wordreference-definition-api/src/classes/Scrap.js

Scrap word definitions from http://www.wordreference.com

const { textContent } = require('../utils/jquery.js')
const cheerio = require('cheerio')

class Scrap {
    HTML(html) {
        let $ = cheerio.load(html)
        let rhs = $('.entryRH')
        let currentClazzName = ''
        let data = []
        const getKey = () => data.length
        rhs.each((rhKey, rh) => {
            let clazz = $(rh).find('.rh_empos, .rh_pos')
            let ols = $(rh).find('ol')
            clazz.each((clazzKey, clazzEL) => {
                currentClazzName = $(clazzEL).text()
                //each grammar class has only one ol's
                //if you remove it, everything will get desynchronized
                if ((clazzKey + 1) > ols.length) {
                    return;
                }
                let key = getKey()
                data[key] = { class: '', defs: [{ def: '', examples: [] }] }
                data[key].class = currentClazzName
                let d = $(ols[clazzKey]).find('.rh_def')
                d.each((defKey, defEL) => {
                    defEL = $(defEL)
                    data[key].defs[defKey] = { def: '', examples: [] }
                    data[key].defs[defKey].def = textContent(defEL, $)
                    defEL.find('.rh_ex').each((_, exampleEL) => {
                        data[key].defs[defKey].examples.push($(exampleEL).text())
                    })
                })
            })
        })
        return data
    }
}

module.exports = new Scrap