From 933558521062191e854c45a25bc3cbc11350fcc7 Mon Sep 17 00:00:00 2001
From: Lei Zhang <leigoto@gmail.com>
Date: Tue, 29 Aug 2023 20:54:06 +0800
Subject: [PATCH] fix: V2EX Route Exclude Duplicate Posts (#13144)

* update: V2EX Route Exclude Duplicate Posts

* use ctx.query.limit

* refactor: migrate to v2

---------

Co-authored-by: pull[bot] <39814207+pull[bot]@users.noreply.github.com>
---
 lib/router.js                     |  6 ++---
 lib/routes/v2ex/post.js           | 39 -------------------------------
 lib/v2/v2ex/maintainer.js         |  5 ++++
 lib/v2/v2ex/post.js               | 35 +++++++++++++++++++++++++++
 lib/v2/v2ex/radar.js              | 35 +++++++++++++++++++++++++++
 lib/v2/v2ex/router.js             |  5 ++++
 lib/{routes => v2}/v2ex/tab.js    | 38 ++++++++++++++----------------
 lib/{routes => v2}/v2ex/topics.js |  2 +-
 8 files changed, 101 insertions(+), 64 deletions(-)
 delete mode 100644 lib/routes/v2ex/post.js
 create mode 100644 lib/v2/v2ex/maintainer.js
 create mode 100644 lib/v2/v2ex/post.js
 create mode 100644 lib/v2/v2ex/radar.js
 create mode 100644 lib/v2/v2ex/router.js
 rename lib/{routes => v2}/v2ex/tab.js (62%)
 rename lib/{routes => v2}/v2ex/topics.js (96%)

diff --git a/lib/router.js b/lib/router.js
index 8f859a2b69d801..e61ec26abce1bc 100644
--- a/lib/router.js
+++ b/lib/router.js
@@ -77,9 +77,9 @@ router.get('/huya/live/:id', lazyloadRouteHandler('./routes/huya/live'));
 router.get('/showroom/room/:id', lazyloadRouteHandler('./routes/showroom/room'));
 
 // v2ex
-router.get('/v2ex/topics/:type', lazyloadRouteHandler('./routes/v2ex/topics'));
-router.get('/v2ex/post/:postid', lazyloadRouteHandler('./routes/v2ex/post'));
-router.get('/v2ex/tab/:tabid', lazyloadRouteHandler('./routes/v2ex/tab'));
+// router.get('/v2ex/topics/:type', lazyloadRouteHandler('./routes/v2ex/topics'));
+// router.get('/v2ex/post/:postid', lazyloadRouteHandler('./routes/v2ex/post'));
+// router.get('/v2ex/tab/:tabid', lazyloadRouteHandler('./routes/v2ex/tab'));
 
