@phunky/scrape-channel-listings
Version:
A TypeScript library for scraping TV channel listings from various providers
52 lines (51 loc) • 1.72 kB
JavaScript
;
/**
* Sky Satellite Channel Listing Scraper
* Extracts channel numbers and names from Sky Q or Sky+HD channel lineup
*/
Object.defineProperty(exports, "__esModule", { value: true });
const scraper_1 = require("../utils/scraper");
/**
* Extracts channel information from the Sky channel listing page
* @param page - Playwright page instance
* @returns Array of partial channel objects containing number and name
*/
const scrapeFunction = async (page) => {
return await page.$$eval('table tbody tr', (rows) => {
return rows.map((row) => {
const number = row.querySelector('.column-1')?.textContent?.trim() || '';
const name = row.querySelector('.column-2')?.textContent?.trim() || '';
// Skip those with dashes in numbers as these are category definitions
if (!name || !number || number.includes('-')) {
return {};
}
return { number, name };
});
});
};
/**
* Channel name overrides to standardize naming across providers
*/
const overrides = {
'Sky Cinema Sci-Fi/Horror HD': 'SKY CINEMA SCFI/HORROR HD',
'RTÉ One HD': 'RTE One',
'RTÉ 2 HD': 'RTE 2',
'RTÉjr': 'RTE Junior',
'RTÉ News': 'RTE News',
'Sky Sports UHD 1': 'Sky Sports 1 UHD',
'Sky Sports UHD 2': 'Sky Sports 2 UHD ',
};
/**
* Sky UK scraper configuration
*/
const config = {
url: 'https://rxtvinfo.com/sky-satellite-channel-list-uk/',
scrapeFunction,
overrides,
outputFile: 'sky-ireland.json'
};
// Run scraper if this file is executed directly
if (require.main === module) {
(0, scraper_1.runScraperCLI)(config).catch(() => process.exit(1));
}
exports.default = config;