fix: 知识分子资讯 (#6804)

lib/routes/zhishifenzi/news.js

@@ -1,67 +1,67 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
 const url = require('url');
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
 const date = require('@/utils/date');
 
-const base = 'http://zhishifenzi.com';
+const titles = {
+    multiple: '综合',
+    biology: '生物',
+    medicine: '医药',
+    ai: '人工智能',
+    physics: '物理',
+    chymistry: '化学',
+    astronomy: '天文',
+    other: '其他',
+};
 
 module.exports = async (ctx) => {
-    let link = `${base}/news/`;
-
-    const type = ctx.params.type;
     // type can be on of:
-    // ["biology", "medicine", "ai", "physics", "chymistry", "astronomy", "others"]
-    if (typeof type !== 'undefined') {
-        link = `${base}/news/${type}`;
-    }
-    const response = await got.get(encodeURI(link));
-    const pageCapture = cheerio.load(response.data);
+    // ["biology", "medicine", "ai", "physics", "chymistry", "astronomy", "other"]
+    const type = ctx.params.type || 'multiple';
 
-    let title = pageCapture('div.all_categories > ul > li.category').text().split('：').pop();
-    if (title === '') {
-        title = '全部資訊';
-    } else {
-        title = `資訊「${title}」`;
-    }
-    const list = pageCapture('div.inner_news_list > ul > li.clearfix').get();
-    const out = await Promise.all(
-        list.map(async (elem) => {
-            const $ = cheerio.load(elem);
-            const title = $('div > h5 > a').text();
-            const partial = $('div > h5 > a').attr('href');
-            const address = url.resolve(base, partial);
-            const pubDate = date($('span').first().text());
+    const rootUrl = 'http://zhishifenzi.com';
+    const currentUrl = url.resolve(rootUrl, '/news/ajaxarticles');
+    const response = await got({
+        method: 'post',
+        url: currentUrl,
+        form: {
+            category: type,
+        },
+    });
 
-            // const keywords = $('span')
-            // .last()
-            // .html();
-            const item = {
-                title,
-                pubDate,
-                link: encodeURI(address),
+    const $ = cheerio.load(response.data);
+    const list = $('h5 a')
+        .map((_, item) => {
+            item = $(item);
+            return {
+                title: item.text(),
+                link: url.resolve(rootUrl, item.attr('href')),
             };
-
-            const value = await ctx.cache.get(address);
-            if (value) {
-                item.description = value;
-            } else {
-                const detail = await got.get(item.link);
-                const detailCapture = cheerio.load(detail.data);
-
-                item.description = detailCapture('.main div.inner_content')
-                    .html()
-                    .replace(/src="\//g, `src="${base}/`);
-                ctx.cache.set(address, item.description);
-            }
-
-            return Promise.resolve(item);
         })
+        .get();
+
+    const out = await Promise.all(
+        list.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+                    const content = cheerio.load(detailResponse.data);
+
+                    item.description = content('.inner_content').html();
+                    item.pubDate = date(content('.inner_view_from').text());
+
+                    return item;
+                })
+        )
     );
 
     ctx.state.data = {
-        title: `知識分子 | ${title}`,
+        title: `知識分子 | 資訊「${titles[type] || '综合'}」`,
         description: `知識分子 | 科學 文明 智慧`,
-        link: link,
+        link: url.resolve(rootUrl, '/news'),
         item: out,
     };
 };