Cycript.l.in

   1 /* Cycript - Optimizing JavaScript Compiler/Runtime
   2  * Copyright (C) 2009-2010  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 #define YY_EXTRA_TYPE CYDriver *
  30
  31 #define F(value) do { \
  32     int token(value); \
  33     return token; \
  34 } while (false)
  35
  36 #define A new($pool)
  37 #define Y apr_pstrmemdup($pool, yytext, yyleng)
  38
  39 #define I(type, Type, value) do { \
  40     yylval->type ## _ = A CY ## Type; \
  41     F(value); \
  42 } while (false)
  43
  44 #define T yylval->newline_ = yyextra->state_ == CYNewLine; BEGIN(Div);
  45 #define C T yyextra->state_ = CYClear;
  46 #define R T yyextra->state_ = CYRestricted;
  47
  48 #define E L C I(literal, RegEx(Y), tk::RegularExpressionLiteral);
  49
  50 #define N \
  51     if (yyextra->state_ != CYNewLine) { \
  52         if (yyextra->state_ != CYRestricted) \
  53             yyextra->state_ = CYNewLine; \
  54         else { \
  55             yyextra->state_ = CYClear; \
  56             F(tk::NewLine); \
  57         } \
  58     }
  59
  60 #define V(more) { \
  61     if (const char *nl = reinterpret_cast<const char *>(memchr(yytext, '\n', yyleng))) { \
  62         unsigned lines(0); \
  63         size_t left; \
  64         do { \
  65             ++lines; \
  66             left = yyleng - (nl - yytext) - 1; \
  67             nl = reinterpret_cast<const char *>(memchr(nl + 1, '\n', left)); \
  68         } while (nl != NULL); \
  69         yylloc->end.lines(lines); \
  70         yylloc->end.columns(left); \
  71         yylloc->step(); \
  72         more \
  73     } else L \
  74 }
  75
  76 #define L { \
  77     yylloc->step(); \
  78     yylloc->columns(yyleng); \
  79 }
  80
  81 int H(char c) {
  82     if (c >= '0' && c <= '9')
  83         return c - '0';
  84     if (c >= 'a' && c <= 'f')
  85         return c - 'a' + 10;
  86     if (c >= 'A' && c <= 'F')
  87         return c - 'A' + 10;
  88     return -1;
  89 }
  90
  91 #define YY_INPUT(data, value, size) { \
  92     if (yyextra->file_ != NULL) { \
  93         size_t copy(fread(data, 1, size, yyextra->file_)); \
  94         value = copy == 0 ? YY_NULL : copy; \
  95     } else if (yyextra->size_ == 0) \
  96         value = YY_NULL; \
  97     else { \
  98         size_t copy(size); \
  99         copy = (std::min(copy, yyextra->size_)); \
 100         memcpy(data, yyextra->data_, copy); \
 101         yyextra->data_ += copy; \
 102         yyextra->size_ -= copy; \
 103         value = copy; \
 104     } \
 105 }
 106
 107 %}
 108
 109 %option prefix="cy"
 110 %option bison-bridge
 111 %option bison-locations
 112 %option noyywrap
 113 %option yylineno
 114 %option nounput
 115 %option interactive
 116 %option reentrant
 117 %option stack
 118
 119 Exponent [eE][+-]?[0-9]+
 120 Escape   \\[\\'"bfnrtv]|\\0|\\x[0-9a-fA-F]{2}|\\u[0-9a-fA-F]{4}|\\\n
 121
 122 IdentifierStart [a-zA-Z$_]
 123 IdentifierPart  [a-zA-Z$_0-9]
 124
 125 NonTerminator [^\n]
 126 BackslashSequence \\{NonTerminator}
 127 RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence}
 128 RegularExpressionChar [^\n\\/]|{BackslashSequence}
 129 RegularExpressionFlags {IdentifierPart}*
 130 RegularExpressionChars {RegularExpressionChar}*
 131 RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}
 132
 133 @begin E4X
 134 XMLNameStart [a-zA-Z_:]
 135 XMLNamePart [a-zA-Z0-9.-_:]
 136 XMLName {XMLNameStart}{XMLNamePart}*
 137 @end
 138
 139 %s Div
 140 %s RegExp
 141
 142 @begin E4X
 143 %x XMLContent
 144 %x XMLTag
 145 @end
 146
 147 %%
 148
 149 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} E
 150
 151 \/\/[^\n]* L
 152
 153         /* http://ostermiller.org/findcomment.html */
 154         /* XXX: unify these two rules using !? */
 155 \/\*!([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V() C I(comment, Comment(Y), tk::Comment);
 156 \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V(N)
 157
 158 @begin E4X
