re2-js-generator/main.js

/**
 * Created by Aleksey Chichenkov <a.chichenkov@initi.ru> on 1/23/19.
 */
var js_beautify = require("./libs/js-beautifier");
var args = require("./libs/args-parser")(process.argv);

var fs = require("fs");
var exec = require('child_process').exec;

var inp = args["in"] ? args["in"] : "lexer.l";
var enable_logs = args["logs"] ? args["logs"] : false;

exec("re2c -i " + inp, function(err, stdout, stderr) {
    err && console.log("ERROR: ", err);
    err && process.exit(1);

    post_process_lexer(stdout);
});

var post_process_lexer = function (_string) {
    var output = args["o"] || "lexer.js";

    // replace start and end fbrackets
    _string = _string.replace(/START\n\{/gm, "");
    _string = _string.replace(/\}\nEND/gm, "");

    _string = _string.replace(/^.*(_r2c_var_.*;|unsigned int yyaccept = 0;)\n/gm, ""); // replace var yych;
    _string = _string.replace(/(yych = \*this._yy_cursor);\n/gm, "\tcase 1:\n yych = this._string[this._yy_cursor];\n"); // insert "case 1:" before;

    _string = _string.replace(/switch \(\(yych = \*this\._yy_cursor\)\) \{/gm, "this._yy_char = this._string[this._yy_cursor]; \n switch (yych) {");  // замена разыменовываний
    _string = _string.replace(/\*\((.*?)\);/gm, "this._string[($1)];");  // замена разыменовываний
    _string = _string.replace(/\*(.* ?);/gm, "this._string[$1];");  // замена разыменовываний
    _string = _string.replace(/\*(.* ?)\)/gm, "this._string[$1])");  // замена разыменовываний

    _string = _string.replace(/^yy(\d*?):/gm, "case $1:"); // replace goto marker onto case
    _string = _string.replace(/\) goto yy(\d*?);/gm, ") { id = $1; break; }"); // replace goto inside if
    _string = _string.replace(/goto yy(\d*?);/gm, "id = $1; break;"); // replace goto outside if
    _string = _string.replace(/\{ (addLexeme.*break;) \}/gm, "$1"); // replace addLexeme
    _string = _string.replace(/\{ (unknownSymbol.*break;) \}/gm, "$1"); // replace unknownSymbol
    _string = _string.replace(/0x00/gm, 'undefined'); // replace 0x00

    _string = _string.replace(/(switch \(yych\) \{[\s\S]*?})/gm, "(function(){$1}.bind(this))(); break;"); // добавим замыкание что бы обработать свиче в свиче

    _string = _string.replace(/switch\((id)\)/gm, "switch(this._state)"); // replace id to this._state
    _string = _string.replace(/id = (\d.*?);/gm, "this._state = $1;"); // replace id = n to this._state = n

    _string = _string.replace(/yyaccept/gm, "this._yy_accept"); // replace yyaccept to this._yy_accept
    _string = _string.replace(/yych/gm, "this._yy_char"); // replace yych to this._yy_char

    if(args["t"] !== undefined) {
        switch (args["t"]) {
            case "web":

                _string = "(function () {var deps = [];define(deps, function(){\n" + _string + "return Lexer; \n});})();";
                break;
            case "node":
                _string += "\n module.exports = Lexer";
                break;
        }
    }


    _string = process_metatags(_string);

    if( !(args["no-beautify"] || args["nb"]) ) {
        _string = js_beautify(_string, {
            indent_size: 4,
            indent_char: ' ',
            preserve_newlines: true,
            space_after_anon_function: true,
            keep_array_indentation: false,
            braces_on_own_line: false
        });
    }

    if(!enable_logs){
        _string = _string.replace(/^ *console\.log[\s\S]*?$\n/gm, "");
    }

    fs.writeFileSync(output, _string);
};

var process_metatags = function (_string) {
    var metatags = {};

    _string = _string.replace(/&&>([A-Z_][A-Z0-9_]+)([\s\S]*?)&&/gm, function (_full, _metatag, _text) {
        if(!metatags[_metatag]){
            metatags[_metatag] = {
                includes: []
            };
        }
        metatags[_metatag].includes.push(_text);
        return "";
    });

    _string = _string.replace(/&&<([A-Z_][A-Z0-9_]+)&&/gm, function (_full, _metatag) {
        var metatag = metatags[_metatag];
        if(!metatag){
            throw "not found metatag for include"
        }

        return metatag.includes.join("");
    });

    return _string;
};