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