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

package/lib/v2/readhub/radar.js

@@ -12,6 +12,12 @@ module.exports = {
                     return `/readhub${category ? `/${category}` : ''}`;
                 },
             },
+            {
+                title: '每日早报',
+                docs: 'https://docs.rsshub.app/routes/new-media#readhub-mei-ri-zao-bao',
+                source: ['/daily'],
+                target: '/readhub/daily',
+            },
         ],
     },
 };

package/lib/v2/readhub/router.js

@@ -1,3 +1,4 @@
 module.exports = (router) => {
+    router.get('/daily', require('./daily'));
     router.get('/:category?', require('./'));
 };

package/lib/v2/readhub/util.js

@@ -0,0 +1,69 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
+const dayjs = require('dayjs');
+
+const domain = 'readhub.cn';
+const rootUrl = `https://${domain}`;
+const apiRootUrl = `https://api.${domain}`;
+const apiTopicUrl = new URL('topic/list', apiRootUrl).href;
+
+const formatDate = (date, format) => dayjs(date).format(format);
+const toTopicUrl = (id) => new URL(`topic/${id}`, rootUrl).href;
+
+art.defaults.imports = {
+    ...art.defaults.imports,
+    ...{
+        formatDate,
+        toTopicUrl,
+    },
+};
+
+/**
+ * Process items asynchronously.
+ *
+ * @param {Array<Object>} items - The array of items to process.
+ * @param {function} tryGet - The tryGet function that handles the retrieval process.
+ * @returns {Promise<Array<Object>>} Returns a Promise that resolves to an array of processed items.
+ */
+const processItems = async (items, tryGet) =>
+    await Promise.all(
+        items.map((item) =>
+            tryGet(item.link, async () => {
+                try {
+                    if (!item.link.startsWith(rootUrl)) {
+                        throw Error(`"${item.link}" is an external URL`);
+                    }
+
+                    const { data: detailResponse } = await got(item.link);
+
+                    const data = JSON.parse(detailResponse.match(/\{\\"topic\\":(.*?)\}\]\\n"\]\)<\/script>/)[1].replace(/\\"/g, '"'));
+
+                    item.title = data.title;
+                    item.link = data.url ?? new URL(`topic/${data.uid}`, rootUrl).href;
+                    item.description = art(path.join(__dirname, 'templates/description.art'), {
+                        description: data.summary,
+                        news: data.newsAggList,
+                        timeline: data.timeline,
+                    });
+                    item.author = data.siteNameDisplay;
+                    item.category = [...(data.entityList.map((c) => c.name) ?? []), ...(data.tagList.map((c) => c.name) ?? [])];
+                    item.guid = `readhub-${data.uid}`;
+                    item.pubDate = parseDate(data.publishDate.replace(/\s/g, ''));
+                } catch (e) {
+                    item.guid = `readhub-${item.guid}`;
+                }
+
+                return item;
+            })
+        )
+    );
+
+module.exports = {
+    rootUrl,
+    apiRootUrl,
+    apiTopicUrl,
+    art,
+    processItems,
+};

package/lib/v2/reuters/common.js

@@ -68,60 +68,67 @@ module.exports = async (ctx) => {
         title: e.title,
         link: new URL(e.canonical_url, rootUrl).href,
         guid: e.id,
+        pubDate: parseDate(e.published_time),
+        updated: parseDate(e.updated_time),
+        author: e.authors.map((e) => e.name).join(', '),
+        category: e.kicker.names,
+        description: e.description,
     }));
 
     items = items.filter((e, i) => items.findIndex((f) => e.guid === f.guid) === i);
 
