rsshub 1.0.0-master.f71451d → 1.0.0-master.f72af1b

package/lib/routes/theblockbeats/index.ts

@@ -1,67 +1,103 @@
 import { Route } from '@/types';
 import cache from '@/utils/cache';
-import got from '@/utils/got';
+import ofetch from '@/utils/ofetch';
 import { load } from 'cheerio';
 import { parseDate } from '@/utils/parse-date';
+import path from 'path';
+import { art } from '@/utils/render';
+import { getCurrentPath } from '@/utils/helpers';
+
+const __dirname = getCurrentPath(import.meta.url);
 
 const domain = 'theblockbeats.info';
 const rootUrl = `https://www.${domain}`;
 const apiBase = `https://api.${domain}`;
 
+const render = (data) => {
+    const html = art(path.join(__dirname, 'templates/description.art'), data);
+    const $ = load(html, null, false);
+
+    $('img').each((_, e) => {
+        const $e = $(e);
+        const src = $e.attr('src');
+        $e.attr('src', src?.split('?')[0]);
+    });
+
+    return $.html();
+};
+
 const channelMap = {
     newsflash: {
         title: '快讯',
         link: `${rootUrl}/newsflash`,
-        api: `${apiBase}/v5/newsflash/select`,
+        api: `${apiBase}/v2/newsflash/list`,
     },
     article: {
         title: '文章',
         link: `${rootUrl}/article`,
-        api: `${apiBase}/v5/Information/newsall`,
+        api: `${apiBase}/v2/article/list`,
     },
 };
 
 export const route: Route = {
-    path: '/:channel?',
+    path: '/:channel?/:original?',
     categories: ['finance'],
     example: '/theblockbeats/newsflash',
-    parameters: { channel: '类型，见下表，默认为快讯' },
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
-    },
+    parameters: { channel: '类型，见下表，默认为快讯', original: '文章类型，仅 `channel` 为 `article` 时有效，见下表，留空为全部' },
     name: '新闻快讯',
     maintainers: ['Fatpandac', 'jameshih'],
     handler,
+    radar: [
+        {
+            title: '文章',
+            source: ['www.theblockbeats.info/article'],
+            target: '/article',
+        },
+        {
+            title: '快讯',
+            source: ['www.theblockbeats.info/newsflash'],
+            target: '/newsflash',
+        },
+    ],
     description: `|    快讯   |   文章  |
   | :-------: | :-----: |
-  | newsflash | article |`,
+  | newsflash | article |
+
+  | 全部 | 深度 | 精选 | 热点追踪 |
+  | :--: | :--: | :--: | :---: |
+  |     | -2  | 1    |  2     |`,
 };
 
 async function handler(ctx) {
-    const { channel = 'newsflash' } = ctx.req.param();
+    const { channel = 'newsflash', original } = ctx.req.param();
 
-    const { data: response } = await got(channelMap[channel].api);
+    const response = await ofetch(channelMap[channel].api, {
+        query: {
+            limit: ctx.req.query('limit') ? Number.parseInt(ctx.req.query('limit')) : 20,
+            original: channel === 'article' ? original : undefined,
+        },
+    });
 
-    const { data } = channel === 'newsflash' ? response.data : response;
-    let list = data.map((item) => ({
+    let list = response.data.list.map((item) => ({
         title: item.title,
-        link: `${rootUrl}/${channel === 'newsflash' ? 'flash' : 'news'}/${item.id}`,
-        description: item.content ?? item.im_abstract,
+        link: `${rootUrl}/${channel === 'newsflash' ? 'flash' : 'news'}/${item.article_id}`,
+        description: item.content ?? item.abstract,
         pubDate: parseDate(item.add_time, 'X'),
+        author: item.author?.nickname,
+        category: item.tag_list,
+        imgUrl: item.img_url,
     }));
 
