Cycript.l.in

   1 /* Cycript - Optimizing JavaScript Compiler/Runtime
   2  * Copyright (C) 2009-2012  Jay Freeman (saurik)
   3 */
   4
   5 /* GNU Lesser General Public License, Version 3 {{{ */
   6 /*
   7  * Cycript is free software: you can redistribute it and/or modify it under
   8  * the terms of the GNU Lesser General Public License as published by the
   9  * Free Software Foundation, either version 3 of the License, or (at your
  10  * option) any later version.
  11  *
  12  * Cycript is distributed in the hope that it will be useful, but WITHOUT
  13  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
  15  * License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser 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     yyextra->no_.AtImplementation = false; \
  37     yyextra->no_.Function = false; \
  38     yyextra->no_.OpenBrace = false; \
  39     yylval->highlight_ = highlight; \
  40     return token; \
  41 } while (false)
  42
  43 #define A new($pool)
  44 #define Y apr_pstrmemdup($pool, yytext, yyleng)
  45
  46 #define I(type, Type, value, highlight) do { \
  47     yylval->type ## _ = A CY ## Type; \
  48     F(value, highlight); \
  49 } while (false)
  50
  51 #define T yylval->newline_ = yyextra->state_ == CYNewLine; BEGIN(Div);
  52 #define C T yyextra->state_ = CYClear;
  53 #define R T yyextra->state_ = CYRestricted;
  54
  55 #define N \
  56     if (yyextra->state_ != CYNewLine) { \
  57         if (yyextra->state_ != CYRestricted) \
  58             yyextra->state_ = CYNewLine; \
  59         else { \
  60             yyextra->state_ = CYClear; \
  61             F(tk::NewLine, hi::Nothing); \
  62         } \
  63     }
  64
  65 #define V(more) { \
  66     if (const char *nl = reinterpret_cast<const char *>(memchr(yytext, '\n', yyleng))) { \
  67         unsigned lines(0); \
  68         size_t left; \
  69         do { \
  70             ++lines; \
  71             left = yyleng - (nl - yytext) - 1; \
  72             nl = reinterpret_cast<const char *>(memchr(nl + 1, '\n', left)); \
  73         } while (nl != NULL); \
  74         yylloc->end.lines(lines); \
  75         yylloc->end.columns(left); \
  76         yylloc->step(); \
  77         more \
  78     } else L \
  79 }
  80
  81 #define L { \
  82     yylloc->step(); \
  83     yylloc->columns(yyleng); \
  84 }
  85
  86 #define M { \
  87     if (yyextra->commented_) { \
  88         I(comment, Comment(Y), tk::Comment, hi::Comment); \
  89     } \
  90 }
  91
  92 int H(char c) {
  93     if (c >= '0' && c <= '9')
  94         return c - '0';
  95     if (c >= 'a' && c <= 'f')
  96         return c - 'a' + 10;
  97     if (c >= 'A' && c <= 'F')
  98         return c - 'A' + 10;
  99     return -1;
 100 }
 101
 102 #define YY_INPUT(data, value, size) { \
 103     if (yyextra->data_.eof()) \
 104         value = YY_NULL; \
 105     else { \
 106         yyextra->data_.read(data, size); \
 107         size_t copy(yyextra->data_.gcount()); \
 108         value = copy == 0 ? YY_NULL : copy; \
 109     } \
 110 }
 111
 112 %}
 113
 114 %option prefix="cy"
 115 %option bison-bridge
 116 %option bison-locations
 117 %option noyywrap
 118 %option yylineno
 119 %option nounput
 120 %option interactive
 121 %option reentrant
 122 %option stack
 123
 124 Exponent [eE][+-]?[0-9]+
 125 Escape   \\[\\'"bfnrtv]|\\0|\\x[0-9a-fA-F]{2}|\\u[0-9a-fA-F]{4}|\\\n
 126
 127 IdentifierStart [a-zA-Z$_]
 128 IdentifierPart  [a-zA-Z$_0-9]
 129
 130 NonTerminator [^\n]
 131 BackslashSequence \\{NonTerminator}
 132 RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence}
 133 RegularExpressionChar [^\n\\/]|{BackslashSequence}
