!="   L C F(tk::ExclamationEqual, hi::Operator);
 318 "!=="  L C F(tk::ExclamationEqualEqual, hi::Operator);
 319 "-"    L C F(tk::Hyphen, hi::Operator);
 320 "-="   L C F(tk::HyphenEqual, hi::Operator);
 321 "--"   L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
 322 "->"   L C F(tk::HyphenRight, hi::Operator);
 323 "<"    L C F(tk::Left, hi::Operator);
 324 "<="   L C F(tk::LeftEqual, hi::Operator);
 325 "<<"   L C F(tk::LeftLeft, hi::Operator);
 326 "<<="  L C F(tk::LeftLeftEqual, hi::Operator);
 327 "%"    L C F(tk::Percent, hi::Operator);
 328 "%="   L C F(tk::PercentEqual, hi::Operator);
 329 "."    L C F(tk::Period, hi::Operator);
 330 "|"    L C F(tk::Pipe, hi::Operator);
 331 "|="   L C F(tk::PipeEqual, hi::Operator);
 332 "||"   L C F(tk::PipePipe, hi::Operator);
 333 "+"    L C F(tk::Plus, hi::Operator);
 334 "+="   L C F(tk::PlusEqual, hi::Operator);
 335 "++"   L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
 336 ">"    L C F(tk::Right, hi::Operator);
 337 ">="   L C F(tk::RightEqual, hi::Operator);
 338 ">>"   L C F(tk::RightRight, hi::Operator);
 339 ">>="  L C F(tk::RightRightEqual, hi::Operator);
 340 ">>>"  L C F(tk::RightRightRight, hi::Operator);
 341 ">>>=" L C F(tk::RightRightRightEqual, hi::Operator);
 342 "*"    L C F(tk::Star, hi::Operator);
 343 "*="   L C F(tk::StarEqual, hi::Operator);
 344 "~"    L C F(tk::Tilde, hi::Operator);
 345
 346 <Div>"/"  L C F(tk::Slash, hi::Operator);
 347 <Div>"/=" L C F(tk::SlashEqual, hi::Operator);
 348
 349 ":"    L C F(tk::Colon, hi::Structure);
 350 ","    L C F(tk::Comma, hi::Structure);
 351 "?"    L C F(tk::Question, hi::Structure);
 352 ";"    L C F(tk::SemiColon, hi::Structure);
 353
 354 "("    L C F(tk::OpenParen, hi::Structure);
 355 ")"    L C F(tk::CloseParen, hi::Structure);
 356
 357 "{"    L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
 358 "}"    L C F(tk::CloseBrace, hi::Structure);
 359
 360 "["    L C F(tk::OpenBracket, hi::Structure);
 361 "]"    L C F(tk::CloseBracket, hi::Structure);
 362     /* }}} */
 363     /* Keyword {{{ */
 364 "@error"          L C F(tk::AtError, hi::Error);
 365
 366 @begin Java
 367 "@class"          L C F(tk::AtClass, hi::Meta);
 368 @end
 369
 370 @begin C
 371 "@encode"         L C F(tk::AtEncode, hi::Meta);
 372 @end
 373
 374 @begin ObjectiveC
 375 "@end"            L C F(tk::AtEnd, hi::Meta);
 376 "@false"          L C F(tk::AtFalse, hi::Constant);
 377 "@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta);
 378 "@import"         L C F(tk::AtImport, hi::Special);
 379 "@NO"             L C F(tk::AtNo, hi::Constant);
 380 "@null"           L C F(tk::AtNull, hi::Constant);
 381 "@selector"       L C F(tk::AtSelector, hi::Meta);
 382 "@true"           L C F(tk::AtTrue, hi::Constant);
 383 "@YES"            L C F(tk::AtYes, hi::Constant);
 384 @end
 385
 386 @({UnicodeStart}{UnicodeScrap}|{UnicodeFail}) L E("invalid keyword")
 387     /* }}} */
 388     /* Highlight {{{ */
 389 "undefined"       L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator);
 390
 391 @begin ObjectiveC
 392 "bool"            L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type);
 393 "BOOL"            L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type);
 394 "id"              L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type);
 395 "nil"             L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant);
 396 "NULL"            L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant);
 397 "SEL"             L C I(identifier, Identifier("SEL"), tk::Identifier_, hi::Type);
 398 @end
 399     /* }}} */
 400     /* Reserved {{{ */
 401 "abstract"        L C /*FII*/ I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta);
 402 "await"           L C /*II?*/ I(identifier, Identifier("await"), tk::Await, hi::Meta);
