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 interactive
 123 %option reentrant
 124 %option stack
 125
 126 Exponent [eE][+-]?[0-9]+
 127 Escape   \\[\\'"bfnrtv]|\\0|\\x[0-9a-fA-F]{2}|\\u[0-9a-fA-F]{4}|\\\n
 128
 129 IdentifierStart [a-zA-Z$_]
 130 IdentifierPart  [a-zA-Z$_0-9]
 131
 132 NonTerminator [^\n]
 133 BackslashSequence \\{NonTerminator}
 134 RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence}
 135 RegularExpressionChar [^\n\\/]|{BackslashSequence}
 136 RegularExpressionFlags {IdentifierPart}*
 137 RegularExpressionChars {RegularExpressionChar}*
 138 RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}
 139
 140 @begin E4X
 141 XMLNameStart [a-zA-Z_:]
 142 XMLNamePart [a-zA-Z0-9.-_:]
 143 XMLName {XMLNameStart}{XMLNamePart}*
 144 @end
 145
 146 %s Div
 147 %s RegExp
 148
 149 @begin E4X
 150 %x XMLContent
 151 %x XMLTag
 152 @end
 153
 154 %%
 155
 156 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);
 157
 158 \/\/[^\n]* L M
 159
 160         /* http://ostermiller.org/findcomment.html */
 161         /* XXX: unify these two rules using !? */
 162 \/\*!([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V() C I(comment, Comment(Y), tk::Comment, hi::Comment);
 163 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V(N) M
 164
 165 @begin E4X
 166 <RegExp>"<>"      L F(tk::LeftRight, hi::Structure);
 167 <XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
 168
 169 <RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
 170 <RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
 171 <RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);
 172
 173 <XMLTag>"="  L F(tk::Equal, hi::Structure);
 174 <XMLTag>">"  L F(tk::Right, hi::Structure);
 175 <XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
 176 <XMLTag>"{"  L F(tk::OpenBrace, hi::Structure);
 177
 178 <XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue, hi::Constant);
 179 <XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
 180 <XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);
 181
 182 <XMLContent>"{"  L F(tk::OpenBrace, hi::Structure);
 183 <XMLContent>"<"  L F(tk::Left, hi::Structure);
 184 <XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
 185 @end
 186
 187 "..."  L C F(tk::PeriodPeriodPeriod, hi::Meta);
 188
 189 @begin E4X
 190 "::"   L C F(tk::ColonColon, hi::Operator);
 191 ".."   L C F(tk::PeriodPeriod, hi::Operator);
 192 @end
 193
 194 @begin E4X ObjectiveC
 195 "@"    L C F(tk::At, hi::Operator);
 196 "#"    L C F(tk::Pound, hi::Operator);
 197 @end
 198
 199 "&"    L C F(tk::Ampersand, hi::Operator);
 200 "&&"   L C F(tk::AmpersandAmpersand, hi::Operator);
 201 "&="   L C F(tk::AmpersandEqual, hi::Operator);
 202 "^"    L C F(tk::Carrot, hi::Operator);
 203 "^="   L C F(tk::CarrotEqual, hi::Operator);
 204 "="    L C F(tk::Equal, hi::Operator);
 205 "=="   L C F(tk::EqualEqual, hi::Operator);
 206 "==="  L C F(tk::EqualEqualEqual, hi::Operator);
 207 "=>"   L C F(tk::EqualRight, hi::Operator);
 208 "!"    L C F(tk::Exclamation, hi::Operator);
 209 "!="   L C F(tk::ExclamationEqual, hi::Operator);
 210 "!=="  L C F(tk::ExclamationEqualEqual, hi::Operator);
 211 "-"    L C F(tk::Hyphen, hi::Operator);
 212 "-="   L C F(tk::HyphenEqual, hi::Operator);
 213 "--"   L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
 214 "->"   L C F(tk::HyphenRight, hi::Operator);
 215 "<"    L C F(tk::Left, hi::Operator);
 216 "<="   L C F(tk::LeftEqual, hi::Operator);
 217 "<<"   L C F(tk::LeftLeft, hi::Operator);
 218 "<<="  L C F(tk::LeftLeftEqual, hi::Operator);
 219 "%"    L C F(tk::Percent, hi::Operator);
 220 "%="   L C F(tk::PercentEqual, hi::Operator);
 221 "."    L C F(tk::Period, hi::Operator);
 222 "|"    L C F(tk::Pipe, hi::Operator);
 223 "|="   L C F(tk::PipeEqual, hi::Operator);
 224 "||"   L C F(tk::PipePipe, hi::Operator);
