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

package/lib/routes/afr/latest.ts

@@ -0,0 +1,69 @@
+import { Route } from '@/types';
+import type { Context } from 'hono';
+
+import cache from '@/utils/cache';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+import { assetsConnectionByCriteriaQuery } from './query';
+import { getItem } from './utils';
+
+export const route: Route = {
+    path: '/latest',
+    categories: ['traditional-media'],
+    example: '/afr/latest',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['www.afr.com/latest', 'www.afr.com/'],
+        },
+    ],
+    name: 'Latest',
+    maintainers: ['TonyRL'],
+    handler,
+    url: 'www.afr.com/latest',
+};
+
+async function handler(ctx: Context) {
+    const limit = Number.parseInt(ctx.req.query('limit') ?? '10');
+    const response = await ofetch('https://api.afr.com/graphql', {
+        query: {
+            query: assetsConnectionByCriteriaQuery,
+            operationName: 'assetsConnectionByCriteria',
+            variables: {
+                brand: 'afr',
+                first: limit,
+                render: 'web',
+                types: ['article', 'bespoke', 'featureArticle', 'liveArticle', 'video'],
+                after: '',
+            },
+        },
+    });
+
+    const list = response.data.assetsConnectionByCriteria.edges.map(({ node }) => ({
+        title: node.asset.headlines.headline,
+        description: node.asset.about,
+        link: `https://www.afr.com${node.urls.published.afr.path}`,
+        pubDate: parseDate(node.dates.firstPublished),
+        updated: parseDate(node.dates.modified),
+        author: node.asset.byline,
+        category: [node.tags.primary.displayName, ...node.tags.secondary.map((tag) => tag.displayName)],
+        image: node.featuredImages && `https://static.ffx.io/images/${node.featuredImages.landscape16x9.data.id}`,
+    }));
+
+    const items = await Promise.all(list.map((item) => cache.tryGet(item.link, () => getItem(item))));
+
+    return {
+        title: 'Latest | The Australian Financial Review | AFR',
+        description: 'The latest news, events, analysis and opinion from The Australian Financial Review',
+        image: 'https://www.afr.com/apple-touch-icon-1024x1024.png',
+        link: 'https://www.afr.com/latest',
+        item: items,
+    };
+}

package/lib/routes/afr/namespace.ts

@@ -0,0 +1,7 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'The Australian Financial Review',
+    url: 'afr.com',
+    lang: 'en',
+};

package/lib/routes/afr/navigation.ts

@@ -0,0 +1,75 @@
+import { Route } from '@/types';
+import type { Context } from 'hono';
+
+import cache from '@/utils/cache';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+import { pageByNavigationPathQuery } from './query';
+import { getItem } from './utils';
+
+export const route: Route = {
+    path: '/navigation/:path{.+}',
+    categories: ['traditional-media'],
+    example: '/afr/navigation/markets',
+    parameters: {
+        path: 'Navigation path, can be found in the URL of the page',
+    },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['www.afr.com/path*'],
+        },
+    ],
+    name: 'Navigation',
+    maintainers: ['TonyRL'],
+    handler,
+    url: 'www.afr.com',
+};
+
+async function handler(ctx: Context) {
+    const { path } = ctx.req.param();
+    const limit = Number.parseInt(ctx.req.query('limit') ?? '10');
+
+    const response = await ofetch('https://api.afr.com/api/content-audience/afr/graphql', {
+        query: {
+            query: pageByNavigationPathQuery,
+            operationName: 'pageByNavigationPath',
+            variables: {
+                input: { brandKey: 'afr', navigationPath: `/${path}`, renderName: 'web' },
+                firstStories: limit,
+                afterStories: '',
+            },
+        },
+    });
+
+    const list = response.data.pageByNavigationPath.page.latestStoriesConnection.edges.map(({ node }) => ({
+        title: node.headlines.headline,
+        description: node.overview.about,
+        link: `https://www.afr.com${node.urls.canonical.path}`,
+        pubDate: parseDate(node.dates.firstPublished),
+        updated: parseDate(node.dates.modified),
+        author: node.byline
+            .filter((byline) => byline.type === 'AUTHOR')
+            .map((byline) => byline.author.name)
+            .join(', '),
+        category: [node.tags.primary.displayName, ...node.tags.secondary.map((tag) => tag.displayName)],
+        image: node.images && `https://static.ffx.io/images/${node.images.landscape16x9.mediaId}`,
+    }));
+
+    const items = await Promise.all(list.map((item) => cache.tryGet(item.link, () => getItem(item))));
+
+    return {
+        title: response.data.pageByNavigationPath.page.seo.title,
+        description: response.data.pageByNavigationPath.page.seo.description,
+        image: 'https://www.afr.com/apple-touch-icon-1024x1024.png',
+        link: `https://www.afr.com/${path}`,
+        item: items,
+    };
+}