 134 RegularExpressionFlags {IdentifierPart}*
 135 RegularExpressionChars {RegularExpressionChar}*
 136 RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}
 137
 138 @begin E4X
 139 XMLNameStart [a-zA-Z_:]
 140 XMLNamePart [a-zA-Z0-9.-_:]
 141 XMLName {XMLNameStart}{XMLNamePart}*
 142 @end
 143
 144 %s Div
 145 %s RegExp
 146
 147 @begin E4X
 148 %x XMLContent
 149 %x XMLTag
 150 @end
 151
 152 %%
 153
 154 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);
 155
 156 \/\/[^\n]* L M
 157
 158         /* http://ostermiller.org/findcomment.html */
 159         /* XXX: unify these two rules using !? */
 160 \/\*!([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V() C I(comment, Comment(Y), tk::Comment, hi::Comment);
 161 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V(N) M
 162
 163 @begin E4X
 164 <RegExp>"<>"      L F(tk::LeftRight, hi::Structure);
 165 <XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
 166
 167 <RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
 168 <RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
 169 <RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);
 170
 171 <XMLTag>"="  L F(tk::Equal, hi::Structure);
 172 <XMLTag>">"  L F(tk::Right, hi::Structure);
 173 <XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
 174 <XMLTag>"{"  L F(tk::OpenBrace, hi::Structure);
 175
 176 <XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue, hi::Constant);
 177 <XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
 178 <XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);
 179
 180 <XMLContent>"{"  L F(tk::OpenBrace, hi::Structure);
 181 <XMLContent>"<"  L F(tk::Left, hi::Structure);
 182 <XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
 183 @end
 184
 185 "..."  L C F(tk::PeriodPeriodPeriod, hi::Meta);
 186
 187 @begin E4X
 188 "::"   L C F(tk::ColonColon, hi::Operator);
 189 ".."   L C F(tk::PeriodPeriod, hi::Operator);
 190 @end
 191
 192 @begin E4X ObjectiveC
 193 "@"    L C F(tk::At, hi::Operator);
 194 @end
 195
 196 "&"    L C F(tk::Ampersand, hi::Operator);
 197 "&&"   L C F(tk::AmpersandAmpersand, hi::Operator);
 198 "&="   L C F(tk::AmpersandEqual, hi::Operator);
 199 "^"    L C F(tk::Carrot, hi::Operator);
 200 "^="   L C F(tk::CarrotEqual, hi::Operator);
 201 "="    L C F(tk::Equal, hi::Operator);
 202 "=="   L C F(tk::EqualEqual, hi::Operator);
 203 "==="  L C F(tk::EqualEqualEqual, hi::Operator);
 204 "=>"   L C F(tk::EqualRight, hi::Operator);
 205 "!"    L C F(tk::Exclamation, hi::Operator);
 206 "!="   L C F(tk::ExclamationEqual, hi::Operator);
 207 "!=="  L C F(tk::ExclamationEqualEqual, hi::Operator);
 208 "-"    L C F(tk::Hyphen, hi::Operator);
 209 "-="   L C F(tk::HyphenEqual, hi::Operator);
 210 "--"   L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
 211 "->"   L C F(tk::HyphenRight, hi::Operator);
 212 "<"    L C F(tk::Left, hi::Operator);
 213 "<="   L C F(tk::LeftEqual, hi::Operator);
 214 "<<"   L C F(tk::LeftLeft, hi::Operator);
 215 "<<="  L C F(tk::LeftLeftEqual, hi::Operator);
 216 "%"    L C F(tk::Percent, hi::Operator);
 217 "%="   L C F(tk::PercentEqual, hi::Operator);
 218 "."    L C F(tk::Period, hi::Operator);
 219 "|"    L C F(tk::Pipe, hi::Operator);
 220 "|="   L C F(tk::PipeEqual, hi::Operator);
 221 "||"   L C F(tk::PipePipe, hi::Operator);
 222 "+"    L C F(tk::Plus, hi::Operator);
 223 "+="   L C F(tk::PlusEqual, hi::Operator);
 224 "++"   L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
 225 ">"    L C F(tk::Right, hi::Operator);
 226 ">="   L C F(tk::RightEqual, hi::Operator);
 227 ">>"   L C F(tk::RightRight, hi::Operator);
