Merge pull request #3163 from tohenk/update-tvpassport.com

Properly generate TV schedule URL when redirected.

sites/tvpassport.com/tvpassport.com.config.js

@@ -11,29 +11,24 @@ dayjs.extend(utc)
 dayjs.extend(timezone)
 dayjs.extend(customParseFormat)
 
+const headers = {}
+
 module.exports = {
   site: 'tvpassport.com',
   days: 3,
-  url({ channel, date }) {
-    return `https://www.tvpassport.com/tv-listings/stations/${channel.site_id}/${date.format(
-      'YYYY-MM-DD'
-    )}`
+  async url({ channel, date }) {
+    return await getSiteUrl(channel, date.format('YYYY-MM-DD'))
   },
-  async request() {
-    return {
-      timeout: 30000,
-      headers: {
-        Cookie: await getCookie()
-      }
-    }
+  request: {
+    timeout: 30000,
+    headers
   },
-  parser: function ({ content }) {
-    let programs = []
-    const currentTimezone = parseCurrentTimezone(content)
-    const items = parseItems(content)
-    for (let item of items) {
-      const $item = cheerio.load(item)
-      const start = parseStart($item, currentTimezone)
+  parser({ content }) {
+    const programs = []
+    const [$, tz, items] = parseItems(content)
+    for (const item of items) {
+      const $item = $(item)
+      const start = parseStart($item, tz)
       const duration = parseDuration($item)
       const stop = start.add(duration, 'm')
       let title = parseTitle($item)
@@ -85,7 +80,7 @@ module.exports = {
     await doFetch(queue, async (url, res) => {
       if (!res) return
 
-      const [, site_id] = url.match(/\/tv-listings\/stations\/(.*)$/)
+      const site_id = getSiteId(url)
 
       console.log(`[${i}/${total}]`, url)
 
@@ -109,63 +104,88 @@ module.exports = {
   }
 }
 
-async function getCookie() {
-  const res = await axios.get('https://www.tvpassport.com/tv-listings')
-  const setCookie = res.headers['set-cookie']
-  if (!setCookie || setCookie.length === 0) return ''
-  const cookies = setCookie.map(cookie => cookie.split(';')[0])
-  return cookies.join('; ')
+async function getSiteUrl(channel, date) {
+  const f = () =>
+    `https://www.tvpassport.com/tv-listings/stations/${channel.site_id}/${date}`
+  let url = f()
+  const res = await axios.head(url, { headers })
+    .then(res => saveCookies(res))
+    .then(res => res?.request?.res?.responseUrl)
+    .catch(console.error)
+  if (res && res !== url) {
+    channel.site_id = getSiteId(res)
+    url = f()
+  }
+
+  return url
+}
+
+function getSiteId(url) {
+  const [, site_id] = url.match(/\/tv-listings\/stations\/(.*)$/)
+
+  return site_id
+}
+
+function saveCookies(res) {
+  if (res.headers && Array.isArray(res.headers['set-cookie'])) {
+    const cookies = []
+    cookies.push(...res.headers['set-cookie']
+      .map(cookie => cookie.split(';')[0].trim()))
+    headers.Cookie = cookies.length ? cookies.join('; ') : null
+  }
+
+  return res
 }
 
 function parseDescription($item) {
-  return $item('*').data('description')
+  return $item.data('description')
 }
 
 function parseImage($item) {
-  const showpicture = $item('*').data('showpicture')
+  const showpicture = $item.data('showpicture')
   const url = new URL(showpicture, 'https://cdn.tvpassport.com/image/show/960x540/')
 
   return url.href
 }
 
 function parseTitle($item) {
-  return $item('*').data('showname').toString()
+  return $item.data('showname').toString()
 }
 
 function parseSubTitle($item) {
-  return $item('*').data('episodetitle')?.toString() || null
+  return $item.data('episodetitle')?.toString() || null
 }
 
 function parseYear($item) {
-  return $item('*').data('year')?.toString() || null
+  return $item.data('year')?.toString() || null
 }
 
 function parseCategory($item) {
-  const showtype = $item('*').data('showtype')
+  const showtype = $item.data('showtype')
 
   return showtype ? showtype.split(', ') : []
 }
 
 function parseActors($item) {
-  const cast = $item('*').data('cast')
+  const cast = $item.data('cast')
 
   return cast ? cast.split(', ') : []
 }
 
 function parseDirector($item) {
-  const director = $item('*').data('director')
+  const director = $item.data('director')
 
   return director ? director.split(', ') : []
 }
 
 function parseGuest($item) {
-  const guest = $item('*').data('guest')
+  const guest = $item.data('guest')
 
   return guest ? guest.split(', ') : []
 }
 
 function parseRating($item) {
-  const rating = $item('*').data('rating')
+  const rating = $item.data('rating')
 
   return rating
     ? {
@@ -176,28 +196,24 @@ function parseRating($item) {
 }
 
 function parseStart($item, currentTimezone) {
-  const time = $item('*').data('st')
+  const time = $item.data('st')
 
   return dayjs.tz(time, 'YYYY-MM-DD HH:mm:ss', currentTimezone)
 }
 
 function parseDuration($item) {
-  const duration = $item('*').data('duration')
+  const duration = $item.data('duration')
 
   return parseInt(duration)
 }
 
 function parseItems(content) {
-  if (!content) return []
+  if (!content) return [null, null, []]
   const $ = cheerio.load(content)
 
-  return $('.station-listings .list-group-item').toArray()
+  return [
+    $,
+    $('#timezone_selector').val() || 'America/New_York',
+    $('.station-listings .list-group-item').toArray()
+  ]
 }
-
-function parseCurrentTimezone(content) {
-  if (!content) return 'America/New_York'
-  const $ = cheerio.load(content)
-
-  return $('#timezone_selector').val()
-}
-

sites/tvpassport.com/tvpassport.com.test.js

@@ -1,20 +1,28 @@
 const { parser, url } = require('./tvpassport.com.config.js')
+const axios = require('axios')
 const fs = require('fs')
 const path = require('path')
 const dayjs = require('dayjs')
 const utc = require('dayjs/plugin/utc')
 const customParseFormat = require('dayjs/plugin/customParseFormat')
+
 dayjs.extend(customParseFormat)
 dayjs.extend(utc)
 
+jest.mock('axios')
+
 const date = dayjs.utc('2022-10-04', 'YYYY-MM-DD').startOf('d')
 const channel = {
   site_id: 'youtoo-america-network/5463',
   xmltv_id: 'YTATV.us'
 }
 
-it('can generate valid url', () => {
-  expect(url({ channel, date })).toBe(
+axios.head.mockImplementation(() => {
+  return Promise.resolve({})
+})
+
+it('can generate valid url', async () => {
+  expect(await url({ channel, date })).toBe(
     'https://www.tvpassport.com/tv-listings/stations/youtoo-america-network/5463/2022-10-04'
   )
 })