     if (channel !== 'newsflash') {
         list = await Promise.all(
             list.map((item) =>
-                cache.tryGet(item.link, async () => {
-                    const { data: response } = await got(item.link);
+                cache.tryGet(`theblockbeats:${item.link}`, async () => {
+                    const response = await ofetch(item.link);
                     const $ = load(response);
-                    item.description = $('div.news-content').html();
+                    item.description = render({
+                        image: item.imgUrl,
+                        description: $('div.news-content').html(),
+                    });
                     return item;
                 })
             )

package/lib/routes/theblockbeats/namespace.ts

@@ -1,6 +1,6 @@
 import type { Namespace } from '@/types';
 
 export const namespace: Namespace = {
-    name: '律动',
-    url: 'rszhaopin.theblockbeats.info',
+    name: '律动 BlockBeats',
+    url: 'www.theblockbeats.info',
 };

package/lib/routes/theblockbeats/templates/description.art

@@ -0,0 +1,5 @@
+{{ if image }}
+    <img src="{{ image }}" /><br>
+{{ /if }}
+
+{{@ description }}

package/lib/routes/theinitium/app.ts

@@ -0,0 +1,137 @@
+import { Route } from '@/types';
+import cache from '@/utils/cache';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { art } from '@/utils/render';
+import path from 'node:path';
+import { config } from '@/config';
+import { getCurrentPath } from '@/utils/helpers';
+
+export const route: Route = {
+    path: '/app/:category?',
+    categories: ['new-media'],
+    example: '/theinitium/app',
+    parameters: {
+        category: 'Category, see below, latest_sc by default',
+    },
+    features: {
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    name: 'App',
+    maintainers: ['quiniapiezoelectricity'],
+    radar: [
+        {
+            source: ['app.theinitium.com/t/latest/:category'],
+            target: '/app/:category',
+        },
+    ],
+    handler,
+    description: `抓取[The Initium App](https://app.theinitium.com/)的文章列表
+
+:::warning
+此路由暂不支持登陆认证
+:::
+
+Category 栏目：
+
+| ----- | 简体中文     | 繁體中文      |
+| ----- | ----------------- | ---------------- |
+| 最新   | latest_sc | latest_tc |
+| 日报   | daily_brief_sc | daily_brief_tc |
+| 速递   | whats_new_sc | whats_new_tc |
+| 专题   | report_sc | report_tc |
+| 评论   | opinion_sc | opinion_tc |
+| 国际   | international_sc | international_tc |
+| 大陆   | mainland_sc | mainland_tc |
+| 香港   | hongkong_sc | hongkong_tc |
+| 台湾   | taiwan_sc | taiwan_tc |
+| 播客   | article_audio_sc | article_audio_tc |`,
+};
+
+async function handler(ctx) {
+    const category = ctx.req.param('category') ?? 'latest_sc';
+    const __dirname = getCurrentPath(import.meta.url);
+
+    const feedListLink = 'https://app.theinitium.com/timelines.json';
+
+    const feedList = await cache.tryGet(
+        feedListLink,
+        async () =>
+            await got({
+                method: 'get',
+                url: feedListLink,
+            }),
+        config.cache.routeExpire,
+        false
+    );
+
+    const feedInfo = feedList.data.timelines.find((timeline) => timeline.id === category);
+
+    const link = `https://app.theinitium.com${feedInfo.feed.slice(1)}`;
+
+    const feedResponse = await got({
+        method: 'get',
+        url: link,
+    });
+
+    const feed = feedResponse.data.stories.filter((item) => item.type === 'article');
+
+    const items = await Promise.all(
+        feed.map((item) =>
+            cache.tryGet('https://app.theinitium.com/' + item.url.replaceAll('../', ''), async () => {
+                item.link = 'https://app.theinitium.com/' + item.url.replaceAll('../', '');
+                item.description = item.summary;
+                item.pubDate = item.published;
+                item.category = [];
+                if (item.section) {
+                    item.category = [...item.category, item.section];
+                }
+                if (item.taxonomy) {
+                    if (item.taxonomy.collection_tag) {
+                        item.category = [...item.category, ...item.taxonomy.collection_tag];
+                    }
+                    if (item.taxonomy.sections) {
+                        item.category = [...item.category, ...item.taxonomy.sections];
+                    }
+                }
+                item.category = [...new Set(item.category)];
+                const response = await got(item.link);
+                const $ = load(response.data);
+                const article = $('.pp-article__body');
+                article.find('.block-related-articles').remove();
+                item.description = art(path.join(__dirname, 'templates/description.art'), {
+                    header: $('.pp-header-group__standfirst').html(),
+                    coverImage: $('.pp-media__image').attr('src'),
+                    coverCaption: $('.pp-media__caption').html(),
+                    article: article.html(),
+                    copyright: $('.copyright').html(),
+                });
+                return item;
+            })
+        )
+    );
+
+    let lang;
+    let titleLoc;
+    if (feedInfo.timeline_sets[0] === 'chinese-simplified') {
+        lang = 'zh-hans';
+        titleLoc = '端传媒';
+    } else if (feedInfo.timeline_sets[0] === 'chinese-traditional') {
+        lang = 'zh-hant';
+        titleLoc = '端傳媒';
+    } else {
+        lang = 'zh-hans';
+        titleLoc = '端传媒';
+    }
+
+    return {
+        title: `${titleLoc} - ${feedInfo.title}`,
+        link: `https://app.theinitium.com/t/latest/${category}/`,
+        language: lang,
+        item: items,
+    };
+}

