Cycript.l.in

   1 /* Cycript - Optimizing JavaScript Compiler/Runtime
   2  * Copyright (C) 2009-2013  Jay Freeman (saurik)
   3 */
   4
   5 /* GNU General Public License, Version 3 {{{ */
   6 /*
   7  * Cycript is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published
   9  * by the Free Software Foundation, either version 3 of the License,
  10  * or (at your option) any later version.
  11  *
  12  * Cycript is distributed in the hope that it will be useful, but
  13  * WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License
  18  * along with Cycript.  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 #define YYLTYPE cy::location
  26 #include "Cycript.tab.hh"
  27 typedef cy::parser::token tk;
  28
  29 #include "Highlight.hpp"
  30
  31 #define YY_EXTRA_TYPE CYDriver *
  32
  33 // do /not/ fold token to the return: this is a macro and the ordering is dependent
  34 #define F(value, highlight) do { \
  35     int token(value); \
  36 @begin ObjectiveC
  37     yyextra->no_.AtImplementation = false; \
  38 @end
  39     yyextra->no_.Function = false; \
  40     yyextra->no_.OpenBrace = false; \
  41     yylval->highlight_ = highlight; \
  42     return token; \
  43 } while (false)
  44
  45 #define A new($pool)
  46 #define Y $pool.strmemdup(yytext, yyleng)
  47
  48 #define I(type, Type, value, highlight) do { \
  49     yylval->type ## _ = A CY ## Type; \
  50     F(value, highlight); \
  51 } while (false)
  52
  53 #define T yylval->newline_ = yyextra->state_ == CYNewLine; BEGIN(Div);
  54 #define C T yyextra->state_ = CYClear;
  55 #define R T yyextra->state_ = CYRestricted;
  56
  57 #define N \
  58     if (yyextra->state_ != CYNewLine) { \
  59         if (yyextra->state_ != CYRestricted) \
  60             yyextra->state_ = CYNewLine; \
  61         else { \
  62             yyextra->state_ = CYClear; \
  63             F(tk::NewLine, hi::Nothing); \
  64         } \
  65     }
  66
  67 #define V(more) { \
  68     if (const char *nl = reinterpret_cast<const char *>(memchr(yytext, '\n', yyleng))) { \
  69         unsigned lines(0); \
  70         size_t left; \
  71         do { \
  72             ++lines; \
  73             left = yyleng - (nl - yytext) - 1; \
  74             nl = reinterpret_cast<const char *>(memchr(nl + 1, '\n', left)); \
  75         } while (nl != NULL); \
  76         yylloc->end.lines(lines); \
  77         yylloc->end.columns(left); \
  78         yylloc->step(); \
  79         more \
  80     } else L \
  81 }
  82
  83 #define L { \
  84     yylloc->step(); \
  85     yylloc->columns(yyleng); \
  86 }
  87
  88 #define M { \
  89     if (yyextra->commented_) { \
  90         I(comment, Comment(Y), tk::Comment, hi::Comment); \
  91     } \
  92 }
  93
  94 int H(char c) {
  95     if (c >= '0' && c <= '9')
  96         return c - '0';
  97     if (c >= 'a' && c <= 'f')
  98         return c - 'a' + 10;
  99     if (c >= 'A' && c <= 'F')
 100         return c - 'A' + 10;
 101     return -1;
 102 }
 103
 104 #define YY_INPUT(data, value, size) { \
 105     if (yyextra->data_.eof()) \
 106         value = YY_NULL; \
 107     else { \
 108         yyextra->data_.read(data, size); \
 109         size_t copy(yyextra->data_.gcount()); \
 110         value = copy == 0 ? YY_NULL : copy; \
 111     } \
 112 }
 113
 114 %}
 115
 116 %option prefix="cy"
 117 %option bison-bridge
 118 %option bison-locations
 119 %option noyywrap
 120 %option yylineno
 121 %option nounput
 122 %option batch
 123 %option never-interactive
 124 %option reentrant
 125 %option stack
 126
 127 Exponent [eE][+-]?[0-9]+
 128 Escape   \\[\\'"bfnrtv]|\\0|\\x[0-9a-fA-F]{2}|\\u[0-9a-fA-F]{4}|\\\n
 129
 130 IdentifierStart [a-zA-Z$_]
 131 IdentifierPart  [a-zA-Z$_0-9]
 132
 133 NonTerminator [^\n]
