rsshub 1.0.0-master.f9ec483 → 1.0.0-master.fa1dcc5

package/lib/routes/dw/rss.ts

@@ -0,0 +1,62 @@
+import { Route } from '@/types';
+import { config } from '@/config';
+import Parser from 'rss-parser';
+import { processItems } from './utils';
+
+export const route: Route = {
+    path: '/rss/:channel?',
+    categories: ['traditional-media'],
+    example: '/dw/rss/rss-en-all',
+    parameters: {
+        category: 'RSS Feed Channel, see below, `rss-en-all` by default',
+    },
+    features: {
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+        requireConfig: false,
+    },
+    name: 'RSS',
+    maintainers: ['quiniapiezoelectricity'],
+    handler,
+    description: `
+For a full list of RSS Feed Channels in English, please refer to [DW RSS Feeds](https://corporate.dw.com/en/rss-feeds/a-68693346).
+RSS Feed Channels in other languages are also available, for example: \`rss-chi-all\` renders the RSS feed in Chinese and \`rss-de-all\` for the RSS Feed in German 
+`,
+};
+
+async function handler(ctx) {
+    const category = ctx.req.param('channel') ?? 'rss-en-all';
+
+    const parser = new Parser({
+        customFields: {
+            item: ['dwsyn:contentID'],
+        },
+        headers: {
+            'User-Agent': config.ua,
+        },
+    });
+
+    const feed = await parser.parseURL(`https://rss.dw.com/rdf/${category}`);
+    const items = await processItems(
+        feed.items.map((item) => {
+            item.id = item['dwsyn:contentID'];
+            item.pubDate = item.isoDate;
+            item.description = item.content;
+            const link = new URL(item.link);
+            link.search = '';
+            item.link = link.href;
+            item.type = link.pathname.substring(link.pathname.lastIndexOf('/') + 1).startsWith('live-') ? 'liveblog' : 'article'; // dw rss feed only includes liveblogs and articles
+            return item;
+        })
+    );
+
+    return {
+        title: feed.title,
+        link: feed.link,
+        description: feed.description,
+        item: items,
+    };
+}

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

@@ -0,0 +1,23 @@
+{{ if teaser }}
+    <blockquote><p><em>{{ teaser }}</em></p></blockquote>
+{{ /if }}
+{{ if video }}
+    {{@ video }}
+{{ else if mainImage }}
+    <figure>
+        <img 
+            src="https://static.dw.com/image/{{ mainImage.targetId }}_{{ formatId }}.jpg" 
+            alt="{{ mainImage.additionalInformation }}" 
+        >
+        <figcaption>
+            {{ mainImage.description }}
+            <small>{{ imageI18n }}: {{ mainImage.target.licenserSupplement }}</small>
+        </figcaption>
+    </figure>
+{{ /if }}
+{{ if text }}
+   {{@ text }}
+{{ /if }}
+{{ if liveblog }}
+    {{@ liveblog }}
+{{ /if }}

package/lib/routes/dw/templates/liveblog.art

@@ -0,0 +1,13 @@
+{{ if posts }}
+{{ each posts }}
+    <hr>
+    {{ if $value.localizedContentDate }}<p><i>{{ $value.localizedContentDate }}</i></p>{{ /if }}
+    {{ if $value.title }}<h2>{{ $value.title }}</h2>{{ /if }}
+    {{ if $value.persons }}
+        {{ each $value.persons }}
+        <p><i>{{ $value.fullName }}</i></p>
+        {{ /each }}
+    {{ /if }}
+    {{ if $value.text }}{{@ $value.text }}{{ /if }}
+{{ /each }}
+{{ /if }}

package/lib/routes/dw/templates/video.art

@@ -0,0 +1,14 @@
+{{ if hlsVideoSrc }}
+<video
+{{ if posterImageUrl }} poster="{{ posterImageUrl }}" {{ /if }}
+controls preload="metadata">
+    <source
+    src="{{ hlsVideoSrc }}"
+    type="application/x-mpegURL"
+    >
+    <source
+    src="{{ mp4VideoSrc }}"
+    type="video/mp4"
+    >
+</video>
+{{ /if }}