package/lib/routes/afr/query.ts

@@ -0,0 +1,349 @@
+export const pageByNavigationPathQuery = `query pageByNavigationPath(
+    $input: PageByNavigationPathInput!
+    $firstStories: Int
+    $afterStories: Cursor
+  ) {
+    pageByNavigationPath(input: $input) {
+      error {
+        message
+        type {
+          class
+          ... on ErrorTypeInvalidRequest {
+            fields {
+              field
+              message
+            }
+          }
+        }
+      }
+      page {
+        ads {
+          suppress
+        }
+        description
+        id
+        latestStoriesConnection(first: $firstStories, after: $afterStories) {
+          edges {
+            node {
+              byline {
+                ...AssetBylineFragment
+              }
+              headlines {
+                headline
+              }
+              ads {
+                sponsor {
+                  name
+                }
+              }
+              overview {
+                about
+                label
+              }
+              type
+              dates {
+                firstPublished
+                published
+              }
+              id
+              publicId
+              images {
+                ...AssetImagesFragmentAudience
+              }
+              tags {
+                primary {
+                  ...TagFragmentAudience
+                }
+                secondary {
+                  ...TagFragmentAudience
+                }
+              }
+              urls {
+                ...AssetUrlsAudienceFragment
+              }
+            }
+          }
+          pageInfo {
+            endCursor
+            hasNextPage
+          }
+        }
+        name
+        seo {
+          canonical {
+            brand {
+              key
+            }
+          }
+          description
+          title
+        }
+        social {
+          image {
+            height
+            url
+            width
+          }
+        }
+      }
+      redirect
+    }
+  }
+  fragment AssetBylineFragment on AssetByline {
+    type
+    ... on AssetBylineAuthor {
+      author {
+        name
+        publicId
+        profile {
+          avatar
+          bio
+          body
+          canonical {
+            brand {
+              key
+            }
+          }
+          email
+          socials {
+            facebook {
+              publicId
+            }
+            twitter {
+              publicId
+            }
+          }
+          title
+        }
+      }
+    }
+    ... on AssetBylineName {
+      name
+    }
+  }
+  fragment AssetImagesFragmentAudience on ImageRenditions {
+    landscape16x9 {
+      ...ImageFragmentAudience
+    }
+    landscape3x2 {
+      ...ImageFragmentAudience
+    }
+    portrait2x3 {
+      ...ImageFragmentAudience
+    }
+    square1x1 {
+      ...ImageFragmentAudience
+    }
+  }
+  fragment ImageFragmentAudience on ImageRendition {
+    altText
+    animated
+    caption
+    credit
+    crop {
+      offsetX
+      offsetY
+      width
+      zoom
+    }
+    mediaId
+    mimeType
+    source
+    type
+  }
+  fragment AssetUrlsAudienceFragment on AssetURLs {
+    canonical {
+      brand {
+        key
+      }
+      path
+    }
+    external {
+      url
+    }
+    published {
+      brand {
+        key
+      }
+      path
+    }
+  }
+  fragment TagFragmentAudience on Tag {
+    company {
+      exchangeCode
+      stockCode
+    }
+    context {
+      name
+    }
+    description
+    displayName
+    externalEntities {
+      google {
+        placeId
+      }
+      wikipedia {
+        publicId
+        url
+      }
+    }
+    id
+    location {
+      latitude
+      longitude
+      postalCode
+      state
+    }
+    name
+    publicId
+    seo {
+      description
+      title
+    }
+    urls {
+      canonical {
+        brand {
+          key
+        }
+        path
+      }
+      published {
+        brand {
+          key
+        }
+        path
+      }
+    }
+  }`;
+
+export const assetsConnectionByCriteriaQuery = `query assetsConnectionByCriteria(
+    $after: ID
+    $brand: Brand!
+    $categories: [Int!]
+    $first: Int!
+    $render: Render!
+    $types: [AssetType!]!
+  ) {
+    assetsConnectionByCriteria(
+      after: $after
+      brand: $brand
+      categories: $categories
+      first: $first
+      render: $render
+      types: $types
+    ) {
+      edges {
+        cursor
+        node {
+          ...AssetFragment
+          sponsor {
+            name
+          }
+        }
+      }
+      error {
+        message
+        type {
+          class
+        }
+      }
+      pageInfo {
+        endCursor
+        hasNextPage
+      }
+    }
+  }
+  fragment AssetFragment on Asset {
+    asset {
+      about
+      byline
+      duration
+      headlines {
+        headline
+      }
+      live
+    }
+    assetType
+    dates {
+      firstPublished
+      modified
+      published
+    }
+    id
+    featuredImages {
+      landscape16x9 {
+        ...ImageFragment
+      }
+      landscape3x2 {
+        ...ImageFragment
+      }
+      portrait2x3 {
+        ...ImageFragment
+      }
+      square1x1 {
+        ...ImageFragment
+      }
+    }
+    label
+    tags {
+      primary: primaryTag {
+        ...AssetTag
+      }
+      secondary {
+        ...AssetTag
+      }
+    }
+    urls {
+      ...AssetURLs
+    }
+  }
+  fragment AssetTag on AssetTagDetails {
+    ...AssetTagAudience
+    shortID
+    slug
+  }
+  fragment AssetTagAudience on AssetTagDetails {
+    company {
+      exchangeCode
+      stockCode
+    }
+    context
+    displayName
+    id
+    name
+    urls {
+      canonical {
+        brand
+        path
+      }
+      published {
+        afr {
+          path
+        }
+      }
+    }
+  }
+  fragment AssetURLs on AssetURLs {
+    canonical {
+      brand
+      path
+    }
+    published {
+      afr {
+        path
+      }
+    }
+  }
+  fragment ImageFragment on Image {
+    data {
+      altText
+      aspect
+      autocrop
+      caption
+      cropWidth
+      id
+      offsetX
+      offsetY
+      zoom
+    }
+  }`;

