Merge pull request #292 from iptv-org/update-programtv-onet-pl

Update programtv.onet.pl

package-lock.json

@@ -21,6 +21,7 @@
         "jsdom": "^16.5.0",
         "lodash": "^4.17.21",
         "markdown-include": "^0.4.3",
+        "mockdate": "^3.0.5",
         "parse-duration": "^1.0.0",
         "pdf-parse": "^1.1.1",
         "srcset": "^4.0.0",
@@ -3554,6 +3555,11 @@
       "resolved": "https://registry.npmjs.org/minimist/-/minimist-1.2.5.tgz",
       "integrity": "sha512-FM9nNUYrRBAELZQT3xeZQ7fmMOBg6nWNmJKTcgsJeaLstP/UODVpGsr5OhXhhXg6f+qtJ8uiZ+PUxkDWcgIXLw=="
     },
+    "node_modules/mockdate": {
+      "version": "3.0.5",
+      "resolved": "https://registry.npmjs.org/mockdate/-/mockdate-3.0.5.tgz",
+      "integrity": "sha512-iniQP4rj1FhBdBYS/+eQv7j1tadJ9lJtdzgOpvsOHng/GbcDh2Fhdeq+ZRldrPYdXvCyfFUmFeEwEGXZB5I/AQ=="
+    },
     "node_modules/ms": {
       "version": "2.1.2",
       "resolved": "https://registry.npmjs.org/ms/-/ms-2.1.2.tgz",
@@ -7432,6 +7438,11 @@
       "resolved": "https://registry.npmjs.org/minimist/-/minimist-1.2.5.tgz",
       "integrity": "sha512-FM9nNUYrRBAELZQT3xeZQ7fmMOBg6nWNmJKTcgsJeaLstP/UODVpGsr5OhXhhXg6f+qtJ8uiZ+PUxkDWcgIXLw=="
     },
+    "mockdate": {
+      "version": "3.0.5",
+      "resolved": "https://registry.npmjs.org/mockdate/-/mockdate-3.0.5.tgz",
+      "integrity": "sha512-iniQP4rj1FhBdBYS/+eQv7j1tadJ9lJtdzgOpvsOHng/GbcDh2Fhdeq+ZRldrPYdXvCyfFUmFeEwEGXZB5I/AQ=="
+    },
     "ms": {
       "version": "2.1.2",
       "resolved": "https://registry.npmjs.org/ms/-/ms-2.1.2.tgz",

package.json

@@ -25,6 +25,7 @@
     "jsdom": "^16.5.0",
     "lodash": "^4.17.21",
     "markdown-include": "^0.4.3",
+    "mockdate": "^3.0.5",
     "parse-duration": "^1.0.0",
     "pdf-parse": "^1.1.1",
     "srcset": "^4.0.0",