package/lib/routes/dw/utils.ts

@@ -0,0 +1,172 @@
+import cache from '@/utils/cache';
+import got from '@/utils/got';
+import { load, type CheerioAPI } from 'cheerio';
+import { art } from '@/utils/render';
+import path from 'node:path';
+import { getCurrentPath } from '@/utils/helpers';
+
+const __dirname = getCurrentPath(import.meta.url);
+const formatId = '605';
+
+const i18n = (word: string, lang: string) => {
+    switch (word) {
+        case 'Image':
+            switch (lang) {
+                case 'sq':
+                    return 'Fotografi';
+                case 'am':
+                    return 'ምስል';
+                case 'ar':
+                    return 'صورة من';
+                case 'bn':
+                    return 'ছবি';
+                case 'bs':
+                    return 'Foto';
+                case 'bg':
+                    return 'Снимка';
+                case 'zh':
+                    return '图像来源';
+                case 'zh-hant':
+                    return '圖片來源';
+                case 'hr':
+                    return 'Foto';
+                case 'fa-af':
+                    return 'عکس';
+                case 'en':
+                    return 'Image';
+                case 'fr':
+                    return 'Image';
+                case 'de':
+                    return 'Bild';
+                case 'el':
+                    return 'Εικόνα';
+                case 'ha':
+                    return 'Hoto';
+                case 'hi':
+                    return 'तस्वीर';
+                case 'id':
+                    return 'Foto';
+                case 'sw':
+                    return 'Picha';
+                case 'mk':
+                    return 'Фотографија';
+                case 'ps':
+                    return 'انځور';
+                case 'fa-ir':
+                    return 'عکس';
+                case 'pl':
+                    return 'Zdjęcie';
+                case 'pt-002':
+                    return 'Foto';
+                case 'pt-br':
+                    return 'Foto';
+                case 'ro':
+                    return 'Imagine';
+                case 'ru':
+                    return 'Фото';
+                case 'sr':
+                    return 'Foto';
+                case 'es':
+                    return 'Imagen';
+                case 'tr':
+                    return 'Fotoğraf';
+                case 'uk':
+                    return 'Фото';
+                case 'ur':
+                    return 'تصویر';
+                default:
+                    return 'Image';
+            }
+        default:
+            return word;
+    }
+};
+
+const m3u8tomp4 = (src: string) =>
+    src.replace('https://hlsvod.dw.com/i/', 'https://tvdownloaddw-a.akamaihd.net/').replace(',AVC_480x270,AVC_512x288,AVC_640x360,AVC_960x540,AVC_1280x720,AVC_1920x1080,.mp4.csmil/master.m3u8', 'AVC_1920x1080.mp4');
+
+const processHtml = ($: CheerioAPI, contentLinks) => {
+    $('img').each((_, elem) => {
+        try {
+            const id = $(elem).attr('data-id');
+            const contentLink = contentLinks.find((item) => String(item.targetId) === id);
+            $(elem).attr({
+                title: contentLink?.name,
+                alt: contentLink?.description,
+                src: `https://static.dw.com/image/${id}_${formatId}.jpg`,
+            });
+            $(elem).removeAttr('style');
+        } catch {
+            // no-empty
+        }
+    });
+    $('video').each((_, elem) => {
+        try {
+            $(elem).attr('poster', $(elem).attr('data-posterurl'));
+        } catch {
+            // no-empty
+        }
+    });
+    $('iframe').each((_, elem) => {
+        try {
+            $(elem).attr('src', $(elem).attr('data-src'));
+        } catch {
+            // no-empty
+        }
+    });
+    $('svg').remove(); // svg will screw up in a lot of rss readers
+};
+
+const processContent = (item, content) => {
+    const $text = load(content.text);
+    processHtml($text, content.contentLinks);
+    const liveblog =
+        item.type === 'liveblog' && content.posts
+            ? art(path.join(__dirname, 'templates/liveblog.art'), {
+                  posts: content.posts.map((post) => {
+                      const $post = load(post.text);
+                      processHtml($post, content.contentLinks);
+                      post.text = $post.html();
+                      return post;
+                  }),
+              })
+            : undefined;
+    const video =
+        item.type === 'video' && content.hlsVideoSrc
+            ? art(path.join(__dirname, 'templates/video.art'), {
+                  hlsVideoSrc: content.hlsVideoSrc,
+                  mp4VideoSrc: m3u8tomp4(content.hlsVideoSrc),
+                  posterImageUrl: content.posterImageUrl,
+              })
+            : undefined;
+    item.description = art(path.join(__dirname, 'templates/description.art'), {
+        teaser: content.teaser,
+        video,
+        mainImage: $text(`[data-id="${content.mainContentImageLink?.targetId}"]`).length === 0 ? content.mainContentImageLink : undefined,
+        // occasionally the text html already includes the main image, testing to see if an image with the same id exists
+        text: $text.html(),
+        liveblog,
+        imageI18n: i18n('Image', item.language),
+        formatId,
+    });
+    if (content.trackingCategories) {
+        item.category = content.trackingCategories;
+    }
+    if (content.firstPersonArray) {
+        item.author = content.firstPersonArray.map((person) => person.fullName).join(', ');
+    }
+    return item;
+};
+
+export const processItems = async (items) => {
+    items = await Promise.all(
+        items.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const response = await got(`https://www.dw.com/graph-api/${item.language}/content/${item.type}/${item.id}`);
+                const content = response.data.data.content;
+                return processContent(item, content);
+            })
+        )
+    );
+    return items;
+};

