rsshub 1.0.0-master.f8ba9fc → 1.0.0-master.f8c1415

package/lib/routes/thepaper/utils.ts

@@ -1,9 +1,6 @@
-import { getCurrentPath } from '@/utils/helpers';
-const __dirname = getCurrentPath(import.meta.url);
-
 import cache from '@/utils/cache';
 import { load } from 'cheerio';
-import { parseDate } from '@/utils/parse-date';
+import { parseDate, parseRelativeDate } from '@/utils/parse-date';
 import got from '@/utils/got';
 import { art } from '@/utils/render';
 import path from 'node:path';
@@ -27,7 +24,7 @@ export default {
             // external link
             return defaultRssItem(item);
         }
-        const itemUrl = `https://m.thepaper.cn/detail/${item.contId}`;
+        const itemUrl = `https://m.thepaper.cn/${item.cornerLabelDesc && item.cornerLabelDesc === '短剧' ? 'series' : 'detail'}/${item.contId}`;
         return cache.tryGet(`${itemUrl}${useOldMode ? ':old' : ''}`, async () => {
             const res = await got(itemUrl);
             const data = JSON.parse(load(res.data)('#__NEXT_DATA__').html());
@@ -58,12 +55,17 @@ export default {
                 }
             }
 
+            let pubDate = parseDate(item.pubTimeLong || contentDetail.publishTime);
+            if (Number.isNaN(pubDate)) {
+                pubDate = parseRelativeDate(contentDetail.pubTime);
+            }
+
             const rss_item = {
                 title: contentDetail.name || contentDetail.shareName,
                 link: itemUrl,
                 description,
                 category: [...(contentDetail.tagList?.map((t) => t.tag) ?? []), contentDetail?.nodeInfo?.name ?? []],
-                pubDate: parseDate(item.pubTimeLong || contentDetail.pubTime || contentDetail.publishTime),
+                pubDate,
                 author: contentDetail.author || '',
                 media: {
                     content: {

package/lib/routes/thepetcity/index.ts

@@ -8,7 +8,7 @@ const baseUrl = 'https://thepetcity.co';
 
 export const route: Route = {
     path: '/:term?',
-    categories: ['new-media', 'popular'],
+    categories: ['new-media'],
     example: '/thepetcity',
     parameters: { term: '見下表，留空為全部文章' },
     radar: Object.entries(termsMap).map(([key, value]) => ({
@@ -21,12 +21,12 @@ export const route: Route = {
     handler,
     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      |`,
+| -------------------- | ------ |
+| 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) {

package/lib/routes/theverge/index.ts

@@ -1,12 +1,18 @@
 import { Route } from '@/types';
 import cache from '@/utils/cache';
-import got from '@/utils/got';
+import ofetch from '@/utils/ofetch';
 import parser from '@/utils/rss-parser';
 import { load } from 'cheerio';
+import path from 'node:path';
+import { art } from '@/utils/render';
+
+const excludeTypes = new Set(['NewsletterBlockType', 'RelatedPostsBlockType', 'ProductsTableBlockType', 'TableOfContentsBlockType']);
+
+const shouldKeep = (b: any) => !excludeTypes.has(b.__typename.trim());
 
 export const route: Route = {
     path: '/:hub?',
-    categories: ['new-media', 'popular'],
+    categories: ['new-media'],
     example: '/theverge',
     parameters: { hub: 'Hub, see below, All Posts by default' },
     features: {
@@ -22,102 +28,118 @@ export const route: Route = {
             source: ['theverge.com/:hub', 'theverge.com/'],
         },
     ],
-    name: 'The Verge',
+    name: 'Category',
     maintainers: ['HenryQW', 'vbali'],
     handler,
     description: `| Hub         | Hub name            |
-  | ----------- | ------------------- |
-  |             | All Posts           |
-  | android     | Android             |
-  | apple       | Apple               |
-  | apps        | Apps & Software     |
-  | blackberry  | BlackBerry          |
-  | culture     | Culture             |
-  | gaming      | Gaming              |
-  | hd          | HD & Home           |
-  | microsoft   | Microsoft           |
-  | photography | Photography & Video |
-  | policy      | Policy & Law        |
-  | web         | Web & Social        |
+| ----------- | ------------------- |
+|             | All Posts           |
+| android     | Android             |
+| apple       | Apple               |
+| apps        | Apps & Software     |
+| blackberry  | BlackBerry          |
+| culture     | Culture             |
+| gaming      | Gaming              |
+| hd          | HD & Home           |
+| microsoft   | Microsoft           |
+| photography | Photography & Video |
+| policy      | Policy & Law        |
+| web         | Web & Social        |
 
