feat(route): 新增 bing 搜索; 迁移 bing 每日图片到 v2; 新增 百度搜索; 迁移 搜狗特色LOGO 到 v2 规范；添加 搜狗搜索；添加 Google Search (#13936)

* fix(route): 修复 米游社 公告栏 template 错误

* feat(route): 新增 bing 搜索

* docs: Update other.mdx

* docs: fix docs

* feat(route): 新增 百度搜索

* fix(route): 修复 pubDate 解析错误

* fix(route): 优化 百度搜索的缓存，减轻反爬问题

* feat(route): 新增 360 搜索

* feat(route): 迁移 搜狗特色LOGO 到 v2 规范；添加 搜狗搜索

* fix(route): 百度搜索增加图片

* feat(route): 新增 Google Search

* fix(route): 修复 百度搜索相关问题

* fix(route): 修复 Google 相关问题

* fix(route): 修复 360 搜索

* fix(route): 修复 搜狗搜索

* fix(route): 修复 await 问题

* fix: 移除 google sites

* fix(route): 修复 缓存和过滤逻辑问题

* fix(route): 修复 360 搜索缺少 cookie 的问题

* fix(route): 修复 360 搜索 cookie 的问题

* feat(route): 移除 so.com 路由

* fix: merge conflict

---------

lib/router.js

@@ -376,7 +376,7 @@ router.get('/autotrader/:query', lazyloadRouteHandler('./routes/autotrader'));
 router.get('/geekpark/breakingnews', lazyloadRouteHandler('./routes/geekpark/breakingnews'));
 
 // 搜狗
-router.get('/sogou/doodles', lazyloadRouteHandler('./routes/sogou/doodles'));
+// router.get('/sogou/doodles', lazyloadRouteHandler('./routes/sogou/doodles'));
 
 // 香港天文台
 router.get('/hko/weather', lazyloadRouteHandler('./routes/hko/weather'));
@@ -690,7 +690,7 @@ router.get('/digitaling/articles/:category/:subcate?', lazyloadRouteHandler('./r
 router.get('/digitaling/projects/:category', lazyloadRouteHandler('./routes/digitaling/project'));
 
 // Bing壁纸
-router.get('/bing', lazyloadRouteHandler('./routes/bing/index'));
+// router.get('/bing', lazyloadRouteHandler('./routes/bing/index'));
 
 // AlgoCasts
 router.get('/algocasts', lazyloadRouteHandler('./routes/algocasts/all'));

lib/v2/baidu/maintainer.js

@@ -1,5 +1,6 @@
 module.exports = {
     '/gushitong/index': ['CaoMeiYouRen'],
+    '/search/:keyword': ['CaoMeiYouRen'],
     '/tieba/forum/good/:kw/:cid?/:sortBy?': ['u3u'],
     '/tieba/forum/:kw/:sortBy?': ['u3u'],
     '/tieba/post/:id': ['u3u'],

lib/v2/baidu/radar.js

@@ -1,6 +1,17 @@
 module.exports = {
     'baidu.com': {
         _name: '百度',
+        www: [
+            {
+                title: '搜索',
+                docs: 'https://docs.rsshub.app/routes/other#bai-du-sou-suo',
+                source: ['/'],
+                target: (params, url) => {
+                    const keyword = new URL(url).searchParams.get('wd');
+                    return `/baidu/search/${keyword}`;
+                },
+            },
+        ],
         gushitong: [
             {
                 title: '首页指数',
@@ -73,7 +84,7 @@ module.exports = {
         top: [
             {
                 title: '热搜榜单',
-                docs: 'https://docs.rsshub.app/routes/other#bai-du-re-sou',
+                docs: 'https://docs.rsshub.app/routes/other#bai-du-re-sou-bang-dan',
                 source: ['/board'],
                 target: (_, url) => `/baidu/top/${new URL(url).searchParams.get('tab')}`,
             },

