rsshub 1.0.0-master.f7cdf8b → 1.0.0-master.f812e91

package/lib/routes/boston/index.js

@@ -1,73 +0,0 @@
-const cheerio = require('cheerio');
-const parser = require('@/utils/rss-parser');
-const got = require('@/utils/got');
-
-module.exports = async (ctx) => {
-    const tag = ctx.params.tag;
-    const rssUrl = tag ? `https://www.boston.com/tag/${tag}/?feed=atom` : `https://www.boston.com/?feed=atom`;
-    const feed = await parser.parseURL(rssUrl);
-
-    const items = await Promise.all(
-        feed.items.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const response = await got({
-                    url: item.link,
-                    method: 'get',
-                });
-                const html = response.body;
-                const $ = cheerio.load(html);
-                const content = $('div.content-text-article');
-
-                // Disable image lazyloading
-                content.find('img').each((i, e) => {
-                    e = $(e);
-                    e.attr('src', e.attr('srcset'));
-                });
-
-                // Categories
-                const categories = [];
-                $('meta[property="article:tag"]').each((i, e) => {
-                    categories.push(e.attribs.content);
-                });
-
-                // Updated date
-                const updatedAt = $('meta[property="article:modified_time"]').attr().content;
-
-                // Summary
-                const summary = $('meta[property="og:description"]').attr().content;
-
-                // Remove unwanted DOMs
-                const unwanted_element_selectors = ['.ad-container', '.content-toaster--newsletter'];
-                unwanted_element_selectors.forEach((selector) => {
-                    content.find(selector).each((i, e) => {
-                        $(e).remove();
-                    });
-                });
-
-                return {
-                    title: item.title.trim(),
-                    id: item.link,
-                    pubDate: new Date(item.pubDate).toUTCString(),
-                    updated: new Date(updatedAt).toUTCString(),
-                    author: item.author.trim(),
-                    link: item.link,
-                    summary: summary.trim(),
-                    description: content.html(),
-                    category: categories,
-                    icon: 'https://www.boston.com/wp-content/themes/bdc/images/favicons/largetile.png',
-                    logo: 'https://www.boston.com/wp-content/themes/bdc/images/favicons/largetile.png',
-                };
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: feed.title,
-        link: feed.link,
-        description: feed.description,
-        item: items,
-        language: 'en-us',
-        icon: 'https://www.boston.com/wp-content/themes/bdc/images/favicons/largetile.png',
-        logo: 'https://www.boston.com/wp-content/themes/bdc/images/favicons/largetile.png',
-    };
-};

package/lib/routes/buaq/index.js

@@ -1,31 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const baseUrl = 'https://buaq.net';
-
-    const response = await got({
-        method: 'get',
-        url: baseUrl,
-    });
-    const $ = cheerio.load(response.data); // 使用 cheerio 加载返回的 HTML
-    const list = $('tr[class]');
-
-    ctx.state.data = {
-        title: `不安全文章 ~ 资讯`,
-        link: `https://buaq.net/`,
-        description: '不安全文章 ~ 资讯',
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-                    return {
-                        title: $('font[size="4"]', item).children('a').text(),
-                        pubDate: new Date($('font[size="1"]', item).children().first().text()).toUTCString(),
-                        link: baseUrl + $('font[size="4"]', item).children('a').attr('href'),
-                    };
-                })
-                .get(),
-    };
-};

package/lib/routes/cgtn/pick.js

@@ -1,45 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const rootUrl = 'https://www.cgtn.com/';
-    const response = await got({
-        method: 'get',
-        url: rootUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-    const list = $('.titleArea')
-        .map((_, item) => {
-            item = $(item);
-            const a = item.find('a');
-            return {
-                title: a.text(),
-                link: a.attr('href'),
-                pubDate: new Date(parseInt(a.attr('data-time'))).toUTCString(),
-            };
-        })
-        .get();
-
-    const items = await Promise.all(
-        list.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const detailResponse = await got({
-                    method: 'get',
-                    url: item.link,
-                });
-                const content = cheerio.load(detailResponse.data);
-
-                item.description = content('#cmsMainContent').html();
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: "CGTN - Editors' Pick",
-        link: rootUrl,
-        item: items,
-    };
-};