   Provides a better reading experience (full text articles) over the official one.`,
 };
 
+const renderBlock = (b) => {
+    if (!shouldKeep(b)) {
+        return '';
+    }
+    switch (b.__typename) {
+        case 'CoreEmbedBlockType':
+            return b.embedHtml;
+        case 'CoreGalleryBlockType':
+            return b.images.map((i) => `<figure><img src="${i.image.thumbnails.horizontal.url.split('?')[0]}" alt="${i.alt}" /><figcaption>${i.caption.html}</figcaption></figure>`).join('');
+        case 'CoreHeadingBlockType':
+            return `<h${b.level}>${b.contents.html}</h${b.level}>`;
+        case 'CoreHTMLBlockType':
+            return b.markup;
+        case 'CoreImageBlockType':
+            return `<figure><img src="${b.thumbnail.url.split('?')[0]}" alt="${b.alt}" /><figcaption>${b.caption.html}</figcaption></figure>`;
+        case 'CoreListBlockType':
+            return `${b.ordered ? '<ol>' : '<ul>'}${b.items.map((i) => `<li>${i.contents.html}</li>`).join('')}${b.ordered ? '</ol>' : '</ul>'}`;
+        case 'CoreParagraphBlockType':
+            return b.tempContents.map((c) => c.html).join('');
+        case 'CorePullquoteBlockType':
+            return `<blockquote>${b.contents.html}</blockquote>`;
+        case 'CoreQuoteBlockType':
+            return `<blockquote>${b.children.map((child) => renderBlock(child)).join('')}</blockquote>`;
+        case 'CoreSeparatorBlockType':
+            return '<hr>';
+        case 'HighlightBlockType':
+            return b.children.map((c) => renderBlock(c)).join('');
+        case 'ImageCompareBlockType':
+            return `<figure><img src="${b.leftImage.thumbnails.horizontal.url.split('?')[0]}" alt="${b.leftImage.alt}" /><img src="${b.rightImage.thumbnails.horizontal.url.split('?')[0]}" alt="${b.rightImage.alt}" /><figcaption>${b.caption.html}</figcaption></figure>`;
+        case 'ImageSliderBlockType':
+            return b.images.map((i) => `<figure><img src="${i.image.originalUrl.split('?')[0]}" alt="${i.alt}" /><figcaption>${i.caption.html}</figcaption></figure>`).join('');
+        case 'MethodologyAccordionBlockType':
+            return `<h2>${b.heading.html}</h2>${b.sections.map((s) => `<h3>${s.heading.html}</h3>${s.content.html}`).join('')}`;
+        case 'ProductBlockType': {
+            const product = b.product;
+            return `<div><figure><img src="${product.image.thumbnails.horizontal.url.split('?')[0]}" alt="${product.image.alt}" /><figcaption>${product.image.alt}</figcaption></figure><br><a href="${product.bestRetailLink.url}">${product.title} $${product.bestRetailLink.price}</a><br>${product.description.html}${product.pros.html ? `<br>The Good${product.pros.html}The Bad${product.cons.html}` : ''}</div>`;
+        }
+        case 'TableBlockType':
+            return `<table><tr>${b.header.map((cell) => `<th>${cell}</th>`).join('')}</tr>${b.rows.map((row) => `<tr>${row.map((cell) => `<td>${cell}</td>`).join('')}</tr>`).join('')}</table>`;
+        default:
+            throw new Error(`Unsupported block type: ${b.__typename}`);
+    }
+};
+
 async function handler(ctx) {
-    const link = ctx.req.param('hub') ? `https://www.theverge.com/${ctx.req.param('hub')}/rss/index.xml` : 'https://www.theverge.com/rss/index.xml';
+    const link = ctx.req.param('hub') ? `https://www.theverge.com/rss/${ctx.req.param('hub')}/index.xml` : 'https://www.theverge.com/rss/index.xml';
 
