Cycript.l

   1 %{
   2 #define YYLTYPE cy::location
   3 #include "Cycript.tab.hh"
   4 typedef cy::parser::token tk;
   5
   6 #define YY_EXTRA_TYPE CYDriver *
   7
   8 #define T yylval->newline_ = yyextra->state_ == CYNewLine;
   9 #define C T yyextra->state_ = CYClear;
  10 #define R T yyextra->state_ = CYRestricted;
  11
  12 #define N \
  13     if (yyextra->state_ != CYNewLine) { \
  14         bool restricted(yyextra->state_ == CYRestricted); \
  15         if (restricted) { \
  16             yyextra->state_ = CYClear; \
  17             return tk::NewLine; \
  18         } else \
  19             yyextra->state_ = CYNewLine; \
  20     }
  21
  22 #define L { \
  23     yylloc->step(); \
  24     yylloc->columns(yyleng); \
  25 }
  26
  27 int H(char c) {
  28     if (c >= '0' && c <= '9')
  29         return c - '0';
  30     if (c >= 'a' && c <= 'f')
  31         return c - 'a' + 10;
  32     if (c >= 'A' && c <= 'F')
  33         return c - 'A' + 10;
  34     return -1;
  35 }
  36
  37 #define YY_INPUT(data, value, size) { \
  38     if (yyextra->file_ != NULL) { \
  39         size_t copy(fread(data, 1, size, yyextra->file_)); \
  40         value = copy == 0 ? YY_NULL : copy; \
  41     } else if (yyextra->size_ == 0) \
  42         value = YY_NULL; \
  43     else { \
  44         size_t copy(std::min(size, yyextra->size_)); \
  45         memcpy(data, yyextra->data_, copy); \
  46         yyextra->data_ += copy; \
  47         yyextra->size_ -= copy; \
  48         value = copy; \
  49     } \
  50 }
  51
  52 %}
  53
  54 %option prefix="cy"
  55 %option bison-bridge
  56 %option bison-locations
  57 %option noyywrap
  58 %option yylineno
  59 %option nounput
  60 %option interactive
  61 %option reentrant
  62
  63 Exponent [eE][+-]?[0-9]+
  64 Escape   \\[\\'"bfnrtv]|\\0|\\x[0-9a-fA-F]{2}|\\u[0-9a-fA-F]{4}|\\\n
  65
  66 %%
  67
  68 \/\/[^\n]* ;
  69 \/\*(\n|[^\*]|\*[^/])\*\/ if (memchr(yytext, '\n', yyleng) != NULL) N // XXX: supposedly I will be screwed on very very long multi-line comments and need to replace this with a manual lexer. http://websrv.cs.fsu.edu/~engelen/courses/COP5621/Pr2.pdf ; XXX: this rule doesn't work anyway, fucking A :(
  70
  71 "&"    L C return tk::Ampersand;
  72 "&&"   L C return tk::AmpersandAmpersand;
  73 "&="   L C return tk::AmpersandEqual;
  74 "^"    L C return tk::Carrot;
  75 "^="   L C return tk::CarrotEqual;
  76 "="    L C return tk::Equal;
  77 "=="   L C return tk::EqualEqual;
  78 "==="  L C return tk::EqualEqualEqual;
  79 "!"    L C return tk::Exclamation;
  80 "!="   L C return tk::ExclamationEqual;
  81 "!=="  L C return tk::ExclamationEqualEqual;
  82 "-"    L C return tk::Hyphen;
  83 "-="   L C return tk::HyphenEqual;
  84 "--"   L C return yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen;
  85 "->"   L C return tk::HyphenRight;
  86 "<"    L C return tk::Left;
  87 "<="   L C return tk::LeftEqual;
  88 "<<"   L C return tk::LeftLeft;
  89 "<<="  L C return tk::LeftLeftEqual;
  90 "%"    L C return tk::Percent;
  91 "%="   L C return tk::PercentEqual;
  92 "."    L C return tk::Period;
  93 "|"    L C return tk::Pipe;
  94 "|="   L C return tk::PipeEqual;
  95 "||"   L C return tk::PipePipe;
  96 "+"    L C return tk::Plus;
  97 "+="   L C return tk::PlusEqual;
  98 "++"   L C return yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus;
  99 ">"    L C return tk::Right;
 100 ">="   L C return tk::RightEqual;
 101 ">>"   L C return tk::RightRight;
 102 ">>="  L C return tk::RightRightEqual;
 103 ">>>"  L C return tk::RightRightRight;
 104 ">>>=" L C return tk::RightRightRightEqual;
 105 "/"    L C return tk::Slash;
 106 "/="   L C return tk::SlashEqual;
 107 "*"    L C return tk::Star;
 108 "*="   L C return tk::StarEqual;
 109 "~"    L C return tk::Tilde;
 110
 111 ":"    L C return tk::Colon;
 112 ","    L C return tk::Comma;
 113 "?"    L C return tk::Question;
 114 ";"    L C return tk::SemiColon;
 115
 116 "("    L C return tk::OpenParen;
