ECMAScript 표준문서 부록에서 문법만 추출하는 node 앱

getEcmaApp.js

// 이거 만들고 까였데시타... AST는 이렇게 만드는게 아니데시타..

//var jsdom = require('jsdom');

var fs = require('fs');
var sprintf = require('sprintf-js').sprintf;
var jquery = fs.readFileSync("./js/jquery-2.1.3.js", "utf-8");
var cheerio = require('cheerio'), $ = cheerio.load(fs.readFileSync("./ecmafile/ecmaspec.htm", "utf-8"));

console.log("start!");








var shortener = {
  origShort : new Map(),
  shortOrig : new Map(),
  shorten : function (str, len) {
    if (this.origShort.get(str) !== undefined)
      return this.origShort.get(str);

    //console.log("before : " + str);

    var camels = str.replace(/([a-z])([A-Z])/,'$1,$2' ).split(",");

    // for condition purpose
    camels.getStrLenSum = function () {
      var sum = 0;
      this.forEach(function (elem) {
        sum += elem.length;
      });
      return sum;
    }

    // phase 1 : remove vowel
    var index = camels.length-1;
    while((camels.getStrLenSum() > len ) && (index >= 0) ) {
      camels[index] = camels[index].replace(/[aeiou]/g,'');
      index--;
    }

    //phase 2 : remove letters
    index = camels.length-1;
    while((camels.getStrLenSum() > len ) || (this.shortOrig.get(camels.join('')) !== undefined) ) {
      if (camels[index].length > 1)
        camels[index] = camels[index].slice(0,-1);
      index--;
      if (index < 0) {
        index = camels.length-1;
      }
    }

    var ret = camels.join('');

    this.shortOrig.set(ret, str);
    this.origShort.set(str, ret);

    //console.log("after : " + ret + " : " + ret.length);
    return ret;
  }

}


exc($);

function exc($) {
  const LHS_MAX_LEN = 16;
  const RHS_MAX_LEN = 39;

  var tokenMap = new Map();

  console.log("response:");

  var str = "";
  $("#sec-A div.gp").each(function (idx, gpElem) {

      // lhs
      var lhs = "";
      var lhsElem = $(".lhs", gpElem);
      lhs = $(".nt", lhsElem).html();

      if (lhs.length > LHS_MAX_LEN) {
        lhs = shortener.shorten(lhs, LHS_MAX_LEN);
      }

      // increase state of keyword
      tokenMap.set(lhs, (tokenMap.get(lhs) !=0 ) ? 1 : (tokenMap.get(lhs)+1) )

      var grhsmodElem = $(".grhsmod", lhsElem);

      // normally there are rhs part
      var rhsElems = $(".rhs", gpElem);
      if ($(rhsElems).length > 0) {
        $(rhsElems).each(function (idx2, rhsElem) {
          var curStr = "";

          // rhs
          var rhs = "";

          var grhsmodElemAtRhs = $(".grhsmod", rhsElem);

          // for each rhs terminal / non-terminal
          $(".rhs *", rhsElem).each(function (idx3, rhsSubElem) {
            if ($(rhsSubElem).hasClass("nt") || $(rhsSubElem).hasClass("t")) {
              var rhs_text = $(rhsSubElem).html();

              // shorten
              if (rhs_text.length > LHS_MAX_LEN) {
                rhs_text = shortener.shorten(rhs_text, LHS_MAX_LEN);
              }

              // decode html
              rhs_text = $('<div/>').html(rhs_text).text();

              // terminal symbol
              if ($(rhsSubElem).hasClass("t")) {
                rhs_text = "'" + rhs_text + "'";
              }
              rhs += " " + rhs_text;

              // if it is one-of...
              if ($(grhsmodElem).html() == "one of") {
                //compose
                curStr += sprintf("%-"+LHS_MAX_LEN+"s -> %-"+"s;", lhs, rhs);

                // newline
                curStr += "\r\n";
                str += curStr;

                //console.log(curStr);
                lhs = "";
                rhs = "";
                curStr = "";
              }
            }

            else if (String($(grhsmodElemAtRhs).html()).match(/but not.*/)) {
              console.log("matches " + $(grhsmodElemAtRhs).html());
              return false; // jQuery .each() break;
            }
          });

          // if rhs is valid (normal route)
          if (rhs.length > 0) {
            //compose
            curStr += sprintf("%-"+LHS_MAX_LEN+"s -> %-"+"s;", lhs, rhs);

            // newline
            curStr += "\r\n";
            str += curStr;
            //console.log(curStr);
            lhs = "";
          }
        });
      }


      // it is one-of type, by table
      else if ($(grhsmodElem).html() == "one of") {

        console.log("one of");



        $("td", $(gpElem).next()).each(function (idx, oneOfElem) {

          var curStr = "";

          // rhs
          var rhs = "";

          //////////////////////////////////////////////

          var rhs_text = $(oneOfElem).html();

          if (rhs_text != "") {
            //console.log(rhs_text);
            if (rhs_text.length > LHS_MAX_LEN) {
              rhs_text = shortener.shorten(rhs_text, LHS_MAX_LEN);
            }

            // decode html
            rhs_text = $('<div/>').html(rhs_text).text();

            //if ($(oneOfElem).hasClass("t")) {
              rhs_text = "'" + rhs_text + "'";
            //}
            rhs += " " + rhs_text;

            //////////////////////////////////////////////

            //compose
            curStr += sprintf("%-"+LHS_MAX_LEN+"s -> %-"+"s;", lhs, rhs);

            // newline
            curStr += "\r\n";
            str += curStr;
            //console.log(curStr);
            lhs = "";
          }
        });


      }

  });

  var state = "";
  // tokenMap.forEach(function (val, key) {
  //   var curState = "key " + key + " : " + val;
  //   state += curState;
  //   console.log(state);
  //   state += "\r\n";
  // });

  fs.writeFile("ECMAScript.gr", str, "UTF-8", function () {
    console.log("grammar written to file.");
  })
  fs.writeFile("ECMAScript.gr.state", state, "UTF-8", function () {
    console.log("state written to file.");
  })
}