fix: pubmed trending articles (DIYgod#6225)

lord63-forks · Nov 29, 2020 · 77f1f57 · 77f1f57
1 parent 481773c
commit 77f1f57
Showing 1 changed file with 34 additions and 37 deletions.
diff --git a/lib/routes/pubmed/trending.js b/lib/routes/pubmed/trending.js
@@ -3,61 +3,58 @@ const cheerio = require('cheerio');
 const url = require('url');
 const date = require('@/utils/date');
 
-const base = 'https://www.ncbi.nlm.nih.gov';
+const base = 'https://pubmed.ncbi.nlm.nih.gov';
 
 module.exports = async (ctx) => {
-    const link = `${base}/pubmed/trending/`;
+    const link = `${base}/trending/`;
     const response = await got.get(encodeURI(link));
     const pageCapture = cheerio.load(response.data);
 
-    const list = pageCapture('.content div.rprt > div.rslt').get();
-    const out = await Promise.all(
-        list.map(async (elem) => {
+    const list = pageCapture('.docsum-content')
+        .map((_, elem) => {
             const $ = cheerio.load(elem);
-            const title = $('p > a').text();
-            const partial = $('p > a').attr('href');
-            const address = url.resolve(base, partial);
-            const author = $('div.supp > p.desc').text();
-            const pubDate = date($('div.supp > p.details').text().split('. ')[1]);
-
-            const item = {
-                title,
-                author,
-                pubDate,
-                link: encodeURI(address),
+            const partial = $('a').attr('href');
+            return {
+                title: $('a').text().trim(),
+                link: url.resolve(base, partial),
+                author: $('.full-authors').text(),
             };
+        })
+        .get();
 
-            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);
+    const out = await Promise.all(
+        list.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const detail = await got.get(item.link);
+                    const detailCapture = cheerio.load(detail.data);
+
+                    item.doi = detailCapture('meta[name="citation_doi"]').attr('content');
+                    item.pubDate = date(detailCapture('meta[name="citation_date"]').attr('content'));
 
-                let authorContents = '';
-                if (author !== '') {
-                    authorContents = `
+                    let authorContents = '';
+                    if (item.author !== '') {
+                        authorContents = `
                     <div id="author-content">
-                      <span style="color: grey">${author}</span>
+                      <span style="color: grey">${item.author}</span>
                     </div>
                     `;
-                }
-                const abs = detailCapture('div.abstr > div').html();
-                let absContents = '';
-                if (abs !== null) {
-                    absContents = `
+                    }
+                    const abs = detailCapture('#enc-abstract').html();
+                    let absContents = '';
+                    if (abs !== null) {
+                        absContents = `
                     <div id="abstract-content">
                       <h2 align="left">Abstract</h2>
                       <p>${abs}</p>
                     </div>
                     `;
-                }
-                item.description = authorContents + absContents;
-                ctx.cache.set(address, item.description);
-            }
+                    }
+                    item.description = authorContents + absContents;
 
-            return Promise.resolve(item);
-        })
+                    return item;
+                })
+        )
     );
 
     ctx.state.data = {