 403 "boolean"         L C /*FII*/ I(identifier, Identifier("boolean"), tk::Boolean, hi::Type);
 404 "break"           L R /*KKK*/ F(tk::Break, hi::Control);
 405 "byte"            L C /*FII*/ I(identifier, Identifier("byte"), tk::Byte, hi::Type);
 406 "case"            L C /*KKK*/ F(tk::Case, hi::Control);
 407 "catch"           L C /*KKK*/ F(tk::Catch, hi::Control);
 408 "char"            L C /*FII*/ I(identifier, Identifier("char"), tk::Char, hi::Type);
 409 "class"           L C /*FFK*/ F(tk::Class, hi::Meta);
 410 "const"           L C /*FFK*/ F(tk::Const, hi::Meta);
 411 "continue"        L R /*KKK*/ F(tk::Continue, hi::Control);
 412 "debugger"        L C /*FKK*/ F(tk::Debugger, hi::Meta);
 413 "default"         L C /*KKK*/ F(tk::Default, hi::Control);
 414 "delete"          L C /*KKK*/ F(tk::Delete, hi::Operator);
 415 "do"              L C /*KKK*/ F(tk::Do, hi::Control);
 416 "double"          L C /*FII*/ I(identifier, Identifier("double"), tk::Double, hi::Type);
 417 "else"            L C /*KKK*/ F(tk::Else, hi::Control);
 418 "enum"            L C /*FFF*/ F(tk::Enum, hi::Meta);
 419 "export"          L C /*FFK*/ F(tk::Export, hi::Meta);
 420 "extends"         L C /*FFK*/ F(tk::Extends, hi::Meta);
 421 "false"           L C /*LLL*/ F(tk::False, hi::Constant);
 422 "final"           L C /*FII*/ I(identifier, Identifier("final"), tk::Final, hi::Meta);
 423 "finally"         L C /*KKK*/ F(tk::Finally, hi::Control);
 424 "float"           L C /*FII*/ I(identifier, Identifier("float"), tk::Float, hi::Type);
 425 "for"             L C /*KKK*/ F(tk::For, hi::Control);
 426 "function"        L C /*KKK*/ F(yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta);
 427 "goto"            L C /*FII*/ I(identifier, Identifier("goto"), tk::Goto, hi::Control);
 428 "if"              L C /*KKK*/ F(tk::If, hi::Control);
 429 "implements"      L C /*FSS*/ I(identifier, Identifier("implements"), tk::Implements, hi::Meta);
 430 "import"          L C /*FFK*/ F(tk::Import, hi::Meta);
 431 "in"              L C /*KKK*/ F(yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator);
 432 "instanceof"      L C /*KKK*/ F(tk::InstanceOf, hi::Operator);
 433 "int"             L C /*FII*/ I(identifier, Identifier("int"), tk::Int, hi::Type);
 434 "interface"       L C /*FSS*/ I(identifier, Identifier("interface"), tk::Interface, hi::Meta);
 435 "let"             L C /*IS?*/ I(identifier, Identifier("let"), tk::Let, hi::Meta);
 436 "long"            L C /*FII*/ I(identifier, Identifier("long"), tk::Long, hi::Type);
 437 "native"          L C /*FII*/ I(identifier, Identifier("native"), tk::Native, hi::Meta);
 438 "new"             L C /*KKK*/ F(tk::New, hi::Operator);
 439 "null"            L C /*LLL*/ F(tk::Null, hi::Constant);
 440 "package"         L C /*FSS*/ I(identifier, Identifier("package"), tk::Package, hi::Meta);
 441 "private"         L C /*FSS*/ I(identifier, Identifier("private"), tk::Private, hi::Meta);
 442 "protected"       L C /*FSS*/ I(identifier, Identifier("protected"), tk::Protected, hi::Meta);
 443 "public"          L C /*FSS*/ I(identifier, Identifier("public"), tk::Public, hi::Meta);
 444 "return"          L R /*KKK*/ F(tk::Return, hi::Control);
 445 "short"           L C /*FII*/ I(identifier, Identifier("short"), tk::Short, hi::Type);
 446 "static"          L C /*FS?*/ I(identifier, Identifier("static"), tk::Static, hi::Meta);
 447 "super"           L C /*FFK*/ F(tk::Super, hi::Constant);
 448 "switch"          L C /*KKK*/ F(tk::Switch, hi::Control);
 449 "synchronized"    L C /*FII*/ I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta);
 450 "this"            L C /*KKK*/ F(tk::This, hi::Constant);
 451 "throw"           L R /*KKK*/ F(tk::Throw, hi::Control);
 452 "throws"          L C /*FII*/ I(identifier, Identifier("throws"), tk::Throws, hi::Meta);
