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

package/lib/routes/jou/utils/index.ts

@@ -1,26 +1,22 @@
 import cache from '@/utils/cache';
-// 导入got库，该库用来请求网页数据
-import got from '@/utils/got';
-// 导入cheerio库，该库用来解析网页数据
+import ofetch from '@/utils/ofetch'; // 使用ofetch库
 import { load } from 'cheerio';
-// 导入parseDate函数，该函数用于日期处理
 import { parseDate } from '@/utils/parse-date';
-// 导入timezone库，该库用于时区处理
 import timezone from '@/utils/timezone';
 
 async function getItems(ctx, url, host, tableClass, timeStyleClass1, titleStyleClass, timeStyleClass2) {
-    // 发起Http请求，获取网页数据
-    const response = await got({ url, https: { rejectUnauthorized: false } });
-    // 解析网页数据
-    const $ = load(response.data);
+    const response = await ofetch(url).catch(() => null);
+    if (!response) {
+        return [];
+    }
+    const $ = load(response);
 
-    // 通知公告的items的标题、url链接、发布日期
     const list = $(`table.${tableClass} > tbody > tr[height=20]`)
         .toArray()
         .map((item) => {
-            const currentItem = $(item); // 获取当前tr元素的jQuery对象
-            const item1 = currentItem.find('td:eq(1)'); // 获取当前tr元素下的第二个td
-            const item2 = currentItem.find('td:eq(2)'); // 获取当前tr元素下的第三个td
+            const currentItem = $(item);
+            const item1 = currentItem.find('td:eq(1)');
+            const item2 = currentItem.find('td:eq(2)');
             const link = new URL(item1.find('a').attr('href'), host).href;
 
             return {
@@ -32,22 +28,36 @@ async function getItems(ctx, url, host, tableClass, timeStyleClass1, titleStyleC
 
     const out = await Promise.all(
         list.map((item) =>
-            // 使用缓存
             cache.tryGet(item.link, async () => {
-                const response = await got({ url: item.link, https: { rejectUnauthorized: false } });
-                if (response.redirectUrls.length) {
-                    item.link = response.redirectUrls[0];
+                const response = await ofetch(item.link).catch(() => null);
+                if (!response || (response.status >= 300 && response.status < 400)) {
+                    // 响应为空或状态码表明发生了重定向
+                    return {
+                        ...item,
+                        description: '该通知无法直接预览，请点击原文链接↑查看',
+                    };
+                }
+                const $ = load(response);
+
+                item.title = $(`.${titleStyleClass}`).text();
+                const hasEmbeddedPDFScript = $('script:contains("showVsbpdfIframe")').length > 0;
+
+                if (hasEmbeddedPDFScript) {
                     item.description = '该通知无法直接预览，请点击原文链接↑查看';
                 } else {
-                    const $ = load(response.data);
-                    item.title = $(`.${titleStyleClass}`).text();
-                    item.description = $('.v_news_content')
-                        .html()
-                        .replaceAll('src="/', `src="${new URL('.', host).href}`)
-                        .replaceAll('href="/', `href="${new URL('.', host).href}`)
-                        .trim();
-                    item.pubDate = timezone(parseDate($(`.${timeStyleClass2}`).text().replace('发布时间：', '')), +8);
+                    const contentHtml = $('.v_news_content').html();
+                    const $content = load(contentHtml);
+                    $content('a').each(function () {
+                        const a = $(this);
+                        const href = a.attr('href');
+                        if (href && !href.startsWith('http')) {
+                            a.attr('href', new URL(href, host).href);
+                        }
+                    });
+                    item.description = $content.html();
                 }
+                item.pubDate = timezone(parseDate($(`.${timeStyleClass2}`).text().replace('发布时间：', '')), +8);
+
                 return item;
             })
         )