package/lib/routes/cgtn/top.js

@@ -1,45 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const rootUrl = 'https://www.cgtn.com/';
-    const response = await got({
-        method: 'get',
-        url: rootUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-    const list = $('div.topNews-item')
-        .map((_, item) => {
-            item = $(item);
-            const a = item.find('div.topNews-item-content-title a');
-            return {
-                title: a.text(),
-                link: a.attr('href'),
-                pubDate: new Date(parseInt(a.attr('data-time'))).toUTCString(),
-            };
-        })
-        .get();
-
-    const items = await Promise.all(
-        list.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const detailResponse = await got({
-                    method: 'get',
-                    url: item.link,
-                });
-                const content = cheerio.load(detailResponse.data);
-
-                item.description = content('#cmsMainContent').html();
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: 'CGTN - Top News',
-        link: rootUrl,
-        item: items,
-    };
-};

package/lib/routes/chinadialogue/column.js

@@ -1,58 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const column = ctx.params.column;
-
-    const response = await got({
-        method: 'get',
-        url: `https://www.chinadialogue.net/${column}`,
-    });
-
-    const data = response.data;
-
-    const $ = cheerio.load(data);
-    const list = $('article.main-listing');
-    let itemPicUrl;
-    let title;
-    let description;
-
-    ctx.state.data = {
-        title: $('title').text(),
-        link: `https://www.chinadialogue.net/${column}`,
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-                    itemPicUrl = item.find('img.img-responsive').attr('src');
-
-                    title = [];
-                    description = [];
-
-                    item.find('h1').each((index, element) => {
-                        title.push(
-                            $(element)
-                                .text()
-                                .replace(/(\r\n|\n|\r)/gm, '')
-                        );
-                    });
-
-                    item.find('p').each((index, element) => {
-                        description.push(
-                            $(element)
-                                .text()
-                                .replace(/(\r\n|\n|\r)/gm, '')
-                        );
-                    });
-
-                    return {
-                        title: title.join(' | '),
-                        description: `<p>${description.join('</p><p>')}</p><img src="${itemPicUrl}">`,
-                        pubDate: item.find('time.blog').attr('datetime'),
-                        link: item.find('a').attr('href'),
-                    };
-                })
-                .get(),
-    };
-};

package/lib/routes/chinadialogue/topics.js

@@ -1,55 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const topic = ctx.params.topic;
-
-    const response = await got({
-        method: 'get',
-        url: `https://www.chinadialogue.net/topics/${topic}`,
-    });
-
-    const data = response.data;
-
-    const $ = cheerio.load(data);
-    const list = $('article.excerpt.col-md-6');
-    let itemPicUrl;
-    let title;
-    let description;
-
-    ctx.state.data = {
-        title: $('title')
-            .text()
-            .replace(/( 主题 >| Topics >)/gm, ''),
-        link: `https://www.chinadialogue.net/topics/${topic}`,
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-                    itemPicUrl = item.find('img.img-responsive').attr('src');
-
-                    title = [];
-                    description = [];
-
-                    item.find('h1').each((index, element) => {
-                        title.push(
-                            $(element)
-                                .text()
-                                .replace(/(\r\n|\n|\r)/gm, '')
-                        );
-                    });
-
-                    item.find('p').each((index, element) => {
-                        description.push($(element).text());
-                    });
-
-                    return {
-                        title: title.join(' | '),
-                        description: `<p>${description.join('</p><p>')}</p><img src="${itemPicUrl}">`,
-                        link: item.find('a').attr('href'),
-                    };
-                })
-                .get(),
-    };
-};

