Rinser/bayes.js

/**
 * Created by mdonnel on 18/05/2017.
 */
let outputFile = 'jobs-special';

let log4js = require('log4js');
let logger = log4js.getLogger();
const stopwords = require('stopwords-en');
let jsonfile = require('jsonfile');
var striptags = require('striptags');

let strippedArray = [];
let corpus = {};

function getNoneStopWords(sentence) {
        var common = getStopWords();
        var wordArr = sentence.match(/\w+/g),
            commonObj = {},
            uncommonArr = [],
            word, i;

        for (i = 0; i < common.length; i++) {
            commonObj[ common[i].trim() ] = true;
        }

        for (i = 0; i < wordArr.length; i++) {
            word = wordArr[i].trim().toLowerCase();
            if (!commonObj[word]) {
                uncommonArr.push(word);
            }
        }
        return uncommonArr;
    }

    function getStopWords() {
        return stopwords;
    }




function processFile(data) {
//  console.log(stopwords);
  let fileName = ".\\dist\\stripped.json";
  console.log(fileName);
  for (let item of data.items) {
    let description = striptags(item.description);
    let stripped = getNoneStopWords(description);

    strippedArray.push(stripped);
  }

  //let o = JSON.stringify({s:strippedArray});
  //console.log(strippedArray);
  jsonfile.writeFile(fileName, strippedArray,function(err, obj) {
    console.error(err);
    console.log(obj);
  });
}


function buildCorpus() {
  console.log('buildCorpus');
  console.log(strippedArray.length);
  let fileName = ".\\dist\\corpus.json";

  for(let i of strippedArray) {
    //console.log(i);
    for(let t of i) {
      //console.log(t);

      if (!corpus.hasOwnProperty(t)) {
        corpus[t] = {score:0, common:1};
      } else {
        corpus[t].common++;
      }
    }

  }

  jsonfile.writeFile(fileName, corpus,function(err, obj) {
      console.error(err);
      console.log(obj);
    });

}

function loadCorpus() {
  let fileName = ".\\dist\\corpus.json";
 console.log(fileName);

 let data = jsonfile.readFile(fileName, function(err, obj) {
  corpus = obj;
  buildCorpus();
 });
}


function go() {

//let fileName = __dirname + "\\dist\\" + outputFile + ".json";
  let fileName = ".\\dist\\" + outputFile + ".json";
console.log(fileName);

let data = jsonfile.readFile(fileName, function(err, obj) {
 processFile(obj);
  loadCorpus();
});

}



go();