package/lib/routes/ecnu/jwc.ts

@@ -13,7 +13,11 @@ export const route: Route = {
     example: '/ecnu/jwc',
     radar: [
         {
-            source: ['www.jwc.ecnu.edu.cn', 'www.ecnu.edu.cn'],
+            source: ['www.jwc.ecnu.edu.cn'],
+            target: '/tzgg',
+        },
+        {
+            source: ['www.ecnu.edu.cn'],
             target: '/tzgg',
         },
     ],

package/lib/routes/espn/news.ts

@@ -2,7 +2,7 @@ import { Route } from '@/types';
 import cache from '@/utils/cache';
 import ofetch from '@/utils/ofetch';
 import * as cheerio from 'cheerio';
-import path from 'path';
+import path from 'node:path';
 import { getCurrentPath } from '@/utils/helpers';
 import { art } from '@/utils/render';
 

package/lib/routes/fashionnetwork/namespace.ts

@@ -3,7 +3,7 @@ import type { Namespace } from '@/types';
 export const namespace: Namespace = {
     name: 'FashionNetwork',
     url: 'fashionnetwork.cn',
-    categories: ['new-media'],
+    categories: ['new-media', 'popular'],
     description: '',
     lang: 'zh-CN',
 };

package/lib/routes/flyert/forum.ts

@@ -0,0 +1,99 @@
+import { Route } from '@/types';
+
+import cache from '@/utils/cache';
+import got from '@/utils/got';
+import iconv from 'iconv-lite';
+import { load } from 'cheerio';
+import { rootUrl, parseArticleList, parsePostList, parseArticle, parsePost } from './util';
+
+export const handler = async (ctx) => {
+    const { params } = ctx.req.param();
+    const limit = ctx.req.query('limit') ? Number.parseInt(ctx.req.query('limit'), 10) : 5;
+
+    const decodedParams = params
+        ? decodeURIComponent(params)
+              .split(/&/)
+              .filter((p) => p.split(/=/).length === 2)
+              .join('&')
+        : undefined;
+
+    const currentUrl = new URL(`forum.php${decodedParams ? `?${decodedParams}` : ''}`, rootUrl).href;
+
+    const { data: response } = await got(currentUrl, {
+        responseType: 'buffer',
+    });
+
+    const $ = load(iconv.decode(response, 'gbk'));
+
+    const language = $('meta[http-equiv="Content-Language"]').prop('content');
+
+    let items = $('table#threadlisttableid').length === 0 ? parseArticleList($, limit) : parsePostList($, limit);
+
+    items = await Promise.all(
+        items.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const { data: detailResponse } = await got(item.link, {
+                    responseType: 'buffer',
+                });
+
+                const $$ = load(iconv.decode(detailResponse, 'gbk').replaceAll(/<\/?ignore_js_op>/g, ''));
+
+                item = $$('div.firstpost').length === 0 ? parseArticle($$, item) : parsePost($$, item);
+
+                item.language = language;
+
+                return item;
+            })
+        )
+    );
+
+    const image = `https:${$('div.wp h2 a img').prop('src')}`;
+
+    return {
+        title: `飞客 - ${$('a.forum_name, li.a, li.cur, li.xw1, div.z > a.xw1')
+            .toArray()
+            .map((a) => $(a).text())
+            .join(' - ')}`,
+        description: $('meta[name="description"]').prop('content'),
+        link: currentUrl,
+        item: items,
+        allowEmpty: true,
+        image,
+        author: $('meta[name="application-name"]').prop('content'),
+    };
+};
+
+export const route: Route = {
+    path: '/forum/:params{.+}?',
+    name: '会员说',
+    url: 'www.flyert.com.cn',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/flyert/forum',
+    parameters: { params: '参数，默认为空，可在对应分类页 URL 中找到' },
+    description: `:::tip
+  若订阅 [酒店集团优惠](https://www.flyert.com.cn/forum.php?mod=forumdisplay&sum=all&fid=all&catid=322&filter=sortid&sortid=144&searchsort=1&youhui_type=19)，网址为 \`https://www.flyert.com.cn/forum.php?mod=forumdisplay&sum=all&fid=all&catid=322&filter=sortid&sortid=144&searchsort=1&youhui_type=19\`。截取 \`https://www.flyert.com.cn/forum.php?\` 到末尾的部分 \`mod=forumdisplay&sum=all&fid=all&catid=322&filter=sortid&sortid=144&searchsort=1&youhui_type=19\` **进行 UrlEncode 编码** 后作为参数填入，此时路由为 [\`/flyert/forum/mod%3Dforumdisplay%26sum%3Dall%26fid%3Dall%26catid%3D322%26filter%3Dsortid%26sortid%3D144%26searchsort%3D1%26youhui_type%3D226\`](https://rsshub.app/flyert/forum/mod%3Dforumdisplay%26sum%3Dall%26fid%3Dall%26catid%3D322%26filter%3Dsortid%26sortid%3D144%26searchsort%3D1%26youhui_type%3D226)。
+  :::
+    `,
+    categories: ['bbs'],
+
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['www.flyert.com.cn/forum.php'],
+            target: (_, url) => {
+                const params = [...url.searchParams.entries()].map(([key, value]) => key + '=' + value).join('&');
+
+                return `/forum${params ? `/${encodeURIComponent(params)}` : ''}`;
+            },
+        },
+    ],
+};

package/lib/routes/flyert/namespace.ts

@@ -3,5 +3,6 @@ import type { Namespace } from '@/types';
 export const namespace: Namespace = {
     name: '飞客茶馆',
     url: 'flyert.com',
+    description: '',
     lang: 'zh-CN',
 };

