From 1c9016d12ee352f3e66f1a13f80a7a8e2ce78b9e Mon Sep 17 00:00:00 2001
From: SUEPbot <114787369+SUEPbot@users.noreply.github.com>
Date: Thu, 19 Jan 2023 15:41:22 +0800
Subject: [PATCH] feat(route): get full text for shiep (#11645)

* feat(route): get full text for shiep

* refactor(route): use WHATWG URL API instead of deprecated url.resolve in shiep

* refactor(route): cache the whole return result

* feat(route): leave a notice if item is redirected to the intranet

Co-authored-by: SUEPbot <SUEPbot@users.noreply.github.com>
---
 lib/v2/shiep/index.js                  | 72 ++++++++++++++++++--------
 lib/v2/shiep/templates/description.art |  1 +
 2 files changed, 50 insertions(+), 23 deletions(-)
 create mode 100644 lib/v2/shiep/templates/description.art

diff --git a/lib/v2/shiep/index.js b/lib/v2/shiep/index.js
index 0d3d28971b..815a588be8 100644
--- a/lib/v2/shiep/index.js
+++ b/lib/v2/shiep/index.js
@@ -1,6 +1,9 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
+const dayjs = require('dayjs');
+const path = require('path');
 const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
 
 module.exports = async (ctx) => {
     const type = ctx.params.type;
@@ -34,32 +37,55 @@ module.exports = async (ctx) => {
         throw Error('Invalid type');
     }
 
-    const id = ctx.params.id ?? config[type].id;
-
-    const response = await got({
-        method: 'get',
-        url: `https://${type}.shiep.edu.cn/${id}/list.htm`,
-    });
-    const data = response.data;
-    const $ = cheerio.load(data);
-    const class_name = 'list_item';
+    const host = `https://${type}.shiep.edu.cn`;
+    const id = ctx.params.id || config[type].id;
     const pubDateClass = type === 'dxxy' ? 'div[class="article-publishdate"]' : 'span[class="Article_PublishDate"]'; // 适配电子与信息工程学院网站改版
-    const list = $(`.${class_name}`);
+
+    const response = await got(`${host}/${id}/list.htm`);
+
+    const $ = cheerio.load(response.data);
+
+    const items = $('.list_item')
+        .toArray()
+        .filter((item) => {
+            const date = dayjs($(item).find(pubDateClass).text().trim());
+            return date.isValid();
+        })
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('a').attr('title') || item.find('a').text(),
+                link: new URL(item.find('a').attr('href'), host).href,
+                pubDate: parseDate(item.find(pubDateClass).text().trim(), 'YYYY-MM-DD'),
+            };
+        });
+
+    await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                try {
+                    const response = await got(item.link);
+                    const $ = cheerio.load(response.data);
+
+                    if ($('.wp_articlecontent').length > 0) {
+                        item.description = art(path.resolve(__dirname, 'templates/description.art'), {
+                            description: $('.wp_articlecontent').html(),
+                        });
+                    } else {
+                        item.description = '请进行统一身份认证后查看内容';
+                    }
+                } catch (e) {
+                    item.description = '请在校内或通过校园VPN查看内容';
+                }
+                return item;
+            })
+        )
+    );
 
     ctx.state.data = {
         title: '上海电力大学-' + config[type].title,
-        link: `https://${type}.shiep.edu.cn/${id}/list.htm`,
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-                    return {
-                        title: item.find('a').text(),
-                        link: item.find('a').attr('href'),
-                        pubDate: parseDate(item.find(pubDateClass).text().trim(), 'YYYY-MM-DD'),
-                    };
-                })
-                .get(),
+        link: `${host}/${id}/list.htm`,
+        description: '上海电力大学-' + config[type].title,
+        item: items,
     };
 };
diff --git a/lib/v2/shiep/templates/description.art b/lib/v2/shiep/templates/description.art
new file mode 100644
index 0000000000..d1254e8e79
--- /dev/null
+++ b/lib/v2/shiep/templates/description.art
@@ -0,0 +1 @@
+{{@ description }}