diff --git a/sites/mi.tv/mi.tv.config.js b/sites/mi.tv/mi.tv.config.js index acab9768..315f81fe 100644 --- a/sites/mi.tv/mi.tv.config.js +++ b/sites/mi.tv/mi.tv.config.js @@ -1,5 +1,4 @@ -const jsdom = require('jsdom') -const { JSDOM } = jsdom +const cheerio = require('cheerio') const dayjs = require('dayjs') const utc = require('dayjs/plugin/utc') const customParseFormat = require('dayjs/plugin/customParseFormat') @@ -15,27 +14,32 @@ module.exports = { return `https://mi.tv/${country}/async/channel/${id}/${date.format('YYYY-MM-DD')}/0` }, logo({ content }) { - const dom = new JSDOM(content) - const img = dom.window.document.querySelector('#listings > div.channel-info > img') - return img ? img.src : null + const $ = cheerio.load(content) + const imgSrc = $('#listings > div.channel-info > img').attr('src') + + return imgSrc || null }, parser({ content, date }) { - let PM = false const programs = [] const items = parseItems(content) items.forEach(item => { - const title = parseTitle(item) - let start = parseStart(item, date) + const prev = programs[programs.length - 1] + const $item = cheerio.load(item) + let start = parseStart($item, date) if (!start) return - if (start.hour() > 11) PM = true - if (start.hour() < 12 && PM) start = start.add(1, 'd') - const stop = start.add(1, 'h') - if (programs.length) { - programs[programs.length - 1].stop = start + if (prev) { + if (start.isBefore(prev.start)) { + start = start.add(1, 'd') + date = date.add(1, 'd') + } + prev.stop = start } - + const stop = start.add(1, 'h') programs.push({ - title, + title: parseTitle($item), + category: parseCategory($item), + description: parseDescription($item), + icon: parseIcon($item), start, stop }) @@ -45,20 +49,35 @@ module.exports = { } } -function parseStart(item, date) { - let time = (item.querySelector('a > div.content > span.time') || { textContent: '' }).textContent - if (!time) return null - time = `${date.format('MM/DD/YYYY')} ${time}` +function parseStart($item, date) { + const timeString = $item('a > div.content > span.time').text() + if (!timeString) return null + const dateString = `${date.format('MM/DD/YYYY')} ${timeString}` - return dayjs.utc(time, 'MM/DD/YYYY HH:mm') + return dayjs.utc(dateString, 'MM/DD/YYYY HH:mm') } -function parseTitle(item) { - return (item.querySelector('a > div.content > h2') || { textContent: '' }).textContent +function parseTitle($item) { + return $item('a > div.content > h2').text().trim() +} + +function parseCategory($item) { + return $item('a > div.content > span.sub-title').text().trim() +} + +function parseDescription($item) { + return $item('a > div.content > p.synopsis').text().trim() +} + +function parseIcon($item) { + const backgroundImage = $item('a > div.image-parent > div.image').css('background-image') + const [_, icon] = backgroundImage.match(/url\(\'(.*)'\)/) || [null, null] + + return icon } function parseItems(content) { - const dom = new JSDOM(content) + const $ = cheerio.load(content) - return dom.window.document.querySelectorAll('#listings > ul > li') + return $('#listings > ul > li').toArray() } diff --git a/sites/mi.tv/mi.tv.test.js b/sites/mi.tv/mi.tv.test.js new file mode 100644 index 00000000..58ac52ac --- /dev/null +++ b/sites/mi.tv/mi.tv.test.js @@ -0,0 +1,71 @@ +// npx epg-grabber --config=sites/mi.tv/mi.tv.config.js --channels=sites/mi.tv/mi.tv_ar.channels.xml --output=.gh-pages/guides/ar/mi.tv.epg.xml --days=2 + +const { parser, url, logo } = require('./mi.tv.config.js') +const dayjs = require('dayjs') +const utc = require('dayjs/plugin/utc') +const customParseFormat = require('dayjs/plugin/customParseFormat') +dayjs.extend(customParseFormat) +dayjs.extend(utc) + +const date = dayjs.utc('2021-11-24', 'YYYY-MM-DD').startOf('d') +const channel = { + site_id: 'ar#24-7-canal-de-noticias', + xmltv_id: '247CanaldeNoticias.ar' +} +const content = `
Programación 24/7 Canal de Noticias

Programación 24/7 Canal de Noticias Miércoles 24 de noviembre

` + +it('can generate valid url', () => { + expect(url({ channel, date })).toBe( + 'https://mi.tv/ar/async/channel/24-7-canal-de-noticias/2021-11-24/0' + ) +}) + +it('can generate valid logo url', () => { + expect(logo({ content })).toBe( + 'https://cdn.mitvstatic.com/channels/ar_24-7-canal-de-noticias_m.png' + ) +}) + +it('can parse response', () => { + const result = parser({ content, date }).map(p => { + p.start = p.start.toJSON() + p.stop = p.stop.toJSON() + return p + }) + + expect(result).toMatchObject([ + { + start: '2021-11-24T03:00:00.000Z', + stop: '2021-11-24T23:00:00.000Z', + title: `Trasnoche de 24/7`, + category: 'Interés general', + description: 'Lo más visto de la semana en nuestra pantalla.', + icon: 'https://cdn.mitvstatic.com/programs/fallback_other_l_m.jpg' + }, + { + start: '2021-11-24T23:00:00.000Z', + stop: '2021-11-25T01:00:00.000Z', + title: `Noticiero central - Segunda edición`, + category: 'Noticiero', + description: `Cerramos el día con un completo resumen de los temas más relevantes con columnistas y análisis especiales para terminar el día.`, + icon: 'https://cdn.mitvstatic.com/programs/fallback_other_l_m.jpg' + }, + { + start: '2021-11-25T01:00:00.000Z', + stop: '2021-11-25T02:00:00.000Z', + title: `Plus energético`, + category: 'Cultural', + description: `La energía tiene mucho para mostrar. Este programa reúne a las principales empresas y protagonistas de la actividad que esta revolucionando la región.`, + icon: 'https://cdn.mitvstatic.com/programs/fallback_other_l_m.jpg' + } + ]) +}) + +it('can handle empty guide', () => { + const result = parser({ + date, + channel, + content: `` + }) + expect(result).toMatchObject([]) +})