Chinese-Chdict/Programmes/ch2xml.js

const fs = require('fs');
const readline = require('readline');

// librairie de traitement du pinyin
const pinyinizer = require('pinyinizer')

var stringify = require('json-stringify');

// librairie de gestion des fichiers xml
const builder = require('xmlbuilder');

// les noms de fichiers
var fichierXml = 'ch.xml';

// on crée le fichier xml
var feed = builder.create('Chdict', {
    version: '1.0',
    encoding: 'UTF-8',
    standalone: true
});

// utilitaires
var tab = '\t'
var endLine = '\n'

var ligne = {
  hanzi: '',
  traditional: '',
  pinyin: '',
  translations: [],
  origin: ''
}

function resetLigne () {
  ligne.hanzi = ''
  ligne.traditional = ''
  ligne.pinyin = ''
  ligne.translations = []
  ligne.origin = 'Chdict'
}

var outputFile = fichierXml;

const rl = readline.createInterface({
    input: fs.createReadStream('chdict.u8')
});

// Each new line emits an event - every time the stream receives \r, \n, or \r\n
rl.on('line', (line) => {
    //console.log(line);
    if (line[0] != '#') {
      if (line.length > 2) {
        chinois = line.split(' ', 2)
        // console.log(chinois[0], " ## ", chinois[1]);
        ligne.traditional = chinois[0]
        ligne.hanzi = chinois[1]
        i = line.search(/\[/)
        j = line.search(/\]/)
        phonetic = line.substr(i + 1, j - i - 1)
        // traitement des champs pinyin avec des chiffres
        phonetic = phonetic.toLowerCase()
        // on va tester ça ...
        try {
          var sauve = phonetic
          phonetic = pinyinizer.pinyinize(phonetic)
          phonetic = phonetic.replace(/5/g, '')
        } catch (err) {
          console.log('Erreur: ' + pinyin)
          phonetic = sauve
        }
        ligne.pinyin = phonetic
        // console.log(phonetic);
        ligne.translations = line.split(/\//g)
        //var l = ligne.translations.length
        ligne.translations.pop()
        ligne.translations.shift()
        var temp = ligne.translations.join(' / ')
        ligne.translations.forEach(function (latraduction, index) {
            var tableau = latraduction.match(/\[.*?\]/g)
            if (tableau != null) {
              tableau.forEach(function (t) {
                t = t.replace('u:', 'v')
                var t1 = pinyinizer.pinyinize(t)
                console.log(t, '-', t1)
                latraduction = latraduction.replace(t, t1)
                latraduction = latraduction.replace('5', '')
                console.log(latraduction)
                ligne.translations[index] = latraduction;
              })
            }
        })
        var ele = feed.ele({ligne});
        resetLigne()
      }
    }
});

rl.on('close', () => {
    // on écrit le fichier xml
    fs.writeFileSync(fichierXml, feed.end({pretty: true}));
    console.log('Done reading file');
});