Merge pull request #2494 from tohenk/fix/programme.tvb.com

Update programme.tvb.com guide.

sites/programme.tvb.com/__data__/content.json

@@ -0,0 +1 @@
+{"code":0,"message":"","data":{"total":1,"list":[{"network_code":"j","ref_date":20241206,"schedules":[{"box_set_id":203,"network_code":"J","video_ratio":"2","video_definition":"1","event_time":1733500200,"event_datetime":"2024-12-06 23:50:00","programme_title":"天氣報告[粵] 及 潮流生活誌[粵]","en_programme_title":"Weather Report[Can] and Pop Lifestyle Guide (Ad Mag)[Can]","synopsis":"","en_synopsis":"","programme_path":"weatherreport_203","default_language":1,"play_available":1,"mytv_super_url":"https://www.mytvsuper.com/tc/programme/weatherreport_203/天氣報告[粵] 及 潮流生活誌[粵]","on_air_code":"L2,","tag":[{"icon":"hd","title":"HD"}]},{"box_set_id":0,"network_code":"J","video_ratio":"2","video_definition":"1","event_time":1733500500,"event_datetime":"2024-12-06 23:55:00","programme_title":"使徒行者3#16[粵][PG]","en_programme_title":"Line Walker: Bull Fight#16[Can][PG]","synopsis":"文鼎從淑梅手上救走大聖爺兒子，大聖爺還恩於歡喜，答允支持九指強。崇聯社定下選舉日子，恰巧是韋傑出獄之日，頭目們顧念舊日恩義，紛紛轉投浩洋。浩洋帶亞希逛傢俬店，憧憬二人未來。亞希向家強承認愛上浩洋，要求退出臥底任務。作榮與歡喜暗中會面，將國際犯罪組織「永恆幫」情報交給他。阿火遭家強出賣，到沐足店搶錢。家強逮住阿火，惟被合星誤會而受拘捕。家強把正植遺下的頸鏈和學生證交還，合星意識到家強已知悉正植身世。","en_synopsis":"","programme_path":"","default_language":0,"play_available":0,"mytv_super_url":"","on_air_code":"S2,C2,C3,C1,A,PG11L,","tag":[{"icon":"s","title":"繁/簡中文字幕"},{"icon":"e","title":"英文字幕"},{"icon":"pg","title":"PG11L"},{"icon":"hd","title":"HD"}]},{"box_set_id":0,"network_code":"J","video_ratio":"1","video_definition":"2","event_time":1733504100,"event_datetime":"2024-12-07 00:55:00","programme_title":"宣傳易[粵]","en_programme_title":"TV Easy[Can]","synopsis":"","en_synopsis":"","programme_path":"","default_language":0,"play_available":0,"mytv_super_url":"","on_air_code":"D2,","tag":[]}]}]}}

sites/programme.tvb.com/__data__/no-content.html

@@ -1,3 +0,0 @@
-
-	<ul class="clearfix" date="2022-11-15">
-			</ul>

sites/programme.tvb.com/programme.tvb.com.channels.xml

@@ -1,12 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<channels>
-  <channel site="programme.tvb.com" lang="zh" xmltv_id="J2.hk" site_id="B">J2</channel>
-  <channel site="programme.tvb.com" lang="zh" xmltv_id="Jade.hk" site_id="J">翡翠台</channel>
-  <channel site="programme.tvb.com" lang="zh" xmltv_id="Pearl.hk" site_id="P">Pearl</channel>
-  <channel site="programme.tvb.com" lang="zh" xmltv_id="TVBChineseDrama.hk" site_id="U">華語劇台</channel>
-  <channel site="programme.tvb.com" lang="zh" xmltv_id="TVBClassic.hk" site_id="E">TVB經典台</channel>
-  <channel site="programme.tvb.com" lang="zh" xmltv_id="TVBClassicMovies.hk" site_id="W">粵語片台</channel>
-  <channel site="programme.tvb.com" lang="zh" xmltv_id="TVBFinanceSportsInformationChannel.hk" site_id="A">無綫財經體育資訊台</channel>
-  <channel site="programme.tvb.com" lang="zh" xmltv_id="TVBNewsChannel.hk" site_id="C">無綫新聞台</channel>
-  <channel site="programme.tvb.com" lang="zh" xmltv_id="TVBXingHe.hk" site_id="X">TVB星河頻道</channel>
-</channels>

sites/programme.tvb.com/programme.tvb.com.config.js

@@ -1,5 +1,4 @@
 const dayjs = require('dayjs')
-const cheerio = require('cheerio')
 const utc = require('dayjs/plugin/utc')
 const timezone = require('dayjs/plugin/timezone')
 const customParseFormat = require('dayjs/plugin/customParseFormat')
@@ -8,57 +7,88 @@ dayjs.extend(utc)
 dayjs.extend(timezone)
 dayjs.extend(customParseFormat)
 
