feat: ChinaFile (#6068)

docs/en/traditional-media.md

@@ -108,6 +108,20 @@ Refer to [Chicago Tribune's feed page](https://www.chicagotribune.com/about/ct-c
 
 </RouteEn>
 
+## ChinaFile
+
+### Reporting & Opinion
+
+<RouteEn author="oppilate" example="/chinafile/all" path="/chinafile/:category?" :paramsDesc="['Category, by default `all`']">
+
+Generates full-text feeds that the official feed doesn't provide.
+
+| All | The China NGO Project |
+| --- | --------------------- |
+| all | ngo                   |
+
+</RouteEn>
+
 ## NHK
 
 ### News Web Easy

docs/traditional-media.md

@@ -91,6 +91,18 @@ pageClass: routes
 
 </Route>
 
+## ChinaFile
+
+<Route author="oppilate" example="/chinafile/all" path="/chinafile/:category?" :paramsDesc="['分类，默认 `all`']">
+
+通过提取文章全文，以提供比官方源更佳的阅读体验。
+
+| 全部 | The China NGO Project |
+| ---- | --------------------- |
+| all  | ngo                   |
+
+</Route>
+
 ## e 公司
 
 ### 快讯

lib/router.js

@@ -3444,4 +3444,7 @@ router.get('/gov/harbin/kjj', require('./routes/gov/harbin/kjj'));
 // WSJ
 router.get('/wsj/:lang/:category', require('./routes/wsj/index'));
 
+// China File
+router.get('/chinafile/:category?', require('./routes/chinafile/index'));
+
 module.exports = router;

lib/routes/chinafile/index.js

@@ -0,0 +1,69 @@
+const Parser = require('rss-parser');
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category || 'all';
+    const rssUrl = `https://feeds.feedburner.com/chinafile/${category}`;
+
+    // The custom UA in @/utils/parser (mimic browser) results in a HTML page
+    // which cannot be parsed
+    const parser = new Parser();
+    const feed = await parser.parseURL(rssUrl);
+
+    const items = await Promise.all(
+        feed.items.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    let url = item.link;
+                    const response = await got({ url });
+                    const html = response.body;
+                    const $ = cheerio.load(html);
+                    const content = $('article');
+
+                    // Cover
+                    const cover = $('.view-featured-photo');
+
+                    if (cover.length > 0) {
+                        cover.insertBefore(content[0].childNodes[0]);
+                        $(cover).remove();
+                    }
+
+                    // Summary
+                    const summary = $('meta[name="description"]').attr('content');
+                    const updatedAt = $('meta[name="og:updated_time"]').attr('content');
+
+                    const categories = $('meta[name="news_keywords"]')
+                        .attr('content')
+                        .split(',')
+                        .map((c) => c.trim());
+
+                    url = $('link[rel="canonical"]').attr('href');
+
+                    return {
+                        title: item.title,
+                        id: item.guid,
+                        pubDate: item.pubDate,
+                        updated: updatedAt,
+                        author: item.creator,
+                        link: url,
+                        summary: summary,
+                        description: content.html(),
+                        category: categories,
+                        icon: 'https://www.chinafile.com/sites/default/files/chinafile_favicon.png',
+                        logo: 'https://www.chinafile.com/sites/all/themes/cftwo/assets/images/logos/logo-large.png',
+                    };
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: feed.title,
+        link: feed.link,
+        description: feed.description,
+        item: items,
+        language: 'en-us',
+        icon: 'https://www.chinafile.com/sites/default/files/chinafile_favicon.png',
+        logo: 'https://www.chinafile.com/sites/all/themes/cftwo/assets/images/logos/logo-large.png',
+    };
+};