Update musor.tv selectors

This commit is contained in:
Kristof Bach
2025-10-12 14:19:40 +02:00
parent eebfcc8278
commit 883779fa72

View File

@@ -3,6 +3,7 @@ const axios = require('axios')
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')
const uniqBy = require('lodash.uniqby')
dayjs.extend(utc) dayjs.extend(utc)
dayjs.extend(customParseFormat) dayjs.extend(customParseFormat)
@@ -49,38 +50,42 @@ module.exports = {
const $ = cheerio.load(html) const $ = cheerio.load(html)
const channels = $('body > div.big_content > div > nav > table > tbody > tr > td > a').toArray() const channels = $('body > div.big_content > div > nav > table > tbody > tr > td > a').toArray()
return channels
.map(item => { return uniqBy(
const $item = $(item) channels
const url = $item.attr('href') .map(item => {
if (!url.startsWith('//musor.tv/mai/tvmusor/')) return null const $item = $(item)
const site_id = url.replace('//musor.tv/mai/tvmusor/', '') const url = $item.attr('href')
return { if (!url.startsWith('//musor.tv/mai/tvmusor/')) return null
lang: 'hu', const site_id = url.replace('//musor.tv/mai/tvmusor/', '')
site_id, return {
name: $item.text() lang: 'hu',
} site_id,
}) name: $item.text()
.filter(i => i) }
})
.filter(i => i),
'site_id'
)
} }
} }
function parseImage($item) { function parseImage($item) {
const imgSrc = $item.find('div.smartpe_screenshot > img').attr('src') const imgSrc = $item.find('div.progentry_screenshot > img').attr('src')
return imgSrc ? `https:${imgSrc}` : null return imgSrc ? `https:${imgSrc}` : null
} }
function parseTitle($item) { function parseTitle($item) {
return $item.find('div:nth-child(2) > div > h3 > a').text().trim() return $item.find('h3 > a').text().trim()
} }
function parseDescription($item) { function parseDescription($item) {
return $item.find('div:nth-child(5) > div > div').text().trim() return $item.find('div.progentrylong').text().trim()
} }
function parseStart($item) { function parseStart($item) {
let datetime = $item.find('div:nth-child(1) > div > div > div > div > time').attr('content') let datetime = $item.find('time').attr('content')
if (!datetime) return null if (!datetime) return null
return dayjs.utc(datetime.replace('GMT', 'T'), 'YYYY-MM-DDTHH:mm:ss') return dayjs.utc(datetime.replace('GMT', 'T'), 'YYYY-MM-DDTHH:mm:ss')
@@ -89,5 +94,5 @@ function parseStart($item) {
function parseItems(content) { function parseItems(content) {
const $ = cheerio.load(content) const $ = cheerio.load(content)
return [$, $('div.multicolumndayprogarea > div.smartpe_progentry').toArray()] return [$, $('div.progarea > div.progentry > div.progentry_internal').toArray()]
} }