package/lib/routes/theinitium/author.ts

@@ -0,0 +1,23 @@
+import { Route } from '@/types';
+import { processFeed } from './utils';
+
+const handler = (ctx) => processFeed('author', ctx);
+
+export const route: Route = {
+    path: '/author/:type/:language?',
+    name: '作者',
+    maintainers: ['AgFlore'],
+    parameters: {
+        type: '作者 ID，可从作者主页 URL 中获取，如 `https://theinitium.com/author/ninghuilulu`',
+        language: '语言，简体`zh-hans`，繁体`zh-hant`，缺省为简体',
+    },
+    radar: [
+        {
+            source: ['theinitium.com/author/:type'],
+            target: '/author/:type',
+        },
+    ],
+    handler,
+    example: '/theinitium/author/ninghuilulu/zh-hans',
+    categories: ['new-media'],
+};

package/lib/routes/theinitium/channel.ts

@@ -0,0 +1,28 @@
+import { Route } from '@/types';
+import { processFeed } from './utils';
+
+const handler = (ctx) => processFeed('channel', ctx);
+
+export const route: Route = {
+    path: '/channel/:type?/:language?',
+    name: '专题・栏目',
+    maintainers: ['prnake', 'mintyfrankie'],
+    parameters: {
+        type: '栏目，缺省为最新',
+        language: '语言，简体`zh-hans`，繁体`zh-hant`，缺省为简体',
+    },
+    radar: [
+        {
+            source: ['theinitium.com/channel/:type'],
+            target: '/channel/:type',
+        },
+    ],
+    handler,
+    example: '/theinitium/channel/latest/zh-hans',
+    categories: ['new-media'],
+    description: `Type 栏目：
+
+  | 最新   | 深度    | What’s New | 广场              | 科技       | 风物    | 特约     | ... |
+  | ------ | ------- | ---------- | ----------------- | ---------- | ------- | -------- | --- |
+  | latest | feature | news-brief | notes-and-letters | technology | culture | pick_up | ... |`,
+};

