feat: add boston.com (#5096)

docs/en/traditional-media.md

@@ -26,6 +26,17 @@ Support major channels, refer to [BBC RSS feeds](https://www.bbc.co.uk/news/1062
 
 </RouteEn>
 
+## Boston.com
+
+### News
+
+<RouteEn author="oppilate" example="/boston/technology" path="/boston/:tag?" :paramsDesc="['Tag']">
+
+Generates full-text feeds that the official feed doesn't provide.
+Refer to [Boston.com's feed page](https://www.boston.com/rss-feeds) for tags. For instance, `https://www.boston.com/tag/local-news/?feed=rss` corresponds to `/boston/local-news`.
+
+</RouteEn>
+
 ## Chicago Tribune
 
 ### News
@@ -35,6 +46,7 @@ Support major channels, refer to [BBC RSS feeds](https://www.bbc.co.uk/news/1062
 Generates full-text that the official feed doesn't provide.
 Refer to [Chicago Tribune's feed page](https://www.chicagotribune.com/about/ct-chicago-tribune-rss-feeds-htmlstory.html) for categories. For instance, `https://www.chicagotribune.com/arcio/rss/category/nation-world/` corresponds to `/chicagotribune/nation-world`.
 
+
 </RouteEn>
 
 ## China Dialogue

docs/traditional-media.md

@@ -32,6 +32,17 @@ pageClass: routes
 
 </Route>
 
+## Boston.com
+
+### 新闻
+
+<Route author="oppilate" example="/boston/technology" path="/boston/:tag?" :paramsDesc="['Tag']">
+
+生成官方未提供的全文订阅点。
+有哪些 tag 请参考 [Boston.com 官网上的订阅页面](https://www.boston.com/rss-feeds)。例如，`https://www.boston.com/tag/local-news/?feed=rss` 对应 RSSHub 路由 `/boston/local-news`。
+
+</Route>
+
 ## Chicago Tribune
 
 ### 新闻

lib/router.js

@@ -2874,6 +2874,9 @@ router.get('/jike/user/:id', require('./routes/jike/user'));
 // 网易新闻
 router.get('/netease/news/rank/:category?/:type?/:time?', require('./routes/netease/news/rank'));
 
+// Boston.com
+router.get('/boston/:tag?', require('./routes/boston/index'));
+
 // 中国邮政速递物流
 router.get('/ems/news', require('./routes/ems/news'));
 

lib/routes/boston/index.js

@@ -0,0 +1,74 @@
+const cheerio = require('cheerio');
+const parser = require('@/utils/rss-parser');
+const got = require('@/utils/got');
+
+module.exports = async (ctx) => {
+    const tag = ctx.params.tag;
+    const rssUrl = tag ? `https://www.boston.com/tag/${tag}/?feed=atom` : `https://www.boston.com/?feed=atom`;
+    const feed = await parser.parseURL(rssUrl);
+
+    const items = await Promise.all(
+        feed.items.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const response = await got({
+                        url: item.link,
+                        method: 'get',
+                    });
+                    const html = response.body;
+                    const $ = cheerio.load(html);
+                    const content = $('div.content-text-article');
+
+                    // Disable image lazyloading
+                    content.find('img').each((i, e) => {
+                        e = $(e);
+                        e.attr('src', e.attr('srcset'));
+                    });
+
+                    // Categories
+                    const categories = [];
+                    $('meta[property="article:tag"]').each((i, e) => {
+                        categories.push(e.attribs.content);
+                    });
+
+                    // Updated date
+                    const updatedAt = $('meta[property="article:modified_time"]').attr().content;
+
+                    // Summary
+                    const summary = $('meta[property="og:description"]').attr().content;
+
+                    // Remove unwanted DOMs
+                    const unwanted_element_selectors = ['.ad-container', '.content-toaster--newsletter'];
+                    unwanted_element_selectors.forEach((selector) => {
+                        content.find(selector).each((i, e) => {
+                            $(e).remove();
+                        });
+                    });
+
+                    return {
+                        title: item.title.trim(),
+                        id: item.link,
+                        pubDate: new Date(item.pubDate).toUTCString(),
+                        updated: new Date(updatedAt).toUTCString(),
+                        author: item.author.trim(),
+                        link: item.link,
+                        summary: summary.trim(),
+                        description: content.html(),
+                        category: categories,
+                        icon: 'https://www.boston.com/wp-content/themes/bdc/images/favicons/largetile.png',
+                        logo: 'https://www.boston.com/wp-content/themes/bdc/images/favicons/largetile.png',
+                    };
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: feed.title,
+        link: feed.link,
+        description: feed.description,
+        item: items,
+        language: 'en-us',
+        icon: 'https://www.boston.com/wp-content/themes/bdc/images/favicons/largetile.png',
+        logo: 'https://www.boston.com/wp-content/themes/bdc/images/favicons/largetile.png',
+    };
+};