Merge pull request #1223 from iptv-org/fix-tvmi.mt

Fix tvmi.mt
This commit is contained in:
Aleksandr Statciuk 2022-10-28 18:36:49 +03:00 committed by GitHub
commit 2d3ab1ca5b
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 1400 additions and 52 deletions

File diff suppressed because it is too large Load diff

View file

@ -8,20 +8,30 @@ dayjs.extend(timezone)
module.exports = { module.exports = {
site: 'tvmi.mt', site: 'tvmi.mt',
url: function ({ date }) { url: function ({ date, channel }) {
return `https://www.tvmi.mt/mt/tvmi/skeda/?sd-date=${date.format('YYYY-MM-DD')}` return `https://tvmi.mt/schedule/${channel.site_id}/${date.format('YYYY-MM-DD')}`
}, },
parser: function ({ content, channel, date }) { parser: function ({ content, date }) {
let programs = [] let programs = []
const items = parseItems(content, channel) const items = parseItems(content)
items.forEach(item => { items.forEach(item => {
const $item = cheerio.load(item) const $item = cheerio.load(item)
const prev = programs[programs.length - 1]
let start = parseStart($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(30, 'm')
programs.push({ programs.push({
title: parseTitle($item), title: parseTitle($item),
description: parseDescription($item), description: parseDescription($item),
icon: parseIcon($item), icon: parseIcon($item),
start: parseStart($item, date), start,
stop: parseStop($item, date) stop
}) })
}) })
@ -30,41 +40,37 @@ module.exports = {
} }
function parseTitle($item) { function parseTitle($item) {
return $item('.programme-title').text().trim() return $item('div > div:nth-child(2) > div:nth-child(2),a > div:nth-child(2) > div:nth-child(2)')
.text()
.trim()
} }
function parseDescription($item) { function parseDescription($item) {
return $item('.description').text().trim() return $item('div > div:nth-child(2) > div:nth-child(3),a > div:nth-child(2) > div:nth-child(3)')
.text()
.trim()
} }
function parseIcon($item) { function parseIcon($item) {
return $item('.image').attr('src') const bg = $item('div > div:nth-child(1) > div > div,a > div:nth-child(1) > div').data('bg')
return bg ? `https:${bg}` : null
} }
function parseStart($item, date) { function parseStart($item, date) {
const times = $item('.title-time > .times').text().trim() const timeString = $item(
const [_, HH, mm] = times.match(/^(\d{2}):(\d{2}) \-/) || [null, null, null] 'div > div:nth-child(2) > div:nth-child(1),a > div:nth-child(2) > div:nth-child(1)'
)
.text()
.trim()
const [_, HH, mm] = timeString.match(/^(\d{2}):(\d{2})/) || [null, null, null]
if (!HH || !mm) return null if (!HH || !mm) return null
return dayjs.tz(`${date.format('YYYY-MM-DD')} ${HH}:${mm}`, 'YYYY-MM-DD HH:mm', 'Europe/Malta') return dayjs.tz(`${date.format('YYYY-MM-DD')} ${HH}:${mm}`, 'YYYY-MM-DD HH:mm', 'Europe/Malta')
} }
function parseStop($item, date) { function parseItems(content) {
const times = $item('.title-time > .times').text().trim()
const [_, HH, mm] = times.match(/\- (\d{2}):(\d{2})$/) || [null, null, null]
if (!HH || !mm) return null
return dayjs.tz(`${date.format('YYYY-MM-DD')} ${HH}:${mm}`, 'YYYY-MM-DD HH:mm', 'Europe/Malta')
}
function parseItems(content, channel) {
const $ = cheerio.load(content) const $ = cheerio.load(content)
const channelSchedules = $(
'#content > div.schedule > div.content > div.schedule-bottom > div.schedules > .channel-schedule'
).toArray()
if (!Array.isArray(channelSchedules)) return []
const index = parseInt(channel.site_id) - 1
if (!channelSchedules[index]) return []
return $(channelSchedules[index], '.programmes').find('.remodal').toArray() return $('body > main > div.mt-8 > div').toArray()
} }