 159 <RegExp>"<>"      L F(tk::LeftRight);
 160 <XMLContent>"</>" L F(tk::LeftSlashRight);
 161
 162 <RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA);
 163 <RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment);
 164 <RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI);
 165
 166 <XMLTag>"="  L F(tk::Equal);
 167 <XMLTag>">"  L F(tk::Right);
 168 <XMLTag>"/>" L F(tk::SlashRight);
 169 <XMLTag>"{"  L F(tk::OpenBrace);
 170
 171 <XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue);
 172 <XMLTag>{XMLName} L F(tk::XMLName);
 173 <XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace);
 174
 175 <XMLContent>"{"  L F(tk::OpenBrace);
 176 <XMLContent>"<"  L F(tk::Left);
 177 <XMLContent>"</" L F(tk::LeftSlash);
 178 @end
 179
 180 "..."  L C F(tk::PeriodPeriodPeriod);
 181
 182 @begin E4X
 183 "::"   L C F(tk::ColonColon);
 184 ".."   L C F(tk::PeriodPeriod);
 185 @end
 186
 187 @begin E4X ObjectiveC
 188 "@"    L C F(tk::At);
 189 @end
 190
 191 "&"    L C F(tk::Ampersand);
 192 "&&"   L C F(tk::AmpersandAmpersand);
 193 "&="   L C F(tk::AmpersandEqual);
 194 "^"    L C F(tk::Carrot);
 195 "^="   L C F(tk::CarrotEqual);
 196 "="    L C F(tk::Equal);
 197 "=="   L C F(tk::EqualEqual);
 198 "==="  L C F(tk::EqualEqualEqual);
 199 "!"    L C F(tk::Exclamation);
 200 "!="   L C F(tk::ExclamationEqual);
 201 "!=="  L C F(tk::ExclamationEqualEqual);
 202 "-"    L C F(tk::Hyphen);
 203 "-="   L C F(tk::HyphenEqual);
 204 "--"   L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen);
 205 "->"   L C F(tk::HyphenRight);
 206 "<"    L C F(tk::Left);
 207 "<="   L C F(tk::LeftEqual);
 208 "<<"   L C F(tk::LeftLeft);
 209 "<<="  L C F(tk::LeftLeftEqual);
 210 "%"    L C F(tk::Percent);
 211 "%="   L C F(tk::PercentEqual);
 212 "."    L C F(tk::Period);
 213 "|"    L C F(tk::Pipe);
 214 "|="   L C F(tk::PipeEqual);
 215 "||"   L C F(tk::PipePipe);
 216 "+"    L C F(tk::Plus);
 217 "+="   L C F(tk::PlusEqual);
 218 "++"   L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus);
 219 ">"    L C F(tk::Right);
 220 ">="   L C F(tk::RightEqual);
 221 ">>"   L C F(tk::RightRight);
 222 ">>="  L C F(tk::RightRightEqual);
 223 ">>>"  L C F(tk::RightRightRight);
 224 ">>>=" L C F(tk::RightRightRightEqual);
 225 "*"    L C F(tk::Star);
 226 "*="   L C F(tk::StarEqual);
 227 "~"    L C F(tk::Tilde);
 228
 229 <Div>"/"  L C F(tk::Slash);
 230 <Div>"/=" L C F(tk::SlashEqual);
 231
 232 ":"    L C F(tk::Colon);
 233 ","    L C F(tk::Comma);
 234 "?"    L C F(tk::Question);
 235 ";"    L C F(tk::SemiColon);
 236
 237 "("    L C F(tk::OpenParen);
 238 ")"    L C F(tk::CloseParen);
 239
 240 "{"    L C F(yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace);
 241 "}"    L C F(tk::CloseBrace);
 242
 243 "["    L C F(tk::OpenBracket);
 244 "]"    L C F(tk::CloseBracket);
 245
 246 @begin Java
 247 "@class"          L C F(tk::AtClass);
 248 @end
 249
 250 @begin ObjectiveC
 251 "@end"            L C F(tk::AtEnd);
 252 "@implementation" L C F(tk::AtImplementation);
 253 "@import"         L C F(tk::AtImport);
 254 "@selector"       L C F(tk::AtSelector);
 255 @end
 256
 257 "false"        L C I(false, False(), tk::False);
 258 "null"         L C I(null, Null(), tk::Null);
 259 "true"         L C I(true, True(), tk::True);
 260
 261 "break"        L R I(word, Word("break"), tk::Break);
 262 "case"         L C I(word, Word("case"), tk::Case);
 263 "catch"        L C I(word, Word("catch"), tk::Catch);
 264 "continue"     L R I(word, Word("continue"), tk::Continue);
 265 "default"      L C I(word, Word("default"), tk::Default);
 266 "delete"       L C I(word, Word("delete"), tk::Delete);
 267 "do"           L C I(word, Word("do"), tk::Do);
 268 "else"         L C I(word, Word("else"), tk::Else);
 269 "finally"      L C I(word, Word("finally"), tk::Finally);
 270 "for"          L C I(word, Word("for"), tk::For);
 271 "function"     L C I(word, Word("function"), tk::Function);
 272 "if"           L C I(word, Word("if"), tk::If);