lib/v2/baidu/router.js

@@ -1,5 +1,6 @@
 module.exports = (router) => {
     router.get('/gushitong/index', require('./gushitong'));
+    router.get('/search/:keyword', require('./search'));
     router.get('/tieba/forum/good/:kw/:cid?/:sortBy?', require('./tieba/forum'));
     router.get('/tieba/forum/:kw/:sortBy?', require('./tieba/forum'));
     router.get('/tieba/post/:id', require('./tieba/post'));

lib/v2/baidu/search.js

@@ -0,0 +1,54 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { art } = require('@/utils/render');
+const path = require('path');
+const renderDescription = (description, images) => art(path.join(__dirname, './templates/description.art'), { description, images });
+const config = require('@/config').value;
+
+module.exports = async (ctx) => {
+    const { keyword } = ctx.params;
+    const url = `https://www.baidu.com/s?wd=${encodeURIComponent(keyword)}`;
+    const key = `baidu-search:${url}`;
+
+    const items = await ctx.cache.tryGet(
+        key,
+        async () => {
+            const response = (await got(url)).data;
+            const visitedLinks = new Set();
+            const $ = cheerio.load(response);
+            const contentLeft = $('#content_left');
+            const containers = contentLeft.find('.c-container');
+            return containers
+                .map((i, el) => {
+                    const element = $(el);
+                    const link = element.find('h3 a').first().attr('href');
+                    if (link && !visitedLinks.has(link)) {
+                        visitedLinks.add(link);
+                        const imgs = element
+                            .find('img')
+                            .map((_j, _el) => $(_el).attr('src'))
+                            .toArray();
+                        const description = element.find('.c-gap-top-small [class^="content-right_"]').first().text() || element.find('.c-row').first().text() || element.find('.cos-row').first().text();
+                        return {
+                            title: element.find('h3').first().text(),
+                            description: renderDescription(description, imgs),
+                            link: element.find('h3 a').first().attr('href'),
+                            author: element.find('.c-row .c-color-gray').first().text() || '',
+                        };
+                    }
+                    return null;
+                })
+                .toArray()
+                .filter((e) => e?.link);
+        },
+        config.cache.routeExpire,
+        false
+    );
+
+    ctx.state.data = {
+        title: `${keyword} - 百度搜索`,
+        description: `${keyword} - 百度搜索`,
+        link: url,
+        item: items,
+    };
+};

lib/v2/baidu/templates/description.art

@@ -0,0 +1,6 @@
+{{@ description }}
+{{if images}}
+    {{each images}}
+        <img src="{{ $value }}">
+    {{/each}}
+{{/if}}

lib/v2/bing/daily-wallpaper.js

@@ -1,22 +1,21 @@
 const got = require('@/utils/got');
-const queryString = require('query-string');
 
 module.exports = async (ctx) => {
     const response = await got({
         method: 'get',
         prefixUrl: 'https://cn.bing.com',
         url: 'HPImageArchive.aspx',
-        searchParams: queryString.stringify({
+        searchParams: {
             format: 'js',
             idx: 0,
-            n: 7,
+            n: ctx.query.limit ? parseInt(ctx.query.limit, 10) : 7,
             mkt: 'zh-CN',
-        }),
+        },
     });
     const data = response.data;
     ctx.state.data = {
         title: 'Bing每日壁纸',
-        link: `https://cn.bing.com/`,
+        link: 'https://cn.bing.com/',
         item: data.images.map((item) => ({
             title: item.copyright,
             description: `<img src="https://cn.bing.com${item.url}">`,

lib/v2/bing/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/': ['FHYunCai'],
+    '/search/:keyword': ['CaoMeiYouRen'],
+};

lib/v2/bing/radar.js

@@ -0,0 +1,22 @@
+module.exports = {
+    'bing.com': {
+        _name: 'Bing',
+        cn: [
+            {
+                title: '每日壁纸',
+                docs: 'https://docs.rsshub.app/routes/picture#bing',
+                source: '/',
+                target: '/bing',
+            },
+            {
+                title: '搜索',
+                docs: 'https://docs.rsshub.app/routes/other#bing',
+                source: '/',
+                target: (params, url) => {
+                    const q = new URL(url).searchParams.get('q');
+                    return `/bing/search/${q}`;
+                },
+            },
+        ],
+    },
+};

