feat: add telecompaper search (#6734)

docs/traditional-media.md

@@ -222,7 +222,7 @@ Solidot 提供的 feed:
 
 ### News
 
-<Route author="nczitzk" example="/telecompaper/news/mobile/2020/China/News" path="/telecompaper/news/:caty/:year?/:country?/:type?/:keyword?" :paramsDesc="['分类，见下表', '年份，可在所选分类页中 Filter 的 `Years` 选择器中选择相应年份，不限年份则填入 `all`，默认为不限', '国家或大洲，可在所选分类页中 Filter 的 `Countries` 选择器中选择相应国家或大洲，不限国家或大洲则填入 `all`，默认为不限', '类型，可在所选分类页中 Filter 的 `Types` 选择器中选择相应类型，不限类型则填入 `all`，默认为不限', '搜索关键字']">
+<Route author="nczitzk" example="/telecompaper/news/mobile/2020/China/News" path="/telecompaper/news/:caty/:year?/:country?/:type?" :paramsDesc="['分类，见下表', '年份，可在所选分类页中 Filter 的 `Years` 选择器中选择相应年份，不限年份则填入 `all`，默认为不限', '国家或大洲，可在所选分类页中 Filter 的 `Countries` 选择器中选择相应国家或大洲，不限国家或大洲则填入 `all`，默认为不限', '类型，可在所选分类页中 Filter 的 `Types` 选择器中选择相应类型，不限类型则填入 `all`，默认为不限']">
 
 可选分类如下
 
@@ -240,6 +240,24 @@ Solidot 提供的 feed:
 
 </Route>
 
+### Search
+
+<Route author="nczitzk" example="/telecompaper/search/Nokia" path="/telecompaper/search/:keyword?/:company?/:sort?/:period?" :paramsDesc="['关键词', '公司名，默认为不限', '排序，见下表，默认为 Date Descending', '发表在时间段内，默认为 12 months']">
+
+排序
+
+| Date Ascending | Date Descending |
+| -------------- | --------------- |
+| 1              | 2               |
+
+发表在时间段内
+
+| 1 month | 3 months | 6 months | 12 months | 24 months |
+| ------- | -------- | -------- | --------- | --------- |
+| 1       | 3        | 6        | 12        | 24        |
+
+</Route>
+
 ## The Economist
 
 ### 分类

lib/router.js

@@ -3224,7 +3224,8 @@ router.get('/141ppv/:type/:key?', require('./routes/141ppv/141ppv'));
 router.get('/curiouscat/user/:id', require('./routes/curiouscat/user'));
 
 // Telecompaper
-router.get('/telecompaper/news/:caty/:year?/:country?/:type?/:keyword?', require('./routes/telecompaper/news'));
+router.get('/telecompaper/news/:caty/:year?/:country?/:type?', require('./routes/telecompaper/news'));
+router.get('/telecompaper/search/:keyword?/:company?/:sort?/:period?', require('./routes/telecompaper/search'));
 
 // 水木社区
 router.get('/newsmth/account/:id', require('./routes/newsmth/account'));

lib/routes/telecompaper/search.js

@@ -0,0 +1,77 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const keyword = ctx.params.keyword || '';
+    const company = ctx.params.company || '';
+    const sort = ctx.params.sort || '2';
+    const period = ctx.params.period || '12';
+
+    const rootUrl = `https://www.telecompaper.com/search/index.aspx?search=${keyword}`;
+
+    let response = await got({
+            method: 'get',
+            url: rootUrl,
+        }),
+        $ = cheerio.load(response.data);
+
+    response = await got({
+        method: 'post',
+        url: rootUrl,
+        data: JSON.stringify({
+            __EVENTTARGET: '',
+            __EVENTARGUMENT: '',
+            __VSTATE: $('#__VSTATE').attr('value'),
+            __VIEWSTATE: '',
+            ctl00$header$searchText: 'Search keywords',
+            ctl00$header$searchTextMobile: 'Search keywords',
+            ctl00$MainPlaceHolder$SearchText: keyword,
+            ctl00$MainPlaceHolder$CompanyText: company,
+            ctl00$MainPlaceHolder$Sort: parseInt(sort),
+            ctl00$MainPlaceHolder$Results: 20,
+            ctl00$MainPlaceHolder$Date: 'Timeframe1',
+            ctl00$MainPlaceHolder$Period: parseInt(period),
+            ctl00$MainPlaceHolder$txtStartDate: '',
+            ctl00$MainPlaceHolder$txtEndDate: '',
+            ctl00$MainPlaceHolder$chkEnglish: 'on',
+            ctl00$MainPlaceHolder$Submit: 'Search',
+        }),
+    });
+    $ = cheerio.load(response.data);
+
+    const list = $('table.details_rows tbody tr')
+        .slice(0, 15)
+        .map((_, item) => {
+            item = $(item);
+            const a = item.find('a');
+            return {
+                title: a.text(),
+                link: a.attr('href'),
+                pubDate: new Date(item.find('span.source').text().split(' | ')[0] + ' GMT+1').toUTCString(),
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+                    const content = cheerio.load(detailResponse.data);
+
+                    item.description = content('#pageContainer').html();
+
+                    return item;
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: `Telecompaper Search - ${keyword}`,
+        link: rootUrl,
+        item: items,
+    };
+};