package/lib/routes/chinalaborwatch/reports.js

@@ -1,53 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const lang = ctx.params.lang || '';
-    const industry = ctx.params.industry || '';
-
-    const rootUrl = 'http://www.chinalaborwatch.org';
-    const currentUrl = `${rootUrl}/${lang === '' ? '' : 'cn/'}reports${industry === '' ? '' : `/industry/${industry}`}`;
-
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-    const $ = cheerio.load(response.data);
-
-    const list = $('.Alt1 a, .Alt2 a')
-        .slice(0, 10)
-        .map((_, item) => {
-            item = $(item);
-            return {
-                title: item.text(),
-                link: `${rootUrl}${item.attr('href')}`,
-            };
-        })
-        .get();
-
-    const items = await Promise.all(
-        list.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const detailResponse = await got({
-                    method: 'get',
-                    url: item.link,
-                });
-                const content = cheerio.load(detailResponse.data);
-
-                item.pubDate = new Date(content('#ContentPlaceHolder1_LabelDate').text().replace(/年|月/g, '-').replace('日', '')).toUTCString();
-
-                content('h1, #ContentPlaceHolder1_LabelDate').remove();
-
-                item.description = content('.mainContent').html();
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: `${$('title').text()} - ${lang === '' ? 'China Labour Watch' : '中国劳工观察'}`,
-        link: currentUrl,
-        item: items,
-    };
-};

package/lib/routes/chinatimes/index.js

@@ -1,115 +0,0 @@
-const url = require('url');
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-const root_url = 'https://www.chinatimes.com';
-
-const config = {
-    realtimenews: {
-        link: '/realtimenews/?chdtv',
-        title: 'China Times - Real Time News',
-    },
-    politic: {
-        link: '/politic/?chdtv',
-        title: 'China Times - Politic',
-    },
-    opinion: {
-        link: '/opinion/?chdtv',
-        title: 'China Times - Opinion',
-    },
-    life: {
-        link: '/life/?chdtv',
-        title: 'China Times - Life',
-    },
-    star: {
-        link: '/star/?chdtv',
-        title: 'China Times - Showbiz',
-    },
-    money: {
-        link: '/money/?chdtv',
-        title: 'China Times - Finance',
-    },
-    society: {
-        link: '/society/?chdtv',
-        title: 'China Times - Society',
-    },
-    hottopic: {
-        link: '/hottopic/?chdtv',
-        title: 'China Times - Hot Topics',
-    },
-    tube: {
-        link: '/tube/?chdtv',
-        title: 'China Times - Videos',
-    },
-    world: {
-        link: '/world/?chdtv',
-        title: 'China Times - World',
-    },
-    armament: {
-        link: '/armament/?chdtv',
-        title: 'China Times - Military',
-    },
-    chinese: {
-        link: '/chinese/?chdtv',
-        title: 'China Times - Mainland & Taiwan',
-    },
-    fashion: {
-        link: '/fashion/?chdtv',
-        title: 'China Times - Fashion',
-    },
-    sports: {
-        link: '/sports/?chdtv',
-        title: 'China Times - Sports',
-    },
-    technologynews: {
-        link: '/technologynews/?chdtv',
-        title: 'China Times - Technology News',
-    },
-    travel: {
-        link: '/travel/?chdtv',
-        title: 'China Times - Travel',
-    },
-    album: {
-        link: '/album/?chdtv',
-        title: 'China Times - Columns',
-    },
-};
-
-module.exports = async (ctx) => {
-    const cfg = config[ctx.params.caty];
-    if (!cfg) {
-        throw Error('Bad category. See <a href="https://docs.rsshub.app/routes/traditional-media#zhong-shi-dian-zi-bao">https://docs.rsshub.app/routes/traditional-media#zhong-shi-dian-zi-bao</a>');
-    }
-
-    const current_url = url.resolve(root_url, cfg.link);
-    const response = await got({
-        method: 'get',
-        url: current_url,
-    });
-
-    const $ = cheerio.load(response.data);
-    const list = $('section.article-list ul div.row')
-        .map((_, item) => {
-            item = $(item);
-            const right = item.find('div.col');
-            const a = right.find('h3.title a');
-            const img = item.find('img.photo').attr('src');
-            let desc = '<p>' + right.find('p.intro').text() + '...</p>';
-            if (img) {
-                desc = `<p><img src="${img}"></p>` + desc;
-            }
-            return {
-                title: a.text(),
-                link: a.attr('href'),
-                pubDate: new Date(right.find('div.meta-info time').attr('datetime') + ' GMT+8').toUTCString(),
-                description: desc,
-            };
-        })
-        .get();
-
-    ctx.state.data = {
-        title: cfg.title,
-        link: current_url,
-        item: list,
-    };
-};

