re2-js-generator/example/lexer.l

var types = [
    "SLASH",
    "LSB",
    "RSB",
    "LCB",
    "RCB",
    "COLON",
    "COMMA",
    "DOT",
    "REM",
    "GT",
    "GTE",
    "LT",
    "LTE",
    "EQ",
    "NEQ",
    "LIKE",
    "NLIKE",
    "AND",
    "OR",
    "NOT",
    "ADDRESS",
    "OID",
    "OID_LITERAL",
    "TIME",
    "TIMEDIFF",
    "INTEGER_LITERAL",
    "FLOAT_LITERAL",
    "BOOL_LITERAL",
    "ID"
];

var errors = {
    "-2": "end of search",
    "-1": "not found any lexemes or errors or anything else",
    "0": "success",
    "1": "found unknown symbol",
    "2": "not found close quote or singleQuote",
    "3": "Unexpected symbol in oid structure",
    "4": "Not found close bracket for Oid",
};

var Lexer = function(_string){
    this._last_found_lexeme = {error: -1};
    this._end = false;
    this._error = false;
    this._string = _string;
    this._state = 1;
    this._yy_char = null;
    this._yy_lex_start = 0;
    this._yy_cursor = 0;
    this._yy_marker = 0;
    this._yy_accept = 0;
};

Lexer.prototype = {
    types: types,
    errors: errors,
    _notFoundCloseQuote: function() {
         this._error = true;
         this._last_found_lexeme = {
             error: 2,
             start: this._yy_lex_start,
             end: this._yy_cursor
         };

         console.log( print_f("Not found close quote start: %s", this._yy_cursor));
    },
    _unknownSymbol: function(){
        this._error = true;
        this._last_found_lexeme = {
            error: 1,
            start: this._yy_lex_start,
            end: this._yy_cursor
        };

        console.log( print_f("Found unknown symbol on position: %s", this._yy_cursor));
    },
    _oidUnexpectedSymbol: function(){
        this._error = true;
        this._last_found_lexeme = {
            error: 3,
            start: this._yy_lex_start,
            end: this._yy_cursor
        };

        console.log( print_f("Found unknown symbol in Oid on position: %s", this._yy_cursor));
    },
    _oidNotFoundCloseBracket: function(){
         this._error = true;
         this._last_found_lexeme = {
             error: 4,
             start: this._yy_lex_start,
             end: this._yy_cursor
         };

         console.log( print_f("Not found close bracket for Oid"));
    },
    _foundLexeme: function(_lexeme) {
        console.log(print_f("found lex: %s; start: %s; end: %s; result => %s", _lexeme, this._yy_lex_start, this._yy_cursor, this._string.substring(this._yy_lex_start, this._yy_cursor)));
        this._last_found_lexeme = {
            error: 0,
             lexeme: _lexeme,
             value: this._string.substring(this._yy_lex_start, this._yy_cursor),
             start: this._yy_lex_start,
             end: this._yy_cursor
        };
      },
    _foundOidLexeme: function(_lexeme, _lsb, _rsb) {
        console.log(print_f("found lex: %s; start: %s; end: %s; result => %s", _lexeme, this._yy_lex_start, this._yy_cursor, this._string.substring(this._yy_lex_start, this._yy_cursor)));
        this._last_found_lexeme = {
            error: 0,
             lexeme: _lexeme,
             value: this._string.substring(this._yy_lex_start, this._yy_cursor),
             start: this._yy_lex_start,
             end: this._yy_cursor,
             lsb: _lsb,
             rsb: _rsb
        };
      },
      _endOfString: function(){
        console.log(print_f("search end\n"));
        this._end = true;
        this._last_found_lexeme = {
            error: -2
        };
      },
      _searchString: function () {
          var _quote = this._string[this._yy_cursor - 1];
          var found_back_slash = false;
          while(this._yy_cursor < this._string.length){
             this._yy_char = this._string[this._yy_cursor];
             if(_quote == '"') {
                 switch (this._yy_char) {
                     case "\\":
                         found_back_slash = true;
                         break;
                     case '"':
                         if(!found_back_slash) {
                             this._yy_cursor++;
                             this._foundLexeme("STRING_LITERAL");
                             return;
                         }
                         found_back_slash = false;
                         break;
                 }
             } else if(_quote == "'") {
                 switch (this._yy_char) {
                     case "\\":
                         found_back_slash = true;
                         break;
                     case "'":
                         if(!found_back_slash) {
                             this._yy_cursor++;
                             this._foundLexeme("STRING_LITERAL");
                             return;
                         }
                         found_back_slash = false;
                         break;
                 }
             }
             this._yy_cursor++;
         }