 225 "+"    L C F(tk::Plus, hi::Operator);
 226 "+="   L C F(tk::PlusEqual, hi::Operator);
 227 "++"   L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
 228 ">"    L C F(tk::Right, hi::Operator);
 229 ">="   L C F(tk::RightEqual, hi::Operator);
 230 ">>"   L C F(tk::RightRight, hi::Operator);
 231 ">>="  L C F(tk::RightRightEqual, hi::Operator);
 232 ">>>"  L C F(tk::RightRightRight, hi::Operator);
 233 ">>>=" L C F(tk::RightRightRightEqual, hi::Operator);
 234 "*"    L C F(tk::Star, hi::Operator);
 235 "*="   L C F(tk::StarEqual, hi::Operator);
 236 "~"    L C F(tk::Tilde, hi::Operator);
 237
 238 <Div>"/"  L C F(tk::Slash, hi::Operator);
 239 <Div>"/=" L C F(tk::SlashEqual, hi::Operator);
 240
 241 ":"    L C F(tk::Colon, hi::Structure);
 242 ","    L C F(tk::Comma, hi::Structure);
 243 "?"    L C F(tk::Question, hi::Structure);
 244 ";"    L C F(tk::SemiColon, hi::Structure);
 245
 246 "("    L C F(tk::OpenParen, hi::Structure);
 247 ")"    L C F(tk::CloseParen, hi::Structure);
 248
 249 "{"    L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
 250 "}"    L C F(tk::CloseBrace, hi::Structure);
 251
 252 "["    L C F(tk::OpenBracket, hi::Structure);
 253 "]"    L C F(tk::CloseBracket, hi::Structure);
 254
 255 "@error"          L C F(tk::AtError, hi::Error);
 256
 257 @begin Java
 258 "@class"          L C F(tk::AtClass, hi::Meta);
 259 @end
 260
 261 @begin C
 262 "typedef"         L C I(identifier, Identifier("typedef"), tk::Typedef, hi::Meta);
 263 @end
 264
 265 @begin ObjectiveC
 266 "@encode"         L C F(tk::AtEncode, hi::Meta);
 267 "@end"            L C F(tk::AtEnd, hi::Meta);
 268 "@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta);
 269 "@import"         L C F(tk::AtImport, hi::Special);
 270 "@selector"       L C F(tk::AtSelector, hi::Meta);
 271
 272 "NULL"         L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant);
 273 "nil"          L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant);
 274 "YES"          L C I(identifier, Identifier("YES"), tk::Yes, hi::Constant);
 275 "NO"           L C I(identifier, Identifier("NO"), tk::No, hi::Constant);
 276
 277 "bool"         L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type);
 278 "BOOL"         L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type);
 279 "id"           L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type);
 280 "SEL"          L C I(identifier, Identifier("SEL"), tk::Identifier_, hi::Type);
 281 @end
 282
 283 "undefined"    L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator);
 284
 285 "false"        L C I(false, False(), tk::False, hi::Constant);
 286 "null"         L C I(null, Null(), tk::Null, hi::Constant);
 287 "true"         L C I(true, True(), tk::True, hi::Constant);
 288
 289 "auto"         L C I(word, Word("auto"), tk::Auto, hi::Meta);
 290 "break"        L R I(word, Word("break"), tk::Break, hi::Control);
 291 "case"         L C I(word, Word("case"), tk::Case, hi::Control);
 292 "catch"        L C I(word, Word("catch"), tk::Catch, hi::Control);
 293 "continue"     L R I(word, Word("continue"), tk::Continue, hi::Control);
 294 "default"      L C I(word, Word("default"), tk::Default, hi::Control);
 295 "delete"       L C I(word, Word("delete"), tk::Delete, hi::Operator);
 296 "do"           L C I(word, Word("do"), tk::Do, hi::Control);
 297 "else"         L C I(word, Word("else"), tk::Else, hi::Control);
 298 "finally"      L C I(word, Word("finally"), tk::Finally, hi::Control);
 299 "for"          L C I(word, Word("for"), tk::For, hi::Control);
 300 "function"     L C I(word, Word("function"), yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta);
 301 "if"           L C I(word, Word("if"), tk::If, hi::Control);
 302 "in"           L C I(word, Word("in"), yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator);