 134 BackslashSequence \\{NonTerminator}
 135 RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence}
 136 RegularExpressionChar [^\n\\/]|{BackslashSequence}
 137 RegularExpressionFlags {IdentifierPart}*
 138 RegularExpressionChars {RegularExpressionChar}*
 139 RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}
 140
 141 @begin E4X
 142 XMLNameStart [a-zA-Z_:]
 143 XMLNamePart [a-zA-Z0-9.-_:]
 144 XMLName {XMLNameStart}{XMLNamePart}*
 145 @end
 146
 147 %s Div
 148 %s RegExp
 149
 150 @begin E4X
 151 %x XMLContent
 152 %x XMLTag
 153 @end
 154
 155 %%
 156
 157 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);
 158
 159 #![^\n]* L M
 160
 161 \/\/[^\n]* L M
 162
 163         /* http://ostermiller.org/findcomment.html */
 164         /* XXX: unify these two rules using !? */
 165 \/\*!([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V() C I(comment, Comment(Y), tk::Comment, hi::Comment);
 166 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V(N) M
 167
 168 @begin E4X
 169 <RegExp>"<>"      L F(tk::LeftRight, hi::Structure);
 170 <XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
 171
 172 <RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
 173 <RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
 174 <RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);
 175
 176 <XMLTag>"="  L F(tk::Equal, hi::Structure);
 177 <XMLTag>">"  L F(tk::Right, hi::Structure);
 178 <XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
 179 <XMLTag>"{"  L F(tk::OpenBrace, hi::Structure);
 180
 181 <XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue, hi::Constant);
 182 <XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
 183 <XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);
 184
 185 <XMLContent>"{"  L F(tk::OpenBrace, hi::Structure);
 186 <XMLContent>"<"  L F(tk::Left, hi::Structure);
 187 <XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
 188 @end
 189
 190 "..."  L C F(tk::PeriodPeriodPeriod, hi::Meta);
 191
 192 @begin E4X
 193 "::"   L C F(tk::ColonColon, hi::Operator);
 194 ".."   L C F(tk::PeriodPeriod, hi::Operator);
 195 @end
 196
 197 @begin E4X ObjectiveC
 198 "@"    L C F(tk::At, hi::Operator);
 199 "#"    L C F(tk::Pound, hi::Operator);
 200 @end
 201
 202 "&"    L C F(tk::Ampersand, hi::Operator);
 203 "&&"   L C F(tk::AmpersandAmpersand, hi::Operator);
 204 "&="   L C F(tk::AmpersandEqual, hi::Operator);
 205 "^"    L C F(tk::Carrot, hi::Operator);
 206 "^="   L C F(tk::CarrotEqual, hi::Operator);
 207 "="    L C F(tk::Equal, hi::Operator);
 208 "=="   L C F(tk::EqualEqual, hi::Operator);
 209 "==="  L C F(tk::EqualEqualEqual, hi::Operator);
 210 "=>"   L C F(tk::EqualRight, hi::Operator);
 211 "!"    L C F(tk::Exclamation, hi::Operator);
 212 "!="   L C F(tk::ExclamationEqual, hi::Operator);
 213 "!=="  L C F(tk::ExclamationEqualEqual, hi::Operator);
 214 "-"    L C F(tk::Hyphen, hi::Operator);
 215 "-="   L C F(tk::HyphenEqual, hi::Operator);
 216 "--"   L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
 217 "->"   L C F(tk::HyphenRight, hi::Operator);
 218 "<"    L C F(tk::Left, hi::Operator);
 219 "<="   L C F(tk::LeftEqual, hi::Operator);
 220 "<<"   L C F(tk::LeftLeft, hi::Operator);
 221 "<<="  L C F(tk::LeftLeftEqual, hi::Operator);
 222 "%"    L C F(tk::Percent, hi::Operator);
 223 "%="   L C F(tk::PercentEqual, hi::Operator);
 224 "."    L C F(tk::Period, hi::Operator);
 225 "|"    L C F(tk::Pipe, hi::Operator);
 226 "|="   L C F(tk::PipeEqual, hi::Operator);
 227 "||"   L C F(tk::PipePipe, hi::Operator);
 228 "+"    L C F(tk::Plus, hi::Operator);
 229 "+="   L C F(tk::PlusEqual, hi::Operator);
 230 "++"   L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
 231 ">"    L C F(tk::Right, hi::Operator);
 232 ">="   L C F(tk::RightEqual, hi::Operator);
 233 ">>"   L C F(tk::RightRight, hi::Operator);