 453 "transient"       L C /*FII*/ I(identifier, Identifier("transient"), tk::Transient, hi::Meta);
 454 "true"            L C /*LLL*/ F(tk::True, hi::Constant);
 455 "try"             L C /*KKK*/ F(tk::Try, hi::Control);
 456 "typeof"          L C /*KKK*/ F(tk::TypeOf, hi::Operator);
 457 "var"             L C /*KKK*/ F(tk::Var, hi::Meta);
 458 "void"            L C /*KKK*/ F(tk::Void, hi::Operator);
 459 "volatile"        L C /*FII*/ I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta);
 460 "while"           L C /*KKK*/ F(tk::While, hi::Control);
 461 "with"            L C /*KKK*/ F(tk::With, hi::Control);
 462 "yield"           L R /*IS?*/ I(identifier, Identifier("yield"), tk::Yield, hi::Control);
 463
 464 "auto"            L C F(tk::Auto, hi::Meta);
 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 C
 469 "extern"          L C I(identifier, Identifier("extern"), tk::Extern, hi::Type);
 470 "signed"          L C I(identifier, Identifier("signed"), tk::Signed, hi::Type);
 471 "typedef"         L C I(identifier, Identifier("typedef"), tk::Typedef, hi::Meta);
 472 "unsigned"        L C I(identifier, Identifier("unsigned"), tk::Unsigned, hi::Type);
 473 @end
 474
 475 @begin ObjectiveC
 476 "NO"              L C I(identifier, Identifier("NO"), tk::No, hi::Constant);
 477 "YES"             L C I(identifier, Identifier("YES"), tk::Yes, hi::Constant);
 478 @end
 479
 480 @begin E4X
 481 "namespace"       L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta);
 482 "xml"             L C I(identifier, Identifier("xml"), tk::XML, hi::Meta);
 483 @end
 484     /* }}} */
 485     /* Identifier {{{ */
 486 {UnicodeStart}{UnicodePart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);
 487
 488 {IdentifierStart}{IdentifierPart}* L C {
 489     char *value(A char[yyleng + 1]);
 490     char *local(value);
 491
 492     for (yy_size_t i(0), e(yyleng); i != e; ++i) {
 493         char next(yytext[i]);
 494         if (next != '\\')
 495             *local++ = next;
 496         else
 497             U(local, yytext, ++i);
 498     }
 499
 500     *local = '\0';
 501     I(identifier, Identifier(value), tk::Identifier_, hi::Identifier);
 502 }
 503
 504 ({IdentifierStart}{IdentifierPart}*)?{IdentifierFail} L E("invalid identifier")
 505     /* }}} */
 506     /* Number {{{ */
 507 0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant);
 508 0[0-9]+ L C I(number, Number(strtoull(yytext + 1, NULL, 10)), tk::NumericLiteral, hi::Constant);
 509
 510 0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant);
 511 0[oO][0-7]+ L C I(number, Number(strtoull(yytext + 2, NULL, 8)), tk::NumericLiteral, hi::Constant);
 512 0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant);
 513
 514 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)([eE][+-]?[0-9]+)? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
 515 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)[eE][+-]?{IdentifierScrap} L E("invalid exponent")
 516 (\.?[0-9]|(0|[1-9][0-9]*)\.){IdentifierScrap} L E("invalid number")
 517     /* }}} */
 518     /* String {{{ */
 519 '{SingleString}'|\"{DoubleString}\" L C {
 520     char *value(A char[yyleng]);
 521     char *local(value);
 522
 523     for (yy_size_t i(1), e(yyleng - 1); i != e; ++i) {
 524         char next(yytext[i]);
 525
 526         if (yytext[i] == '\\')
 527             // XXX: support more line continuation characters
 528             if (false) line: {
 529                 yylloc->end.lines(1);
 530                 yylloc->end.columns(yyleng - i);
 531             } else switch (next = yytext[++i]) {
 532                 case '\n': goto line;
 533
 534                 case '\\': next = '\\'; break;
 535                 case '\'': next = '\''; break;
 536                 case '"': next = '"'; break;
 537                 case 'b': next = '\b'; break;