lib/v2/bing/router.js

@@ -0,0 +1,4 @@
+module.exports = function (router) {
+    router.get('/', require('./daily-wallpaper'));
+    router.get('/search/:keyword', require('./search'));
+};

lib/v2/bing/search.js

@@ -0,0 +1,28 @@
+const parser = require('@/utils/rss-parser');
+const { parseDate } = require('@/utils/parse-date');
+const dayjs = require('dayjs');
+const localizedFormat = require('dayjs/plugin/localizedFormat');
+require('dayjs/locale/zh-cn');
+dayjs.extend(localizedFormat);
+
+module.exports = async (ctx) => {
+    const q = ctx.params.keyword;
+    const searchParams = new URLSearchParams({
+        format: 'rss',
+        q,
+    });
+    const url = new URL('https://cn.bing.com/search');
+    url.search = searchParams.toString();
+    const data = await parser.parseURL(url.toString());
+    ctx.state.data = {
+        title: data.title,
+        link: data.link,
+        description: data.description + ' - ' + data.copyright,
+        image: data.image.url,
+        item: data.items.map((e) => ({
+            ...e,
+            description: e.content,
+            pubDate: parseDate(e.pubDate, 'dddd, DD MMM YYYY HH:mm:ss [GMT]', 'zh-cn'),
+        })),
+    };
+};

lib/v2/google/maintainer.js

@@ -6,6 +6,7 @@ module.exports = {
     '/fonts/:sort?': ['Fatpandac'],
     '/news/:category/:locale': ['zoenglinghou'],
     '/scholar/:query': ['HenryQW'],
+    '/search/:keyword/:language?': ['CaoMeiYouRen'],
     '/sites/:id': ['hoilc'],
     '/sites/recentChanges/:id': ['nczitzk'],
 };

lib/v2/google/radar.js

@@ -1,6 +1,18 @@
 module.exports = {
     'google.com': {
         _name: '谷歌',
+        www: [
+            {
+                title: '搜索',
+                docs: 'https://docs.rsshub.app/routes/other#google',
+                source: '/',
+                target: (params, url, document) => {
+                    const q = new URL(url).searchParams.get('q');
+                    const lang = document.documentElement.lang;
+                    return `/google/search/${q}/${lang}`;
+                },
+            },
+        ],
         chrome: [
             {
                 title: '插件更新',

lib/v2/google/router.js

@@ -6,4 +6,5 @@ module.exports = function (router) {
     router.get('/fonts/:sort?', require('./fonts'));
     router.get('/news/:category/:locale', require('./news'));
     router.get('/scholar/:query', require('./scholar'));
+    router.get('/search/:keyword/:language?', require('./search'));
 };

lib/v2/google/search.js

@@ -0,0 +1,62 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { art } = require('@/utils/render');
+const path = require('path');
+const config = require('@/config').value;
+
+const renderDescription = (description, images) => art(path.join(__dirname, './templates/description.art'), { description, images });
+
+module.exports = async (ctx) => {
+    const { keyword, language } = ctx.params;
+    const searchParams = new URLSearchParams({
+        q: keyword,
+    });
+    const tempUrl = new URL('https://www.google.com/search');
+    tempUrl.search = searchParams.toString();
+    const url = tempUrl.toString();
+    const key = `google-search:${language}:${url}`;
+    const items = await ctx.cache.tryGet(
+        key,
+        async () => {
+            const response = (
+                await got(url, {
+                    headers: {
+                        'Accept-Language': language,
+                    },
+                })
+            ).data;
+            const $ = cheerio.load(response);
+            const content = $('#rso');
+            return content
+                .find('> div')
+                .map((i, el) => {
+                    const element = $(el);
+                    const link = element.find('div > div > div > div > div > span > a').first().attr('href');
+                    const title = element.find('div > div > div> div > div > span > a > h3').first().text();
+                    const imgs = element
+                        .find('img')
+                        .map((_j, _el) => $(_el).attr('src'))
+                        .toArray();
+                    const description = element.find('div[style="-webkit-line-clamp:2"]').first().text() || element.find('div[role="heading"]').first().text();
+                    const author = element.find('div > div > div > div > div > span > a > div > div > span').first().text() || '';
+                    return {
+                        link,
+                        title,
+                        description: renderDescription(description, imgs),
+                        author,
+                    };
+                })
+                .toArray()
+                .filter((e) => e?.link);
+        },
+        config.cache.routeExpire,
+        false
+    );
+
+    ctx.state.data = {
+        title: `${keyword} - Google Search`,
+        description: `${keyword} - Google Search`,
+        link: url,
+        item: items,
+    };
+};