 234 ">>="  L C F(tk::RightRightEqual, hi::Operator);
 235 ">>>"  L C F(tk::RightRightRight, hi::Operator);
 236 ">>>=" L C F(tk::RightRightRightEqual, hi::Operator);
 237 "*"    L C F(tk::Star, hi::Operator);
 238 "*="   L C F(tk::StarEqual, hi::Operator);
 239 "~"    L C F(tk::Tilde, hi::Operator);
 240
 241 <Div>"/"  L C F(tk::Slash, hi::Operator);
 242 <Div>"/=" L C F(tk::SlashEqual, hi::Operator);
 243
 244 ":"    L C F(tk::Colon, hi::Structure);
 245 ","    L C F(tk::Comma, hi::Structure);
 246 "?"    L C F(tk::Question, hi::Structure);
 247 ";"    L C F(tk::SemiColon, hi::Structure);
 248
 249 "("    L C F(tk::OpenParen, hi::Structure);
 250 ")"    L C F(tk::CloseParen, hi::Structure);
 251
 252 "{"    L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
 253 "}"    L C F(tk::CloseBrace, hi::Structure);
 254
 255 "["    L C F(tk::OpenBracket, hi::Structure);
 256 "]"    L C F(tk::CloseBracket, hi::Structure);
 257
 258 "@error"          L C F(tk::AtError, hi::Error);
 259
 260 @begin Java
 261 "@class"          L C F(tk::AtClass, hi::Meta);
 262 @end
 263
 264 @begin C
 265 "typedef"         L C I(identifier, Identifier("typedef"), tk::Typedef, hi::Meta);
 266 "unsigned"        L C I(identifier, Identifier("unsigned"), tk::Unsigned, hi::Type);
 267 "signed"          L C I(identifier, Identifier("signed"), tk::Signed, hi::Type);
 268 @end
 269
 270 @begin ObjectiveC
 271 "@encode"         L C F(tk::AtEncode, hi::Meta);
 272 "@end"            L C F(tk::AtEnd, hi::Meta);
 273 "@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta);
 274 "@import"         L C F(tk::AtImport, hi::Special);
 275 "@selector"       L C F(tk::AtSelector, hi::Meta);
 276
 277 "NULL"         L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant);
 278 "nil"          L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant);
 279 "YES"          L C I(identifier, Identifier("YES"), tk::Yes, hi::Constant);
 280 "NO"           L C I(identifier, Identifier("NO"), tk::No, hi::Constant);
 281
 282 "bool"         L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type);
 283 "BOOL"         L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type);
 284 "id"           L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type);
 285 "SEL"          L C I(identifier, Identifier("SEL"), tk::Identifier_, hi::Type);
 286 @end
 287
 288 "undefined"    L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator);
 289
 290 "false"        L C I(false, False(), tk::False, hi::Constant);
 291 "null"         L C I(null, Null(), tk::Null, hi::Constant);
 292 "true"         L C I(true, True(), tk::True, hi::Constant);
 293
 294 "auto"         L C I(word, Word("auto"), tk::Auto, hi::Meta);
 295 "break"        L R I(word, Word("break"), tk::Break, hi::Control);
 296 "case"         L C I(word, Word("case"), tk::Case, hi::Control);
 297 "catch"        L C I(word, Word("catch"), tk::Catch, hi::Control);
 298 "continue"     L R I(word, Word("continue"), tk::Continue, hi::Control);
 299 "default"      L C I(word, Word("default"), tk::Default, hi::Control);
 300 "delete"       L C I(word, Word("delete"), tk::Delete, hi::Operator);
 301 "do"           L C I(word, Word("do"), tk::Do, hi::Control);
 302 "else"         L C I(word, Word("else"), tk::Else, hi::Control);
 303 "finally"      L C I(word, Word("finally"), tk::Finally, hi::Control);
 304 "for"          L C I(word, Word("for"), tk::For, hi::Control);
 305 "function"     L C I(word, Word("function"), yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta);
 306 "if"           L C I(word, Word("if"), tk::If, hi::Control);
 307 "in"           L C I(word, Word("in"), yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator);
 308 "instanceof"   L C I(word, Word("instanceof"), tk::InstanceOf, hi::Operator);