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

sites/programtv.onet.pl/programtv.onet.pl.test.js

@@ -0,0 +1,81 @@
+// npx epg-grabber --config=sites/programtv.onet.pl/programtv.onet.pl.config.js --channels=sites/programtv.onet.pl/programtv.onet.pl_pl.channels.xml --output=.gh-pages/guides/pl/programtv.onet.pl.epg.xml --days=2
+
+const MockDate = require('mockdate')
+const { parser, url, logo } = require('./programtv.onet.pl.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: '13th-street-250',
+  xmltv_id: '13thStreetDeutschland.us'
+}
+const content = `<!DOCTYPE html><html lang="pl"> <head></head> <body class="withFilters pageChannel"> <div id="channelPage"> <div id="channelTV" class="nextToMenu"> <section class="channelEmissions"> <header> <span class="logoTV"> <img src="//ocdn.eu/ptv2-images-transforms/1/zB4kr1sb2dvLW1pZ3JhdGVkLzEzdGgtc3RyZWV0LnBuZ5KVAmQAwsOVAgAowsM" alt="13th Street"/> </span> </header> <div class="emissions"> <ul> <li class="hh03 hh04 fltrSerie"> <div class="hours"> <span class="hour">03:20</span> </div><div class="titles"> <a href="/tv/law-and-order-odcinek-15/rlmzu?entry=21970867" >Law &amp; Order, odc. 15: Letzte Worte</a > <span class="type">Krimiserie</span> <p> Bei einer Reality-TV-Show stirbt einer der Teilnehmer. Zunächst tappen Briscoe (Jerry Orbach) und Green (Jesse L.... </p></div></li><li class="hh23 hh00 fltrSerie"> <div class="hours"> <span class="hour">23:30</span> </div><div class="titles"> <a href="/tv/navy-cis-odcinek-1/73vbw?entry=22035734" >Navy CIS, odc. 1: New Orleans</a > <span class="type">Krimiserie</span> <p> Der Abgeordnete Dan McLane, ein ehemaliger Vorgesetzter von Gibbs, wird in New Orleans ermordet. In den 90er Jahren... </p></div></li><li class="hh01 fltrSerie"> <div class="hours"> <span class="hour">01:00</span> </div><div class="titles"> <a href="/tv/navy-cis-la-odcinek-13/tuc34?entry=22035821" >Navy CIS: L.A, odc. 13: High Society</a > <span class="type">Krimiserie</span> <p> Die Zahl der Drogentoten ist gestiegen. Das Team des NCIS glaubt, dass sich Terroristen durch den zunehmenden... </p></div></li></ul> </div></section> </div></div></body></html>`
+
+it('can generate valid url', () => {
+  MockDate.set(new Date('2021-11-24'))
+  expect(url({ channel, date })).toBe(
+    'https://programtv.onet.pl/program-tv/13th-street-250?dzien=0'
+  )
+  MockDate.reset()
+})
+
+it('can generate valid url for next day', () => {
+  MockDate.set(new Date('2021-11-25'))
+  expect(url({ channel, date })).toBe(
+    'https://programtv.onet.pl/program-tv/13th-street-250?dzien=1'
+  )
+  MockDate.reset()
+})
+
+it('can generate valid logo url', () => {
+  expect(logo({ content })).toBe(
+    'https://ocdn.eu/ptv2-images-transforms/1/zB4kr1sb2dvLW1pZ3JhdGVkLzEzdGgtc3RyZWV0LnBuZ5KVAmQAwsOVAgAowsM'
+  )
+})
+
+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: '2021-11-24T02:20:00.000Z',
+      stop: '2021-11-24T22:30:00.000Z',
+      title: `Law & Order, odc. 15: Letzte Worte`,
+      category: 'Krimiserie',
+      description: `Bei einer Reality-TV-Show stirbt einer der Teilnehmer. Zunächst tappen Briscoe (Jerry Orbach) und Green (Jesse L....`
+    },
+    {
+      start: '2021-11-24T22:30:00.000Z',
+      stop: '2021-11-25T00:00:00.000Z',
+      title: `Navy CIS, odc. 1: New Orleans`,
+      category: 'Krimiserie',
+      description:
+        'Der Abgeordnete Dan McLane, ein ehemaliger Vorgesetzter von Gibbs, wird in New Orleans ermordet. In den 90er Jahren...'
+    },
+    {
+      start: '2021-11-25T00:00:00.000Z',
+      stop: '2021-11-25T01:00:00.000Z',
+      title: `Navy CIS: L.A, odc. 13: High Society`,
+      category: 'Krimiserie',
+      description:
+        'Die Zahl der Drogentoten ist gestiegen. Das Team des NCIS glaubt, dass sich Terroristen durch den zunehmenden...'
+    }
+  ])
+})
+
+it('can handle empty guide', () => {
+  const result = parser({
+    date,
+    channel,
+    content: `<!DOCTYPE html><html><head></head><body></body></html>`
+  })
+  expect(result).toMatchObject([])
+})