Merge pull request #2479 from DIYgod/master

[pull] master from diygod:master

lib/routes/dehenglaw/index.ts

@@ -0,0 +1,128 @@
+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 { language = 'CN', category = 'paper' } = ctx.req.param();
+    const limit = ctx.req.query('limit') ? Number.parseInt(ctx.req.query('limit'), 10) : 6;
+
+    const rootUrl = 'https://www.dehenglaw.com';
+    const currentUrl = new URL(`${language}/${category}/0008/000901.aspx`, rootUrl).href;
+
+    const { data: response } = await got(currentUrl);
+
+    const $ = load(response);
+
+    let items = $('div.news_box ul li')
+        .slice(0, limit)
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            const title = item.find('h2').text();
+            const description = art(path.join(__dirname, 'templates/description.art'), {
+                intro: item.find('div.deheng_newscontent p').text(),
+            });
+
+            return {
+                title,
+                description,
+                pubDate: parseDate(item.find('span').text(), 'YYYY/M/D'),
+                link: item.find('a').first().prop('href'),
+            };
+        });
+
+    items = await Promise.all(
+        items.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const { data: detailResponse } = await got(item.link);
+
+                const $$ = load(detailResponse);
+
+                const description =
+                    item.description +
+                    art(path.join(__dirname, 'templates/description.art'), {
+                        description: $$('div.news_content').html(),
+                    });
+                const image = $$('div.news_content img').prop('src');
+
+                item.description = description;
+                item.author = $$('div.name h4 a').text();
+                item.content = {
+                    html: description,
+                    text: $$('div.news_content').text(),
+                };
+                item.image = image;
+                item.banner = image;
+
+                return item;
+            })
+        )
+    );
+
+    const image = $('div.logo_content a img').prop('src');
+
+    return {
+        title: $('title')
+            .text()
+            .replace(/\|.*?$/, `| ${$('li.onthis').text()}`),
+        description: $('meta[name="Description"]').prop('content'),
+        link: currentUrl,
+        item: items,
+        allowEmpty: true,
+        image,
+        author: $('meta[name="Description"]').prop('content'),
+    };
+};
+
+export const route: Route = {
+    path: '/:language?/:category?',
+    name: '德恒探索',
+    url: 'dehenglaw.com',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/dehenglaw/CN/paper',
+    parameters: { language: '语言，默认为中文，即 CN，可在对应分类页 URL 中找到，可选 CN 和 EN', category: '分类，默认为专业文章，即 paper，可在对应分类页 URL 中找到' },
+    description: `:::tip
+  若订阅 [专业文章](https://dehenglaw.com/)，网址为 \`https://www.dehenglaw.com/CN/paper/0008/000902.aspx\`。截取 \`https://dehenglaw.com/\` 到末尾 \`/0008/000902.aspx\` 的部分 \`CN/paper\` 作为参数填入，此时路由为 [\`/dehenglaw/CN/paper\`](https://rsshub.app/dehenglaw/CN/paper)。
+
+  | 专业文章 | 出版物  | 德恒论坛 |
+  | -------- | ------- | -------- |
+  | paper    | publish | luntan   |
+  :::`,
+    categories: ['new-media'],
+
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            title: '专业文章',
+            source: ['dehenglaw.com/:language/paper/0008/000902.aspx'],
+            target: '/:language/paper',
+        },
+        {
+            title: '出版物',
+            source: ['dehenglaw.com/:language/publish/0008/000903.aspx'],
+            target: '/:language/publish',
+        },
+        {
+            title: '德恒论坛',
+            source: ['dehenglaw.com/:language/luntan/0008/000901.aspx'],
+            target: '/:language/luntan',
+        },
+    ],
+};

lib/routes/dehenglaw/namespace.ts

@@ -0,0 +1,8 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: '德恒律师事务所',
+    url: 'dehenglaw.com',
+    categories: ['new-media'],
+    description: '',
+};

lib/routes/dehenglaw/templates/description.art

@@ -0,0 +1,7 @@
+{{ if intro }}
+  <blockquote>{{ intro }}</blockquote>
+{{ /if }}
+
+{{ if description }}
+  {{@ description }}
+{{ /if }}

lib/routes/reuters/common.ts

@@ -64,13 +64,13 @@ export const route: Route = {
 };
 