 228 ">>="  L C F(tk::RightRightEqual, hi::Operator);
 229 ">>>"  L C F(tk::RightRightRight, hi::Operator);
 230 ">>>=" L C F(tk::RightRightRightEqual, hi::Operator);
 231 "*"    L C F(tk::Star, hi::Operator);
 232 "*="   L C F(tk::StarEqual, hi::Operator);
 233 "~"    L C F(tk::Tilde, hi::Operator);
 234
 235 <Div>"/"  L C F(tk::Slash, hi::Operator);
 236 <Div>"/=" L C F(tk::SlashEqual, hi::Operator);
 237
 238 ":"    L C F(tk::Colon, hi::Structure);
 239 ","    L C F(tk::Comma, hi::Structure);
 240 "?"    L C F(tk::Question, hi::Structure);
 241 ";"    L C F(tk::SemiColon, hi::Structure);
 242
 243 "("    L C F(tk::OpenParen, hi::Structure);
 244 ")"    L C F(tk::CloseParen, hi::Structure);
 245
 246 "{"    L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
 247 "}"    L C F(tk::CloseBrace, hi::Structure);
 248
 249 "["    L C F(tk::OpenBracket, hi::Structure);
 250 "]"    L C F(tk::CloseBracket, hi::Structure);
 251
 252 @begin Java
 253 "@class"          L C F(tk::AtClass, hi::Meta);
 254 @end
 255
 256 @begin ObjectiveC
 257 "@encode"         L C F(tk::AtEncode, hi::Meta);
 258 "@end"            L C F(tk::AtEnd, hi::Meta);
 259 "@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta);
 260 "@import"         L C F(tk::AtImport, hi::Meta);
 261 "@selector"       L C F(tk::AtSelector, hi::Meta);
 262
 263 "NULL"         L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant);
 264 "nil"          L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant);
 265 "YES"          L C I(identifier, Identifier("YES"), tk::Identifier_, hi::Constant);
 266 "NO"           L C I(identifier, Identifier("NO"), tk::Identifier_, hi::Constant);
 267
 268 "bool"         L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type);
 269 "BOOL"         L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type);
 270 "id"           L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type);
 271 @end
 272
 273 "undefined"    L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator);
 274
 275 "false"        L C I(false, False(), tk::False, hi::Constant);
 276 "null"         L C I(null, Null(), tk::Null, hi::Constant);
 277 "true"         L C I(true, True(), tk::True, hi::Constant);
 278
 279 "auto"         L C I(word, Word("auto"), tk::Auto, hi::Meta);
 280 "break"        L R I(word, Word("break"), tk::Break, hi::Control);
 281 "case"         L C I(word, Word("case"), tk::Case, hi::Control);
 282 "catch"        L C I(word, Word("catch"), tk::Catch, hi::Control);
 283 "continue"     L R I(word, Word("continue"), tk::Continue, hi::Control);
 284 "default"      L C I(word, Word("default"), tk::Default, hi::Control);
 285 "delete"       L C I(word, Word("delete"), tk::Delete, hi::Operator);
 286 "do"           L C I(word, Word("do"), tk::Do, hi::Control);
 287 "else"         L C I(word, Word("else"), tk::Else, hi::Control);
 288 "finally"      L C I(word, Word("finally"), tk::Finally, hi::Control);
 289 "for"          L C I(word, Word("for"), tk::For, hi::Control);
 290 "function"     L C I(word, Word("function"), yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta);
 291 "if"           L C I(word, Word("if"), tk::If, hi::Control);
 292 "in"           L C I(word, Word("in"), yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator);
 293 "instanceof"   L C I(word, Word("instanceof"), tk::InstanceOf, hi::Operator);
 294 "new"          L C I(word, Word("new"), tk::New, hi::Operator);
 295 "return"       L R I(word, Word("return"), tk::Return, hi::Control);
 296 "switch"       L C I(word, Word("switch"), tk::Switch, hi::Control);
 297 "this"         L C I(this, This(), tk::This, hi::Constant);
