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