 // f-droid
 router.get('/fdroid/apprelease/:app', lazyloadRouteHandler('./routes/fdroid/apprelease'));
diff --git a/lib/routes/v2ex/post.js b/lib/routes/v2ex/post.js
deleted file mode 100644
index 5b4d26c4d723af..00000000000000
--- a/lib/routes/v2ex/post.js
+++ /dev/null
@@ -1,39 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const { parseDate } = require('@/utils/parse-date');
-
-module.exports = async (ctx) => {
-    const postid = ctx.params.postid;
-    const pageUrl = `https://www.v2ex.com/t/${postid}`;
-
-    const response = await got({
-        method: 'get',
-        url: pageUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-    const list = $('[id^="r_"]').get();
-
-    ctx.state.data = {
-        title: `V2EX-${$('.header h1').text()}`,
-        link: pageUrl,
-        description: $('.topic_content').text(),
-        item: list
-            .map((item) => {
-                const post = $(item);
-                const reply_content = post.find('.reply_content').first();
-                const no = post.find('.no').first();
-
-                return {
-                    title: `#${no.text()} ${reply_content.text()}`,
-                    description: reply_content.html(),
-                    guid: post.attr('id'),
-                    link: `${pageUrl}#${post.attr('id')}`,
-                    author: post.find('.dark').first().text(),
-                    pubDate: parseDate(post.find('.ago').attr('title')),
-                };
-            })
-            .reverse(),
-        allowEmpty: true,
-    };
-};
diff --git a/lib/v2/v2ex/maintainer.js b/lib/v2/v2ex/maintainer.js
new file mode 100644
index 00000000000000..01c685653ac57d
--- /dev/null
+++ b/lib/v2/v2ex/maintainer.js
@@ -0,0 +1,5 @@
+module.exports = {
+    '/post/:postid': ['kt286'],
+    '/tab/:tabid': ['liyefox'],
+    '/topics/:type': ['WhiteWorld'],
+};
diff --git a/lib/v2/v2ex/post.js b/lib/v2/v2ex/post.js
new file mode 100644
index 00000000000000..b933692cfbf113
--- /dev/null
+++ b/lib/v2/v2ex/post.js
@@ -0,0 +1,35 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const { postid } = ctx.params;
+    const pageUrl = `https://www.v2ex.com/t/${postid}`;
+
+    const { data: topicResponse } = await got('https://www.v2ex.com/api/topics/show.json', {
+        searchParams: {
+            id: postid,
+        },
+    });
+
+    const { data: replies } = await got('https://www.v2ex.com/api/replies/show.json', {
+        searchParams: {
+            topic_id: postid,
+        },
+    });
+
+    const topic = topicResponse[0];
+
+    ctx.state.data = {
+        title: `V2EX-${topic.title}`,
+        link: pageUrl,
+        description: topic.content,
+        item: replies.map((item, index) => ({
+            title: `#${index + 1} ${item.content}`,
+            description: item.content_rendered,
+            link: `${pageUrl}#r_${item.id}`,
+            author: item.member.username,
+            pubDate: parseDate(item.created, 'X'),
+        })),
+        allowEmpty: true,
+    };
+};
diff --git a/lib/v2/v2ex/radar.js b/lib/v2/v2ex/radar.js
new file mode 100644
index 00000000000000..f3a9d43bbfcdf6
--- /dev/null
+++ b/lib/v2/v2ex/radar.js
@@ -0,0 +1,35 @@
+module.exports = {
+    'v2ex.com': {
+        _name: 'V2EX',
+        '.': [
+            {
+                title: '最热 / 最新主题',
+                docs: 'https://docs.rsshub.app/routes/v2ex',
+                source: ['/'],
+                target: (_, url) => {
+                    const { searchParams } = new URL(url);
+                    if (searchParams.get('tab') === 'all' || searchParams.get('tab') === 'hot') {
+                        return `/v2ex/topics/${searchParams.get('tab')?.replace('all', 'latest')}`;
+                    }
+                },
+            },
+            {
+                title: '帖子',
+                docs: 'https://docs.rsshub.app/routes/v2ex',
+                source: ['/t/:postid'],
+                target: '/v2ex/post/:postid',
+            },
+            {
+                title: '标签',
+                docs: 'https://docs.rsshub.app/routes/v2ex',
+                source: ['/'],
+                target: (_, url) => {
+                    const { searchParams } = new URL(url);
+                    if (searchParams.get('tab') && searchParams.get('tab') !== 'all' && searchParams.get('tab') !== 'hot') {
+                        return `/v2ex/tab/${searchParams.get('tab')}`;
+                    }
+                },
+            },
+        ],
+    },
+};
diff --git a/lib/v2/v2ex/router.js b/lib/v2/v2ex/router.js
new file mode 100644
index 00000000000000..c17828c761aec2
--- /dev/null
+++ b/lib/v2/v2ex/router.js
@@ -0,0 +1,5 @@
+module.exports = (router) => {
+    router.get('/post/:postid', require('./post'));
+    router.get('/tab/:tabid', require('./tab'));
+    router.get('/topics/:type', require('./topics'));
+};
diff --git a/lib/routes/v2ex/tab.js b/lib/v2/v2ex/tab.js
similarity index 62%
rename from lib/routes/v2ex/tab.js
rename to lib/v2/v2ex/tab.js
index 140eca922d4750..71e141dc32eabe 100644
--- a/lib/routes/v2ex/tab.js
+++ b/lib/v2/v2ex/tab.js
@@ -13,24 +13,21 @@ module.exports = async (ctx) => {
 
     const $ = cheerio.load(response.data);
     const links = $('span.item_title > a')
-        .map((i, link) => `${host}${$(link).attr('href')}`)
-        .slice(0, 10)
-        .get();
+        .toArray()
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 10)
+        .map((link) => `${host}${$(link).attr('href').replace(/#.*$/, '')}`);
+
     const items = await Promise.all(
-        links.map(async (pageUrl) => {
-            const cacheKey = `v2ex-${pageUrl}`;
-            const cacheValue = await ctx.cache.get(cacheKey);
-            let post = {};
-            if (cacheValue) {
-                post = cacheValue;
-            } else {
+        links.map((link) =>
+            ctx.cache.tryGet(`v2ex-${link}`, async () => {
                 const response = await got({
                     method: 'get',
-                    url: pageUrl,
+                    url: link,
                 });
+
                 const $ = cheerio.load(response.data);
-                const list = $('[id^="r_"]').get();
-                const reply_content = list
+                const list = $('[id^="r_"]').toArray();
+                const replyContent = list
                     .map((item) => {
                         const post = $(item);
                         const content = post.find('.reply_content').html();
@@ -39,18 +36,17 @@ module.exports = async (ctx) => {
                         return `<p><div>#${no}: <i>${author}</i></div><div>${content}</div></p>`;
                     })
                     .join('');
-                post = {
+
+                return {
                     title: $('.header h1').text(),
-                    link: pageUrl,
-                    guid: pageUrl,
-                    description: $('div.topic_content').html() + `<div>${reply_content}</div>`,
+                    link,
+                    description: `${$('div.topic_content').html()}<div>${replyContent}</div>`,
                     author: $('div.header > small > a').text(),
                 };
-                ctx.cache.set(cacheKey, post);
-            }
-            return Promise.resolve(post);
-        })
+            })
+        )
     );
+
     ctx.state.data = {
         title: `V2EX-${tabid}`,
         link: pageUrl,
diff --git a/lib/routes/v2ex/topics.js b/lib/v2/v2ex/topics.js
similarity index 96%
rename from lib/routes/v2ex/topics.js
rename to lib/v2/v2ex/topics.js
index db3097d9c15531..28062030b8d85f 100644
--- a/lib/routes/v2ex/topics.js
+++ b/lib/v2/v2ex/topics.js
@@ -2,7 +2,7 @@ const got = require('@/utils/got');
 const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
-    const type = ctx.params.type;
+    const { type } = ctx.params;
 
     const { data } = await got(`https://www.v2ex.com/api/topics/${type}.json`);