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

@@ -0,0 +1,21 @@
+{{ if images }}
+  {{ each images image }}
+    {{ if image?.src }}
+      <figure>
+        <img
+          {{ if image.alt }}
+            alt="{{ image.alt }}"
+          {{ /if }}
+        src="{{ image.src }}">
+      </figure>
+    {{ /if }}
+  {{ /each }}
+{{ /if }}
+
+{{ if intro }}
+  <blockquote>{{ intro }}</blockquote>
+{{ /if }}
+
+{{ if description }}
+  {{@ description }}
+{{ /if }}

package/lib/routes/flyert/util.ts

@@ -0,0 +1,188 @@
+import { getCurrentPath } from '@/utils/helpers';
+const __dirname = getCurrentPath(import.meta.url);
+
+import { CheerioAPI } from 'cheerio';
+import timezone from '@/utils/timezone';
+import { parseDate } from '@/utils/parse-date';
+import { art } from '@/utils/render';
+import path from 'node:path';
+
+const rootUrl = 'https://www.flyert.com.cn';
+
+/**
+ * Parses a list of articles based on a CheerioAPI object and a limit.
+ * @param $ The CheerioAPI object.
+ * @param limit The maximum number of articles to parse.
+ * @returns An array of parsed article objects.
+ */
+const parseArticleList = ($: CheerioAPI, limit: number) =>
+    $('div.comiis_wzli')
+        .slice(0, limit)
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            const title = item.find('div.wzbt').text().trim();
+            const image = item.find('div.wzpic img').prop('src');
+            const description = art(path.join(__dirname, 'templates/description.art'), {
+                images: image
+                    ? [
+                          {
+                              src: image,
+                              alt: title,
+                          },
+                      ]
+                    : undefined,
+                description: item.find('div.wznr').html(),
+            });
+            const pubDate = item.find('div.subcat span.y').contents()?.eq(2)?.text().trim() ?? undefined;
+            const link = new URL(item.find('div.wzbt a').prop('href'), rootUrl).href;
+
+            return {
+                title,
+                description,
+                pubDate: pubDate ? parseDate(pubDate) : undefined,
+                link,
+                author: item.find('div.subcat span.y a').first().text(),
+                content: {
+                    html: description,
+                    text: item.find('div.wznr').text(),
+                },
+                image,
+                banner: image,
+            };
+        });
+
+/**
+ * Parses a list of posts based on a CheerioAPI object and a limit.
+ * @param $ The CheerioAPI object.
+ * @param limit The maximum number of posts to parse.
+ * @returns An array of parsed post objects.
+ */
+const parsePostList = ($: CheerioAPI, limit: number) =>
+    $('div.comiis_postlist')
+        .toArray()
+        .filter((item) => {
+            item = $(item);
+
+            return item
+                .find('span.comiis_common a[data-track]')
+                .toArray()
+                .some((a) => {
+                    a = $(a);
+
+                    const dataTrack = a.attr('data-track') || '';
+                    return dataTrack.endsWith('文章');
+                });
+        })
+        .slice(0, limit)
+        .map((item) => {
+            item = $(item);
+
+            const aEl = $(
+                item
+                    .find('span.comiis_common a[data-track]')
