Update programtv.onet.pl.config.js

This commit is contained in:
Aleksandr Statciuk 2021-11-25 00:03:02 +03:00
parent 27533d399b
commit 13be2f2aab

View file

@ -1,5 +1,4 @@
const jsdom = require('jsdom') const cheerio = require('cheerio')
const { JSDOM } = jsdom
const dayjs = require('dayjs') const dayjs = require('dayjs')
const utc = require('dayjs/plugin/utc') const utc = require('dayjs/plugin/utc')
const timezone = require('dayjs/plugin/timezone') const timezone = require('dayjs/plugin/timezone')
@ -13,36 +12,36 @@ module.exports = {
delay: 5000, delay: 5000,
site: 'programtv.onet.pl', site: 'programtv.onet.pl',
url: function ({ date, channel }) { url: function ({ date, channel }) {
const today = dayjs().utc().startOf('d') const currDate = dayjs.utc().startOf('d')
const day = date.diff(today, 'd') const day = currDate.diff(date, 'd')
return `https://programtv.onet.pl/program-tv/${channel.site_id}?dzien=${day}` return `https://programtv.onet.pl/program-tv/${channel.site_id}?dzien=${day}`
}, },
logo: function ({ content }) { logo: function ({ content }) {
const dom = new JSDOM(content) const $ = cheerio.load(content)
const img = dom.window.document.querySelector('#channelTV > section > header > span > img') const imgSrc = $('#channelTV > section > header > span > img').attr('src')
return img ? 'https:' + img.src : null return imgSrc ? `https:${imgSrc}` : null
}, },
parser: function ({ content, date }) { parser: function ({ content, date }) {
let PM = false
const programs = [] const programs = []
const items = parseItems(content) const items = parseItems(content)
items.forEach(item => { items.forEach(item => {
const title = parseTitle(item) const prev = programs[programs.length - 1]
const description = parseDescription(item) const $item = cheerio.load(item)
const category = parseCategory(item) let start = parseStart($item, date)
let start = parseStart(item, date) if (prev) {
if (start.hour() > 11) PM = true if (start.isBefore(prev.start)) {
if (start.hour() < 12 && PM) start = start.add(1, 'd') start = start.add(1, 'd')
const stop = start.add(1, 'h') date = date.add(1, 'd')
if (programs.length) {
programs[programs.length - 1].stop = start
} }
prev.stop = start
}
const stop = start.add(1, 'h')
programs.push({ programs.push({
title, title: parseTitle($item),
description, description: parseDescription($item),
category, category: parseCategory($item),
start, start,
stop stop
}) })
@ -52,27 +51,27 @@ module.exports = {
} }
} }
function parseStart(item, date) { function parseStart($item, date) {
let time = (item.querySelector('.hours > .hour') || { textContent: '' }).textContent const timeString = $item('.hours > .hour').text()
time = `${date.format('MM/DD/YYYY')} ${time}` const dateString = `${date.format('MM/DD/YYYY')} ${timeString}`
return dayjs.tz(time, 'MM/DD/YYYY HH:mm', 'Europe/Warsaw') return dayjs.tz(dateString, 'MM/DD/YYYY HH:mm', 'Europe/Warsaw')
} }
function parseCategory(item) { function parseCategory($item) {
return (item.querySelector('.titles > .type') || { textContent: '' }).textContent return $item('.titles > .type').text()
} }
function parseDescription(item) { function parseDescription($item) {
return (item.querySelector('.titles > p') || { textContent: '' }).textContent return $item('.titles > p').text().trim()
} }
function parseTitle(item) { function parseTitle($item) {
return (item.querySelector('.titles > a') || { textContent: '' }).textContent return $item('.titles > a').text().trim()
} }
function parseItems(content) { function parseItems(content) {
const dom = new JSDOM(content) const $ = cheerio.load(content)
return dom.window.document.querySelectorAll('#channelTV > section > div.emissions > ul > li') return $('#channelTV > section > div.emissions > ul > li').toArray()
} }