1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889 |
- const fs = require('fs');
- const readline = require('readline');
- // librairie de traitement du pinyin
- const pinyinizer = require('pinyinizer')
- var stringify = require('json-stringify');
- // 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 = ''
- }
- var outputFile = 'ch.json';
- 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]
- ligne.origin = 'Chdict'
- 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)
- ligne.translations.pop()
- ligne.translations.shift()
- 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;
- })
- }
- })
- fs.appendFileSync(outputFile, stringify(ligne, null, 2, {offset: 4}) + ',\n');
- resetLigne()
- }
- }
- });
- rl.on('close', () => {
- fs.appendFileSync(outputFile, '{}]}');
- console.log('Done reading file');
- });
|