package/lib/routes/cmes/news.js

@@ -1,64 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const category = ctx.params.category || 'Information';
-
-    const rootUrl = 'https://www.cmes.org';
-    const currentUrl = `${rootUrl}/News/${category}/index.html`;
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-
-    const list = $('.r_ct')
-        .find('a')
-        .slice(0, 10)
-        .map((_, item) => {
-            item = $(item);
-
-            const link = item.attr('href');
-
-            let date = '00000000';
-            if (link.split('/')[2] !== 'mkc.ckcest.cn') {
-                date = link.split('/')[5];
-            }
-
-            return {
-                link,
-                title: item.text(),
-                pubDate: new Date(`${date.substr(0, 4)}-${date.substr(5, 2)}-${date.substr(7, 2)}`).toDateString(),
-            };
-        })
-        .get();
-
-    const items = await Promise.all(
-        list.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const detailResponse = await got({
-                    method: 'get',
-                    url: item.link,
-                });
-                const content = cheerio.load(detailResponse.data);
-
-                content('h3, script, .bshare-custom').remove();
-
-                item.description = content('.xhjj').html() || content('.detail-page').html();
-
-                if (content('#articleTime').html()) {
-                    item.pubDate = new Date(content('#articleTime').text()).toUTCString();
-                }
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: $('title').text(),
-        link: currentUrl,
-        item: items,
-    };
-};

package/lib/routes/commonapp/blog.js

@@ -1,42 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const { parseDate } = require('@/utils/parse-date');
-
-module.exports = async (ctx) => {
-    const rootUrl = 'https://www.commonapp.org';
-    const jsonUrl = `${rootUrl}/page-data/sq/d/4283793559.json`;
-    const response = await got({
-        method: 'get',
-        url: jsonUrl,
-    });
-
-    const list = response.data.data.allNodeBlogPost.edges.map((item) => ({
-        title: item.node.title,
-        author: item.node.field_blog_author,
-        pubDate: parseDate(item.node.created),
-        link: `${rootUrl}${item.node.path.alias}`,
-    }));
-
-    const items = await Promise.all(
-        list.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const detailResponse = await got({
-                    method: 'get',
-                    url: item.link,
-                });
-
-                const content = cheerio.load(detailResponse.data);
-
-                item.description = content('.inner-page').html();
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: 'Blog - Common App',
-        link: `${rootUrl}/blog`,
-        item: items,
-    };
-};

package/lib/routes/cria/news.js

@@ -1,57 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const id = ctx.params.id || '';
-
-    const rootUrl = 'http://www.cria.org.cn';
-    const currentUrl = `${rootUrl}/${id ? `newslist/${id}.html` : 'news'}`;
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-
-    const list = $('.ch_news,.enterprise_news_c')
-        .find('ul li')
-        .map((_, item) => {
-            item = $(item);
-            const a = item.find('a');
-
-            return {
-                link: `${rootUrl}${a.attr('href')}`,
-            };
-        })
-        .get();
-
-    const items = await Promise.all(
-        list.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                try {
-                    const detailResponse = await got({
-                        method: 'get',
-                        url: item.link,
-                    });
-                    const content = cheerio.load(detailResponse.data);
-
-                    item.title = content('title').text();
-                    item.description = content('.news_details_c').html();
-                    item.pubDate = new Date(detailResponse.data.match(/<span>发布时间：(.*)<\/span><span>来源:/)[1]).toUTCString();
-
-                    return item;
-                } catch (e) {
-                    return Promise.resolve('');
-                }
-            })
-        )
-    );
-
-    const title = id ? response.data.match(/<h3><span style="margin: 10px; font-size:15px;">当前位置：中国橡胶网 > (.*)<\/span><\/h3>/)[1] : '新闻资讯';
-
-    ctx.state.data = {
-        title: `${title} - 中国橡胶网`,
-        link: currentUrl,
-        item: items,
-    };
-};