fix(route): Readhub (#13567)

* fix(route): Readhub * fix docs
2023-10-20 04:51:34 +08:00 · 2023-10-20 04:51:34 +08:00 · b1f206fcdf
parent 9ef53d597d
commit b1f206fcdf
6 changed files with 147 additions and 97 deletions
--- a/lib/v2/readhub/index.js
+++ b/lib/v2/readhub/index.js
@ -1,106 +1,115 @@
 const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const timezone = require('@/utils/timezone');
 const { parseDate } = require('@/utils/parse-date');
-
-const titles = {
-    topic: '热门话题',
-    news: '科技动态',
-    tech: '技术资讯',
-    technews: '技术资讯',
-    blockchain: '区块链快讯',
-    daily: '每日早报',
-};
+const { art } = require('@/utils/render');
+const path = require('path');
+const dayjs = require('dayjs');

 module.exports = async (ctx) => {
-    let category = 'topic';
-    let overview = false;
-    if (titles[ctx.params.category]) {
-        category = ctx.params.category;
-        overview = !!ctx.params.overview;
-        category = category === 'tech' ? 'technews' : category;
-    } else if (ctx.params.category) {
-        overview = true;
-    }
+    const { category } = ctx.params;
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit, 10) : 20;

-    const rootUrl = 'https://readhub.cn';
-    const apiRootUrl = 'https://api.readhub.cn';
-    const currentUrl = `${rootUrl}/topic/${category}`;
-    const apiUrl = `${apiRootUrl}/${category === 'daily' ? 'topic/daily' : category}`;
+    const domain = 'readhub.cn';
+    const rootUrl = `https://${domain}`;
+    const apiRootUrl = `https://api.${domain}`;
+    const currentUrl = new URL(category ?? '', rootUrl).href;

-    const response = await got({
-        method: 'get',
-        url: apiUrl,
+    const isCategory = !category || category === 'daily';
+
+    const formatDate = (date, format) => dayjs(date).format(format);
+    const toTopicUrl = (id) => new URL(`topic/${id}`, rootUrl).href;
+
+    art.defaults.imports = {
+        ...art.defaults.imports,
+        ...{
+            formatDate,
+            toTopicUrl,
+        },
+    };
+
+    const { data: currentResponse } = await got(currentUrl);
+
+    const buildId = currentResponse.match(/"buildId":"(\w+)"/)[1];
+
+    const apiUrl = new URL(category ? (category === 'daily' ? `_next/data/${buildId}/daily.json` : 'news/list') : 'topic/list', category === 'daily' ? rootUrl : apiRootUrl).href;
+
+    const { data: response } = await got(apiUrl, {
+        searchParams: {
+            ...(isCategory
+                ? {}
+                : {
+                      type: currentResponse.match(/"query":\{"type":"(\d+)"\}/)[1],
+                  }),
+            ...{
+                page: 1,
+                size: limit,
+            },
+        },
    });

-    const list = response.data.data.map((item) => ({
-        id: item.id,
+    let items = (response.data?.items ?? response.pageProps.daily).slice(0, limit).map((item) => ({
        title: item.title,
-        pubDate: parseDate(item.publishDate),
-        description: item.summaryAuto || item.summary || '',
-        link: item.url || item.mobileUrl || `${rootUrl}/topic/${item.id}`,
-        author: item.authorName || item.siteName || '',
-        hasInstantView: item.hasInstantView || false,
+        link: item.url ?? new URL(`topic/${item.uid}`, rootUrl).href,
+        description: art(path.join(__dirname, 'templates/description.art'), {
+            description: item.summary,
+            news: item.newsAggList,
+            timeline: item.timeline,
+        }),
+        author: item.siteNameDisplay,
+        category: item.tagList?.map((c) => c.name) ?? [],
+        guid: isCategory ? item.uid : `readhub-${item.guid}`,
+        pubDate: timezone(parseDate(item.publishDate), +8),
    }));

-    const items = await Promise.all(
-        list.map((item) =>
-            ctx.cache.tryGet(item.id + overview, async () => {
-                if (item.hasInstantView || category === 'daily') {
-                    const { data } = await got({
-                        method: 'get',
-                        url: `${apiRootUrl}/topic/instantview?topicId=${item.id}`,
+    if (isCategory) {
+        items = await Promise.all(
+            items.map((item) =>
+                ctx.cache.tryGet(`readhub-${item.guid}`, async () => {
+                    const { data: detailResponse } = await got(new URL(`_next/data/${buildId}/topic/${item.guid}.json`, rootUrl).href);
+
+                    const data = detailResponse.pageProps.topic;
+
+                    item.title = data.title;
+                    item.link = data.url ?? new URL(`topic/${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.entityList.map((a) => a.name).join('/');
+                    item.category = data.tagList.map((c) => c.name);
+                    item.guid = `readhub-${data.uid}`;
+                    item.pubDate = timezone(parseDate(data.publishDate), +8);

-                    item.description = `<a href="${data.url}">访问原网址</a><br>${!overview ? data.content : ''}`;
-                }
+                    return item;
+                })
+            )
+        );
+    }

-                if (isNaN(item.id)) {
-                    const { data } = await got({
-                        method: 'get',
-                        url: `${apiRootUrl}/topic/${item.id}`,
-                    });
+    const $ = cheerio.load(currentResponse);