     const results = await Promise.allSettled(
         items.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const detailResponse = await got(item.link);
-                const content = cheerio.load(detailResponse.data);
-
-                if (detailResponse.url.startsWith('https://www.reuters.com/investigates/')) {
-                    const ldJson = JSON.parse(content('script[type="application/ld+json"]').text());
-                    content('.special-report-article-container .container, #slide-dek, #slide-end, .share-in-article-container').remove();
-
-                    item.title = ldJson.headline;
-                    item.pubDate = parseDate(ldJson.dateCreated);
-                    item.author = ldJson.creator;
-                    item.category = ldJson.keywords;
-                    item.description = content('.special-report-article-container').html();
-
-                    return item;
-                }
-
-                const matches = content('script#fusion-metadata')
-                    .text()
-                    .match(/Fusion.globalContent=(\{[\s\S]*?});/);
-
-                if (matches) {
-                    const data = JSON.parse(matches[1]);
-
-                    item.title = data.result.title || item.title;
-                    item.description = art(path.join(__dirname, 'templates/description.art'), {
-                        result: data.result,
-                    });
-                    item.pubDate = parseDate(data.result.display_time);
-                    item.author = data.result.authors.map((author) => author.name).join(', ');
-                    item.category = data.result.taxonomy.keywords;
-
-                    return item;
-                }
-
-                content('.title').remove();
-                content('.article-metadata').remove();
-
-                item.title = content('meta[property="og:title"]').attr('content');
-                item.pubDate = parseDate(detailResponse.data.match(/"datePublished":"(.*?)","dateModified/)[1]);
-                item.author = detailResponse.data
-                    .match(/{"@type":"Person","name":"(.*?)"}/g)
-                    .map((p) => p.match(/"name":"(.*?)"/)[1])
-                    .join(', ');
-                item.description = content('article').html();
-
-                return item;
-            })
+            ctx.query.mode !== 'fulltext'
+                ? item
+                : ctx.cache.tryGet(item.link, async () => {
+                      const detailResponse = await got(item.link);
+                      const content = cheerio.load(detailResponse.data);
+
+                      if (detailResponse.url.startsWith('https://www.reuters.com/investigates/')) {
+                          const ldJson = JSON.parse(content('script[type="application/ld+json"]').text());
+                          content('.special-report-article-container .container, #slide-dek, #slide-end, .share-in-article-container').remove();
+
+                          item.title = ldJson.headline;
+                          item.pubDate = parseDate(ldJson.dateCreated);
+                          item.author = ldJson.creator;
+                          item.category = ldJson.keywords;
+                          item.description = content('.special-report-article-container').html();
+
+                          return item;
+                      }
+
+                      const matches = content('script#fusion-metadata')
+                          .text()
+                          .match(/Fusion.globalContent=(\{[\s\S]*?});/);
+
+                      if (matches) {
+                          const data = JSON.parse(matches[1]);
+
+                          item.title = data.result.title || item.title;
+                          item.description = art(path.join(__dirname, 'templates/description.art'), {
+                              result: data.result,
+                          });
+                          item.pubDate = parseDate(data.result.display_time);
+                          item.author = data.result.authors.map((author) => author.name).join(', ');
+                          item.category = data.result.taxonomy.keywords;
+
+                          return item;
+                      }
+
+                      content('.title').remove();
+                      content('.article-metadata').remove();
+
+                      item.title = content('meta[property="og:title"]').attr('content');
+                      item.pubDate = parseDate(detailResponse.data.match(/"datePublished":"(.*?)","dateModified/)[1]);
+                      item.author = detailResponse.data
+                          .match(/{"@type":"Person","name":"(.*?)"}/g)
+                          .map((p) => p.match(/"name":"(.*?)"/)[1])
+                          .join(', ');
+                      item.description = content('article').html();
+
+                      return item;
+                  })
         )
     );
     items = results.filter((r) => r.status === 'fulfilled').map((r) => r.value);

package/lib/v2/routledge/book-series.js

@@ -0,0 +1,78 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const { join } = require('path');
+
+module.exports = async (ctx) => {
+    const { bookName, bookId } = ctx.params;
+    const baseUrl = 'https://www.routledge.com';
+    const pageUrl = `${baseUrl}/${bookName}/book-series/${bookId}`;
+    const { data: response } = await got(pageUrl, {
+        headers: {
+            accept: 'text/html, */*; q=0.01',
+        },
+        searchParams: {
+            publishedFilter: 'alltitles',
+            pd: 'published,forthcoming',
+            pg: 1,
+            pp: 12,
+            so: 'pub',
+            view: 'list',
+        },
+    });
+    const $ = cheerio.load(response);
+
+    const list = $('.row.book')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            const title = item.find('h3 a');
+            const description = item.find('p.description');
+            const meta = item.find('p.description').prev().text().split('\n');
+            return {
+                title: title.text(),
+                link: title.attr('href'),
+                description: description.text(),
+                pubDate: parseDate(meta.pop().trim(), 'MMMM DD, YYYY'),
+                author: meta
+                    .map((i) => i.trim())
+                    .filter((i) => i)
+                    .join(', '),
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const { data } = await got(item.link);
+                const $ = cheerio.load(data);
+                const isbn = $('meta[property="books:isbn"]').attr('content');
+                const { data: image } = await got('https://productimages.routledge.com', {
+                    searchParams: {
+                        isbn,
+                        size: 'amazon',
+                        ext: 'jpg',
+                    },
+                });
+
+                const description = $('.sticky-div');
+                description.find('button.accordion-button').contents().unwrap();
+                description.find('.fa-shopping-cart').parent().parent().remove();
+
+                item.description = art(join(__dirname, 'templates/description.art'), {
+                    image,
+                    description: description.html(),
+                });
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('head title').text(),
+        description: $('head meta[name="description"]').attr('content'),
+        link: pageUrl,
+        item: items,
+    };
+};

package/lib/v2/routledge/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/:bookName/book-series/:bookId': ['TonyRL'],
+};

package/lib/v2/routledge/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'routledge.com': {
+        _name: 'Routledge',
+        '.': [
+            {
+                title: 'Book Series',
+                docs: 'https://docs.rsshub.app/routes/journals#routledge',
+                source: ['/:bookName/book-series/:bookId'],
+                target: '/routledge/:bookName/book-series/:bookId',
+            },
+        ],
+    },
+};

package/lib/v2/routledge/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/:bookName/book-series/:bookId', require('./book-series'));
+};

package/lib/v2/{qdaily/templates/article.art → routledge/templates/description.art}