package/lib/routes/theinitium/follow.ts

@@ -0,0 +1,43 @@
+import { Route } from '@/types';
+import { processFeed } from './utils';
+
+const handler = (ctx) => processFeed('follow', ctx);
+
+export const route: Route = {
+    path: '/follow/articles/:language?',
+    name: '个人订阅追踪动态',
+    maintainers: ['AgFlore'],
+    parameters: {
+        language: '语言，简体`zh-hans`，繁体`zh-hant`，缺省为简体',
+    },
+    radar: [
+        {
+            title: '作者',
+            source: ['theinitium.com/author/:type'],
+            target: '/author/:type',
+        },
+    ],
+    handler,
+    example: '/theinitium/author/ninghuilulu/zh-hans',
+    categories: ['new-media'],
+    description: '需填入 Web 版认证 token, 也可选择直接在环境设置中填写明文的用户名和密码',
+    features: {
+        requireConfig: [
+            {
+                name: 'INITIUM_BEARER_TOKEN',
+                optional: true,
+                description: `端传媒 Web 版认证 token。获取方式：登陆后打开端传媒站内任意页面，打开浏览器开发者工具中 “网络”(Network) 选项卡，筛选 URL 找到任一个地址为 \`api.initium.com\` 开头的请求，点击检查其 “消息头”，在 “请求头” 中找到Authorization字段，将其值复制填入配置即可。你的配置应该形如 \`INITIUM_BEARER_TOKEN: 'Bearer eyJxxxx......xx_U8'\`。使用 token 部署的好处是避免占据登陆设备数的额度，但这个 token 一般有效期为两周，因此只可作临时测试使用。`,
+            },
+            {
+                name: 'INITIUM_USERNAME',
+                optional: true,
+                description: `端传媒用户名 （邮箱）`,
+            },
+            {
+                name: 'INITIUM_PASSWORD',
+                optional: true,
+                description: `端传媒密码`,
+            },
+        ],
+    },
+};

package/lib/routes/theinitium/tags.ts

@@ -0,0 +1,23 @@
+import { Route } from '@/types';
+import { processFeed } from './utils';
+
+const handler = (ctx) => processFeed('tags', ctx);
+
+export const route: Route = {
+    path: '/tags/:type/:language?',
+    name: '话题・标签',
+    maintainers: ['AgFlore'],
+    parameters: {
+        type: '话题 ID，可从话题页 URL 中获取，如 `https://theinitium.com/tags/2019_10/`',
+        language: '语言，简体`zh-hans`，繁体`zh-hant`，缺省为简体',
+    },
+    radar: [
+        {
+            source: ['theinitium.com/tags/:type'],
+            target: '/tags/:type',
+        },
+    ],
+    handler,
+    example: '/theinitium/tags/2019_10/zh-hans',
+    categories: ['new-media'],
+};

package/lib/routes/theinitium/templates/description.art

@@ -0,0 +1,17 @@
+{{ if header }}
+    <h4> {{ header }} </h4>
+{{ /if }}
+{{ if coverImage}}
+<figure>
+    <img src={{ coverImage }}>
+    {{ if coverCaption }}
+        <figcaption>{{ coverCaption }}</figcaption>
+    {{ /if }}
+</figure>
+{{ /if }}
+{{ if article }}
+   {{@ article }}
+{{ /if }}
+{{ if copyright }}
+    <figure><small>{{@ copyright }}</small></figure>
+{{ /if }}

package/lib/routes/theinitium/{full.ts → utils.ts}

@@ -1,22 +1,16 @@
-import { Route } from '@/types';
+import { Context } from 'hono';
 import cache from '@/utils/cache';
 import got from '@/utils/got';
 import { config } from '@/config';
 import { load } from 'cheerio';
 import { parseDate } from '@/utils/parse-date';
+import InvalidParameterError from '@/errors/types/invalid-parameter';
+import { FetchError } from 'ofetch';
 
 const TOKEN = 'Basic YW5vbnltb3VzOkdpQ2VMRWp4bnFCY1ZwbnA2Y0xzVXZKaWV2dlJRY0FYTHY=';
 
