fix: some bugs in discuz (#3951)

This commit is contained in:
junfengP
2020-02-15 19:26:24 +08:00
committed by GitHub
parent c2bd5ea283
commit 2e22386781
4 changed files with 36 additions and 13 deletions

View File

@@ -16,10 +16,20 @@ pageClass: routes
## Discuz
### 通用子版块
### 通用子版块-自动检测
<Route author="junfengP" example="/discuz/http%3a%2f%2fwww.u-share.cn%2fforum.php%3fmod%3dforumdisplay%26fid%3d56" path="/discuz/:link" :paramsDesc="['子版块链接, 需要手动Url编码']"/>
### 通用子版块-指定版本
<Route author="junfengP" example="/discuz/x/https%3a%2f%2fwww.52pojie.cn%2fforum-16-1.html" path="/discuz/:ver/:link" :paramsDesc="['discuz版本类型见下表','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

View File

@@ -3,3 +3,19 @@ pageClass: routes
---
# BBS
## Discuz
### General Subforum - Auto detection
<Route author="junfengP" example="/discuz/http%3a%2f%2fwww.u-share.cn%2fforum.php%3fmod%3dforumdisplay%26fid%3d56" path="/discuz/:link" :paramsDesc="['link of subforum, require url encoded ']"/>
### General Subforum - Manual version
<Route author="junfengP" example="/discuz/x/https%3a%2f%2fwww.52pojie.cn%2fforum-16-1.html" path="/discuz/:ver/:link" :paramsDesc="['discuz versionsee below table','link of subforum, require url encoded']" >
| Discuz X Series | Discuz 7.x Series |
| --------------- | ----------------- |
| x | 7 |
</Route>

View File

@@ -2152,6 +2152,7 @@ router.get('/gbcc/trust', require('./routes/gbcc/trust'));
router.get('/apnews/topics/:topic', require('./routes/apnews/topics'));
// discuz
router.get('/discuz/:ver([7|x])/:link(.*)', require('./routes/discuz/discuz'));
router.get('/discuz/:link(.*)', require('./routes/discuz/discuz'));
// China Dialogue 中外对话

View File

@@ -44,6 +44,7 @@ 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;
link = link.replace(/:\/\//, ':/').replace(/:\//, '://');
const response = await got.get(link);
@@ -52,18 +53,18 @@ module.exports = async (ctx) => {
let charset = 'utf-8';
for (const attr of contentType.split(';')) {
if (attr.indexOf('charset=') >= 0) {
charset = attr.split('=').pop();
charset = attr
.split('=')
.pop()
.toLowerCase();
}
}
const responseData = charset === 'utf-8' ? response.data : iconv.decode((await got.get({ url: link, responseType: 'buffer' })).data, charset);
const $ = cheerio.load(responseData);
const title = $('head > title').text();
const version = ver ? 'DISCUZ! ' + ver : $('head > meta[name=generator]').attr('content');
let process;
if (
$('div#footer p em')
.text()
.startsWith('7')
) {
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')
@@ -82,12 +83,7 @@ module.exports = async (ctx) => {
return Promise.resolve(Object.assign({}, single, detail));
})
);
} else if (
$('div#frt p em')
.text()
.toUpperCase()
.startsWith('X')
) {
} 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')