View file

@ -1,48 +1,47 @@
// npx epg-grabber --config=sites/tvmi.mt/tvmi.mt.config.js --channels=sites/tvmi.mt/tvmi.mt_mt.channels.xml --output=guide.xml --days=2 // npx epg-grabber --config=sites/tvmi.mt/tvmi.mt.config.js --channels=sites/tvmi.mt/tvmi.mt_mt.channels.xml --output=guide.xml --days=2
const { parser, url } = require('./tvmi.mt.config.js') const { parser, url } = require('./tvmi.mt.config.js')
const fs = require('fs')
const path = require('path')
const dayjs = require('dayjs') const dayjs = require('dayjs')
const utc = require('dayjs/plugin/utc') const utc = require('dayjs/plugin/utc')
const customParseFormat = require('dayjs/plugin/customParseFormat') const customParseFormat = require('dayjs/plugin/customParseFormat')
dayjs.extend(customParseFormat) dayjs.extend(customParseFormat)
dayjs.extend(utc) dayjs.extend(utc)
const date = dayjs.utc('2022-04-21', 'YYYY-MM-DD').startOf('d') const date = dayjs.utc('2022-10-29', 'YYYY-MM-DD').startOf('d')
const channel = { const channel = {
site_id: '1', site_id: '2',
xmltv_id: 'TVM.mt' xmltv_id: 'TVM.mt'
} }
it('can generate valid url', () => { it('can generate valid url', () => {
expect(url({ date })).toBe('https://www.tvmi.mt/mt/tvmi/skeda/?sd-date=2022-04-21') expect(url({ date, channel })).toBe('https://tvmi.mt/schedule/2/2022-10-29')
}) })
it('can parse response', () => { it('can parse response', () => {
const content = `<!doctype html><html lang="en-US" xmlns:fb="https://www.facebook.com/2008/fbml" xmlns:addthis="https://www.addthis.com/help/api-spec" > <head></head> <body class="page-template page-template-page-templates page-template-new-schedule page-template-page-templatesnew-schedule-php page page-id-457931 page-child parent-pageid-554693 blog-1 tvmi-page"> <div id="page" class="site relative "> <div id="content" class="site-content relative"> <div class="schedule"> <div class="content"> <div class="schedule-bottom"> <div class="schedules"> <div class="channel-schedule"> <div class="programmes"> <a href="#sorelle8211ittieletstaun" class="programme-wrapper" style="left: 300px; min-width: 300px; max-width: 300px;"> <div class="programme " style="min-width: 294px; max-width: 294px;"> <div class="programme-info"> <p class="title">Sorelle &#8211; It-Tielet Staġun </p><p class="times">06:30 - 07:00</p></div></div></a> <div class="remodal" data-remodal-id="sorelle8211ittieletstaun" data-remodal-options="modifier: skeda-programme-popup"> <button data-remodal-action="close" class="remodal-close"></button> <img class="image" src="https://www.tvmi.mt/mt/wp-content/uploads/sites/1/2019/09/pink-leather-11-for-Airing.png" alt=""> <div class="bottom-modal"> <div class="modal-top"> <div class="title-time"> <p class="programme-title">Sorelle &#8211; It-Tielet Staġun</p><p class="times">06:30 - 07:00</p></div><img class="logo" src="https://www.tvmi.mt/mt/wp-content/themes/tvm/dist/images/tvm-logo-2021.png" alt=""> </div><p class="description">Fit-tielet staġun ta <em>Sorelle</em> se naraw lill-aħwa, flimkien ma dawk kollha li sofrew taħt idejn Leo, inklużi martu Bella u bintu Janice, ikomplu jagħmlu minn kollox biex jeħilsu darba għal dejjem minnu. Il-fatti se juruna li dan mhu se jkun faċli xejn għax Leo għandu għeruq fil-fond u għalhekk għad baqagħlu ħafna sorpriżi għalihom ilkoll. Dan waqt li flimkien ikomplu jaraw kif jesponu lil dawk li bxi mod jew ieħor ikun qed jirnexxilhom jaħarbu mill-awtoritajiet.