+const tz = 'Asia/Hong_Kong'
+
 module.exports = {
   site: 'programme.tvb.com',
   days: 2,
-  url: function ({ channel, date }) {
+  url({ channel, date, time = null }) {
-    return `https://programme.tvb.com/ajax.php?action=channellist&code=${
+    return `https://programme.tvb.com/api/schedule?input_date=${
-      channel.site_id
+      date.format('YYYYMMDD')
-    }&date=${date.format('YYYY-MM-DD')}`
+    }&network_code=${channel.site_id}&_t=${time ? time : parseInt(Date.now() / 1000)}`
   },
-  parser: function ({ content, date }) {
+  parser({ content, channel, date }) {
-    let programs = []
+    const programs = []
-    const items = parseItems(content)
+    const data = content ? JSON.parse(content) : {}
-    items.forEach(item => {
+    if (Array.isArray(data.data?.list)) {
-      const prev = programs[programs.length - 1]
+      const dt = date.format('YYYY-MM-DD')
-      const $item = cheerio.load(item)
+      for (const d of data.data.list) {
-      let start = parseStart($item, date)
+        if (Array.isArray(d.schedules)) {
-      if (prev) {
+          const schedules = d.schedules
-        if (start.isBefore(prev.start)) {
+            .filter(s => s.network_code === channel.site_id)
-          start = start.add(1, 'd')
+          schedules.forEach((s, i) => {
-          date = date.add(1, 'd')
+            const start = dayjs.tz(s.event_datetime, 'YYYY-MM-DD HH:mm:ss', tz)
+            let stop
+            if (i < schedules.length - 1) {
+              stop = dayjs.tz(schedules[i + 1].event_datetime, 'YYYY-MM-DD HH:mm:ss', tz)
+            } else {
+              stop = date.add(1, 'd')
             }
-        prev.stop = start
-      }
-      const stop = start.add(30, 'm')
             programs.push({
-        title: parseTitle($item),
+              title: channel.lang === 'en' ? s.en_programme_title : s.programme_title,
-        description: parseDescription($item),
+              description: channel.lang === 'en' ? s.en_synopsis : s.synopsis,
               start,
               stop
             })
           })
+        }
+      }
+    }
 
     return programs
+  },
+  async channels({ lang = 'en' }) {
+    const channels = []
+    const axios = require('axios')
+    const base = 'https://programme.tvb.com'
+    const queues = [base]
+    while (true) {
+      if (queues.length) {
+        const url = queues.shift()
+        const content = await axios
+          .get(url)
+          .then(response => response.data)
+          .catch(console.error)
+        if (content) {
+          const assets = content.match(/assets\/index\.([a-z0-9]+)\.js/g)
+          if (assets) {
+            queues.push(...assets.map(a => base + '/' + a))
+          } else {
+            const metadata = content.match(/e\=(\[(.*?)\])/)
+            if (metadata) {
+              const infos = eval(metadata[1])
+              if (Array.isArray(infos)) {
+                infos
+                  .filter(a => a.code.length)
+                  .map(a => {
+                    channels.push({
+                      lang,
+                      site_id: a.code,
+                      name: lang === 'en' ? a.nameEn : a.name
+                    })
+                  })
+                break
+              }
+            }
+          }
+          if (queues.length) {
+            continue
+          }
+        }
+      }
+      break
+    }
+
+    return channels
   }
 }
-
-function parseTitle($item) {
-  return $item('.ftit').text().trim()
-}
-
-function parseDescription($item) {
-  return $item('.full').text().trim()
-}
-
-function parseStart($item, date) {
-  const time = $item('.time').text()
-
-  return dayjs.tz(`${date.format('YYYY-MM-DD')} ${time}`, 'YYYY-MM-DD hh:mmA', 'Asia/Hong_Kong')
-}
-
-function parseItems(content) {
-  const $ = cheerio.load(content)
-
-  return $('ul > li.item').toArray()
-}

sites/programme.tvb.com/programme.tvb.com.test.js

@@ -4,41 +4,60 @@ 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)
 
-const date = dayjs.utc('2022-11-15', 'YYYY-MM-DD').startOf('d')
+const content = fs.readFileSync(path.resolve(__dirname, '__data__/content.json'))
+const date = dayjs.utc('2024-12-06', 'YYYY-MM-DD').startOf('d')
 const channel = {
-  site_id: 'B',
+  site_id: 'J',
-  xmltv_id: 'J2.hk'
+  xmltv_id: 'Jade.hk',
+  lang: 'en'
 }
 
 it('can generate valid url', () => {
-  expect(url({ channel, date })).toBe(
+  const time = 1733491000
-    'https://programme.tvb.com/ajax.php?action=channellist&code=B&date=2022-11-15'
+  expect(url({ channel, date, time })).toBe(
+    'https://programme.tvb.com/api/schedule?input_date=20241206&network_code=J&_t=1733491000'
   )
 })
 
