fix: bbc

docs/en/traditional-media.md

@@ -28,7 +28,7 @@ Site
 
 ### BBC
 
-<RouteEn author="HenryQW" example="/bbc/world-asia" path="/bbc/:channel?" :paramsDesc="['channel, default to `top stories`']">
+<RouteEn author="HenryQW DIYgod" example="/bbc/world-asia" path="/bbc/:channel?" :paramsDesc="['channel, default to `top stories`']">
 
 Provides a better reading experience (full text articles) over the official ones.
 

docs/traditional-media.md

@@ -34,7 +34,7 @@ pageClass: routes
 
 ### BBC 英文
 
-<Route author="HenryQW" example="/bbc/world-asia" path="/bbc/:channel" :paramsDesc="['频道，缺省为热门']">
+<Route author="HenryQW DIYgod" example="/bbc/world-asia" path="/bbc/:channel" :paramsDesc="['频道，缺省为热门']">
 
 通过提取文章全文，以提供比官方源更佳的阅读体验。
 

lib/routes/bbc/index.js

@@ -47,7 +47,7 @@ module.exports = async (ctx) => {
     }
 
     const items = await Promise.all(
-        feed.items.splice(0, 10).map(async (item) => {
+        feed.items.splice(0, 30).map(async (item) => {
             const response = await got({
                 method: 'get',
                 url: item.link,
@@ -58,13 +58,21 @@ module.exports = async (ctx) => {
             let description;
 
             if (response.request.options.url.pathname.startsWith('/news/av')) {
-                description = utils.ProcessAVFeed($, item.link);
+                description = item.content;
             } else {
-                description = utils.ProcessFeed($, item.link);
+                description = utils.ProcessFeed($);
             }
 
+            let section = 'sport';
+            const urlSplit = item.link.split('/');
+            const sectionSplit = urlSplit[urlSplit.length - 1].split('-');
+            if (sectionSplit.length > 1) {
+                section = sectionSplit[0];
+            }
+            section = section[0].toUpperCase() + section.slice(1);
+
             const single = {
-                title: item.title,
+                title: `[${section}] ${item.title}`,
                 description,
                 pubDate: item.pubDate,
                 link: item.link,

lib/routes/bbc/utils.js

@@ -1,83 +1,6 @@
-const cheerio = require('cheerio');
-const ProcessImage = ($, e, c, noscript = false) => {
-    const img = $(e).find(c);
-
-    if (img.length > 0) {
-        let message;
-        // special handling for img wrapped in <noscript>
-        if (noscript) {
-            // eslint-disable-next-line no-unused-vars
-            const i = cheerio.load(img[0].children[0].data)('img')[0];
-
-            message = `<figure><img alt='${i.attribs.alt}' src='${i.attribs.src}'><br><figcaption>`;
-        } else {
-            if (img[0].attribs.src) {
-                // handle cover image and lazy-loading images
-                message = `<figure><img alt='${img[0].attribs.alt ? img[0].attribs.alt : ''}' src='${img[0].attribs.src.replace(/(?<=\/news\/).*?(?=\/cpsprodpb)/, '600')}'><br><figcaption>`;
-            } else {
-                message = `<figure><img alt='${img[0].attribs['data-alt'] ? img[0].attribs['data-alt'] : ''}' src='${img[0].attribs['data-src'].replace(/(?<=\/news\/).*?(?=\/cpsprodpb)/, '600')}'><br><figcaption>`;
-            }
-
-            // add image caption
-            const figcaption = $(e).find('.media-caption__text');
-            if (figcaption.length > 0) {
-                message += `${$(figcaption[0]).text().trim()}`;
-            }
-
-            // add image copyright
-            const copyright = $(e).find('.story-image-copyright');
-            if (copyright.length > 0) {
-                message += ` ©${$(copyright[0]).text().trim()}`;
-            }
-        }
-
-        message += '</figcaption></figure>';
-
-        $(message).insertAfter(e);
-    }
-};
-
-const ProcessVideo = ($, e, c) => {
-    const video = $(e).find(c);
-
-    video.each((i, f) => {
-        const cover = $(f).children('img.player-with-placeholder__image');
-        if (cover.length > 0) {
-            $(cover[0]).insertBefore(e);
-        }
-        const link = $(f).find('figure.js-media-player-unprocessed');
-        if (link.length > 0) {
-            const url = JSON.parse(link[0].attribs['data-playable']).settings.externalEmbedUrl;
-            const message = `<br><a href='${url}'>使用浏览器播视频/view video in browser</a>`;
-            $(message).insertAfter(e);
-        }
-    });
-};
-
-const ProcessAVFeed = ($) => {
-    const content = $('div.vxp-media__summary');
-
-    const video = $('div.vxp-media__player');
-
-    video.each((i, e) => {
-        const cover = $(e).children('img.vxp-media__placeholder-image');
-        if (cover.length > 0) {
-            $(cover[0]).insertBefore(content[0].firstChild);
-        }
-        const link = $(e).find('figure.js-media-player-unprocessed');
-        if (link.length > 0) {
-            const url = JSON.parse(link[0].attribs['data-playable']).settings.externalEmbedUrl;
-            const message = `<br><a href='${url}'>使用浏览器播视频/view video in browser</a>`;
-            $(message).insertAfter(content[0].firstChild);
-        }
-    });
-
-    return content.html();
-};
-
-const ProcessFeed = ($, link) => {
+const ProcessFeed = ($) => {
     // by default treat it as a hybrid news with video and story-body__inner
-    let content = $('div.story-body__inner');
+    let content = $('#main-content article');
 
     if (content.length === 0) {
         // it's a video news with video and story-body
@@ -90,45 +13,25 @@ const ProcessFeed = ($, link) => {
     }
 
     // remove useless DOMs
-    content.find('div.bbccom_slot, .off-screen, .embed-report-link').each((i, e) => {
-        $(e).remove();
+    content.find('header, section, [data-testid="bbc-logo-wrapper"]').remove();
+
+    content.find('noscript').each((i, e) => {
+        $(e).parent().html($(e).html());
     });
 
-    if ($('#comp-media-player').length > 0) {
-        // there is a cover video
-
-        let message = `<br><a href='${link}'>使用浏览器播视频/view video in browser</a>`;
-
-        const cover = $('#comp-media-player').find('img.player-with-placeholder__image');
-
-        if (cover.length > 0) {
-            // there is a cover image
-            message = `<br><img src='${cover[0].attribs.src}'>` + message;
+    content.find('img').each((i, e) => {
+        if (!$(e).attr('src') && $(e).attr('srcSet')) {
+            const srcs = $(e).attr('srcSet').split(', ');
+            const lastSrc = srcs[srcs.length - 1];
+            $(e).attr('src', lastSrc.split(' ')[0]);
         }
-
-        $(message).insertAfter(content[0].lastChild);
-    }
-
-    // resize all images and videos
-    content.find('figure').each((i, e) => {
-        ProcessImage($, e, '.image-and-copyright-container > img');
-
-        // handle lazy-loading images
-        ProcessImage($, e, '.js-delayed-image-load');
-
-        ProcessImage($, e, 'div > div > div > img');
-
-        ProcessImage($, e, 'div > div > div > noscript', true);
-
-        ProcessVideo($, e, 'figure.media-with-caption > div.player-with-placeholder');
-
-        $(e).remove();
     });
 
+    content.find('[data-component="media-block"] figcaption').prepend('<span>View video in browser: </span>');
+
     return content.html();
 };
 
 module.exports = {
     ProcessFeed,
-    ProcessAVFeed,
 };