fix(route): thepaper changes its site to nextjs based backend, so change the feeder (#10850)

* fix(route) thepaper now use nextjs as backend * Id to Name for channel and list * summary ok * determine the detail by null check instead of type check * list_url for better reading * add failback of thumbnail * fix video article * add list for thepaper * add list for thepaper * del the dup line * small fix and optimize * add maintainer
2025-12-05 12:21:31 +08:00 · 2022-09-21 22:53:50 +08:00
parent 00d6a689d9
commit fe50f9cb18
8 changed files with 217 additions and 71 deletions
--- a/lib/v2/thepaper/utils.js
+++ b/lib/v2/thepaper/utils.js
@@ -1,55 +1,70 @@
 const cheerio = require('cheerio');
-const date = require('@/utils/date');
 const { parseDate } = require('@/utils/parse-date');
-const timezone = require('@/utils/timezone');
 const got = require('@/utils/got');
+const { art } = require('@/utils/render');
+const path = require('path');

 module.exports = {
-    ProcessFeed: async (query, link, ctx) => {
-        const res = await got(link);
+    ProcessItem: async (item, ctx) => {
+        if (item.link) {
+            // external link
+            return {
+                title: item.name,
+                link: item.link,
+                description: item.name,
+                pubDate: parseDate(item.pubTimeLong),
+                media: {
+                    content: {
+                        url: item.pic,
+                    },
+                },
+            };
+        } else {
+            const itemUrl = `https://m.thepaper.cn/detail/${item.contId}`;
+            return await ctx.cache.tryGet(itemUrl, async () => {
+                const res = await got(itemUrl);
+                const data = JSON.parse(cheerio.load(res.data)('#__NEXT_DATA__').html());
+                const detailData = data.props.pageProps.detailData;

-        const $ = cheerio.load(res.data);
-        const list = $(query).slice(0, 10).get();
+                const contentDetail = detailData.contentDetail || detailData.liveDetail;
+                let description = contentDetail.content || contentDetail.summary;

-        return Promise.all(
-            list.map((item) => {
-                const $ = cheerio.load(item);
-                const itemUrl = `https://m.thepaper.cn/${$(item).find('a').eq(0).attr('href')}`;
-                return ctx.cache.tryGet(itemUrl, async () => {
-                    const res = await got(itemUrl);
-                    const content = cheerio.load(res.data);
+                if (contentDetail.videos) {
+                    description =
+                        art(path.join(__dirname, 'templates/video_detail.art'), {
+                            videos: contentDetail.videos,
+                        }) + description;
+                }

-                    let description, pubDate;
-
-                    if (content('div.news_video_msg').length > 0) {
-                        description = content('#vdetail_sum').html();
-                        pubDate = timezone(
-                            parseDate(
-                                content('div.news_video_msg')
-                                    .html()
-                                    .replace(/&nbsp;/gi, '')
-                                    .split('<br>')[0]
-                            ),
-                            +8
-                        );
-                    } else if (content('#slider_wrapper_ul').length > 0) {
-                        description = '';
-                        pubDate = new Date(date($(item).find('div.list_item_extra span').eq(1).text())).toUTCString();
-                    } else {
-                        description = content('div.newsdetail_content').html();
-                        pubDate = timezone(parseDate(content('div.date').text().trim().split('来源：')[0].trim()), +8);
+                return {
+                    title: contentDetail.name,
+                    link: itemUrl,
+                    description,
+                    pubDate: parseDate(contentDetail.pubTime),
+                    author: contentDetail.author,
+                    media: {
+                        content: {
+                            url: item.pic || contentDetail.sharePic || (contentDetail.videos && contentDetail.videos.coverUrl),
+                        },
+                        thumbnails: {
+                            url: item.pic || contentDetail.sharePic,
+                        },
+                    },
+                };
+            });
+        }
+    },
+    ChannelIdToName: (nodeId, next_data) => next_data.props.appProps.menu.channelList.find((c) => c.nodeId.toString() === nodeId.toString()).name,
+    ListIdToName: (listId, next_data) => {
+        const channelList = next_data.props.appProps.menu.channelList;
+        for (const c of channelList) {
+            if (c.childNodeList && c.childNodeList.length > 0) {
+                for (const l of c.childNodeList) {
+                    if (l.nodeId.toString() === listId.toString()) {
+                        return l.name;
                    }
-
-                    const single = {
-                        title: content('title').text(),
-                        link: itemUrl,
-                        description,
-                        pubDate,
-                        author: content('div.author').text(),
-                    };
-                    return single;
-                });
-            })
-        );
+                }
+            }
+        }
    },
 };