 298 "throw"        L R I(word, Word("throw"), tk::Throw, hi::Control);
 299 "try"          L C I(word, Word("try"), tk::Try, hi::Control);
 300 "typeof"       L C I(word, Word("typeof"), tk::TypeOf, hi::Operator);
 301 "var"          L C I(word, Word("var"), tk::Var, hi::Meta);
 302 "void"         L C I(word, Word("void"), tk::Void, hi::Operator);
 303 "while"        L C I(word, Word("while"), tk::While, hi::Control);
 304 "with"         L C I(word, Word("with"), tk::With, hi::Control);
 305
 306 "debugger"     L C I(word, Word("debugger"), tk::Debugger, hi::Meta);
 307
 308 "const"        L C I(word, Word("const"), tk::Const, hi::Meta);
 309
 310 "class"        L C I(word, Word("class"), tk::Class, hi::Meta);
 311 "enum"         L C I(word, Word("enum"), tk::Enum, hi::Meta);
 312 "export"       L C I(word, Word("export"), tk::Export, hi::Meta);
 313 "extends"      L C I(word, Word("extends"), tk::Extends, hi::Meta);
 314 "import"       L C I(word, Word("import"), tk::Import, hi::Meta);
 315 "super"        L C I(word, Word("super"), tk::Super, hi::Constant);
 316
 317 "implements"   L C I(identifier, Identifier("implements"), tk::Implements, hi::Meta);
 318 "interface"    L C I(identifier, Identifier("interface"), tk::Interface, hi::Meta);
 319 "package"      L C I(identifier, Identifier("package"), tk::Package, hi::Meta);
 320 "private"      L C I(identifier, Identifier("private"), tk::Private, hi::Meta);
 321 "protected"    L C I(identifier, Identifier("protected"), tk::Protected, hi::Meta);
 322 "public"       L C I(identifier, Identifier("public"), tk::Public, hi::Meta);
 323 "static"       L C I(identifier, Identifier("static"), tk::Static, hi::Meta);
 324
 325 "abstract"     L C I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta);
 326 "boolean"      L C I(identifier, Identifier("boolean"), tk::Boolean, hi::Type);
 327 "byte"         L C I(identifier, Identifier("byte"), tk::Byte, hi::Type);
 328 "char"         L C I(identifier, Identifier("char"), tk::Char, hi::Type);
 329 "double"       L C I(identifier, Identifier("double"), tk::Double, hi::Type);
 330 "final"        L C I(identifier, Identifier("final"), tk::Final, hi::Meta);
 331 "float"        L C I(identifier, Identifier("float"), tk::Float, hi::Type);
 332 "goto"         L C I(identifier, Identifier("goto"), tk::Goto, hi::Control);
 333 "int"          L C I(identifier, Identifier("int"), tk::Int, hi::Type);
 334 "long"         L C I(identifier, Identifier("long"), tk::Long, hi::Type);
 335 "native"       L C I(identifier, Identifier("native"), tk::Native, hi::Meta);
 336 "short"        L C I(identifier, Identifier("short"), tk::Short, hi::Type);
 337 "synchronized" L C I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta);
 338 "throws"       L C I(identifier, Identifier("throws"), tk::Throws, hi::Meta);
 339 "transient"    L C I(identifier, Identifier("transient"), tk::Transient, hi::Meta);
 340 "volatile"     L C I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta);
 341
 342 "let"          L C I(identifier, Identifier("let"), tk::Let, hi::Meta);
 343 "yield"        L C I(identifier, Identifier("yield"), tk::Yield, hi::Control);
 344
 345 "each"         L C I(identifier, Identifier("each"), tk::Each, hi::Control);
 346 "of"           L C I(identifier, Identifier("of"), tk::Of, hi::Operator);
 347
 348 @begin E4X
 349 "namespace"    L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta);
 350 "xml"          L C I(identifier, Identifier("xml"), tk::XML, hi::Meta);
 351 @end
 352
 353 {IdentifierStart}{IdentifierPart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);
 354