-export const route: Route = {
-    path: '/:model?/:type?/:language?',
-    name: 'Unknown',
-    maintainers: [],
-    handler,
-};
-
-async function handler(ctx) {
+export const processFeed = async (model: string, ctx: Context) => {
     // model是channel/tag/etc.，而type是latest/feature/quest-academy这些一级栏目/标签/作者名的slug名。如果是追踪的话，那就是model是follow，type是articles。
-    const model = ctx.req.param('model') ?? 'channel';
     const type = ctx.req.param('type') ?? 'latest';
     const language = ctx.req.param('language') ?? 'zh-hans';
     let listUrl;
@@ -38,12 +32,13 @@ async function handler(ctx) {
             listUrl = `https://api.theinitium.com/api/v2/tag/articles/?language=${language}&slug=${type}`;
             listLink = `https://theinitium.com/tags/${type}/`;
             break;
+        default:
+            throw new InvalidParameterError('wrong model');
     }
 
     const key = {
         email: config.initium.username,
         password: config.initium.password,
-        iapReceipt: config.initium.iap_receipt,
     };
     const body = JSON.stringify(key);
 
@@ -63,25 +58,11 @@ async function handler(ctx) {
                 Accept: 'application/json',
                 Connection: 'keep-alive',
                 Authorization: TOKEN,
-                Origin: `https://theinitium.com/`,
-                Referer: `https://theinitium.com/`,
-                'X-Client-Name': 'Web',
             },
             body,
         });
 
-        /*
-        const devices = login.data.access.devices;
-
-        for (const key in devices) {
-            const device = devices[key];
-            if (device.status === 'logged_in' && !device.logout_at && device.platform === 'web') {
-                token = 'Bearer ' + device.device_id;
-                break;
-            }
-        }
-        */
-        token = 'Bearer ' + login.data.token;
+        token = 'token ' + login.data.token;
         cache.set('initium:token', token);
     }
 
@@ -89,12 +70,20 @@ async function handler(ctx) {
         Accept: '*/*',
         Connection: 'keep-alive',
         Authorization: token,
-        'X-IAP-Receipt': key.iapReceipt || '',
     };
 
-    const response = await got(listUrl, {
-        headers,
-    });
+    let response;
+    try {
+        response = await got(listUrl, {
+            headers,
+        });
+    } catch (error) {
+        if (error instanceof FetchError && error.statusCode === 401) {
+            // 401 说明 token 过期了，将它删掉
+            await cache.set('initium:token', '');
+        }
+        throw error;
+    }
 
     const name = response.data.name || (response.data[model] && response.data[model].name) || '追踪';
     // 从v1直升的channel和tags里面是digests，v2新增的author和follow出来都是results
@@ -149,7 +138,7 @@ async function handler(ctx) {
     const items = await Promise.all(
         articles
             .filter((a) => a.article)
-            .slice(0, token === TOKEN && key.iapReceipt === undefined ? 25 : articles.length)
+            .slice(0, token === TOKEN ? 25 : articles.length)
             .map(async (item) => {
                 item.article.date = parseDate(item.article.date);
                 item.article.updated = parseDate(item.article.updated);
@@ -175,4 +164,4 @@ async function handler(ctx) {
         item: items,
         image,
     };
-}
+};

package/lib/routes/{thecatcity → thepetcity}/index.ts

@@ -1,47 +1,38 @@
 import { Route } from '@/types';
 import cache from '@/utils/cache';
-import got from '@/utils/got';
+import ofetch from '@/utils/ofetch';
 import { parseDate } from '@/utils/parse-date';
 import { termsMap } from './terms-map';
 