lib/v2/google/templates/description.art

@@ -0,0 +1,6 @@
+{{@ description }}
+{{if images}}
+    {{each images}}
+        <img src="{{ $value }}">
+    {{/each}}
+{{/if}}

lib/v2/sogou/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/doodles': ['xyqfer'],
+    '/search/:keyword': ['CaoMeiYouRen'],
+};

lib/v2/sogou/radar.js

@@ -0,0 +1,16 @@
+module.exports = {
+    'sogou.com': {
+        _name: '搜狗',
+        www: [
+            {
+                title: '搜索',
+                docs: 'https://docs.rsshub.app/routes/other#sou-gou-sou-suo',
+                source: '/',
+                target: (params, url) => {
+                    const keyword = new URL(url).searchParams.get('query');
+                    return `/sogou/search/${keyword}`;
+                },
+            },
+        ],
+    },
+};

lib/v2/sogou/router.js

@@ -0,0 +1,4 @@
+module.exports = function (router) {
+    router.get('/doodles', require('./doodles'));
+    router.get('/search/:keyword', require('./search'));
+};

lib/v2/sogou/search.js

@@ -0,0 +1,54 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { art } = require('@/utils/render');
+const path = require('path');
+const { parseDate } = require('@/utils/parse-date');
+const config = require('@/config').value;
+
+const renderDescription = (description, images) => art(path.join(__dirname, './templates/description.art'), { description, images });
+
+module.exports = async (ctx) => {
+    const { keyword } = ctx.params;
+    const url = `https://www.sogou.com/web?query=${encodeURIComponent(keyword)}`;
+    const key = `sogou-search:${url}`;
+    const items = await ctx.cache.tryGet(
+        key,
+        async () => {
+            const response = (await got(url)).data;
+            const $ = cheerio.load(response);
+            const result = $('#main');
+            return result
+                .find('.vrwrap')
+                .map((i, el) => {
+                    const element = $(el);
+                    const imgs = element
+                        .find('img')
+                        .map((j, el2) => $(el2).attr('src'))
+                        .toArray();
+                    const link = element.find('h3 a').first().attr('href');
+                    const title = element.find('h3').first().text();
+                    const description = element.find('.text-layout').first().text() || element.find('.space-txt').first().text() || element.find('[class^="translate"]').first().text();
+                    const author = element.find('.citeurl span').first().text() || '';
+                    const pubDate = parseDate(element.find('.citeurl .cite-date').first().text().trim());
+                    return {
+                        link,
+                        title,
+                        description: renderDescription(description, imgs),
+                        author,
+                        pubDate,
+                    };
+                })
+                .toArray()
+                .filter((e) => e?.link);
+        },
+        config.cache.routeExpire,
+        false
+    );
+
+    ctx.state.data = {
+        title: `${keyword} - 搜狗搜索`,
+        description: `${keyword} - 搜狗搜索`,
+        link: url,
+        item: items,
+    };
+};

lib/v2/sogou/templates/description.art

@@ -0,0 +1,6 @@
+{{@ description }}
+{{if images}}
+    {{each images}}
+        <img src="{{ $value }}">
+    {{/each}}
+{{/if}}

website/docs/routes/other.mdx

@@ -91,6 +91,12 @@ See [#app-store-mac-app-store](/routes/program-update#app-store-mac-app-store)
   2.  Copy everything in the URL after `?`, for example: `https://www.autotrader.co.uk/car-search?radius=50&postcode=sw1a1aa&onesearchad=Used&onesearchad=Nearly%20New&onesearchad=New&price-to=9000&year-from=2012&body-type=Hatchback&transmission=Automatic&exclude-writeoff-categories=on` will produce `radius=50&postcode=sw1a1aa&onesearchad=Used&onesearchad=Nearly%20New&onesearchad=New&price-to=9000&year-from=2012&body-type=Hatchback&transmission=Automatic&exclude-writeoff-categories=on`
 </Route>
 