     const feed = await parser.parseURL(link);
 
     const items = await Promise.all(
         feed.items.map((item) =>
             cache.tryGet(item.link, async () => {
-                const response = await got(item.link);
-
-                const $ = load(response.data);
+                const response = await ofetch(item.link);
 
-                const content = $('#content');
-                const body = $('.duet--article--article-body-component-container');
+                const $ = load(response);
 
-                // 处理封面图片
+                const nextData = JSON.parse($('script#__NEXT_DATA__').text());
+                const node = nextData.props.pageProps.hydration.responses.find((x) => x.operationName === 'PostLayoutQuery' || x.operationName === 'StreamLayoutQuery').data.node;
 
-                const cover = $('meta[property="og:image"]');
-
-                if (cover.length > 0) {
-                    $(`<img src=${cover[0].attribs.content}>`).insertBefore(body[0].childNodes[0]);
-                }
-
-                // 处理封面视频
-                $('div.l-col__main > div.c-video-embed, div.c-entry-hero > div.c-video-embed').each((i, e) => {
-                    const src = `https://volume.vox-cdn.com/embed/${e.attribs['data-volume-uuid']}?autoplay=false`;
-
-                    $(`<iframe src="${src}" style="border: 0; top: 0; left: 0; width: 100%; height: 100%; position: absolute;" allowfullscreen scrolling="no"></iframe>`).insertBefore(body[0].childNodes[0]);
+                let description = art(path.join(__dirname, 'templates/header.art'), {
+                    featuredImage: node.featuredImage,
+                    ledeMediaData: node.ledeMediaData,
                 });
 
-                // 处理封面视频
-                $('div.l-col__main > div.c-video-embed--media iframe').each((i, e) => {
-                    $(e).insertBefore(body[0].childNodes[0]);
-                });
-
-                // 处理文章图片
-                content.find('figure.e-image').each((i, e) => {
-                    let src, caption;
-
-                    // 处理 jpeg, png
-                    if ($(e).find('picture > source').length > 0) {
-                        src = $(e)
-                            .find('picture > img')[0]
-                            .attribs.srcset.match(/(?<=320w,).*?(?=520w)/g)[0]
-                            .trim();
-                    } else if ($(e).find('img.c-dynamic-image').length > 0) {
-                        // 处理 gif
-                        src = $(e).find('span.e-image__image')[0].attribs['data-original'];
-                    }
-
-                    // 处理 caption
-                    if ($(e).find('span.e-image__meta').length > 0) {
-                        caption = $(e).find('span.e-image__meta').text();
-                    }
-
-                    const figure = `<figure><img src=${src}>${caption ? `<br><figcaption>${caption}</figcaption>` : ''}</figure>`;
-
-                    $(figure).insertBefore(e);
-
-                    $(e).remove();
-                });
-
-                const lede = $('.duet--article--lede h2:first');
-                if (lede[0]) {
-                    lede.insertBefore(body[0].childNodes[0]);
+                description += node.blocks.map((b) => renderBlock(b)).join('<br><br>');
+
+                if (node.__typename === 'StreamResourceType') {
+                    description += node.posts.edges
+                        .map(({ node: n }) => {
+                            let d =
+                                `<h2><a href="${n.permalink}">${n.promo.headline || n.title}</a></h2>` +
+                                art(path.join(__dirname, 'templates/header.art'), {
+                                    ledeMediaData: n.ledeMediaData,
+                                });
+                            switch (n.__typename) {
+                                case 'PostResourceType':
+                                    d += n.excerpt.map((e) => e.contents.html).join('<br>');
+                                    break;
+                                case 'QuickPostResourceType':
+                                    d += n.blocks.map((b) => renderBlock(b)).join('<br>');
+                                    break;
+                                default:
+                                    break;
+                            }
+                            return d;
+                        })
+                        .join('<br>');
                 }
 