-it('can parse response', () => {
+it('can parse response (en)', () => {
-  const content = fs.readFileSync(path.resolve(__dirname, '__data__/content.html'))
+  const results = parser({ content, channel, date }).map(p => {
-  const results = parser({ content, date }).map(p => {
     p.start = p.start.toJSON()
     p.stop = p.stop.toJSON()
     return p
   })
 
-  expect(results[0]).toMatchObject({
+  expect(results.length).toBe(3)
-    start: '2022-11-14T22:00:00.000Z',
+  expect(results[1]).toMatchObject({
-    stop: '2022-11-14T23:00:00.000Z',
+    start: '2024-12-06T15:55:00.000Z',
-    title: '想見你#3[粵/普][PG]',
+    stop: '2024-12-06T16:55:00.000Z',
+    title: 'Line Walker: Bull Fight#16[Can][PG]',
+  })
+})
+
+it('can parse response (zh)', () => {
+  const results = parser({ content, channel: { ...channel, lang: 'zh' }, date }).map(p => {
+    p.start = p.start.toJSON()
+    p.stop = p.stop.toJSON()
+    return p
+  })
+
+  expect(results.length).toBe(3)
+  expect(results[1]).toMatchObject({
+    start: '2024-12-06T15:55:00.000Z',
+    stop: '2024-12-06T16:55:00.000Z',
+    title: '使徒行者3#16[粵][PG]',
     description:
-      '韻如因父母離婚都不要自己而跑出家門，遇到子維，兩人互吐心事。雨萱順著照片上的唱片行線索，找到一家同名咖啡店，從文磊處得知照片中人是已經過世的韻如，從而推測那個男生也不是詮勝，但她內心反而更加痛苦。'
+      '文鼎從淑梅手上救走大聖爺兒子，大聖爺還恩於歡喜，答允支持九指強。崇聯社定下選舉日子，恰巧是韋傑出獄之日，頭目們顧念舊日恩義，紛紛轉投浩洋。浩洋帶亞希逛傢俬店，憧憬二人未來。亞希向家強承認愛上浩洋，要求退出臥底任務。作榮與歡喜暗中會面，將國際犯罪組織「永恆幫」情報交給他。阿火遭家強出賣，到沐足店搶錢。家強逮住阿火，惟被合星誤會而受拘捕。家強把正植遺下的頸鏈和學生證交還，合星意識到家強已知悉正植身世。',
   })
 })
 
 it('can handle empty guide', () => {
   const result = parser({
-    content: fs.readFileSync(path.resolve(__dirname, '__data__/no-content.html')),
+    content: '',
     date
   })
   expect(result).toMatchObject([])

sites/programme.tvb.com/programme.tvb.com_en.channels.xml

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<channels>
+  <channel site="programme.tvb.com" lang="en" xmltv_id="" site_id="B">tvb plus</channel>
+  <channel site="programme.tvb.com" lang="en" xmltv_id="Jade.hk" site_id="J">jade</channel>
+  <channel site="programme.tvb.com" lang="en" xmltv_id="Pearl.hk" site_id="P">pearl</channel>
+  <channel site="programme.tvb.com" lang="en" xmltv_id="TVBNewsChannel.hk" site_id="C">tvb news channel</channel>
+</channels>

sites/programme.tvb.com/programme.tvb.com_zh.channels.xml

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<channels>
+  <channel site="programme.tvb.com" lang="zh" xmltv_id="" site_id="B">TVB Plus</channel>
+  <channel site="programme.tvb.com" lang="zh" xmltv_id="Jade.hk" site_id="J">翡翠台</channel>
+  <channel site="programme.tvb.com" lang="zh" xmltv_id="Pearl.hk" site_id="P">明珠台</channel>
+  <channel site="programme.tvb.com" lang="zh" xmltv_id="TVBNewsChannel.hk" site_id="C">無綫新聞台</channel>
+</channels>

sites/programme.tvb.com/readme.md

@@ -1,11 +1,33 @@
 # programme.tvb.com
 
-https://programme.tvb.com/
+https://www.programme.tvb.com/
 
 ### Download the guide
 
+English:
+
 ```sh
-npm run grab -- --site=programme.tvb.com
+npm run grab -- --site=programme.tvb.com --lang=en
+```
+
+Chinese:
+
+```sh
+npm run grab -- --site=programme.tvb.com --lang=zh
+```
+
+### Update channel list
+
+English:
+
+```sh
+npm run channels:parse -- --config=sites/programme.tvb.com/programme.tvb.com.config.js --output=sites/programme.tvb.com/programme.tvb.com_en.channels.xml --set=lang:en
+```
+
+Chinese:
+
+```sh
+npm run channels:parse -- --config=sites/programme.tvb.com/programme.tvb.com.config.js --output=sites/programme.tvb.com/programme.tvb.com_zh.channels.xml --set=lang:zh
 ```
 
 ### Test