Merge pull request #6634 from zphw/feature/voa

feat: add Voice of America

docs/traditional-media.md

@@ -268,6 +268,26 @@ Solidot 提供的 feed:
 
 <Route author="emdoe" example="/udn/global/鏡頭背後" path="/udn/global/:tid" :paramsDesc="['標籤名稱，請在轉角國際首頁獲取；如果選擇輸入 `newest` 則輸出最新文章']">
 
+## Voice of America (VOA)
+
+透過提取全文，以獲得更好的閱讀體驗
+
+<Route author="zphw" example="/voa/cantonese/zprtie-ttp" path="/voa/:language/:channel?" :paramsDesc="['語言','頻道，可於官網獲取']">
+
+`语言`
+
+| 粵語      | 中文    | 藏語    |
+| --------- | ------- | ------- |
+| cantonese | chinese | tibetan |
+
+`频道`
+
+可於各語言官網聚合新聞處 (如 <https://www.voacantonese.com/rssfeeds>) 獲取
+
+例如 `https://www.voacantonese.com/api/zyrtyequty` 將對應 `/voa/cantonese/zyrtyequty`
+
+</Route>
+
 ## Yahoo
 
 ### 新聞

lib/router.js

@@ -3795,6 +3795,8 @@ router.get('/semiconductors/latest-news', require('./routes/semiconductors/lates
 
 // VOA News
 router.get('/voa/day-photos', require('./routes/voa/day-photos'));
+// Voice of America
+router.get('/voa/:language/:channel?', require('./routes/voa/index'));
 
 // 留园网
 router.get('/6park/:id?/:type?/:keyword?', require('./routes/6park/index'));

lib/routes/voa/index.js

@@ -0,0 +1,55 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+
+module.exports = async (ctx) => {
+    let baseURL;
+    switch (ctx.params.language) {
+        case 'cantonese':
+            baseURL = 'https://www.voacantonese.com/';
+            break;
+        case 'chinese':
+            baseURL = 'https://www.voachinese.com/';
+            break;
+        case 'tibetan':
+            baseURL = 'https://www.voatibetan.com/';
+            break;
+        default:
+            throw Error('The language specified does not exist');
+    }
+
+    const response = await got.get(baseURL + 'api/' + (ctx.params.channel || ''));
+    const $ = cheerio.load(response.data);
+
+    const list = [];
+    $('item').each(function (_, e) {
+        const item = {};
+        item.title = $(e).find('title').first().text();
+        item.link = $(e).find('guid').first().text();
+        item.pubDate = $(e).find('pubDate').first().text();
+        item.description = $(e).find('description').first().text();
+        list.push(item);
+    });
+
+    const result = await Promise.all(
+        list
+            .filter((item) => item.link)
+            .map(
+                async (item) =>
+                    await ctx.cache.tryGet(item.link, async () => {
+                        const content = await got.get(item.link);
+                        const description = cheerio.load(content.data);
+                        description('.wsw__embed').remove();
+
+                        item.description = description('div#article-content div').first().html() ? description('div.cover-media') + description('div#article-content div').first().html() : item.description;
+
+                        return item;
+                    })
+            )
+    );
+
+    ctx.state.data = {
+        title: $('channel title').first().text(),
+        link: baseURL,
+        item: result,
+    };
+};