-const baseUrl = 'https://thecatcity.com';
+const baseUrl = 'https://thepetcity.co';
 
 export const route: Route = {
     path: '/:term?',
     categories: ['new-media'],
-    example: '/thecatcity',
+    example: '/thepetcity',
     parameters: { term: '見下表，留空為全部文章' },
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
-    },
-    radar: [
-        {
-            source: ['thecatcity.com/'],
-            target: '',
-        },
-    ],
+    radar: Object.entries(termsMap).map(([key, value]) => ({
+        title: value.title,
+        source: [...new Set([`thepetcity.co${value.slug}`, 'thepetcity.co/'])],
+        target: key ? `/${key}` : '',
+    })),
     name: '分類',
-    maintainers: ['TonyRL'],
+    maintainers: ['TonyRL', 'bigfei'],
     handler,
-    url: 'thecatcity.com/',
-    description: `| 貓物分享 | 貓咪新聞 | 養貓大全 | 貓奴景點 | 新手養貓教學 |
-  | -------- | -------- | -------- | -------- | ------------ |
-  | 1        | 2        | 3        | 4        | 5            |`,
+    url: 'thepetcity.co/',
+    description: `| Column Name       | TermID |
+    | -------------------- | ------ |
+    | Knowledge飼養大全     | 3      |
+    | Funny News毛孩趣聞    | 2      |
+    | Raise Pets 養寵物新手  | 5      |
+    | Hot Spot 毛孩打卡點    | 4      |
+    | Pet Staff 毛孩好物    | 1      |`,
 };
 
 async function handler(ctx) {
     const term = ctx.req.param('term');
-    const { data } = await got(`${baseUrl}/node_api/v1/articles/posts`, {
-        searchParams: {
-            pageId: 977_080_509_047_743,
-            term,
-        },
-    });
+    const searchParams = term ? { pageId: 977_080_509_047_743, term } : { pageId: 977_080_509_047_743 };
+    const data = await ofetch(`${baseUrl}/node_api/v1/articles/posts`, { query: { ...searchParams } });
 
     const list = data.data.posts.map((post) => ({
         title: post.title,
@@ -55,8 +46,8 @@ async function handler(ctx) {
     const items = await Promise.all(
         list.map((item) =>
             cache.tryGet(item.guid, async () => {
-                const { data } = await got(item.api, {
-                    searchParams: {
+                const data = await ofetch(item.api, {
+                    query: {
                         pageId: 977_080_509_047_743,
                     },
                 });
@@ -70,9 +61,9 @@ async function handler(ctx) {
 
     return {
         title: termsMap[term] ? termsMap[term].title : termsMap[''].title,
-        description: '提供貓咪日常照顧、新手準備、貓用品、貓咪醫療、貓飲食與行為等相關知識，以及療癒貓影片、貓趣聞、貓小物流行資訊，不論你是貓奴、還是貓控，一切所需都在貓奴日常找到',
+        description: '專屬毛孩愛好者的資訊平台，不論你是貓奴、狗奴，還是其他動物控，一起發掘最新的萌寵趣聞、有趣的寵物飼養知識、訓練動物、竉物用品推介、豐富多樣的寵物可愛影片。',
         link: baseUrl,
-        image: 'https://assets.presslogic.com/presslogic-hk-tc/static/favicon.ico',
+        image: 'https://assets.presslogic.com/presslogic-hk-pc/static/favicon.ico',
         item: items,
     };
 }

package/lib/routes/thepetcity/namespace.ts

@@ -0,0 +1,6 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'PetCity 毛孩日常',
+    url: 'thepetcity.com',
+};

package/lib/routes/thepetcity/terms-map.ts

@@ -0,0 +1,28 @@
+const termsMap = {
+    '': {
+        title: 'PetCity 毛孩日常 | 飼養竉物、竉物用品、萌寵趣聞',
+        slug: '/',
+    },
+    1: {
+        title: 'Pet Staff 毛孩好物',
+        slug: '/category/cute-item',
+    },
+    2: {
+        title: 'Funny News毛孩趣聞',
+        slug: '/category/funny-news',
+    },
+    3: {
+        title: 'Knowledge飼養大全',
+        slug: '/category/knowledge',
+    },
+    4: {
+        title: 'Hot Spot 毛孩打卡點',
+        slug: '/category/hot-spot',
+    },
+    5: {
+        title: 'Raise Pets 養寵物新手',
+        slug: '/category/raise-cats',
+    },
+};
+
+export { termsMap };