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