 async function handler(ctx) {
-    const MUST_FETCH_BY_TOPICS = new Set(['authors']);
+    const MUST_FETCH_BY_TOPICS = new Set(['authors', 'tags']);
     const CAN_USE_SOPHI = ['world'];
 
     const category = ctx.req.param('category');
     const topic = ctx.req.param('topic') ?? (category === 'authors' ? 'reuters' : '');
     const limit = ctx.req.query('limit') ? Number.parseInt(ctx.req.query('limit')) : 20;
-    const useSophi = ctx.req.query('sophi') === 'true' && 'topic' !== '' && CAN_USE_SOPHI.includes(category);
+    const useSophi = ctx.req.query('sophi') === 'true' && topic !== '' && CAN_USE_SOPHI.includes(category);
 
     const section_id = `/${category}/${topic ? `${topic}/` : ''}`;
     const { title, description, rootUrl, response } = await (async () => {

lib/utils/wechat-mp.ts

@@ -156,7 +156,7 @@ class ExtractMetadata {
             $,
             (script) => {
                 const scriptText = $(script).text();
-                const metadataExtracted = <Record<string, string>>this.doExtract(this.commonMetadataToBeExtracted, scriptText);
+                const metadataExtracted = <Record<string, string>> this.doExtract(this.commonMetadataToBeExtracted, scriptText);
                 const showType = showTypeMapReverse[metadataExtracted.showType];
                 const realShowType = showTypeMapReverse[metadataExtracted.realShowType];
                 metadataExtracted.sourceUrl = metadataExtracted.sourceUrl && fixUrl(metadataExtracted.sourceUrl);
@@ -191,7 +191,7 @@ class ExtractMetadata {
             $,
             (script) => {
                 const scriptText = $(script).text();
-                const metadataExtracted = <Record<string, string>>this.doExtract(this.audioMetadataToBeExtracted, scriptText);
+                const metadataExtracted = <Record<string, string>> this.doExtract(this.audioMetadataToBeExtracted, scriptText);
                 throw new LoopReturn(metadataExtracted);
             },
             {},
@@ -207,7 +207,7 @@ class ExtractMetadata {
             $,
             (script) => {
                 const scriptText = $(script).text();
-                const metadataExtracted = <Record<string, string[]>>this.doExtract(this.imgMetadataToBeExtracted, scriptText);
+                const metadataExtracted = <Record<string, string[]>> this.doExtract(this.imgMetadataToBeExtracted, scriptText);
                 if (Array.isArray(metadataExtracted.imgUrls)) {
                     metadataExtracted.imgUrls = metadataExtracted.imgUrls.map((url) => fixUrl(url));
                 }

package.json

@@ -25,8 +25,8 @@
         "build:docs": "tsx scripts/workflow/build-docs.ts",
         "dev": "cross-env NODE_ENV=dev tsx watch --no-cache lib/index.ts",
         "dev:cache": "cross-env NODE_ENV=production tsx watch lib/index.ts",
-        "format": "eslint --cache --fix \"**/*.{ts,tsx,js,yml}\" && prettier \"**/*.{ts,tsx,js,json}\" --write",
-        "format:check": "eslint --cache \"**/*.{ts,tsx,js,yml}\" && prettier \"**/*.{ts,tsx,js,json}\" --check",
+        "format": "prettier \"**/*.{ts,tsx,js,json}\" --write && eslint --cache --fix \"**/*.{ts,tsx,js,yml}\"",
+        "format:check": "prettier \"**/*.{ts,tsx,js,json}\" --check && eslint --cache \"**/*.{ts,tsx,js,yml}\"",
         "format:staged": "lint-staged",
         "lint": "eslint --cache .",
         "prepare": "husky || true",
@@ -55,7 +55,7 @@
         "@hono/zod-openapi": "0.11.0",
         "@notionhq/client": "2.2.15",
         "@postlight/parser": "2.2.3",
-        "@sentry/node": "7.110.1",
+        "@sentry/node": "7.111.0",
         "@tonyrl/rand-user-agent": "2.0.59",
         "aes-js": "3.1.2",
         "art-template": "4.13.2",
@@ -75,7 +75,7 @@
         "fanfou-sdk": "5.0.0",
         "form-data": "4.0.0",
         "googleapis": "134.0.0",
-        "hono": "4.2.4",
+        "hono": "4.2.5",
         "html-to-text": "9.0.5",
         "https-proxy-agent": "7.0.4",
         "iconv-lite": "0.6.3",