Initial Commit

This commit is contained in:
2023-09-10 21:48:48 +02:00
commit 227cca7d31
791 changed files with 165200 additions and 0 deletions

View File

@@ -0,0 +1,6 @@
<?xml version="1.0" encoding="UTF-8"?>
<site site="andorradifusio.ad">
<channels>
<channel lang="ca" xmltv_id="AndorraTV.ad" site_id="atv">Andorra TV</channel>
</channels>
</site>

View File

@@ -0,0 +1,66 @@
const cheerio = require('cheerio')
const dayjs = require('dayjs')
const utc = require('dayjs/plugin/utc')
const timezone = require('dayjs/plugin/timezone')
const customParseFormat = require('dayjs/plugin/customParseFormat')
require('dayjs/locale/ca')
dayjs.extend(utc)
dayjs.extend(timezone)
dayjs.extend(customParseFormat)
module.exports = {
site: 'andorradifusio.ad',
days: 2,
url({ channel }) {
return `https://www.andorradifusio.ad/programacio/${channel.site_id}`
},
parser({ content, date }) {
const programs = []
const items = parseItems(content, date)
items.forEach(item => {
const prev = programs[programs.length - 1]
let start = parseStart(item, date)
if (prev) {
if (start.isBefore(prev.start)) {
start = start.add(1, 'd')
date = date.add(1, 'd')
}
prev.stop = start
}
const stop = start.add(1, 'h')
programs.push({
title: item.title,
start,
stop
})
})
return programs
}
}
function parseStart(item, date) {
const dateString = `${date.format('MM/DD/YYYY')} ${item.time}`
return dayjs.tz(dateString, 'MM/DD/YYYY HH:mm', 'Europe/Madrid')
}
function parseItems(content, date) {
const $ = cheerio.load(content)
const dayOfWeek = dayjs(date).locale('ca').format('dddd').toLowerCase()
const column = $('.programacio-dia > h3')
.filter((i, el) => $(el).text().startsWith(dayOfWeek))
.first()
.parent()
const items = []
const titles = column.find(`p`).toArray()
column.find(`h4`).each((i, time) => {
items.push({
time: $(time).text(),
title: $(titles[i]).text()
})
})
return items
}

View File

@@ -0,0 +1,54 @@
// npx epg-grabber --config=sites/andorradifusio.ad/andorradifusio.ad.config.js --channels=sites/andorradifusio.ad/andorradifusio.ad.channels.xml --output=guide.xml --days=2
const { parser, url } = require('./andorradifusio.ad.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: 'atv',
xmltv_id: 'AndorraTV.ad'
}
it('can generate valid url', () => {
expect(url({ channel })).toBe('https://www.andorradifusio.ad/programacio/atv')
})
it('can parse response', () => {
const content = `<!DOCTYPE html><html lang="ca"> <head></head> <body id="rtvabody"> <div class="container"> <div class="row"> <div class="col-sm-12"> <div class="programacio-dia"> <h3>dimecres <span class="dia">24 nov.</span></h3> <h4>07:00</h4> <p>Club Piolet</p><h4>23:30</h4> <p>Informatiu vespre</p><h4>01:00</h4> <p>&#192;rea Andorra Difusi&#243;</p></div><div class="programacio-dia"> <h3>dijous <span class="dia">25 nov.</span></h3> <h4>07:00</h4> <p>Club Piolet</p></div></div></div></div></body></html>`
const result = parser({ content, date }).map(p => {
p.start = p.start.toJSON()
p.stop = p.stop.toJSON()
return p
})
expect(result).toMatchObject([
{
start: '2021-11-24T06:00:00.000Z',
stop: '2021-11-24T22:30:00.000Z',
title: `Club Piolet`
},
{
start: '2021-11-24T22:30:00.000Z',
stop: '2021-11-25T00:00:00.000Z',
title: `Informatiu vespre`
},
{
start: '2021-11-25T00:00:00.000Z',
stop: '2021-11-25T01:00:00.000Z',
title: `Àrea Andorra Difusió`
}
])
})
it('can handle empty guide', () => {
const result = parser({
date,
channel,
content: `<!DOCTYPE html><html><head></head><body></body></html>`
})
expect(result).toMatchObject([])
})