Cycript.l

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