-                    if (data.summary && overview) {
-                        item.description += `${data.summary}<hr>`;
-                    }
-
-                    if (data.newsArray) {
-                        item.description += '<br><b>媒体报道</b><ul>';
-
-                        for (const news of data.newsArray) {
-                            item.description +=
-                                `<li><a href="${news.url || news.mobileUrl}">${news.title}</a>` +
-                                (news.siteName ? `&nbsp;&nbsp;<small><b>${news.siteName}</b></small>` : '') +
-                                `${news.publishDate ? `&nbsp;&nbsp;<small>${news.publishDate.split('T')[0]}</small>` : ''}</li>`;
-                        }
-
-                        item.description += '</ul>';
-                    }
-
-                    if (data.timeline && data.timeline.topics) {
-                        const hasTimeline = data.timeline.commonEntities && data.timeline.commonEntities.length > 0;
-
-                        item.description += `<br><b>${hasTimeline ? '事件追踪' : '相关事件'}</b><ul>`;
-
-                        for (const news of data.timeline.topics) {
-                            const createdAt = news.createdAt ? `<small>${news.createdAt.split('T')[0]}</small>` : '';
-
-                            item.description += `<li>${hasTimeline ? `${createdAt}&nbsp;&nbsp;` : ''}<a href="${rootUrl}/topic/${news.id}">${news.title}</a>${hasTimeline ? '' : `&nbsp;&nbsp;${createdAt}`}</li>`;
-                        }
-
-                        item.description += '</ul>';
-                    }
-                }
-                delete item.id;
-                delete item.hasInstantView;
-
-                return item;
-            })
-        )
-    );
+    const author = $('meta[property="og:site_name"]').prop('content');
+    const subtitle = $(
+        $('nav div a')
+            .toArray()
+            .filter((a) => /style_active.*/.test($(a).prop('class')))
+            .pop()
+    ).text();
+    const icon = new URL($('link[rel="apple-touch-icon"]').prop('href'), rootUrl).href;

    ctx.state.data = {
-        title: `Readhub - ${titles[category]}`,
-        link: currentUrl,
        item: items,
+        title: `${author} - ${subtitle}`,
+        link: currentUrl,
+        description: $('meta[property="og:description"]').prop('content'),
+        language: $('html').prop('lang'),
+        image: $('meta[property="og:image"]').prop('content'),
+        icon,
+        logo: icon,
+        subtitle,
+        author,
+        allowEmpty: true,
    };
 };
--- a/lib/v2/readhub/maintainer.js
+++ b/lib/v2/readhub/maintainer.js
@ -1,3 +1,3 @@
 module.exports = {
-    '/:category?/:overview?': ['WhiteWorld', 'nczitzk', 'Fatpandac'],
+    '/:category?': ['WhiteWorld', 'nczitzk', 'Fatpandac'],
 };
--- a/lib/v2/readhub/radar.js
+++ b/lib/v2/readhub/radar.js
@ -4,9 +4,13 @@ module.exports = {
        '.': [
            {
                title: '分类',
-                docs: 'https://docs.rsshub.app/routes/new-media#readhub',
+                docs: 'https://docs.rsshub.app/routes/new-media#readhub-fen-lei',
                source: ['/', '/:category'],
-                target: (params) => `/readhub/${params.category ? params.category : ''}`,
+                target: (params) => {
+                    const category = params.category;
+
+                    return `/readhub${category ? `/${category}` : ''}`;
+                },
            },
        ],
    },
--- a/lib/v2/readhub/router.js
+++ b/lib/v2/readhub/router.js
@ -1,6 +1,3 @@
 module.exports = (router) => {
-    // deprecated
-    router.get('/category/:category?/:overview?', require('./index'));
-
-    router.get('/:category?/:overview?', require('./index'));
+    router.get('/:category?', require('./'));
 };
--- a/lib/v2/readhub/templates/description.art
+++ b/lib/v2/readhub/templates/description.art
@ -0,0 +1,40 @@
+{{ if description }}
+  <p>{{ description }}</p>
+{{ /if }}
+
+{{ if news }}
+  <h3>媒体报道</h3>
+  <table cellspacing="8">
+    <tbody>
+      {{ each news n }}
+        <tr>
+          <td>
+            <a href="{{ n.url }}">{{ n.title }}</a>
+          </td>
+          <th align="left">
+            <small>{{ n.siteNameDisplay }}</small>
+          </th>
+        </tr>
+      {{ /each }}
+    </tbody>
+  </table>
+{{ /if }}
+
+{{ if timeline }}
+  <h3>事件追踪</h3>
+  <table cellspacing="10">
+    <tbody>
+      {{ set topics = timeline.topics }}
+      {{ each topics t }}
+        <tr>
+          <th align="left">
+            <small>{{ t.publishDate | formatDate 'YYYY-MM-DD HH:mm:ss' }}</small>
+          </th>
+          <td>
+            <a href="{{ t.uid | toTopicUrl }}">{{ t.title }}</a>
+          </td>
+        </tr>
+      {{ /each }}
+    </tbody>
+  </table>
+{{ /if }}
--- a/website/docs/routes/new-media.md
+++ b/website/docs/routes/new-media.md
@ -1708,11 +1708,11 @@ Compared to the official one, this feed:

 ### 分类 {#readhub-fen-lei}

-<Route author="WhiteWorld nczitzk Fatpandac" example="/readhub" path="/readhub/:category?/:overview?" paramsDesc={['分类，见下表，默认为热门话题', '获取概述，任意值获取概述，默认为不获取']}>
+<Route author="WhiteWorld nczitzk Fatpandac" example="/readhub" path="/readhub/:category?" paramsDesc={['分类，见下表，默认为热门话题']} radar="1" rssbud="1">

-| 热门话题 | 科技动态 | 技术资讯 | 区块链快讯 | 每日早报 |
-| -------- | -------- | -------- | ---------- | -------- |
-| topic    | news     | tech     | blockchain | daily    |
+| 热门话题 | 每日早报 | 科技动态 | 医疗产业 |
+| -------- | -------- | -------- | -------- |
+|          | daily    | news     | medical  |

 </Route>