@@ -2,4 +2,6 @@
 <img src="{{ image }}"><br>
 {{ /if }}
 
+{{ if description }}
 {{@ description }}
+{{ /if }}

package/lib/v2/sehuatang/index.js

@@ -88,20 +88,18 @@ module.exports = async (ctx) => {
                         $(image).replaceWith($(`<img src="${file}">`));
                     }
                 }
-                // if postMessage does not have any images, try to parse image url from `.pattl`
-                if (images.length === 0) {
-                    const pattl = $('.pattl');
-                    const pattlImages = $(pattl).find('img');
-                    for (const pattlImage of pattlImages) {
-                        const file = $(pattlImage).attr('file');
-                        if (!file || file === 'undefined') {
-                            $(pattlImage).replaceWith('');
-                        } else {
-                            $(pattlImage).replaceWith($(`<img src="${file}" />`));
-                        }
+                // also parse image url from `.pattl`
+                const pattl = $('.pattl');
+                const pattlImages = $(pattl).find('img');
+                for (const pattlImage of pattlImages) {
+                    const file = $(pattlImage).attr('file');
+                    if (!file || file === 'undefined') {
+                        $(pattlImage).replaceWith('');
+                    } else {
+                        $(pattlImage).replaceWith($(`<img src="${file}" />`));
                     }
-                    postMessage.append($(pattl));
                 }
+                postMessage.append($(pattl));
                 $('em[onclick]').remove();
 
                 info.description = (postMessage.html() || '抓取原帖失败').replace(/ignore_js_op/g, 'div');

package/lib/v2/sehuatang/user.js

@@ -62,20 +62,19 @@ module.exports = async (ctx) => {
                         $(image).replaceWith($(`<img src="${file}">`));
                     }
                 }
-                // if postMessage does not have any images, try to parse image url from `.pattl`
-                if (images.length === 0) {
-                    const pattl = $('.pattl');
-                    const pattlImages = $(pattl).find('img');
-                    for (const pattlImage of pattlImages) {
-                        const file = $(pattlImage).attr('file');
-                        if (!file || file === 'undefined') {
-                            $(pattlImage).replaceWith('');
-                        } else {
-                            $(pattlImage).replaceWith($(`<img src="${file}" />`));
-                        }
+                // also parse image url from `.pattl`
+                const pattl = $('.pattl');
+                const pattlImages = $(pattl).find('img');
+                for (const pattlImage of pattlImages) {
+                    const file = $(pattlImage).attr('file');
+                    if (!file || file === 'undefined') {
+                        $(pattlImage).replaceWith('');
+                    } else {
+                        $(pattlImage).replaceWith($(`<img src="${file}" />`));
                     }
-                    postMessage.append($(pattl));
                 }
+                postMessage.append($(pattl));
+
                 $('em[onclick]').remove();
 
                 info.description = (postMessage.html() || '抓取原帖失败').replace(/ignore_js_op/g, 'div');

package/lib/v2/showstart/artist.js

@@ -0,0 +1,15 @@
+const { TITLE, HOST } = require('./const');
+const { fetchPerformerInfo } = require('./service');
+
+module.exports = async (ctx) => {
+    const id = ctx.params.id;
+    const artist = await fetchPerformerInfo({
+        performerId: id,
+    });
+    ctx.state.data = {
+        title: `${TITLE} - ${artist.name}`,
+        description: artist.content,
+        link: `${HOST}/artist/${artist.id}`,
+        item: artist.activityList,
+    };
+};

package/lib/v2/showstart/brand.js

@@ -0,0 +1,15 @@
+const { TITLE, HOST } = require('./const');
+const { fetchBrandInfo } = require('./service');
+
+module.exports = async (ctx) => {
+    const id = ctx.params.id;
+    const brand = await fetchBrandInfo({
+        brandId: id,
+    });
+    ctx.state.data = {
+        title: `${TITLE} - ${brand.name}`,
+        description: brand.content,
+        link: `${HOST}/host/${brand.id}`,
+        item: brand.activityList,
+    };
+};

package/lib/v2/showstart/const.js

@@ -0,0 +1,4 @@
+module.exports = {
+    HOST: 'https://www.showstart.com',
+    TITLE: '秀动网',
+};

