diff --git a/sites/znbc.co.zm.config.js b/sites/znbc.co.zm.config.js index 7f078b55..8e24dc80 100644 --- a/sites/znbc.co.zm.config.js +++ b/sites/znbc.co.zm.config.js @@ -4,6 +4,7 @@ const dayjs = require('dayjs') const utc = require('dayjs/plugin/utc') const timezone = require('dayjs/plugin/timezone') const customParseFormat = require('dayjs/plugin/customParseFormat') +const tabletojson = require('tabletojson').Tabletojson dayjs.extend(utc) dayjs.extend(timezone) @@ -11,6 +12,7 @@ dayjs.extend(customParseFormat) module.exports = { lang: 'en', + days: 7, site: 'znbc.co.zm', channels: 'znbc.co.zm.channels.xml', output: '.gh-pages/guides/znbc.co.zm.guide.xml', @@ -29,9 +31,9 @@ module.exports = { const programs = [] const items = parseItems(content, date) items.forEach(item => { - const title = parseTitle(item) + const title = item.title const start = parseStart(item, date) - const stop = parseStop(item, date) + const stop = start.add(30, 'm') if (programs.length) { programs[programs.length - 1].stop = start } @@ -43,35 +45,30 @@ module.exports = { } } -function parseStop(item, date) { - return date.endOf('d') -} - function parseStart(item, date) { - const row = (item.querySelector('td > p') || { textContent: '' }).textContent - let time = row.split(' ').shift() - time = `${date.format('MM/DD/YYYY')} ${time}` + const time = `${date.format('MM/DD/YYYY')} ${item.time}` return dayjs.tz(time, 'MM/DD/YYYY HH:mm', 'Africa/Lusaka') } -function parseTitle(item) { - const row = (item.querySelector('td > p') || { textContent: '' }).textContent - const title = row.split(' ') - title.shift() - - return title - .map(i => i.trim()) - .filter(s => s) - .join(' ') -} - function parseItems(content, date) { + const items = [] const day = date.day() // 0 => Sunday const dom = new JSDOM(content) const tabs = dom.window.document.querySelectorAll( `.elementor-tabs-content-wrapper > div[id*='elementor-tab-content']` ) + const table = tabs[day].querySelector(`table`) + const data = tabletojson.convert(table.outerHTML) + if (!data) return items + const rows = data[0] - return tabs[day].querySelectorAll(`table > tbody > tr:not(:first-child)`) + return rows + .map(row => { + const time = row['0'].slice(0, 5).trim() + const title = row['0'].replace(time, '').replace(/\s\s+/g, ' ').trim() + + return { time, title } + }) + .filter(i => dayjs(i.time, 'HH:mm').isValid()) }