package/lib/routes/juejin/dynamic.ts

@@ -0,0 +1,114 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { parseDate } from '@/utils/parse-date';
+
+export const route: Route = {
+    path: '/dynamic/:id',
+    categories: ['programming'],
+    example: '/juejin/dynamic/3051900006845944',
+    parameters: { id: '用户 id, 可在用户页 URL 中找到' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['juejin.cn/user/:id'],
+        },
+    ],
+    name: '用户动态',
+    maintainers: ['CaoMeiYouRen'],
+    handler,
+};
+
+async function handler(ctx) {
+    const id = ctx.req.param('id');
+
+    const response = await got({
+        method: 'get',
+        url: 'https://api.juejin.cn/user_api/v1/user/dynamic',
+        searchParams: {
+            user_id: id,
+            cursor: 0,
+        },
+    });
+    const list = response.data.data.list;
+
+    const username = list[0].user.user_name;
+
+    const items = list.map((e) => {
+        const { target_type, target_data, action, time } = e; // action: 0.发布文章；1.点赞文章；2.发布沸点；3.点赞沸点；4.关注用户
+        let title: string | undefined;
+        let description: string | undefined;
+        let pubDate: Date | undefined;
+        let author: string | undefined;
+        let link: string | undefined;
+        let category: string[] | undefined;
+        switch (target_type) {
+            case 'short_msg': {
+                // 沸点/点赞等
+                const { msg_Info, author_user_info, msg_id, topic } = target_data;
+                const { content, pic_list, ctime } = msg_Info;
+                title = content;
+                const imgs = pic_list.map((img) => `<img src="${img}"><br>`).join('');
+                description = `${content.replaceAll('\n', '<br>')}<br>${imgs}`;
+                pubDate = parseDate(Number(ctime) * 1000);
+                author = author_user_info.user_name;
+                link = `https://juejin.cn/pin/${msg_id}`;
+                category = topic.title;
+                if (action === 3) {
+                    title = `${username} 赞了这篇沸点//@${author}：${title}`;
+                    description = `${username} 赞了这篇沸点//@${author}：${description}`;
+                }
+                break;
+            }
+            case 'article': {
+                // 文章
+                const { article_id, article_info, author_user_info, tags } = target_data;
+                const { ctime, brief_content } = article_info;
+                title = article_info.title;
+                description = brief_content;
+                pubDate = parseDate(Number(ctime) * 1000);
+                author = author_user_info.user_name;
+                link = `https://juejin.cn/post/${article_id}`;
+                category = [...new Set([target_data.category.category_name, ...tags.map((t) => t.tag_name)])];
+                if (action === 1) {
+                    title = `${username} 赞了这篇文章//@${author}：${title}`;
+                }
+                break;
+            }
+            case 'user': {
+                // 关注用户
+                const { user_name, user_id } = target_data;
+                title = `${username} 关注了 ${user_name}`;
+                description = `${user_name}<br>简介：${target_data.description}`;
+                author = user_name;
+                link = `https://juejin.cn/user/${user_id}`;
+                pubDate = parseDate(time * 1000);
+                break;
+            }
+            default:
+                break;
+        }
+        return {
+            title,
+            description,
+            pubDate,
+            author,
+            link,
+            category,
+            guid: link,
+        };
+    });
+    return {
+        title: `掘金用户动态-${username}`,
+        link: `https://juejin.cn/user/${id}/`,
+        description: `掘金用户动态-${username}`,
+        item: items,
+        author: username,
+    };
+}