-                // 移除无用 DOM
-                content.find('.duet--article--comments-join-the-conversation').remove();
-                content.find('.duet--recirculation--related-list').remove();
-                delete item.content;
-                delete item.contentSnippet;
-                delete item.isoDate;
-
-                item.description = body.html();
+                item.description = description;
+                item.category = node.categories;
 
                 return item;
             })

package/lib/routes/theverge/templates/header.art

@@ -0,0 +1,19 @@
+{{ if featuredImage }}
+    <figure>
+        <img src="{{ featuredImage.image.originalUrl.split('?')[0] }}" alt="{{ featuredImage.image.originalUrl.alt }}">
+        <figcaption>{{ featuredImage.image.title }}</figcaption>
+    </figure>
+{{ /if }}
+
+{{ if ledeMediaData }}
+    {{ if ledeMediaData.__typename === 'LedeMediaEmbedType'}}
+        {{@ ledeMediaData.embedHtml }}
+    {{ else if ledeMediaData.__typename === 'LedeMediaImageType' && !featuredImage }}
+        <figure>
+            <img src="{{ ledeMediaData.image.thumbnails.horizontal.url.split('?')[0] }}" alt="{{ ledeMediaData.image.title }}">
+            <figcaption>{{ ledeMediaData.image.credit.plaintext || ledeMediaData.image.title }}</figcaption>
+        </figure>
+    {{ else if ledeMediaData.__typename === 'LedeMediaVideoType'}}
+        <iframe src="https://volume.vox-cdn.com/embed/{{ ledeMediaData.video.volumeUuid }}" allowfullscreen></iframe>
+    {{ /if }}
+{{ /if }}

package/lib/routes/thewirehindi/category.ts

@@ -0,0 +1,78 @@
+import { Data, Route } from '@/types';
+import got from '@/utils/got';
+import { mapPostToItem } from './utils';
+
+const categories = [
+    { value: 'interview', label: 'इंटरव्यू', id: 92820 },
+    { value: 'audio', label: 'ऑडियो', id: 5153 },
+    { value: 'kala-sahitya', label: 'कला-साहित्य', id: 101061 },
+    { value: 'campus', label: 'कैंपस', id: 5261 },
+    { value: 'covid-19', label: 'कोविड-19', id: 73034 },
+    { value: 'jan-ki-baat', label: 'जन की बात', id: 985 },
+    { value: 'duniya', label: 'दुनिया', id: 33 },
+    { value: 'north-east', label: 'नॉर्थ ईस्ट', id: 5834 },
+    { value: 'prasangik', label: 'प्रासंगिक', id: 3394 },
+    { value: 'bharat', label: 'भारत', id: 30 },
+    { value: 'media', label: 'मीडिया', id: 3338 },
+    { value: 'media-bol', label: 'मीडिया बोल', id: 7963 },
+    { value: 'rajneeti', label: 'राजनीति', id: 31 },
+    { value: 'vichar', label: 'विचार', id: 73061 },
+    { value: 'vigyan', label: 'विज्ञान', id: 32 },
+    { value: 'vishesh', label: 'विशेष', id: 2494 },
+    { value: 'video', label: 'वीडियो', id: 34 },
+    { value: 'samaj', label: 'समाज', id: 28 },
+    { value: 'ham-bhi-bharat', label: 'हम भी भारत', id: 14383 },
+    { value: 'hamare-bare-mein', label: 'हमारे बारे में', id: 29 },
+];
+
+export const route: Route = {
+    path: '/category/:category',
+    categories: ['new-media'],
+    example: '/thewirehindi/category/bharat',
+    parameters: {
+        category: {
+            description: 'Category name',
+            options: categories.map(({ value, label }) => ({ value, label })),
+        },
+    },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['thewirehindi.com/category/*'],
+        },
+    ],
+    name: 'Category',
+    maintainers: ['Rjnishant530'],
+    handler,
+    url: 'thewirehindi.com/',
+};
+
+async function handler(ctx) {
+    const { category } = ctx.req.param();
+    const categoryData = categories.find((cat) => cat.value === category);
+
+    if (!categoryData) {
+        throw new Error(`Category "${category}" not found`);
+    }
+
+    const apiUrl = `https://thewirehindi.com/wp-json/wp/v2/posts?categories=${categoryData.id}&_embed`;
+    const { data } = await got(apiUrl);
+
+    const items = data.map((v) => mapPostToItem(v));
+
+    return {
+        title: `The Wire Hindi - ${categoryData.label}`,
+        link: `https://thewirehindi.com/category/${category}/`,
+        item: items,
+        description: `Latest news from The Wire Hindi - ${categoryData.label} category`,
+        logo: 'https://thewirehindi.com/wp-content/uploads/2023/05/cropped-The-wire-32x32.jpeg',
+        language: 'hi',
+    } as Data;
+}