 309 "new"          L C I(word, Word("new"), tk::New, hi::Operator);
 310 "return"       L R I(word, Word("return"), tk::Return, hi::Control);
 311 "switch"       L C I(word, Word("switch"), tk::Switch, hi::Control);
 312 "this"         L C I(this, This(), tk::This, hi::Constant);
 313 "throw"        L R I(word, Word("throw"), tk::Throw, hi::Control);
 314 "try"          L C I(word, Word("try"), tk::Try, hi::Control);
 315 "typeof"       L C I(word, Word("typeof"), tk::TypeOf, hi::Operator);
 316 "var"          L C I(word, Word("var"), tk::Var, hi::Meta);
 317 "void"         L C I(word, Word("void"), tk::Void, hi::Operator);
 318 "while"        L C I(word, Word("while"), tk::While, hi::Control);
 319 "with"         L C I(word, Word("with"), tk::With, hi::Control);
 320
 321 "debugger"     L C I(word, Word("debugger"), tk::Debugger, hi::Meta);
 322
 323 "const"        L C I(word, Word("const"), tk::Const, hi::Meta);
 324
 325 "class"        L C I(word, Word("class"), tk::Class, hi::Meta);
 326 "enum"         L C I(word, Word("enum"), tk::Enum, hi::Meta);
 327 "export"       L C I(word, Word("export"), tk::Export, hi::Meta);
 328 "extends"      L C I(word, Word("extends"), tk::Extends, hi::Meta);
 329 "import"       L C I(word, Word("import"), tk::Import, hi::Meta);
 330 "super"        L C I(word, Word("super"), tk::Super, hi::Constant);
 331
 332 "implements"   L C I(identifier, Identifier("implements"), tk::Implements, hi::Meta);
 333 "interface"    L C I(identifier, Identifier("interface"), tk::Interface, hi::Meta);
 334 "package"      L C I(identifier, Identifier("package"), tk::Package, hi::Meta);
 335 "private"      L C I(identifier, Identifier("private"), tk::Private, hi::Meta);
 336 "protected"    L C I(identifier, Identifier("protected"), tk::Protected, hi::Meta);
 337 "public"       L C I(identifier, Identifier("public"), tk::Public, hi::Meta);
 338 "static"       L C I(identifier, Identifier("static"), tk::Static, hi::Meta);
 339
 340 "abstract"     L C I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta);
 341 "boolean"      L C I(identifier, Identifier("boolean"), tk::Boolean, hi::Type);
 342 "byte"         L C I(identifier, Identifier("byte"), tk::Byte, hi::Type);
 343 "char"         L C I(identifier, Identifier("char"), tk::Char, hi::Type);
 344 "double"       L C I(identifier, Identifier("double"), tk::Double, hi::Type);
 345 "final"        L C I(identifier, Identifier("final"), tk::Final, hi::Meta);
 346 "float"        L C I(identifier, Identifier("float"), tk::Float, hi::Type);
 347 "goto"         L C I(identifier, Identifier("goto"), tk::Goto, hi::Control);
 348 "int"          L C I(identifier, Identifier("int"), tk::Int, hi::Type);
 349 "long"         L C I(identifier, Identifier("long"), tk::Long, hi::Type);
 350 "native"       L C I(identifier, Identifier("native"), tk::Native, hi::Meta);
 351 "short"        L C I(identifier, Identifier("short"), tk::Short, hi::Type);
 352 "synchronized" L C I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta);
 353 "throws"       L C I(identifier, Identifier("throws"), tk::Throws, hi::Meta);
 354 "transient"    L C I(identifier, Identifier("transient"), tk::Transient, hi::Meta);
 355 "volatile"     L C I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta);
 356
 357 "let"          L C I(identifier, Identifier("let"), tk::Let, hi::Meta);
 358 "yield"        L C I(identifier, Identifier("yield"), tk::Yield, hi::Control);
 359
 360 "each"         L C I(identifier, Identifier("each"), tk::Each, hi::Control);
 361 "of"           L C I(identifier, Identifier("of"), tk::Of, hi::Operator);
 362
 363 @begin E4X
 364 "namespace"    L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta);
 365 "xml"          L C I(identifier, Identifier("xml"), tk::XML, hi::Meta);
