From aadc66ba31c189004aba99604333e830eb4d9fe7 Mon Sep 17 00:00:00 2001
From: fengkx <liangkx8237@gmail.com>
Date: Sun, 12 May 2019 23:55:53 +0800
Subject: [PATCH] fix  natgeo (page structure change) (#2111)

* fix  natgeo (page structure change)

* add cache
---
 docs/travel.md              |  2 +-
 lib/router.js               |  1 -
 lib/routes/natgeo/natgeo.js | 82 ++++++++++++++++++++++++++-----------
 3 files changed, 60 insertions(+), 25 deletions(-)

diff --git a/docs/travel.md b/docs/travel.md
index 49feb8cc47..675ae6eabe 100644
--- a/docs/travel.md
+++ b/docs/travel.md
@@ -37,7 +37,7 @@ IATA 国际航空运输协会机场代码, 参见[维基百科 国际航空运
 
 ## 国家地理
 
-<Route name="分类" author="fengkx" example="/natgeo/news/ngnews" path="/natgeo/:cat/:type?" :paramsDesc="['分类', '类型, 例如`https://www.natgeomedia.com/category/news/ngnews`对应 cat, type 分别为 news, ngnews']"/>
+<Route name="分类" author="fengkx" example="/natgeo/environment/article" path="/natgeo/:cat/:type?" :paramsDesc="['分类', '类型, 例如`https://www.natgeomedia.com/environment/photo/`对应 cat, type 分别为 environment, photo']"/>
 
 ## 马蜂窝
 
diff --git a/lib/router.js b/lib/router.js
index 461f5469c8..4efe0f70ee 100644
--- a/lib/router.js
+++ b/lib/router.js
@@ -422,7 +422,6 @@ router.get('/sexinsex/:id/:type?', require('./routes/sexinsex/index'));
 router.get('/gcores/category/:category', require('./routes/gcores/category'));
 
 // 国家地理
-router.get('/natgeo/dailyphoto', require('./routes/natgeo/dailyphoto'));
 router.get('/natgeo/:cat/:type?', require('./routes/natgeo/natgeo'));
 
 // 一个
diff --git a/lib/routes/natgeo/natgeo.js b/lib/routes/natgeo/natgeo.js
index aca53ed352..9113e0ae09 100644
--- a/lib/routes/natgeo/natgeo.js
+++ b/lib/routes/natgeo/natgeo.js
@@ -1,40 +1,76 @@
 const cheerio = require('cheerio');
 const axios = require('../../utils/axios');
 
+// https://www.natgeomedia.com//article/
+
+const MonthMap = {
+    JAN: 0,
+    FEB: 1,
+    MAR: 2,
+    APR: 3,
+    MAY: 4,
+    JUN: 5,
+    JUL: 6,
+    AUG: 7,
+    SEP: 8,
+    OCT: 9,
+    NOV: 10,
+    DEC: 11,
+};
+
+async function load(link) {
+    const { data } = await axios.get(link);
+    const $ = cheerio.load(data);
+    const dtStr = $('.content-title-area')
+        .find('h6')
+        .first()
+        .text()
+        .split(/[\s.]+/);
+    const dt = new Date();
+    dt.setMonth(MonthMap[dtStr[0].toUpperCase()]);
+    dt.setDate(dtStr[1]);
+    let description =
+        $('article')
+            .first()
+            .html() +
+        $('article')
+            .eq(1)
+            .html();
+    if (/photo/.test(link)) {
+        description = $('#content-album').html() + description;
+    }
+    return {
+        title: $('title').text(),
+        pubDate: dt.toUTCString(),
+        description,
+    };
+}
+
 module.exports = async (ctx) => {
-    const type = `${ctx.params.type ? ctx.params.type : ''}`;
-    const url = `https://www.natgeomedia.com/category/${ctx.params.cat}/${type}`;
+    const type = `${ctx.params.type || ''}`;
+    const url = `https://www.natgeomedia.com/${ctx.params.cat}/${type}`;
     const res = await axios.get(url);
     const $ = cheerio.load(res.data);
 
-    const list = $('.td-ss-main-content').find('.td-animation-stack');
+    const urlList = Array.prototype.map.call($('.article-link-w100').find('.read-btn'), (i) =>
+        $(i)
+            .find('a[href]')
+            .first()
+            .attr('href')
+    );
 
     const count = [];
-    for (let i = 0; i < Math.min(list.length, 10); i++) {
+    for (let i = 0; i < Math.min(urlList.length, 10); i++) {
         count.push(i);
     }
     const out = await Promise.all(
         count.map(async (i) => {
-            const each = $(list[i]);
-            const storyLink = each.find('a[itemprop=url]').attr('href');
-            const item = {
-                title: each.find('a[itemprop=url]').attr('title'),
-                pubDate: each.find('time').attr('datetime'),
-                link: storyLink,
-                guid: storyLink.match(/\d+/g)[0],
+            const link = urlList[i];
+            const single = {
+                link,
             };
-            const key = `${ctx.params.cat}${type}${item.guid}`;
-            const value = await ctx.cache.get(key);
-            if (value) {
-                item.description = value;
-            } else {
-                const storyDetail = await axios.get(item.link);
-                const data = storyDetail.data;
-                const $ = cheerio.load(data);
-                item.description = $('.td-post-content').html();
-                ctx.cache.set(key, item.description);
-            }
-            return Promise.resolve(item);
+            const other = await ctx.cache.tryGet(link, async () => await load(link));
+            return Object.assign({}, single, other);
         })
     );
     ctx.state.data = {