+## Bing {#bing}
+
+### 搜索 {#bing-sou-suo}
+
+<Route author="CaoMeiYouRen" example="/bing/search/rss" path="/bing/search/:keyword" paramsDesc={['搜索关键词']} radar="1" />
+
 ## BOOKWALKERTW {#bookwalkertw}
 
 ### 热门新书 {#bookwalkertw-re-men-xin-shu}
@@ -181,6 +187,10 @@ Official Website: [https://news.yahoo.co.jp/pages/article/20200207](https://news
 
 <Route author="TonyRL" example="/google/alerts/RSSHub" path="/google/alerts/:keyword" paramsDesc={['Keyword']} />
 
+### Search {#google-search}
+
+<Route author="CaoMeiYouRen" example="/google/search/rss/zh-CN,zh" path="/google/search/:keyword/:language?" paramsDesc={['Keyword','Accept-Language. Example: zh-CN,zh;q=0.9,en;q=0.8,ja;q=0.7']} />
+
 ## Grand-Challenge {#grand-challenge}
 
 ### Challenge 列表 {#grand-challenge-challenge-lie-biao}
@@ -597,9 +607,13 @@ Specify options (in the format of query string) in parameter `routeParams` param
 
 <Route author="sanmmm" example="/afdian/dynamic/@afdian" path="/afdian/dynamic/:uid?" paramsDesc={['用户id，用户动态页面url里可找到']} />
 
-## 百度热搜 {#bai-du-re-sou}
+## 百度 {#bai-du}
 
-### 榜单 {#bai-du-re-sou-bang-dan}
+### 搜索 {#bai-du-sou-suo}
+
+<Route author="CaoMeiYouRen" example="/baidu/search/rss" path="/baidu/search/:keyword" paramsDesc={['搜索关键词']} radar="1" anticrawler="1" />
+
+### 热搜榜单 {#bai-du-re-sou-bang-dan}
 
 <Route author="xyqfer" example="/baidu/top" path="/baidu/top/:board?" paramsDesc={['榜单，默认为 `realtime`']} radar="1">
   | 热搜榜   | 小说榜 | 电影榜 | 电视剧榜 | 汽车榜 | 游戏榜 |
@@ -851,10 +865,14 @@ Specify options (in the format of query string) in parameter `routeParams` param
 
 ## 搜狗 {#sou-gou}
 
-### 搜狗特色 LOGO {#sou-gou-sou-gou-te-se-logo}
+### 特色 LOGO {#sou-gou-te-se-logo}
 
 <Route author="xyqfer" example="/sogou/doodles" path="/sogou/doodles" />
 
+### 搜索 {#sou-gou-sou-suo}
+
+<Route author="CaoMeiYouRen" example="/sogou/search/rss" path="/sogou/search/:keyword" paramsDesc={['搜索关键词']} radar="1" />
+
 ## 台灣事實查核中心 {#tai-wan-shi-shi-cha-he-zhong-xin}
 
 ### 最新相關資訊 / 最新查核報告 {#tai-wan-shi-shi-cha-he-zhong-xin-zui-xin-xiang-guan-zi-xun-zui-xin-cha-he-bao-gao}

website/docs/routes/picture.mdx

@@ -165,9 +165,9 @@
 
 <Route author="nczitzk" example="/asiantolick/page/news" path="/asiantolick/page/:id" paramsDesc={['Page id']} radar="1" />
 
-## Bing Wallpaper {#bing-wallpaper}
+## Bing {#bing}
 
-### Daily Wallpaper {#bing-wallpaper-daily-wallpaper}
+### Daily Wallpaper {#bing-daily-wallpaper}
 
 <Route author="FHYunCai" example="/bing" path="/bing" radar="1" />