</p><div class="flex" style="margin-left: -10px; margin-right: -10px;"> <a class="more-button" href="https://www.tvmi.mt/mt/tvmi/programmes/sorelle-it-tielet-stagun/"> ARA IKTAR </a> </div></div></div><a href="#ta8217filgodu" class="programme-wrapper" style="left: 600px; min-width: 1200px; max-width: 1200px;"> <div class="programme " style="min-width: 1194px; max-width: 1194px;"> <div class="programme-info"> <p class="title">Ta Filgħodu </p><p class="times">07:00 - 09:00</p></div></div></a> <div class="remodal" data-remodal-id="ta8217filgodu" data-remodal-options="modifier: skeda-programme-popup"> <button data-remodal-action="close" class="remodal-close"></button> <img class="image" src="https://www.tvmi.mt/mt/wp-content/uploads/sites/1/2021/09/ta-Filghodu-02.jpeg" alt=""> <div class="bottom-modal"> <div class="modal-top"> <div class="title-time"> <p class="programme-title">Ta Filgħodu</p><p class="times">07:00 - 09:00</p></div><img class="logo" src="https://www.tvmi.mt/mt/wp-content/themes/tvm/dist/images/tvm-logo-2021.png" alt=""> </div><p class="description"><strong><em>Ta' Filgħodu</em></strong>&nbsp;se jkun qed jibda jum ġdid fuq nota pożittiva; b'informazzjoni u aġġornamenti regolari dwar it-traffiku, it-temp u l-ewwel faċċati tal-gazzetti mal-ewwel tazza cafè. Dan iwassalna biex flimkien ma Ray Attard niltaqgħu mal-protagonisti li minn wara l-kwinti jkunu qed jagħmlu l-ġurnata ta kulħadd waħda isbaħ fl-oqsma rispettivi tagħhom.&nbsp;<em>Ta' Filgħodu</em>&nbsp;se jkun kurrenti imma din id-darba minn angoli oħra ta interess, li bis-sewwa se jkunu qed jgħarrfu lit-telespettaturi b'dak kollu li jkun qed jiġri madwarna.&nbsp;Preżentazzjoni ta Ray Attard.</p><div class="flex" style="margin-left: -10px; margin-right: -10px;"> <a class="more-button" href="https://www.tvmi.mt/mt/tvmi/programmes/ta-filghodu/"> ARA IKTAR </a> </div></div></div></div></div><div class="channel-schedule"> <div class="programmes"> <a href="#euronews" class="programme-wrapper" style="left: -3600px; min-width: 150px; max-width: 150px;"> <div class="programme " style="min-width: 144px; max-width: 144px;"> <div class="programme-info"> <p class="title">EuroNews </p><p class="times">00:00 - 00:15</p></div></div></a> </div></div></div></div></div></div></div></div></body></html>` const content = fs.readFileSync(path.resolve(__dirname, '__data__/content.html'))
const result = parser({ content, channel, date }).map(p => { const results = parser({ content, date }).map(p => {
p.start = p.start.toJSON() p.start = p.start.toJSON()
p.stop = p.stop.toJSON() p.stop = p.stop.toJSON()
return p return p
}) })
expect(result).toMatchObject([ expect(results[0]).toMatchObject({
