refactor: refactor thepaper utils for optimize rss reading (#10859)

* 1. fix timezone shift 2. add podcast for voice article 3.use list icon for list feed 4. category support

* extract logo using cheerio

* fix the fallback logo image on list feed

* fix the deepscan insufficient null checking

* add support for album type article

lib/v2/thepaper/channel.js

@@ -14,5 +14,7 @@ module.exports = async (ctx) => {
         title: `澎湃新闻频道 - ${utils.ChannelIdToName(id, data)}`,
         link: channel_url,
         item: items,
+        itunes_author: '澎湃新闻',
+        image: utils.ExtractLogo(response),
     };
 };

lib/v2/thepaper/featured.js

@@ -8,10 +8,11 @@ module.exports = async (ctx) => {
     const list = data.props.pageProps.data.list;
 
     const items = await Promise.all(list.map((item) => utils.ProcessItem(item, ctx)));
-
     ctx.state.data = {
         title: '澎湃新闻 - 首页头条',
         link: 'https://m.thepaper.cn',
         item: items,
+        itunes_author: '澎湃新闻',
+        image: utils.ExtractLogo(response),
     };
 };

lib/v2/thepaper/list.js

@@ -7,13 +7,15 @@ module.exports = async (ctx) => {
     const list_url = `https://m.thepaper.cn/list/${id}`;
     const response = await got(list_url);
     const data = JSON.parse(cheerio.load(response.data)('#__NEXT_DATA__').html());
-    const list = data.props.pageProps.data.list;
+    const pagePropsData = data.props.pageProps.data;
+    const list = pagePropsData.list;
 
     const items = await Promise.all(list.map((item) => utils.ProcessItem(item, ctx)));
-
     ctx.state.data = {
         title: `澎湃新闻栏目 - ${utils.ListIdToName(id, data)}`,
         link: list_url,
         item: items,
+        itunes_author: '澎湃新闻',
+        image: pagePropsData.nodeInfo?.pic ?? utils.ExtractLogo(response),
     };
 };

lib/v2/thepaper/templates/image_detail.art

@@ -0,0 +1,6 @@
+{{ each images }}
+<figure class="wp-block-image size-full">
+    <img width="{{ $value.width }}" src="{{ $value.src }}">
+</figure>
+<p>{{ $value.description }}</p>
+{{ /each }}

lib/v2/thepaper/utils.js

@@ -2,6 +2,7 @@ const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
 const got = require('@/utils/got');
 const { art } = require('@/utils/render');
+const timezone = require('@/utils/timezone');
 const path = require('path');
 
 module.exports = {
@@ -12,7 +13,7 @@ module.exports = {
                 title: item.name,
                 link: item.link,
                 description: item.name,
-                pubDate: parseDate(item.pubTimeLong),
+                pubDate: timezone(parseDate(item.pubTimeLong), +8),
                 media: {
                     content: {
                         url: item.pic,
@@ -27,7 +28,7 @@ module.exports = {
                 const detailData = data.props.pageProps.detailData;
 
                 const contentDetail = detailData.contentDetail || detailData.liveDetail;
-                let description = contentDetail.content || contentDetail.summary;
+                let description = contentDetail.content || contentDetail.summary || '';
 
                 if (contentDetail.videos) {
                     description =
@@ -36,21 +37,35 @@ module.exports = {
                         }) + description;
                 }
 
-                return {
+                if (contentDetail.images) {
+                    description =
+                        art(path.join(__dirname, 'templates/image_detail.art'), {
+                            images: contentDetail.images,
+                        }) + description;
+                }
+
+                const rss_item = {
                     title: contentDetail.name,
                     link: itemUrl,
                     description,
-                    pubDate: parseDate(contentDetail.pubTime),
+                    category: contentDetail.tagList?.map((t) => t.tag) ?? [],
+                    pubDate: timezone(parseDate(item.pubTimeLong || contentDetail.pubTime), +8),
                     author: contentDetail.author,
                     media: {
                         content: {
-                            url: item.pic || contentDetail.sharePic || (contentDetail.videos && contentDetail.videos.coverUrl),
+                            url: item.pic || contentDetail.videos?.coverUrl,
                         },
                         thumbnails: {
-                            url: item.pic || contentDetail.sharePic,
+                            url: item.sharePic || contentDetail.sharePic,
                         },
                     },
                 };
+                if (contentDetail.voiceInfo?.isHaveVoice) {
+                    rss_item.enclosure_type = 'audio/mpeg';
+                    rss_item.enclosure_url = contentDetail.voiceInfo.voiceSrc;
+                    rss_item.itunes_item_image = item.pic || contentDetail.videos?.coverUrl;
+                }
+                return rss_item;
             });
         }
     },
@@ -67,4 +82,5 @@ module.exports = {
             }
         }
     },
+    ExtractLogo: (response) => 'https://m.thepaper.cn' + cheerio.load(response.data)('img.imageCover').attr('src'),
 };