80 lines
1.8 KiB
JavaScript
80 lines
1.8 KiB
JavaScript
const dayjs = require('dayjs')
|
|
const cheerio = require('cheerio')
|
|
|
|
const parseDates = (startDate, endDate) => {
|
|
const start = [
|
|
startDate.year(),
|
|
startDate.month() + 1,
|
|
startDate.date(),
|
|
startDate.hour(),
|
|
startDate.minute(),
|
|
]
|
|
|
|
const duration = { hours: 1 }
|
|
return {
|
|
start,
|
|
duration,
|
|
}
|
|
}
|
|
|
|
const parseEventData = (eventData) => {
|
|
const startDate = eventData.startDate && dayjs(eventData.startDate)
|
|
const endDate = eventData.endDate && dayjs(eventData.endDate)
|
|
const { start, duration } = parseDates(startDate, endDate)
|
|
const { location } = eventData || {}
|
|
const { address } = location || {}
|
|
|
|
const locationStr = location && [
|
|
location.name || '',
|
|
address.streetAddress || '',
|
|
address.addressLocality || '',
|
|
address.postalCode || '',
|
|
address.addressCountry || '',
|
|
].join(', ')
|
|
const title = eventData.name
|
|
const url = eventData.url
|
|
const description = eventData.description
|
|
|
|
return {
|
|
start,
|
|
duration,
|
|
location: locationStr,
|
|
title,
|
|
url,
|
|
description,
|
|
}
|
|
}
|
|
|
|
const parseHTML = (html) => {
|
|
try {
|
|
// NOTE: Mobile web should have serialized
|
|
// event info in one of the script tags
|
|
const $ = cheerio.load(html)
|
|
const $scripts = $('head script[type="application/ld+json"]')
|
|
const rawData = $scripts.toArray().reduce((data, node) => {
|
|
const firstNode = node.children[0]
|
|
|
|
if (!firstNode || !firstNode.data) {
|
|
return data
|
|
}
|
|
|
|
if (firstNode.data.startsWith('//<![CDATA')) {
|
|
return firstNode.data
|
|
}
|
|
|
|
return data
|
|
}, null)
|
|
|
|
const eventData = JSON.parse(rawData.slice(12, -5))
|
|
const data = parseEventData(eventData)
|
|
|
|
console.log(data)
|
|
return data
|
|
} catch (err) {
|
|
throw err
|
|
return
|
|
}
|
|
}
|
|
|
|
module.exports = parseHTML
|