bandcamp-fetch/dist/cjs/lib/article/ArticleListParser.js

Scrape Bandcamp content

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
const cheerio_1 = require("cheerio");
const Constants_js_1 = require("../utils/Constants.js");
const Parse_js_1 = require("../utils/Parse.js");
class ArticleListParser {
    static parseList(html, opts) {
        const $ = (0, cheerio_1.load)(html);
        const dailyUrl = Constants_js_1.URLS.DAILY;
        const result = {
            articles: [],
            total: 0,
            start: 0,
            end: 0
        };
        $('articles-list').each((i, list) => {
            $('.list-article', $(list)).each((i, article) => {
                article = $(article);
                const imgSrc = article.find('img').attr('src') || null;
                // Category
                const infoText = article.find('.article-info-text');
                const infoTextCategoryLink = infoText.find('a.franchise');
                const infoTextMiddot = infoText.find('.middot');
                const categoryName = infoTextCategoryLink.text();
                const category = categoryName ? {
                    name: categoryName
                } : null;
                const categoryUrl = infoTextCategoryLink.attr('href');
                if (category && categoryUrl) {
                    category.url = (0, Parse_js_1.isAbsoluteUrl)(categoryUrl) ? categoryUrl : (0, Parse_js_1.normalizeUrl)(categoryUrl, dailyUrl);
                }
                // Date
                infoTextCategoryLink.remove();
                infoTextMiddot.remove();
                const date = (0, Parse_js_1.stripLineBreaks)(infoText.text()).trim();
                // Title and url
                const titleLink = article.find('a.title');
                const title = titleLink.text();
                let url = titleLink.attr('href');
                if (!(0, Parse_js_1.isAbsoluteUrl)(url)) {
                    url = (0, Parse_js_1.normalizeUrl)(url, dailyUrl);
                }
                if (titleLink) {
                    const parsed = {
                        url,
                        title,
                        date
                    };
                    if (category) {
                        parsed.category = category;
                    }
                    const imageUrl = (0, Parse_js_1.reformatImageUrl)(imgSrc, opts.imageFormat);
                    if (imageUrl) {
                        parsed.imageUrl = imageUrl;
                    }
                    result.articles.push(parsed);
                }
            });
        });
        const resultsText = (0, Parse_js_1.stripLineBreaks)($('#num-results').text()).trim();
        const rtm = resultsText.match(/(\d+)(?:\s*to\s*)(\d+)(?:\s*of\s*)(\d+)/);
        if (rtm?.length === 4) {
            result.total = parseInt(rtm[3], 10);
            result.start = parseInt(rtm[1], 10);
            result.end = parseInt(rtm[2], 10);
        }
        return result;
    }
}
exports.default = ArticleListParser;
//# sourceMappingURL=ArticleListParser.js.map