 273 "in"           L C I(word, Word("in"), yyextra->in_.top() ? tk::In_ : tk::In);
 274 "instanceof"   L C I(word, Word("instanceof"), tk::InstanceOf);
 275 "new"          L C I(word, Word("new"), tk::New);
 276 "return"       L R I(word, Word("return"), tk::Return);
 277 "switch"       L C I(word, Word("switch"), tk::Switch);
 278 "this"         L C I(this, This(), tk::This);
 279 "throw"        L R I(word, Word("throw"), tk::Throw);
 280 "try"          L C I(word, Word("try"), tk::Try);
 281 "typeof"       L C I(word, Word("typeof"), tk::TypeOf);
 282 "var"          L C I(word, Word("var"), tk::Var);
 283 "void"         L C I(word, Word("void"), tk::Void);
 284 "while"        L C I(word, Word("while"), tk::While);
 285 "with"         L C I(word, Word("with"), tk::With);
 286
 287 "debugger"     L C I(word, Word("debugger"), tk::Debugger);
 288
 289 "const"        L C I(word, Word("const"), tk::Const);
 290
 291 "class"        L C I(word, Word("class"), tk::Class);
 292 "enum"         L C I(word, Word("enum"), tk::Enum);
 293 "export"       L C I(word, Word("export"), tk::Export);
 294 "extends"      L C I(word, Word("extends"), tk::Extends);
 295 "import"       L C I(word, Word("import"), tk::Import);
 296 "super"        L C I(word, Word("super"), tk::Super);
 297
 298 "implements"   L C I(identifier, Identifier("implements"), tk::Implements);
 299 "interface"    L C I(identifier, Identifier("interface"), tk::Interface);
 300 "package"      L C I(identifier, Identifier("package"), tk::Package);
 301 "private"      L C I(identifier, Identifier("private"), tk::Private);
 302 "protected"    L C I(identifier, Identifier("protected"), tk::Protected);
 303 "public"       L C I(identifier, Identifier("public"), tk::Public);
 304 "static"       L C I(identifier, Identifier("static"), tk::Static);
 305
 306 "abstract"     L C I(identifier, Identifier("abstract"), tk::Abstract);
 307 "boolean"      L C I(identifier, Identifier("boolean"), tk::Boolean);
 308 "byte"         L C I(identifier, Identifier("byte"), tk::Byte);
 309 "char"         L C I(identifier, Identifier("char"), tk::Char);
 310 "double"       L C I(identifier, Identifier("double"), tk::Double);
 311 "final"        L C I(identifier, Identifier("final"), tk::Final);
 312 "float"        L C I(identifier, Identifier("float"), tk::Float);
 313 "goto"         L C I(identifier, Identifier("goto"), tk::Goto);
 314 "int"          L C I(identifier, Identifier("int"), tk::Int);
 315 "long"         L C I(identifier, Identifier("long"), tk::Long);
 316 "native"       L C I(identifier, Identifier("native"), tk::Native);
 317 "short"        L C I(identifier, Identifier("short"), tk::Short);
 318 "synchronized" L C I(identifier, Identifier("synchronized"), tk::Synchronized);
 319 "throws"       L C I(identifier, Identifier("throws"), tk::Throws);
 320 "transient"    L C I(identifier, Identifier("transient"), tk::Transient);
 321 "volatile"     L C I(identifier, Identifier("volatile"), tk::Volatile);
 322
 323 "let"          L C I(identifier, Identifier("let"), tk::Let);
 324 "yield"        L C I(identifier, Identifier("yield"), tk::Yield);
 325
 326 "each"         L C I(identifier, Identifier("each"), tk::Each);
 327
 328 @begin E4X
 329 "namespace"    L C I(identifier, Identifier("namespace"), tk::Namespace);
 330 "xml"          L C I(identifier, Identifier("xml"), tk::XML);
 331 @end
 332
 333 {IdentifierStart}{IdentifierPart}* L C I(identifier, Identifier(Y), tk::Identifier_);
 334
 335 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?){Exponent}? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral);
 336
 337 0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral);
 338 0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral);
 339 0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral);
 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);
 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); } 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 }