feat: add cookie support for discuz (#4033)

docs/bbs.md

@@ -30,6 +30,16 @@ pageClass: routes
 
 </Route>
 
+### 通用子版块-支持 Cookie
+
+<Route author="junfengP" example="/discuz/x/00/https%3a%2f%2fbbs.zdfx.net%2fforum-2-1.html" path="/discuz/:ver/:cid/:link" :paramsDesc="['discuz版本类型，见下表', 'Cookie id，需自建并配置环境变量，详情见部署页面的配置模块','子版块链接， 需要手动Url编码']" >
+
+| Discuz X 系列 | Discuz 7.x 系列 |
+| ------------- | --------------- |
+| x             | 7               |
+
+</Route>
+
 ## MCBBS
 
 ### 版块

docs/en/bbs.md

@@ -19,3 +19,13 @@ pageClass: routes
 | x               | 7                 |
 
 </Route>
+
+### General Subforum - Support cookie
+
+<Route author="junfengP" example="/discuz/x/00/https%3a%2f%2fbbs.zdfx.net%2fforum-2-1.html" path="/discuz/:ver/:cid/:link" :paramsDesc="['discuz version，see below table', 'Cookie id，require self hosted and set environment parameters, see Deploy - Configuration pages for detail','link of subforum, require url encoded']" >
+
+| Discuz X Series | Discuz 7.x Series |
+| --------------- | ----------------- |
+| x               | 7                 |
+
+</Route>

docs/en/install/README.md

@@ -372,3 +372,7 @@ Access control includes a whitelist and a blacklist, support IP and route, use `
 
     -   `NHENTAI_USERNAME`: nhentai username or email
     -   `NHENTAI_PASSWORD`: nhentai password
+
+-   discuz cookies
+
+    -   `DISCUZ_COOKIE_{cid}`: Cookie of a forum powered by discuz, cid can be anything from 00 to 99. When visiting route discuz, using cid to specify this cookie.

docs/install/README.md

@@ -409,3 +409,7 @@ RSSHub 支持 `memory` 和 `redis` 两种缓存方式
 
     -   `NHENTAI_USERNAME`: nhentai 用户名或邮箱
     -   `NHENTAI_PASSWORD`: nhentai 密码
+
+-   discuz cookies 设定
+
+    -   `DISCUZ_COOKIE_{cid}`: 某 Discuz 驱动的论坛，用户注册后的 Cookie 值 , cid 可自由设定，取值范围[00, 99], 使用 discuz 通用路由时, 通过指定 cid 来调用该 cookie

lib/config.js

@@ -6,6 +6,7 @@ const calculateValue = () => {
     const bilibili_cookies = {};
     const twitter_tokens = {};
     const email_config = {};
+    const discuz_cookies = {};
 
     for (const name in envs) {
         if (name.startsWith('BILIBILI_COOKIE_')) {
@@ -17,6 +18,9 @@ const calculateValue = () => {
         } else if (name.startsWith('EMAIL_CONFIG_')) {
             const id = name.slice(13);
             email_config[id] = envs[name];
+        } else if (name.startsWith('DISCUZ_COOKIE_')) {
+            const cid = name.slice(14);
+            discuz_cookies[cid] = envs[name];
         }
     }
 
@@ -117,6 +121,9 @@ const calculateValue = () => {
             username: envs.NHENTAI_USERNAME,
             password: envs.NHENTAI_PASSWORD,
         },
+        discuz: {
+            cookies: discuz_cookies,
+        },
     };
 };
 calculateValue();

lib/router.js

@@ -2166,6 +2166,7 @@ router.get('/gbcc/trust', require('./routes/gbcc/trust'));
 router.get('/apnews/topics/:topic', require('./routes/apnews/topics'));
 
 // discuz
+router.get('/discuz/:ver([7|x])/:cid([0-9]{2})/:link(.*)', require('./routes/discuz/discuz'));
 router.get('/discuz/:ver([7|x])/:link(.*)', require('./routes/discuz/discuz'));
 router.get('/discuz/:link(.*)', require('./routes/discuz/discuz'));
 

lib/routes/discuz/discuz.js

@@ -2,9 +2,10 @@ const got = require('@/utils/got');
 const cheerio = require('cheerio');
 const iconv = require('iconv-lite');
 const dateUtil = require('@/utils/date');
+const config = require('@/config').value;
 