package/lib/routes/afr/utils.ts

@@ -0,0 +1,80 @@
+import * as cheerio from 'cheerio';
+import ofetch from '@/utils/ofetch';
+
+export const getItem = async (item) => {
+    const response = await ofetch(item.link);
+    const $ = cheerio.load(response);
+
+    const reduxState = JSON.parse($('script#__REDUX_STATE__').text().replaceAll(':undefined', ':null').match('__REDUX_STATE__=(.*);')?.[1] || '{}');
+
+    const content = reduxState.page.content;
+    const asset = content.asset;
+
+    switch (content.assetType) {
+        case 'liveArticle':
+            item.description = asset.posts.map((post) => `<h2>${post.asset.headlines.headline}</h2>${post.asset.body}`).join('');
+            break;
+
+        case 'article':
+        case 'featureArticle':
+            item.description = renderArticle(asset, item.link);
+            break;
+
+        default:
+            throw new Error(`Unknown asset type: ${content.assetType} in ${item.link}`);
+    }
+
+    return item;
+};
+
+const renderArticle = (asset, link: string) => {
+    const $ = cheerio.load(asset.body, null, false);
+    $('x-placeholder').each((_, el) => {
+        const $el = $(el);
+        const id = $el.attr('id');
+        if (!id) {
+            $el.replaceWith('');
+        }
+
+        const placeholder = asset.bodyPlaceholders[id!];
+        switch (placeholder?.type) {
+            case 'callout':
+            case 'relatedStory':
+                $el.replaceWith('');
+                break;
+
+            case 'iframe':
+                $el.replaceWith(`<iframe src="${placeholder.data.url}" frameborder="0" allowfullscreen></iframe>`);
+                break;
+
+            case 'image':
+                $el.replaceWith(`<img src="https://static.ffx.io/images/${placeholder.data.fileName}" alt="${placeholder.data.altText}" />`);
+                break;
+
+            case 'linkArticle':
+                $el.replaceWith(placeholder.data.text);
+                break;
+
+            case 'linkExternal':
+                $el.replaceWith(`<a href="${placeholder.data.url}" target="_blank" rel="noopener">${placeholder.data.text}</a>`);
+                break;
+
+            case 'quote':
+                $el.replaceWith(placeholder.data.markup);
+                break;
+
+            case 'scribd':
+                $el.replaceWith(`<a href="${placeholder.data.url}" target="_blank" rel="noopener">View on Scribd</a>`);
+                break;
+
+            case 'twitter':
+                $el.replaceWith(`<a href="${placeholder.data.url}">${placeholder.data.url}</a>`);
+                break;
+
+            default:
+                throw new Error(`Unknown placeholder type: ${placeholder?.type} in ${link}`);
+        }
+    });
+
+    return $.html();
+};