+                    .toArray()
+                    .find((a) => {
+                        a = $(a);
+
+                        const dataTrack = a.attr('data-track') || '';
+                        return dataTrack.endsWith('文章');
+                    })
+            );
+
+            const pubDate = item.find('span.author_b span').prop('title') || undefined;
+
+            return {
+                title: aEl.text().trim(),
+                pubDate: pubDate ? parseDate(pubDate) : undefined,
+                link: new URL(aEl.prop('href'), rootUrl).href,
+                author: item.find('a.author_t').text().trim(),
+            };
+        });
+
+/**
+ * Parses an article based on a CheerioAPI object and an item.
+ * @param $$ The CheerioAPI object.
+ * @param item The item to parse.
+ * @returns The parsed article object.
+ */
+const parseArticle = ($$: CheerioAPI, item) => {
+    const title = $$('h1.ph').text().trim();
+    const description = art(path.join(__dirname, 'templates/description.art'), {
+        intro: $$('div.s').text() || undefined,
+        description: $$('div#artMain').html(),
+    });
+    const pubDate =
+        $$('p.xg1')
+            .contents()
+            .first()
+            .text()
+            .trim()
+            ?.match(/(\d{4}-\d{1,2}-\d{1,2}\s\d{2}:\d{2})/)?.[1] ?? undefined;
+    const guid = `flyert-${item.link.split(/=/).pop()}`;
+
+    item.title = title;
+    item.description = description;
+    item.pubDate = pubDate ? timezone(parseDate(pubDate), +8) : item.pubDate;
+    item.author = $$('p.xg1 a').first().text();
+    item.guid = guid;
+    item.id = guid;
+    item.content = {
+        html: description,
+        text: $$('div#artMain').text(),
+    };
+
+    return item;
+};
+
+/**
+ * Parses a post based on a CheerioAPI object and an item.
+ * @param $$ The CheerioAPI object.
+ * @param item The item to parse.
+ * @returns The parsed post object.
+ */
+const parsePost = ($$: CheerioAPI, item) => {
+    $$('img.zoom').each((_, el) => {
+        el = $$(el);
+
+        el.replaceWith(
+            art(path.join(__dirname, 'templates/description.art'), {
+                images:
+                    el.prop('zoomfile') || el.prop('file')
+                        ? [
+                              {
+                                  src: el.prop('zoomfile') || el.prop('file'),
+                                  alt: el.prop('alt') || el.prop('title'),
+                              },
+                          ]
+                        : undefined,
+            })
+        );
+    });
+
+    $$('i.pstatus').remove();
+    $$('div.tip').remove();
+
+    const title = $$('span#thread_subject').text().trim();
+    const description = $$('div.post_message').first().html();
+    const pubDate = $$('span[title]').first().prop('title');
+
+    const tid = item.link.match(/tid=(\d+)/)?.[1] ?? undefined;
+    const guid = tid ? `flyert-${tid}` : undefined;
+
+    item.title = title;
+    item.description = description;
+    item.pubDate = pubDate ? timezone(parseDate(pubDate), +8) : item.pubDate;
+    item.author = $$('a.kmxi2').first().text();
+    item.guid = guid;
+    item.id = guid;
+    item.content = {
+        html: description,
+        text: $$('div.post_message').first().text(),
+    };
+
+    return item;
+};
+
+export { rootUrl, parseArticleList, parsePostList, parseArticle, parsePost };