package/lib/v2/showstart/event.js

@@ -0,0 +1,18 @@
+const { TITLE, HOST } = require('./const');
+const { fetchActivityList, fetchDictionary } = require('./service');
+
+module.exports = async (ctx) => {
+    const cityCode = parseInt(ctx.params.cityCode);
+    const showStyle = parseInt(ctx.params.showStyle);
+    const items = await fetchActivityList({
+        cityCode,
+        showStyle,
+    });
+    const { cityName, showName } = await fetchDictionary(cityCode, showStyle);
+    const tags = [cityName, showName].filter((item) => Boolean(item)).join(' - ');
+    ctx.state.data = {
+        title: `${TITLE} - ${tags}`,
+        link: HOST,
+        item: items,
+    };
+};

package/lib/v2/showstart/maintainer.js

@@ -0,0 +1,6 @@
+module.exports = {
+    '/artist/:id': ['lchtao26'],
+    '/brand/:id': ['lchtao26'],
+    '/event/:cityCode/:showStyle?': ['lchtao26'],
+    '/search/:type/:keyword?': ['lchtao26'],
+};

package/lib/v2/showstart/radar.js

@@ -0,0 +1,40 @@
+module.exports = {
+    'showstart.com': {
+        _name: '秀动网',
+        www: [
+            {
+                title: '演出更新',
+                docs: 'https://docs.rsshub.app/routes/shopping#xiu-dong-wang-yan-chu-geng-xin',
+                source: ['/event/list'],
+                target: (_, url) => {
+                    const search = new URL(url).searchParams;
+                    const cityCode = search.get('cityCode') || 0;
+                    const showStyle = search.get('showStyle') || 0;
+                    return `/showstart/event/${cityCode}/${showStyle}`;
+                },
+            },
+            {
+                title: '演出搜索',
+                docs: 'https://docs.rsshub.app/routes/shopping#xiu-dong-wang-yan-chu-sou-suo',
+                source: ['/event/list'],
+                target: (_, url) => {
+                    const search = new URL(url).searchParams;
+                    const keyword = search.get('keyword') || '';
+                    return `/showstart/search/event/${keyword}`;
+                },
+            },
+            {
+                title: '音乐人 - 演出更新',
+                docs: 'https://docs.rsshub.app/routes/shopping#yin-yue-ren-yan-chu-geng-xin',
+                source: ['/artist/:id'],
+                target: '/showstart/artist/:id',
+            },
+            {
+                title: '厂牌 - 演出更新',
+                docs: 'https://docs.rsshub.app/routes/shopping#chang-pai-yan-chu-geng-xin',
+                source: ['/host/:id'],
+                target: '/showstart/brand/:id',
+            },
+        ],
+    },
+};

package/lib/v2/showstart/router.js

@@ -0,0 +1,6 @@
+module.exports = (router) => {
+    router.get('/artist/:id', require('./artist'));
+    router.get('/brand/:id', require('./brand'));
+    router.get('/event/:cityCode/:showStyle?', require('./event'));
+    router.get('/search/:type/:keyword?', require('./search'));
+};

package/lib/v2/showstart/search.js

@@ -0,0 +1,51 @@
+const { TITLE, HOST } = require('./const');
+const { fetchActivityList, fetchPerformerList, fetchBrandList, fetchCityList, fetchStyleList } = require('./service');
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type || '';
+    const keyword = ctx.params.keyword || '';
+
+    switch (type) {
+        case 'event':
+            ctx.state.data = {
+                title: `${TITLE} - 搜演出 - ${keyword || '全部'}`,
+                link: HOST,
+                item: await fetchActivityList({ keyword }),
+            };
+            break;
+        case 'artist':
+            ctx.state.data = {
+                title: `${TITLE} - 搜艺人 - ${keyword || '全部'}`,
+                link: HOST,
+                item: await fetchPerformerList({ searchKeyword: keyword }),
+            };
+            break;
+        case 'brand':
+            ctx.state.data = {
+                title: `${TITLE} - 搜厂牌 - ${keyword || '全部'}`,
+                link: HOST,
+                item: await fetchBrandList({ searchKeyword: keyword }),
+            };
+            break;
+        case 'city':
+            ctx.state.data = {
+                title: `${TITLE} - 搜城市 - ${keyword || '全部'}`,
+                link: HOST,
+                item: await fetchCityList(keyword),
+            };
+            break;
+        case 'style':
+            ctx.state.data = {
+                title: `${TITLE} - 搜风格 - ${keyword || '全部'}`,
+                link: HOST,
+                item: await fetchStyleList(keyword),
+            };
+            break;
+        default:
+            ctx.state.data = {
+                title: `${TITLE} - 搜演出 - ${type || '全部'}`,
+                link: HOST,
+                item: await fetchActivityList({ keyword: type }),
+            };
+    }
+};