 // discuz 7.x 与 discuz x系列 通用文章内容抓取
-async function load(baseUrl, itemLink, ctx, charset) {
+async function load(baseUrl, itemLink, ctx, charset, header) {
     // 处理相对链接
     if (itemLink) {
         if (baseUrl && !baseUrl.match(/^https?:\/\//)) {
@@ -25,9 +26,25 @@ async function load(baseUrl, itemLink, ctx, charset) {
     // 处理编码问题
     let responseData;
     if (charset === 'utf-8') {
-        responseData = (await got.get(itemLink)).data;
+        responseData = (
+            await got({
+                method: 'get',
+                url: itemLink,
+                headers: header,
+            })
+        ).data;
     } else {
-        responseData = iconv.decode((await got.get({ url: itemLink, responseType: 'buffer' })).data, charset);
+        responseData = iconv.decode(
+            (
+                await got({
+                    method: 'get',
+                    url: itemLink,
+                    responseType: 'buffer',
+                    headers: header,
+                })
+            ).data,
+            charset
+        );
     }
     if (!responseData) {
         const description = '获取详细内容失败';
@@ -45,9 +62,21 @@ async function load(baseUrl, itemLink, ctx, charset) {
 module.exports = async (ctx) => {
     let link = ctx.params.link;
     const ver = ctx.params.ver ? ctx.params.ver.toUpperCase() : undefined;
+    const cid = ctx.params.cid;
     link = link.replace(/:\/\//, ':/').replace(/:\//, '://');
-
-    const response = await got.get(link);
+    const cookie = cid === undefined ? '' : config.discuz.cookies[cid];
+    if (cookie === undefined) {
+        throw Error('缺少对应论坛的cookie.');
+    }
+    const header = {
+        Cookie: cookie,
+        Referer: link,
+    };
+    const response = await got({
+        method: 'get',
+        url: link,
+        headers: header,
+    });
     const contentType = response.headers['content-type'] || '';
     // 若没有指定编码，则默认utf-8
     let charset = 'utf-8';
@@ -59,7 +88,22 @@ module.exports = async (ctx) => {
                 .toLowerCase();
         }
     }
-    const responseData = charset === 'utf-8' ? response.data : iconv.decode((await got.get({ url: link, responseType: 'buffer' })).data, charset);
+    const responseData =
+        charset === 'utf-8'
+            ? response.data
+            : iconv.decode(
+                  (
+                      await got({
+                          method: 'get',
+                          url: link,
+                          responseType: 'buffer',
+                          headers: {
+                              Cookie: cookie,
+                          },
+                      })
+                  ).data,
+                  charset
+              );
     const $ = cheerio.load(responseData);
     const title = $('head > title').text();
     const version = ver ? 'DISCUZ! ' + ver : $('head > meta[name=generator]').attr('content');
@@ -67,7 +111,7 @@ module.exports = async (ctx) => {
     if (version.toUpperCase().startsWith('DISCUZ! 7')) {
         // discuz 7.x 系列
         // 支持全文抓取，限制抓取页面5个
-        const list = $('tbody[id^="normalthread"] tr')
+        const list = $('tbody[id^="normalthread"] > tr')
             .slice(0, 5)
             .get();
         process = await Promise.all(
@@ -79,14 +123,14 @@ module.exports = async (ctx) => {
                     link: itemLink,
                     pubDate: dateUtil(item.find('td.author em').text()),
                 };
-                const detail = await load(link, itemLink, ctx, charset);
+                const detail = await load(link, itemLink, ctx, charset, header);
                 return Promise.resolve(Object.assign({}, single, detail));
             })
         );
     } else if (version.toUpperCase().startsWith('DISCUZ! X')) {
         // discuz X 系列
         // 支持全文抓取，限制抓取页面5个
-        const list = $('tbody[id^="normalthread"] tr')
+        const list = $('tbody[id^="normalthread"] > tr')
             .slice(0, 5)
             .get();
         process = await Promise.all(
@@ -103,7 +147,7 @@ module.exports = async (ctx) => {
                             .text()
                     ),
                 };
-                const detail = await load(link, itemLink, ctx, charset);
+                const detail = await load(link, itemLink, ctx, charset, header);
                 return Promise.resolve(Object.assign({}, single, detail));
             })
         );