package/lib/routes/kadokawa/blog.ts

@@ -0,0 +1,134 @@
+import { Route } from '@/types';
+import { getCurrentPath } from '@/utils/helpers';
+const __dirname = getCurrentPath(import.meta.url);
+
+import cache from '@/utils/cache';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import { art } from '@/utils/render';
+import path from 'node:path';
+
+export const handler = async (ctx) => {
+    const limit = ctx.req.query('limit') ? Number.parseInt(ctx.req.query('limit'), 10) : 10;
+
+    const rootUrl = 'https://www.kadokawa.com.tw';
+    const currentUrl = new URL('blog/posts', rootUrl).href;
+
+    const { data: response } = await got(currentUrl);
+
+    const $ = load(response);
+
+    const language = $('html').prop('lang');
+
+    let items = $('div.List-item')
+        .slice(0, limit)
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            const image = item.find('div.List-item-excerpt img').prop('src')?.split(/\?/)[0] ?? undefined;
+            const title = item.find('h2.List-item-title').text();
+            const description = art(path.join(__dirname, 'templates/description.art'), {
+                images: image
+                    ? [
+                          {
+                              src: image,
+                              alt: title,
+                          },
+                      ]
+                    : undefined,
+                intro: item.find('div.List-item-preview').text(),
+            });
+
+            return {
+                title,
+                description,
+                pubDate: parseDate(item.find('span.primary-border-color-after').text()),
+                link: new URL(item.find('a').prop('href'), rootUrl).href,
+                content: {
+                    html: description,
+                    text: item.find('div.List-item-preview').text(),
+                },
+                image,
+                banner: image,
+                language,
+                enclosure_url: image,
+                enclosure_type: image ? `image/${image.split(/\./).pop()}` : undefined,
+                enclosure_title: title,
+            };
+        });
+
+    items = await Promise.all(
+        items.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const { data: detailResponse } = await got(item.link);
+
+                const $$ = load(detailResponse);
+
+                const title = $$('h1.Post-title').text().trim();
+                const description = art(path.join(__dirname, 'templates/description.art'), {
+                    description: $$('div.Post-content').html(),
+                });
+                const image = $$('meta[property="og:image"]').prop('content')?.split(/\?/)[0] ?? undefined;
+
+                item.title = title;
+                item.description = description;
+                item.pubDate = parseDate($$('div.Post-date').text().trim());
+                item.content = {
+                    html: description,
+                    text: $$('div.Post-content').text(),
+                };
+                item.image = image;
+                item.banner = image;
+                item.language = language;
+                item.enclosure_url = image;
+                item.enclosure_type = image ? `image/${image.split(/\./).pop()}` : undefined;
+                item.enclosure_title = title;
+
+                return item;
+            })
+        )
+    );
+
+    const image = new URL($('meta[property="og:image"]').prop('content'), rootUrl).href;
+
+    return {
+        title: $('title').text(),
+        description: $('meta[property="og:description"]').prop('content'),
+        link: currentUrl,
+        item: items,
+        allowEmpty: true,
+        image,
+        author: $('meta[property="og:site_name"]').prop('content'),
+        language,
+    };
+};
+
+export const route: Route = {
+    path: '/blog',
+    name: '角編新聞台',
+    url: 'kadokawa.com.tw',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/kadokawa/blog',
+    parameters: undefined,
+    description: '',
+    categories: ['blog'],
+
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['kadokawa.com.tw/blog/posts'],
+            target: '/blog',
+        },
+    ],
+};

package/lib/routes/kadokawa/namespace.ts

@@ -0,0 +1,8 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: '台灣角川',
+    url: 'kadokawa.com.tw',
+    categories: ['shopping'],
+    description: 'TAIWAN KADOKAWA',
+};

package/lib/routes/kadokawa/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/kakuyomu/namespace.ts

@@ -0,0 +1,6 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'カクヨム',
+    url: 'kakuyomu.jp',
+};

package/lib/routes/kakuyomu/types.ts

@@ -0,0 +1,6 @@
+export interface NextDataEpisode {
+    __typename: 'Episode';
+    id: string;
+    title: string;
+    publishedAt: string;
+}