 538                 case 'f': next = '\f'; break;
 539                 case 'n': next = '\n'; break;
 540                 case 'r': next = '\r'; break;
 541                 case 't': next = '\t'; break;
 542                 case 'v': next = '\v'; break;
 543
 544                 case '0': case '1': case '2': case '3':
 545                     if (yytext[i + 1] < '0' || yytext[i + 1] > '7')
 546                         next = H(yytext[i]), i += 0;
 547                     else if (yytext[i + 2] < '0' || yytext[i + 2] > '7')
 548                         next = H(yytext[i]) << 3 | H(yytext[i + 1]), i += 1;
 549                     else
 550                         next = H(yytext[i]) << 6 | H(yytext[i + 1]) << 3 | H(yytext[i + 2]), i += 2;
 551                 break;
 552
 553                 case '4': case '5': case '6': case '7':
 554                     if (yytext[i + 1] < '0' || yytext[i + 1] > '7')
 555                         next = H(yytext[i]), i += 0;
 556                     else
 557                         next = H(yytext[i]) << 3 | H(yytext[i + 1]), i += 1;
 558                 break;
 559
 560                 case 'x':
 561                     U(local, H(yytext[i + 1]) << 4 | H(yytext[i + 2]));
 562                     i += 2;
 563                 continue;
 564
 565                 case 'u':
 566                     U(local, yytext, i);
 567                 continue;
 568             }
 569
 570         *local++ = next;
 571     }
 572
 573     *local = '\0';
 574     I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
 575 }
 576
 577 {StringPrefix}\\(x.{0,2}|u([^{].{0,3}|\{[^}]*)?|{UnicodeFail})? L E("invalid escape")
 578 {StringPrefix} L E("invalid string")
 579     /* }}} */
 580
 581 {LineTerminatorSequence} yylloc->step(); yylloc->end.lines(); N
 582 {WhiteSpace} L
 583
 584 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
 585
 586 . L E("invalid character")
 587
 588 %%
 589
 590 void CYDriver::ScannerInit() {
 591     cylex_init(&scanner_);
 592     cyset_extra(this, scanner_);
 593 }
 594
 595 void CYDriver::ScannerDestroy() {
 596     cylex_destroy(scanner_);
 597 }
 598
 599 CYDriver::Condition CYDriver::GetCondition() {
 600     switch (yy_top_state(scanner_)) {
 601         case RegExp:
 602             return RegExpCondition;
 603 @begin E4X
 604         case XMLContent:
 605             return XMLContentCondition;
 606         case XMLTag:
 607             return XMLTagCondition;
 608 @end
 609         default:
 610             _assert(false);
 611     }
 612 }
 613
 614 void CYDriver::SetCondition(Condition condition) {
 615     struct yyguts_t *yyg(reinterpret_cast<struct yyguts_t *>(scanner_));
 616
 617     switch (condition) {
 618         case RegExpCondition:
 619             BEGIN(RegExp);
 620             break;
 621 @begin E4X
 622         case XMLContentCondition:
 623             BEGIN(XMLContent);
 624             break;
 625         case XMLTagCondition:
 626             BEGIN(XMLTag);
 627             break;
 628 @end
 629         default:
 630             _assert(false);
 631     }
 632 }
 633
 634 void CYDriver::PushCondition(Condition condition) {
 635     switch (condition) {
 636         case RegExpCondition:
 637             yy_push_state(RegExp, scanner_);
 638             break;
 639 @begin E4X
 640         case XMLContentCondition:
 641             yy_push_state(XMLContent, scanner_);
 642             break;
 643         case XMLTagCondition:
 644             yy_push_state(XMLTag, scanner_);
 645             break;
 646 @end
 647         default:
 648             _assert(false);
 649     }
 650 }
 651
 652 void CYDriver::PopCondition() {
 653     yy_pop_state(scanner_);
 654 }
 655
 656 #if defined(__clang__)
 657 #pragma clang diagnostic pop
 658 #else
 659 // must not pop -Wunused-function
 660 //#pragma GCC diagnostic pop
 661 #endif