 366 @end
 367
 368 {IdentifierStart}{IdentifierPart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);
 369
 370 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?){Exponent}? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
 371
 372 0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant);
 373 0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant);
 374 0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant);
 375
 376 \"([^"\\\n]|{Escape})*\"|'([^'\\\n]|{Escape})*' L C {
 377     char *value(A char[yyleng]);
 378     char *local(value);
 379
 380     for (yy_size_t i(1), e(yyleng - 1); i != e; ++i) {
 381         char next(yytext[i]);
 382
 383         if (yytext[i] == '\\')
 384             switch (next = yytext[++i]) {
 385                 case '\n': continue;
 386                 case '\\': next = '\\'; break;
 387                 case '\'': next = '\''; break;
 388                 case '"': next = '"'; break;
 389                 case 'b': next = '\b'; break;
 390                 case 'f': next = '\f'; break;
 391                 case 'n': next = '\n'; break;
 392                 case 'r': next = '\r'; break;
 393                 case 't': next = '\t'; break;
 394                 case 'v': next = '\v'; break;
 395                 case '0': next = '\0'; break;
 396
 397                 case 'x':
 398                     next = H(yytext[i + 1]) << 4 | H(yytext[i + 2]);
 399                     i += 2;
 400                 break;
 401             }
 402
 403         *local++ = next;
 404     }
 405
 406     *local = '\0';
 407     I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
 408 }
 409
 410 \r?\n|\r|\xe2\x80[\xa8\xa9] yylloc->end.lines(); yylloc->step(); N
 411
 412 [ \t] L
 413
 414 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
 415
 416 . L {
 417     CYDriver::Error error;
 418     error.location_ = *yylloc;
 419     error.message_ = "syntax error, unknown token";
 420     yyextra->errors_.push_back(error);
 421     yyterminate();
 422 }
 423
 424 %%
 425
 426 void CYDriver::ScannerInit() {
 427     cylex_init(&scanner_);
 428     cyset_extra(this, scanner_);
 429 }
 430
 431 void CYDriver::ScannerDestroy() {
 432     cylex_destroy(scanner_);
 433 }
 434
 435 CYDriver::Condition CYDriver::GetCondition() {
 436     switch (yy_top_state(scanner_)) {
 437         case RegExp:
 438             return RegExpCondition;
 439 @begin E4X
 440         case XMLContent:
 441             return XMLContentCondition;
 442         case XMLTag:
 443             return XMLTagCondition;
 444 @end
 445         default:
 446             _assert(false);
 447     }
 448 }
 449
 450 void CYDriver::SetCondition(Condition condition) {
 451     struct yyguts_t *yyg(reinterpret_cast<struct yyguts_t *>(scanner_));
 452
 453     switch (condition) {
 454         case RegExpCondition:
 455             BEGIN(RegExp);
 456             break;
 457 @begin E4X
 458         case XMLContentCondition:
 459             BEGIN(XMLContent);
 460             break;
 461         case XMLTagCondition:
 462             BEGIN(XMLTag);
 463             break;
 464 @end
 465         default:
 466             _assert(false);
 467     }
 468 }
 469
 470 void CYDriver::PushCondition(Condition condition) {
 471     switch (condition) {
 472         case RegExpCondition:
 473             yy_push_state(RegExp, scanner_);
 474             break;
 475 @begin E4X
 476         case XMLContentCondition:
 477             yy_push_state(XMLContent, scanner_);
 478             break;
 479         case XMLTagCondition:
 480             yy_push_state(XMLTag, scanner_);
 481             break;
 482 @end
 483         default:
 484             _assert(false);
 485     }
 486 }
 487
 488 void CYDriver::PopCondition() {
 489     yy_pop_state(scanner_);
 490 }