         this._notFoundCloseQuote();
     },
     _searchOid: function (){
         var lsb, rsb;
         var state = 0;
         while(this._yy_cursor < this._string.length){
             switch(state){
                 case 0:
                    this._yy_char = this._string[this._yy_cursor];
                    (function(){
                        switch(this._yy_char){
                            case " ":
                                state = 0;
                                this._yy_cursor++;
                                break;
                            case "[":
                                lsb = {start: this._yy_cursor, end: this._yy_cursor + 1};
                                state = 1;
                                break;
                            default:
                                state = 5;
                        }
                    }.bind(this))();
                    break;
                 case 1:
                     this._yy_char = this._string[++this._yy_cursor];
                     (function(){
                         switch(this._yy_char){
                             case " ":
                                 state = 1;
                                 break;
                             case "0":
                             case "1":
                             case "2":
                             case "3":
                             case "4":
                             case "5":
                             case "6":
                             case "7":
                             case "8":
                             case "9":
                                state = 2;
                                break;
                             case "]":
                                state = 4;
                                break;
                             default:
                                 state = 5;
                         }
                     }.bind(this))();
                     break;
                 case 2:
                     this._yy_char = this._string[++this._yy_cursor];
                     (function(){
                         switch(this._yy_char){
                             case " ":
                             case "0":
                             case "1":
                             case "2":
                             case "3":
                             case "4":
                             case "5":
                             case "6":
                             case "7":
                             case "8":
                             case "9":
                                state = 2;
                                break;
                             case ".":
                                state = 3;
                                break;
                             case "]":
                                state = 4;
                                break;
                             default:
                                state = 5;
                         }
                     }.bind(this))();
                     break;
                 case 3:
                     this._yy_char = this._string[++this._yy_cursor];
                     (function(){
                         switch(this._yy_char){
                             case " ":
                             case "0":
                             case "1":
                             case "2":
                             case "3":
                             case "4":
                             case "5":
                             case "6":
                             case "7":
                             case "8":
                             case "9":
                                state = 2;
                                break;
                             default:
                                state = 5;
                         }
                     }.bind(this))();
                     break;
                 case 4:
                    rsb = {start: this._yy_cursor, end: this._yy_cursor + 1};
                    ++this._yy_cursor;
                    this._foundOidLexeme("OID_LITERAL", lsb, rsb);
                    return;
                 case 5:
                    this._oidUnexpectedSymbol();
                    return;
             }
         }

         this._oidNotFoundCloseBracket();
     },
    _set_next: function(){
        this._yy_accept = 0;
        this._state = 1;
        this._yy_lex_start = this._yy_cursor;
        this._yy_marker = this._yy_cursor;
    },
    next: function(){
        if(this._end || this._error) return null;
        
        this.search();
        return this.token();
    },
    token: function(){
        return this._last_found_lexeme;
    },
    search: function(){
        if(this._end) return false;

        while(true){
            switch(id) {

        START/*!re2c
            re2c:define:YYCTYPE = _r2c_var_;
            re2c:define:YYCURSOR = this._yy_cursor;
            re2c:define:YYMARKER = this._yy_marker;
            re2c:yyfill:enable = 0;

            D                           = [0-9];
            DSEQ                        = D+;

            end                         = "\x00";
            L                           = [A-Za-z_];

            CR                          = "\r";
            LF                          = "\n";
            CRLF                        = CR?LF;
            INTEGER                     = "-"?D+;
            SP                          = " ";
            TAB                         = "\t";

            DELIM                       = SP|TAB|CR|LF;

            LSB                         = "[";
            SLASH                       = "/";
            RSB                         = "]";
            LCB                         = "(";
            RCB                         = ")";
            COLON                       = ":";
            COMMA                       = ",";
            DOT                         = ".";
            REM                         = "%";
            GT                          = ">";
            GTE                         = ">=";
            LT                          = "<";
            LTE                         = "<=";
            EQ                          = "==";
            NEQ                         = "!=";