{ start: '2022-10-29T03:30:00.000Z',
start: '2022-04-21T04:30:00.000Z', stop: '2022-10-29T04:00:00.000Z',
stop: '2022-04-21T05:00:00.000Z', title: 'Bizzilla',
title: 'Sorelle It-Tielet Staġun',
description: description:
'Fit-tielet staġun ta Sorelle se naraw lill-aħwa, flimkien ma dawk kollha li sofrew taħt idejn Leo, inklużi martu Bella u bintu Janice, ikomplu jagħmlu minn kollox biex jeħilsu darba għal dejjem minnu. Il-fatti se juruna li dan mhu se jkun faċli xejn għax Leo għandu għeruq fil-fond u għalhekk għad baqagħlu ħafna sorpriżi għalihom ilkoll. Dan waqt li flimkien ikomplu jaraw kif jesponu lil dawk li bxi mod jew ieħor ikun qed jirnexxilhom jaħarbu mill-awtoritajiet.', 'Storja ta tliet familji, tnejn minnhom miżżewġin bejniethom, u familja oħra li għalkemm mhijiex, bdaqshekk ma jfissirx li mhijiex parti ntegrali fil-kompliċitá li ilha għaddejja bejniethom għal dawn l-aħħar tletin sena.',
icon: 'https://www.tvmi.mt/mt/wp-content/uploads/sites/1/2019/09/pink-leather-11-for-Airing.png' icon: 'https://dist4.tvmi.mt/eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJzdWIiOjMyNjEwNywiYXVkIjoiMTg4LjI0Mi40OC45MyIsImV4cCI6MTY2NzAxNjM1OH0.N4de761te_pRvWwSUnF6httRAzdukup5syejwXTUv8g/vod/663927/image.jpg'
}, })
{
start: '2022-04-21T05:00:00.000Z', expect(results[1]).toMatchObject({
stop: '2022-04-21T07:00:00.000Z', start: '2022-10-29T04:00:00.000Z',
title: 'Ta Filgħodu', stop: '2022-10-29T04:30:00.000Z',
description: title: 'The Adventures of Puss in Boots',
"Ta' Filgħodu se jkun qed jibda jum ġdid fuq nota pożittiva; b'informazzjoni u aġġornamenti regolari dwar it-traffiku, it-temp u l-ewwel faċċati tal-gazzetti mal-ewwel tazza cafè. Dan iwassalna biex flimkien ma Ray Attard niltaqgħu mal-protagonisti li minn wara l-kwinti jkunu qed jagħmlu l-ġurnata ta kulħadd waħda isbaħ fl-oqsma rispettivi tagħhom. Ta' Filgħodu se jkun kurrenti imma din id-darba minn angoli oħra ta interess, li bis-sewwa se jkunu qed jgħarrfu lit-telespettaturi b'dak kollu li jkun qed jiġri madwarna. Preżentazzjoni ta Ray Attard.", icon: 'https://dist4.tvmi.mt/eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJzdWIiOjMyNjEwNywiYXVkIjoiMTg4LjI0Mi40OC45MyIsImV4cCI6MTY2NzAxNjM1OH0.N4de761te_pRvWwSUnF6httRAzdukup5syejwXTUv8g/vod/747336/image.jpg'
icon: 'https://www.tvmi.mt/mt/wp-content/uploads/sites/1/2021/09/ta-Filghodu-02.jpeg' })
}
])
}) })
it('can handle empty guide', () => { it('can handle empty guide', () => {

View file

@ -1,7 +1,8 @@
<?xml version="1.0" encoding="UTF-8"?> <?xml version="1.0" encoding="UTF-8"?>
<site site="tvmi.mt"> <site site="tvmi.mt">
<channels> <channels>
<channel lang="mt" xmltv_id="TVM.mt" site_id="1">TVM</channel> <channel lang="mt" xmltv_id="TVM.mt" site_id="2">TVM</channel>
<channel lang="mt" xmltv_id="TVMNewsPlus.mt" site_id="2">TVMNews+</channel> <channel lang="mt" xmltv_id="TVMnewsPlus.mt" site_id="3">TVMnews+</channel>
<channel lang="mt" xmltv_id="TVMsportPlus.mt" site_id="4">TVMsport+</channel>
</channels> </channels>
</site> </site>