 117 ")"    L C return tk::CloseParen;
 118
 119 "{"    L C return tk::OpenBrace;
 120 "}"    L C return tk::CloseBrace;
 121
 122 "["    L C return tk::OpenBracket;
 123 "]"    L C return tk::CloseBracket;
 124
 125 "@class"       L C return tk::AtClass;
 126 "@end"         L C return tk::AtEnd;
 127 "@selector"    L C return tk::AtSelector;
 128
 129 "false"        L C yylval->false_ = new CYFalse(); return tk::False;
 130 "null"         L C yylval->null_ = new CYNull(); return tk::Null;
 131 "true"         L C yylval->true_ = new CYTrue(); return tk::True;
 132
 133 "break"        L R yylval->word_ = new CYWord("break"); return tk::Break;
 134 "case"         L C yylval->word_ = new CYWord("case"); return tk::Case;
 135 "catch"        L C yylval->word_ = new CYWord("catch"); return tk::Catch;
 136 "continue"     L R yylval->word_ = new CYWord("continue"); return tk::Continue;
 137 "default"      L C yylval->word_ = new CYWord("default"); return tk::Default;
 138 "delete"       L C yylval->word_ = new CYWord("delete"); return tk::Delete;
 139 "do"           L C yylval->word_ = new CYWord("do"); return tk::Do;
 140 "else"         L C yylval->word_ = new CYWord("else"); return tk::Else;
 141 "finally"      L C yylval->word_ = new CYWord("finally"); return tk::Finally;
 142 "for"          L C yylval->word_ = new CYWord("for"); return tk::For;
 143 "function"     L C yylval->word_ = new CYWord("function"); return tk::Function;
 144 "if"           L C yylval->word_ = new CYWord("if"); return tk::If;
 145 "in"           L C yylval->word_ = new CYWord("in"); return tk::In;
 146 "instanceof"   L C yylval->word_ = new CYWord("instanceof"); return tk::InstanceOf;
 147 "new"          L C yylval->word_ = new CYWord("new"); return tk::New;
 148 "return"       L R yylval->word_ = new CYWord("return"); return tk::Return;
 149 "switch"       L C yylval->word_ = new CYWord("switch"); return tk::Switch;
 150 "this"         L C yylval->this_ = new CYThis(); return tk::This;
 151 "throw"        L R yylval->word_ = new CYWord("throw"); return tk::Throw;
 152 "try"          L C yylval->word_ = new CYWord("try"); return tk::Try;
 153 "typeof"       L C yylval->word_ = new CYWord("typeof"); return tk::TypeOf;
 154 "var"          L C yylval->word_ = new CYWord("var"); return tk::Var;
 155 "void"         L C yylval->word_ = new CYWord("void"); return tk::Void;
 156 "while"        L C yylval->word_ = new CYWord("while"); return tk::While;
 157 "with"         L C yylval->word_ = new CYWord("with"); return tk::With;
 158
 159 "debugger"     L C yylval->word_ = new CYWord("debugger"); return tk::Debugger;
 160
 161 "const"        L C yylval->word_ = new CYWord("const"); return tk::Const;
 162
 163 "class"        L C yylval->word_ = new CYWord("class"); return tk::Class;
 164 "enum"         L C yylval->word_ = new CYWord("enum"); return tk::Enum;
 165 "export"       L C yylval->word_ = new CYWord("export"); return tk::Export;
 166 "extends"      L C yylval->word_ = new CYWord("extends"); return tk::Extends;
 167 "import"       L C yylval->word_ = new CYWord("import"); return tk::Import;
 168 "super"        L C yylval->word_ = new CYWord("super"); return tk::Super;
 169
 170 "implements"   L C yylval->identifier_ = new CYIdentifier("implements"); return tk::Implements;
 171 "interface"    L C yylval->identifier_ = new CYIdentifier("interface"); return tk::Interface;
 172 "package"      L C yylval->identifier_ = new CYIdentifier("package"); return tk::Package;
 173 "private"      L C yylval->identifier_ = new CYIdentifier("private"); return tk::Private;
 174 "protected"    L C yylval->identifier_ = new CYIdentifier("protected"); return tk::Protected;
 175 "public"       L C yylval->identifier_ = new CYIdentifier("public"); return tk::Public;
 176 "static"       L C yylval->identifier_ = new CYIdentifier("static"); return tk::Static;
 177
 178 "abstract"     L C yylval->identifier_ = new CYIdentifier("abstract"); return tk::Abstract;