 303 "instanceof"   L C I(word, Word("instanceof"), tk::InstanceOf, hi::Operator);
 304 "new"          L C I(word, Word("new"), tk::New, hi::Operator);
 305 "return"       L R I(word, Word("return"), tk::Return, hi::Control);
 306 "switch"       L C I(word, Word("switch"), tk::Switch, hi::Control);
 307 "this"         L C I(this, This(), tk::This, hi::Constant);
 308 "throw"        L R I(word, Word("throw"), tk::Throw, hi::Control);
 309 "try"          L C I(word, Word("try"), tk::Try, hi::Control);
 310 "typeof"       L C I(word, Word("typeof"), tk::TypeOf, hi::Operator);
 311 "var"          L C I(word, Word("var"), tk::Var, hi::Meta);
 312 "void"         L C I(word, Word("void"), tk::Void, hi::Operator);
 313 "while"        L C I(word, Word("while"), tk::While, hi::Control);
 314 "with"         L C I(word, Word("with"), tk::With, hi::Control);
 315
 316 "debugger"     L C I(word, Word("debugger"), tk::Debugger, hi::Meta);
 317
 318 "const"        L C I(word, Word("const"), tk::Const, hi::Meta);
 319
 320 "class"        L C I(word, Word("class"), tk::Class, hi::Meta);
 321 "enum"         L C I(word, Word("enum"), tk::Enum, hi::Meta);
 322 "export"       L C I(word, Word("export"), tk::Export, hi::Meta);
 323 "extends"      L C I(word, Word("extends"), tk::Extends, hi::Meta);
 324 "import"       L C I(word, Word("import"), tk::Import, hi::Meta);
 325 "super"        L C I(word, Word("super"), tk::Super, hi::Constant);
 326
 327 "implements"   L C I(identifier, Identifier("implements"), tk::Implements, hi::Meta);
 328 "interface"    L C I(identifier, Identifier("interface"), tk::Interface, hi::Meta);
 329 "package"      L C I(identifier, Identifier("package"), tk::Package, hi::Meta);
 330 "private"      L C I(identifier, Identifier("private"), tk::Private, hi::Meta);
 331 "protected"    L C I(identifier, Identifier("protected"), tk::Protected, hi::Meta);
 332 "public"       L C I(identifier, Identifier("public"), tk::Public, hi::Meta);
 333 "static"       L C I(identifier, Identifier("static"), tk::Static, hi::Meta);
 334
 335 "abstract"     L C I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta);
 336 "boolean"      L C I(identifier, Identifier("boolean"), tk::Boolean, hi::Type);
 337 "byte"         L C I(identifier, Identifier("byte"), tk::Byte, hi::Type);
 338 "char"         L C I(identifier, Identifier("char"), tk::Char, hi::Type);
 339 "double"       L C I(identifier, Identifier("double"), tk::Double, hi::Type);
 340 "final"        L C I(identifier, Identifier("final"), tk::Final, hi::Meta);
 341 "float"        L C I(identifier, Identifier("float"), tk::Float, hi::Type);
 342 "goto"         L C I(identifier, Identifier("goto"), tk::Goto, hi::Control);
 343 "int"          L C I(identifier, Identifier("int"), tk::Int, hi::Type);
 344 "long"         L C I(identifier, Identifier("long"), tk::Long, hi::Type);
 345 "native"       L C I(identifier, Identifier("native"), tk::Native, hi::Meta);
 346 "short"        L C I(identifier, Identifier("short"), tk::Short, hi::Type);
 347 "synchronized" L C I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta);
 348 "throws"       L C I(identifier, Identifier("throws"), tk::Throws, hi::Meta);
 349 "transient"    L C I(identifier, Identifier("transient"), tk::Transient, hi::Meta);
 350 "volatile"     L C I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta);
 351
 352 "let"          L C I(identifier, Identifier("let"), tk::Let, hi::Meta);
 353 "yield"        L C I(identifier, Identifier("yield"), tk::Yield, hi::Control);
 354
 355 "each"         L C I(identifier, Identifier("each"), tk::Each, hi::Control);
 356 "of"           L C I(identifier, Identifier("of"), tk::Of, hi::Operator);
 357
 358 @begin E4X
 359 "namespace"    L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta);
 360 "xml"          L C I(identifier, Identifier("xml"), tk::XML, hi::Meta);