package/lib/routes/kakuyomu/works.ts

@@ -0,0 +1,73 @@
+import type { Data, DataItem, Route } from '@/types';
+import { load } from 'cheerio';
+import type { Context } from 'hono';
+import ofetch from '@/utils/ofetch';
+import cache from '@/utils/cache';
+import type { NextDataEpisode } from './types';
+import { parseDate } from '@/utils/parse-date';
+
+export const route: Route = {
+    name: '投稿',
+    categories: ['reading'],
+    path: '/works/:id',
+    example: '/kakuyomu/works/1177354054894027232',
+    parameters: {
+        id: '投稿 ID',
+    },
+    maintainers: ['KarasuShin'],
+    handler,
+    features: {
+        supportRadar: true,
+    },
+    radar: [
+        {
+            source: ['kakuyomu.jp/works/:id'],
+            target: '/works/:id',
+        },
+    ],
+};
+
+async function handler(ctx: Context): Promise<Data> {
+    const id = ctx.req.param('id');
+    const url = `https://kakuyomu.jp/works/${id}`;
+    const limit = Number.parseInt(ctx.req.query('limit') || '10');
+    const $ = load(await ofetch(url));
+
+    const nextData = JSON.parse($('#__NEXT_DATA__').text());
+
+    const {
+        props: {
+            pageProps: { __APOLLO_STATE__ },
+        },
+    } = nextData;
+
+    const {
+        [`Work:${id}`]: { title, catchphrase },
+    } = __APOLLO_STATE__;
+
+    const values = Object.values(__APOLLO_STATE__);
+    const episodes = values.filter((value) => value.__typename === 'Episode') as NextDataEpisode[];
+    const items = (await Promise.all(
+        episodes
+            .sort((a, b) => b.publishedAt.localeCompare(a.publishedAt))
+            .slice(0, limit)
+            .map((item) => {
+                const episodeUrl = `https://kakuyomu.jp/works/${id}/episodes/${item.id}`;
+                return cache.tryGet(episodeUrl, async () => {
+                    const $ = load(await ofetch(episodeUrl));
+                    const description = $('.widget-episodeBody').html();
+                    return {
+                        title: item.title,
+                        description,
+                        pubDate: parseDate(item.publishedAt),
+                    };
+                });
+            })
+    )) as DataItem[];
+
+    return {
+        title,
+        description: catchphrase,
+        item: items,
+    };
+}

package/lib/routes/kaopu/news.ts

@@ -5,7 +5,7 @@ import { parseDate } from '@/utils/parse-date';
 export const route: Route = {
     path: '/news/:language?',
     categories: ['new-media'],
-    example: '/news/zh-hans',
+    example: '/kaopu/news/zh-hans',
     parameters: {
         language: '语言',
     },
@@ -16,8 +16,8 @@ export const route: Route = {
     ],
     name: '全部',
     maintainers: ['fashioncj'],
-    description: `| 简体中文    | 繁体中文     | 
-  | ------- | -------- | 
+    description: `| 简体中文    | 繁体中文     |
+  | ------- | -------- |
   | zh-hans | zh-hant | `,
     handler,
 };

package/lib/routes/kcna/utils.ts

@@ -54,7 +54,7 @@ const fetchVideo = (ctx, url) =>
         const js = $('script[type="text/javascript"]:not([src])').html();
         let sources = js.match(/<[^>]*source[^>]+src[^>]+>/g);
         sources = sources && sources.map((item) => item.replaceAll("'", '"').replaceAll(/src="([^"]+)"/g, `src="${rootUrl}$1"`));
-        return `<video controls preload="none">${sources.join('\n')}</video>`;
+        return `<video controls preload="metadata">${sources.join('\n')}</video>`;
     });
 
 export { parseJucheDate, fixDesc, fetchPhoto, fetchVideo };

package/lib/routes/kelownacapnews/namespace.ts

@@ -0,0 +1,6 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'Kelowna Capital News',
+    url: 'www.kelownacapnews.com',
+};