 179 "boolean"      L C yylval->identifier_ = new CYIdentifier("boolean"); return tk::Boolean;
 180 "byte"         L C yylval->identifier_ = new CYIdentifier("byte"); return tk::Byte;
 181 "char"         L C yylval->identifier_ = new CYIdentifier("char"); return tk::Char;
 182 "double"       L C yylval->identifier_ = new CYIdentifier("double"); return tk::Double;
 183 "final"        L C yylval->identifier_ = new CYIdentifier("final"); return tk::Final;
 184 "float"        L C yylval->identifier_ = new CYIdentifier("float"); return tk::Float;
 185 "goto"         L C yylval->identifier_ = new CYIdentifier("goto"); return tk::Goto;
 186 "int"          L C yylval->identifier_ = new CYIdentifier("int"); return tk::Int;
 187 "long"         L C yylval->identifier_ = new CYIdentifier("long"); return tk::Long;
 188 "native"       L C yylval->identifier_ = new CYIdentifier("native"); return tk::Native;
 189 "short"        L C yylval->identifier_ = new CYIdentifier("short"); return tk::Short;
 190 "synchronized" L C yylval->identifier_ = new CYIdentifier("synchronized"); return tk::Synchronized;
 191 "throws"       L C yylval->identifier_ = new CYIdentifier("throws"); return tk::Throws;
 192 "transient"    L C yylval->identifier_ = new CYIdentifier("transient"); return tk::Transient;
 193 "volatile"     L C yylval->identifier_ = new CYIdentifier("volatile"); return tk::Volatile;
 194
 195 "let"          L C yylval->identifier_ = new CYIdentifier("let"); return tk::Let;
 196 "yield"        L C yylval->identifier_ = new CYIdentifier("yield"); return tk::Yield;
 197
 198 "each"         L C yylval->identifier_ = new CYIdentifier("each"); return tk::Each;
 199
 200 [a-zA-Z$_][a-zA-Z$_0-9]* yylval->identifier_ = new CYIdentifier(apr_pstrmemdup(yyextra->pool_, yytext, yyleng)); L C return tk::Identifier_;
 201
 202 (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?){Exponent}? yylval->number_ = new CYNumber(strtod(yytext, NULL)); L C return tk::NumericLiteral;
 203
 204 0[xX][0-9a-fA-F]+ L C yylval->number_ = new CYNumber(strtoull(yytext + 2, NULL, 16)); return tk::NumericLiteral;
 205
 206 0[bB][0-1]+ L C yylval->number_ = new CYNumber(strtoull(yytext + 2, NULL, 2)); return tk::NumericLiteral;
 207
 208 \"([^"\\\n]|{Escape})*\"|'([^'\\\n]|{Escape})*' L C {
 209     char *value(reinterpret_cast<char *>(apr_palloc(yyextra->pool_, yyleng)));
 210     char *local(value);
 211
 212     for (int i(1); i != yyleng - 1; ++i) {
 213         char next(yytext[i]);
 214
 215         if (yytext[i] == '\\')
 216             switch (next = yytext[++i]) {
 217                 case '\n': continue;
 218                 case '\\': next = '\\'; break;
 219                 case '\'': next = '\''; break;
 220                 case '"': next = '"'; break;
 221                 case 'b': next = '\b'; break;
 222                 case 'f': next = '\f'; break;
 223                 case 'n': next = '\n'; break;
 224                 case 'r': next = '\r'; break;
 225                 case 't': next = '\t'; break;
 226                 case 'v': next = '\v'; break;
 227                 case '0': next = '\0'; break;
 228
 229                 case 'x':
 230                     next = H(yytext[i + 1]) << 4 | H(yytext[i + 2]);
 231                     i += 2;
 232                 break;
 233             }
 234
 235         *local++ = next;
 236     }
 237
 238     *local = '\0';
 239     yylval->string_ = new CYString(value, local - value);
 240     return tk::StringLiteral;
 241 }
 242
 243 \n yylloc->end.lines(); yylloc->step(); N
 244
 245 [ \t] L
 246 <<EOF>> L yyterminate();
 247
 248 . L {
 249     CYDriver::Error error;
 250     error.location_ = *yylloc;
 251     error.message_ = "syntax error, unknown token";
 252     yyextra->errors_.push_back(error);
 253     yyterminate();
 254 }
 255
 256 %%
 257
 258 void CYDriver::ScannerInit() {
 259     cylex_init(&scanner_);
 260     cyset_extra(this, scanner_);
 261 }
 262
 263 void CYDriver::ScannerDestroy() {
 264     cylex_destroy(scanner_);
 265 }