            AND                         = 'AND';
            OR                          = 'OR';
            NOT                         = 'NOT';
            LIKE                        = 'LIKE';
            NLIKE                       = 'NLIKE';

            ADDRESS                     = "Address";
            TIME                        = "Time";
            OID                         = "Oid";
            TIMEDIFF                    = "TimeDiff";

            ID                          = L(L|D)*;
            BOOL_LITERAL                = 'true'|'false';
            INTEGER_LITERAL             = INTEGER;
            FLOAT_LITERAL               = "-"? D* "." D+ ("e" "-"? D+)?;

            QU = "\"";
            SQU = "'";




            end                         { this._endOfString(); return; }

            SLASH                       { this._foundLexeme("SLASH"); this._set_next(); return; }
            LSB                         { this._foundLexeme("LSB"); this._set_next(); return; }
            RSB                         { this._foundLexeme("RSB"); this._set_next(); return; }
            LCB                         { this._foundLexeme("LCB"); this._set_next(); return; }
            RCB                         { this._foundLexeme("RCB"); this._set_next(); return; }
            COLON                       { this._foundLexeme("COLON"); this._set_next(); return; }
            COMMA                       { this._foundLexeme("COMMA"); this._set_next(); return; }
            DOT                         { this._foundLexeme("DOT"); this._set_next(); return; }
            REM                         { this._foundLexeme("REM"); this._set_next(); return; }

            GT                          { this._foundLexeme("GT"); this._set_next(); return; }
            GTE                         { this._foundLexeme("GTE"); this._set_next(); return; }
            LT                          { this._foundLexeme("LT"); this._set_next(); return; }
            LTE                         { this._foundLexeme("LTE"); this._set_next(); return; }
            EQ                          { this._foundLexeme("EQ"); this._set_next(); return; }
            NEQ                         { this._foundLexeme("NEQ"); this._set_next(); return; }
            LIKE                        { this._foundLexeme("LIKE"); this._set_next(); return; }
            NLIKE                       { this._foundLexeme("NLIKE"); this._set_next(); return; }

            AND                         { this._foundLexeme("AND"); this._set_next(); return; }
            OR                          { this._foundLexeme("OR"); this._set_next(); return; }
            NOT                         { this._foundLexeme("NOT"); this._set_next(); return; }

            ADDRESS                     { this._foundLexeme("ADDRESS"); this._set_next(); return; }
            TIME                        { this._foundLexeme("TIME"); this._set_next(); return; }
            TIMEDIFF                    { this._foundLexeme("TIMEDIFF"); this._set_next(); return; }
            OID                         { id = 100000001; break; }

            INTEGER_LITERAL             { this._foundLexeme("INTEGER_LITERAL"); this._set_next(); return; }
            FLOAT_LITERAL               { this._foundLexeme("FLOAT_LITERAL"); this._set_next(); return; }
            BOOL_LITERAL                { this._foundLexeme("BOOL_LITERAL"); this._set_next(); return; }

            ID                          { this._foundLexeme("ID"); this._set_next(); return; }
            DELIM                       { this._set_next(); break;  }

            QU|SQU                      { id = 100000000; break;}

            [^]                         { this._unknownSymbol(); this._set_next(); return;  }
        */END &&<STRING&&
              &&<OID&&
            }
        }
    }
};

&&>STRING
case 100000000:
    this._searchString();
    this._set_next();
    return;
&&

&&>OID
case 100000001:
    this._searchOid();
    this._set_next();
    return;
&&

var print_f = function() {
    var r_str = "";
    var next = arguments[0];

    var rx = /(%[a-zA-Z]{1})/;
    var a = 1,
        match;
    while (match = rx.exec(next)) {
        var prev = next.substring(0, match.index);
        var macro = next.substring(match.index + 1, match.index + 2);
        next = next.substring(match.index + 2, next.length);
        r_str += prev;

        var arg = arguments[a];

        if (arg !== undefined) {
            switch (macro) {
                case "s":
                    r_str += arg.toString();
                    break;
                case "i":
                    r_str += parseInt(arg);
                    break;
                case "f":
                    r_str += parseFloat(arg);
                    break;
            }
        } else {
            r_str += "%" + macro;
        }
        a++;
    }

    r_str += next;

    return r_str;
};