package/lib/routes/kelownacapnews/news.ts

@@ -0,0 +1,93 @@
+import { DataItem, Route } from '@/types';
+import ofetch from '@/utils/ofetch';
+import { load } from 'cheerio';
+import cache from '@/utils/cache';
+import { parseDate } from '@/utils/parse-date';
+
+export const route: Route = {
+    path: '/:type',
+    categories: ['new-media'],
+    example: '/kelownacapnews/local-news',
+    parameters: { type: 'Type of news' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['www.kelownacapnews.com/:type'],
+            target: '/:type',
+        },
+    ],
+    name: 'News',
+    maintainers: ['hualiong'],
+    url: 'www.kelownacapnews.com',
+    description: `\`type\` is as follows:
+  
+| News type     | Value         | News type    | Value        |
+| ------------- | ------------- | ------------ | ------------ |
+| News          | news          | Sports       | sports       |
+| Local News    | local-news    | Business     | business     |
+| Canadian News | national-news | Trending Now | trending-now |
+| World News    | world-news    | Opinion      | opinion      |
+| Entertainment | entertainment |              |              |`,
+    handler: async (ctx) => {
+        const type = ctx.req.param('type');
+        const baseURL = 'https://www.kelownacapnews.com';
+
+        const response = await ofetch(`${baseURL}/${type}`);
+        const $ = load(response);
+
+        const list = $('.media')
+            .toArray()
+            .map((item): DataItem => {
+                const a = $(item);
+                return {
+                    title: a.find('.media-heading').text(),
+                    pubDate: parseDate(a.find('.media-links time').attr('datetime')!),
+                    link: baseURL + a.attr('href'),
+                };
+            });
+
+        const items = await Promise.all(
+            list.map((item) =>
+                cache.tryGet(item.link!, async () => {
+                    const response = await ofetch(item.link!);
+                    const $ = load(response);
+
+                    let image = $('.details-file')!;
+                    if (!image.length) {
+                        image = $('#sliderImgs .tablist-item .galleryWrap');
+                    }
+                    const byline = $('.details-byline');
+                    const profileTitle = byline.find('.profile-title');
+                    if (profileTitle.length) {
+                        item.author = profileTitle.find('a').text();
+                    }
+                    let label = '';
+                    if (image.length > 1) {
+                        for (const e of image.toArray()) {
+                            const img = $(e);
+                            label += `<figure style="margin: 10px 0 0 0"><img src='${img.data('src')}' /><figcaption>${img.attr('title')}</figcaption></figure>`;
+                        }
+                    } else {
+                        label = `<figure style="margin: 0">${image.html()!}</figure>`;
+                    }
+                    item.description = label + $('.details-body').html()!;
+
+                    return item;
+                })
+            )
+        );
+
+        return {
+            title: `${$('.body-title').text()} - Kelowna Capital News`,
+            link: `${baseURL}/${type}`,
+            item: items as DataItem[],
+        };
+    },
+};

package/lib/routes/kemono/index.ts

@@ -140,6 +140,26 @@ async function handler(ctx) {
                     desc += kemonoFile;
                 }
 
+                let enclosureInfo = {};
+                load(desc)('audio source, video source').each(function () {
+                    const src = $(this).attr('src') ?? '';
+                    const mimeType =
+                        {
+                            m4a: 'audio/mp4',
+                            mp3: 'audio/mpeg',
+                            mp4: 'video/mp4',
+                        }[src.replace(/.*\./, '').toLowerCase()] || null;
+
+                    if (mimeType === null) {
+                        return;
+                    }
+
+                    enclosureInfo = {
+                        enclosure_url: new URL(src, rootUrl).toString(),
+                        enclosure_type: mimeType,
+                    };
+                });
+
                 return {
                     title: i.title,
                     description: desc,
@@ -147,6 +167,7 @@ async function handler(ctx) {
                     pubDate: parseDate(i.published),
                     guid: `${apiUrl}/${i.service}/user/${i.user}/post/${i.id}`,
                     link: `${rootUrl}/${i.service}/user/${i.user}/post/${i.id}`,
+                    ...enclosureInfo,
                 };
             });
     }