package/lib/routes/follow/profile.ts

@@ -19,7 +19,7 @@ export const route: Route = {
         },
     ],
     handler,
-    maintainers: ['KarasuShin', 'DIYgod'],
+    maintainers: ['KarasuShin', 'DIYgod', 'DFobain'],
     features: {
         supportRadar: true,
     },
@@ -64,7 +64,7 @@ async function handler(ctx: Context): Promise<Data> {
                 category: subscription.category ? [subscription.category] : undefined,
             };
         }),
-        link: `https://app.follow.is/share/user/${handleOrId}`,
+        link: `https://app.follow.is/share/users/${handleOrId}`,
         image: profile.data.image,
     };
 }

package/lib/routes/foresightnews/article.ts

@@ -3,7 +3,7 @@ import { rootUrl, apiRootUrl, processItems, icon, image } from './util';
 
 export const route: Route = {
     path: '/article',
-    categories: ['new-media'],
+    categories: ['new-media', 'popular'],
     example: '/foresightnews/article',
     parameters: {},
     features: {

package/lib/routes/foresightnews/column.ts

@@ -3,7 +3,7 @@ import { rootUrl, apiRootUrl, processItems, icon, image } from './util';
 
 export const route: Route = {
     path: '/column/:id',
-    categories: ['new-media'],
+    categories: ['new-media', 'popular'],
     example: '/foresightnews/column/1',
     parameters: { id: '专栏 id, 可在对应专栏页 URL 中找到' },
     features: {