package/lib/routes/thewirehindi/index.ts

@@ -0,0 +1,43 @@
+import { Data, Route } from '@/types';
+import got from '@/utils/got';
+import { mapPostToItem } from './utils';
+
+export const route: Route = {
+    path: '/',
+    categories: ['new-media'],
+    example: '/thewirehindi',
+    parameters: {},
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['thewirehindi.com/'],
+        },
+    ],
+    name: 'Latest News',
+    maintainers: ['Rjnishant530'],
+    handler,
+    url: 'thewirehindi.com/',
+};
+
+async function handler() {
+    const apiUrl = 'https://thewirehindi.com/wp-json/wp/v2/posts?_embed';
+    const { data } = await got(apiUrl);
+
+    const items = data.map((v) => mapPostToItem(v));
+
+    return {
+        title: 'The Wire Hindi - Latest News',
+        link: 'https://thewirehindi.com/',
+        item: items,
+        description: 'Latest news from The Wire Hindi',
+        logo: 'https://thewirehindi.com/wp-content/uploads/2023/05/cropped-The-wire-32x32.jpeg',
+        language: 'hi',
+    } as Data;
+}

package/lib/routes/thewirehindi/namespace.ts

@@ -0,0 +1,7 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'The Wire Hindi',
+    url: 'thewirehindi.com',
+    lang: 'hi',
+};

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

@@ -0,0 +1,9 @@
+{{@ excerpt }}
+<br><br>
+{{ if image }}
+<img src="{{ image }}" alt="{{ altText }}" style="max-width: 100%; height: auto;">
+<br><br>
+{{ /if }}
+{{ if content }}
+{{@ content }}
+{{ /if }}

package/lib/routes/thewirehindi/utils.ts

@@ -0,0 +1,33 @@
+import { DataItem } from '@/types';
+import { parseDate } from '@/utils/parse-date';
+import { art } from '@/utils/render';
+import path from 'node:path';
+
+export function mapPostToItem(post): DataItem {
+    const featuredMedia = post._embedded?.['wp:featuredmedia']?.find((v) => v.id === post.featured_media);
+    const image = featuredMedia?.source_url;
+    const altText = featuredMedia?.alt_text || featuredMedia?.title?.rendered || 'Featured Image';
+    return {
+        title: post.title.rendered,
+        description: art(path.join(__dirname, 'templates/description.art'), {
+            excerpt: post.excerpt.rendered,
+            content: post.content.rendered,
+            image,
+            altText,
+        }),
+        link: post.link,
+        pubDate: parseDate(post.date_gmt),
+        updated: parseDate(post.modified_gmt),
+        itunes_item_image: image,
+        image,
+        author:
+            post._embedded?.author
+                ?.map((v) => ({
+                    name: v.name,
+                    url: v.link,
+                    avatar: v.avatar_urls?.['96'] || v.avatar_urls?.['48'] || v.avatar_urls?.['24'],
+                }))
+                .join(', ') || 'The Wire Hindi',
+        category: post._embedded?.['wp:term']?.flat().map((v) => v.name),
+    } as DataItem;
+}