 355 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?){Exponent}? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
 356
 357 0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant);
 358 0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant);
 359 0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant);
 360
 361 \"([^"\\\n]|{Escape})*\"|'([^'\\\n]|{Escape})*' L C {
 362     char *value(A char[yyleng]);
 363     char *local(value);
 364
 365     for (yy_size_t i(1), e(yyleng - 1); i != e; ++i) {
 366         char next(yytext[i]);
 367
 368         if (yytext[i] == '\\')
 369             switch (next = yytext[++i]) {
 370                 case '\n': continue;
 371                 case '\\': next = '\\'; break;
 372                 case '\'': next = '\''; break;
 373                 case '"': next = '"'; break;
 374                 case 'b': next = '\b'; break;
 375                 case 'f': next = '\f'; break;
 376                 case 'n': next = '\n'; break;
 377                 case 'r': next = '\r'; break;
 378                 case 't': next = '\t'; break;
 379                 case 'v': next = '\v'; break;
 380                 case '0': next = '\0'; break;
 381
 382                 case 'x':
 383                     next = H(yytext[i + 1]) << 4 | H(yytext[i + 2]);
 384                     i += 2;
 385                 break;
 386             }
 387
 388         *local++ = next;
 389     }
 390
 391     *local = '\0';
 392     I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
 393 }
 394
 395 \r?\n|\r|\xe2\x80[\xa8\xa9] yylloc->end.lines(); yylloc->step(); N
 396
 397 [ \t] L
 398
 399 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
 400
 401 . L {
 402     CYDriver::Error error;
 403     error.location_ = *yylloc;
 404     error.message_ = "syntax error, unknown token";
 405     yyextra->errors_.push_back(error);
 406     yyterminate();
 407 }
 408
 409 %%
 410
 411 void CYDriver::ScannerInit() {
 412     cylex_init(&scanner_);
 413     cyset_extra(this, scanner_);
 414 }
 415
 416 void CYDriver::ScannerDestroy() {
 417     cylex_destroy(scanner_);
 418 }
 419
 420 CYDriver::Condition CYDriver::GetCondition() {
 421     switch (yy_top_state(scanner_)) {
 422         case RegExp:
 423             return RegExpCondition;
 424 @begin E4X
 425         case XMLContent:
 426             return XMLContentCondition;
 427         case XMLTag:
 428             return XMLTagCondition;
 429 @end
 430         default:
 431             _assert(false);
 432     }
 433 }
 434
 435 void CYDriver::SetCondition(Condition condition) {
 436     struct yyguts_t *yyg(reinterpret_cast<struct yyguts_t *>(scanner_));
 437
 438     switch (condition) {
 439         case RegExpCondition:
 440             BEGIN(RegExp);
 441             break;
 442 @begin E4X
 443         case XMLContentCondition:
 444             BEGIN(XMLContent);
 445             break;
 446         case XMLTagCondition:
 447             BEGIN(XMLTag);
 448             break;
 449 @end
 450         default:
 451             _assert(false);
 452     }
 453 }
 454
 455 void CYDriver::PushCondition(Condition condition) {
 456     switch (condition) {
 457         case RegExpCondition:
 458             yy_push_state(RegExp, scanner_);
 459             break;
 460 @begin E4X
 461         case XMLContentCondition:
 462             yy_push_state(XMLContent, scanner_);
 463             break;
 464         case XMLTagCondition:
 465             yy_push_state(XMLTag, scanner_);
 466             break;
 467 @end
 468         default:
 469             _assert(false);
 470     }
 471 }
 472
 473 void CYDriver::PopCondition() {
 474     yy_pop_state(scanner_);
 475 }