package/lib/routes/aibase/discover.ts

@@ -189,7 +189,7 @@ export const route: Route = {
   | [法律](https://top.aibase.com/discover/138-139) | [138-139](https://rsshub.app/aibase/discover/138-139) |
   </details>
     `,
-    categories: ['new-media'],
+    categories: ['new-media', 'popular'],
 
     features: {
         requireConfig: false,

package/lib/routes/aibase/news.ts

@@ -58,7 +58,7 @@ export const route: Route = {
     },
     example: '/aibase/news',
     description: '获取 AI 资讯列表',
-    categories: ['new-media'],
+    categories: ['new-media', 'popular'],
     features: {
         requireConfig: false,
         requirePuppeteer: false,

package/lib/routes/aibase/topic.ts

@@ -90,7 +90,7 @@ export const route: Route = {
     | [娱乐](https://top.aibase.com/topic/%E5%A8%B1%E4%B9%90)                                             | [简历](https://top.aibase.com/topic/%E7%AE%80%E5%8E%86)                                             | [OpenAI](https://top.aibase.com/topic/OpenAI)                                            | [客户服务](https://top.aibase.com/topic/%E5%AE%A2%E6%88%B7%E6%9C%8D%E5%8A%A1)            | [室内设计](https://top.aibase.com/topic/%E5%AE%A4%E5%86%85%E8%AE%BE%E8%AE%A1) |
   </details>
     `,
-    categories: ['new-media'],
+    categories: ['new-media', 'popular'],
 
     features: {
         requireConfig: false,

package/lib/routes/air-level/index.ts

@@ -0,0 +1,49 @@
+import { Route } from '@/types';
+import ofetch from '@/utils/ofetch'; // 统一使用的请求库
+import { load } from 'cheerio'; // 类似 jQuery 的 API HTML 解析器
+
+export const route: Route = {
+    path: '/air/:area',
+    radar: [
+        {
+            source: ['m.air-level.com/air/:area/'],
+            target: '/air/:area',
+        },
+    ],
+    parameters: {
+        area: '地区',
+    },
+    name: '空气质量',
+    maintainers: ['lifetraveler'],
+    example: '/air-level/air/xian',
+    handler,
+};
+
+async function handler(ctx) {
+    const area = ctx.req.param('area');
+    const currentUrl = `https://m.air-level.com/air/${area}`;
+    const response = await ofetch(currentUrl);
+    const $ = load(response);
+
+    const title = $('body > div.container > div.row.page > div:nth-child(1) > h2').text().replaceAll('[]', '');
+
+    const table = $('body > div.container > div.row.page > div:nth-child(1) > div:nth-child(3) > table');
+
+    const qt = $('body > div.container > div.row.page > div:nth-child(1) > div.aqi-dv > div > span.aqi-bg.aqi-level-2').text();
+    const pubtime = $('body > div.container > div.row.page > div:nth-child(1) > div.aqi-dv > div > span.label.label-info').text();
+
+    const items = [
+        {
+            title: title + ' ' + qt + ' ' + pubtime,
+            link: currentUrl,
+            description: `<table border="1 solid black">${table.html()}</table>`,
+            guid: pubtime,
+        },
+    ];
+    return {
+        title,
+        item: items,
+        description: '订阅每个城市的天气质量',
+        link: currentUrl,
+    };
+}