 361 @end
 362
 363 {IdentifierStart}{IdentifierPart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);
 364
 365 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?){Exponent}? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
 366
 367 0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant);
 368 0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant);
 369 0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant);
 370
 371 \"([^"\\\n]|{Escape})*\"|'([^'\\\n]|{Escape})*' L C {
 372     char *value(A char[yyleng]);
 373     char *local(value);
 374
 375     for (yy_size_t i(1), e(yyleng - 1); i != e; ++i) {
 376         char next(yytext[i]);
 377
 378         if (yytext[i] == '\\')
 379             switch (next = yytext[++i]) {
 380                 case '\n': continue;
 381                 case '\\': next = '\\'; break;
 382                 case '\'': next = '\''; break;
 383                 case '"': next = '"'; break;
 384                 case 'b': next = '\b'; break;
 385                 case 'f': next = '\f'; break;
 386                 case 'n': next = '\n'; break;
 387                 case 'r': next = '\r'; break;
 388                 case 't': next = '\t'; break;
 389                 case 'v': next = '\v'; break;
 390                 case '0': next = '\0'; break;
 391
 392                 case 'x':
 393                     next = H(yytext[i + 1]) << 4 | H(yytext[i + 2]);
 394                     i += 2;
 395                 break;
 396             }
 397
 398         *local++ = next;
 399     }
 400
 401     *local = '\0';
 402     I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
 403 }
 404
 405 \r?\n|\r|\xe2\x80[\xa8\xa9] yylloc->end.lines(); yylloc->step(); N
 406
 407 [ \t] L
 408
 409 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
 410
 411 . L {
 412     CYDriver::Error error;
 413     error.location_ = *yylloc;
 414     error.message_ = "syntax error, unknown token";
 415     yyextra->errors_.push_back(error);
 416     yyterminate();
 417 }
 418
 419 %%
 420
 421 void CYDriver::ScannerInit() {
 422     cylex_init(&scanner_);
 423     cyset_extra(this, scanner_);
 424 }
 425
 426 void CYDriver::ScannerDestroy() {
 427     cylex_destroy(scanner_);
 428 }
 429
 430 CYDriver::Condition CYDriver::GetCondition() {
 431     switch (yy_top_state(scanner_)) {
 432         case RegExp:
 433             return RegExpCondition;
 434 @begin E4X
 435         case XMLContent:
 436             return XMLContentCondition;
 437         case XMLTag:
 438             return XMLTagCondition;
 439 @end
 440         default:
 441             _assert(false);
 442     }
 443 }
 444
 445 void CYDriver::SetCondition(Condition condition) {
 446     struct yyguts_t *yyg(reinterpret_cast<struct yyguts_t *>(scanner_));
 447
 448     switch (condition) {
 449         case RegExpCondition:
 450             BEGIN(RegExp);
 451             break;
 452 @begin E4X
 453         case XMLContentCondition:
 454             BEGIN(XMLContent);
 455             break;
 456         case XMLTagCondition:
 457             BEGIN(XMLTag);
 458             break;
 459 @end
 460         default:
 461             _assert(false);
 462     }
 463 }
 464
 465 void CYDriver::PushCondition(Condition condition) {
 466     switch (condition) {
 467         case RegExpCondition:
 468             yy_push_state(RegExp, scanner_);
 469             break;
 470 @begin E4X
 471         case XMLContentCondition:
 472             yy_push_state(XMLContent, scanner_);
 473             break;
 474         case XMLTagCondition:
 475             yy_push_state(XMLTag, scanner_);
 476             break;
 477 @end
 478         default:
 479             _assert(false);
 480     }
 481 }
 482
 483 void CYDriver::PopCondition() {
 484     yy_pop_state(scanner_);
 485 }