Create commands/load-cluster.js

This commit is contained in:
Aleksandr Statciuk 2022-01-06 16:01:23 +03:00
parent f5dbc9376e
commit a1f0bc2d2a
10 changed files with 29683 additions and 29503 deletions

View file

@ -0,0 +1,65 @@
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',
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 @@
{"lang":"ca","xmltv_id":"AndorraTV.ad","site_id":"atv","name":"Andorra TV","site":"andorradifusio.ad","channelsPath":"tests/__data__/input/site.channels.xml","configPath":"tests/__data__/input/site.config.js","cluster_id":1,"_id":"K1kaxwsWVjsRIZL6"}

View file

@ -1 +0,0 @@
{"lang":"ca","xmltv_id":"AndorraTV.ad","site_id":"atv","name":"Andorra TV","site":"andorradifusio.ad","filepath":"tests/__data__/input/site.channels.xml","cluster_id":1,"_id":"2lf1xEfDjyxPsXhO"}

View file

@ -23,7 +23,8 @@ it('can create database', () => {
xmltv_id: 'AndorraTV.ad',
site_id: 'atv',
site: 'andorradifusio.ad',
filepath: 'tests/__data__/input/site.channels.xml',
channelsPath: 'tests/__data__/input/site.channels.xml',
configPath: 'tests/__data__/input/andorradifusio.ad.config.js',
cluster_id: 1
})
})

View file

@ -0,0 +1,36 @@
const fs = require('fs')
const path = require('path')
const { execSync } = require('child_process')
beforeEach(() => {
fs.rmdirSync('tests/__data__/output', { recursive: true })
fs.mkdirSync('tests/__data__/output')
fs.copyFileSync('tests/__data__/input/test.db', 'tests/__data__/temp/test.db')
})
afterEach(() => {
fs.rmdirSync('tests/__data__/temp', { recursive: true })
fs.mkdirSync('tests/__data__/temp')
})
it('can load cluster', () => {
const result = execSync(
'DB_FILEPATH=tests/__data__/temp/test.db LOGS_PATH=tests/__data__/output/logs node scripts/commands/load-cluster.js --cluster-id=1',
{ encoding: 'utf8' }
)
const logs = fs.readFileSync(
path.resolve('tests/__data__/output/logs/load-cluster/cluster_1.log'),
{
encoding: 'utf8'
}
)
const lines = logs.split('\n')
const parsed = JSON.parse(lines[0])
expect(parsed['K1kaxwsWVjsRIZL6'][0]).toMatchObject({
title: 'InfoNeu ',
start: '2022-01-06T07:00:00.000Z',
stop: '2022-01-06T08:00:00.000Z',
channel: 'AndorraTV.ad',
lang: 'ca'
})
})