feat(route): add ally/rail (#9660)
Signed-off-by: Rongrong <i@rong.moe>
This commit is contained in:
Rongrong
GitHub
parent
d9dd7afbb0
commit
c4ca3e4ee9
|
|
@ -1297,6 +1297,20 @@ Supported sub-sites:
|
|||
|
||||
<Route author="AlexdanerZe TonyRL" example="/zaker/focusread" path="/zaker/focusread" />
|
||||
|
||||
## 艾莱资讯
|
||||
|
||||
### 世界轨道交通资讯网
|
||||
|
||||
<Route author="Rongronggg9" example="/ally/rail/hyzix/chengguijiaotong/" path="/ally/rail/:category?/:topic?" :paramsDesc="['分类，可在 URL 中找到；略去则抓取首页', '话题，可在 URL 中找到；并非所有页面均有此字段']" radar="1" rssbud="1">
|
||||
|
||||
::: tip 提示
|
||||
|
||||
默认抓取前 20 条，可通过 `?limit=` 改变。
|
||||
|
||||
:::
|
||||
|
||||
</Route>
|
||||
|
||||
## 爱范儿 ifanr
|
||||
|
||||
### 爱范儿频道
|
||||
|
|
|
|||
|
|
@ -0,0 +1,3 @@
|
|||
module.exports = {
|
||||
'/rail/:category?/:topic?': ['Rongronggg9'],
|
||||
};
|
||||
|
|
@ -0,0 +1,13 @@
|
|||
module.exports = {
|
||||
'ally.net.cn': {
|
||||
_name: '艾莱资讯',
|
||||
rail: [
|
||||
{
|
||||
title: '世界轨道交通资讯网',
|
||||
docs: 'https://docs.rsshub.app/new-media.html#ai-lai-zi-xun',
|
||||
source: ['/', '/html/:category?/:topic?'],
|
||||
target: '/ally/rail/:category?/:topic?',
|
||||
},
|
||||
],
|
||||
},
|
||||
};
|
||||
|
|
@ -0,0 +1,109 @@
|
|||
const cheerio = require('cheerio');
|
||||
const got = require('@/utils/got');
|
||||
const { parseDate } = require('@/utils/parse-date');
|
||||
const timezone = require('@/utils/timezone');
|
||||
|
||||
module.exports = async (ctx) => {
|
||||
// http://rail.ally.net.cn/sitemap.html
|
||||
const { category, topic } = ctx.params;
|
||||
const rootUrl = 'http://rail.ally.net.cn';
|
||||
const pageUrl = category ? (topic ? `${rootUrl}/html/${category}/${topic}/` : `${rootUrl}/html/${category}/`) : rootUrl;
|
||||
|
||||
const response = await got.get(pageUrl);
|
||||
const $ = cheerio.load(response.data);
|
||||
let title = $('.container .regsiter a') // what a typo...
|
||||
.get()
|
||||
.slice(1) // drop "首页"
|
||||
.reduce((prev, curr) => (prev ? `${prev} - ${$(curr).text()}` : $(curr).text()), '');
|
||||
title = title || (category && topic ? `${category} - ${topic}` : category) || '首页';
|
||||
let links = [
|
||||
// list page: http://rail.ally.net.cn/html/lujuzixun/
|
||||
$('.left .hynewsO h2 a').get(),
|
||||
// multi-sub-topic page: http://rail.ally.net.cn/html/hyzix/
|
||||
$('.left .list_content_c').find('.new_hy_focus_con_tit a, .new_hy_list_name a').get(),
|
||||
// multi-sub-topic page 2: http://rail.ally.net.cn/html/foster/
|
||||
$('.left').find('.nnewslistpic a, .nnewslistinfo dd a').get(),
|
||||
// data list page: http://rail.ally.net.cn/html/tongjigongbao/
|
||||
$('.left .list_con .datacountTit a').get(),
|
||||
// home page: http://rail.ally.net.cn
|
||||
$('.container_left').find('dd a, h1 a, ul.slideshow li a').get(),
|
||||
].reduce((a, b) => a.concat(b), []);
|
||||
if (!links.length) {
|
||||
// try aggressively sniffing links, e.g. http://rail.ally.net.cn/html/InviteTen/
|
||||
links = $('.left a, .container_left a').get();
|
||||
}
|
||||
|
||||
let items = links
|
||||
.map((link) => {
|
||||
link = $(link);
|
||||
const url = link.attr('href');
|
||||
const urlMatch = url && url.match(/\/html\/(\d{4})\/\w+_(\d{4})\/\d+\.html/);
|
||||
if (!urlMatch) {
|
||||
return null;
|
||||
}
|
||||
const title = link.text();
|
||||
return {
|
||||
title,
|
||||
link: url.startsWith('/') ? `${rootUrl}${url}` : url,
|
||||
pubDate: timezone(parseDate(`${urlMatch[1]}${urlMatch[2]}`), 8),
|
||||
};
|
||||
})
|
||||
.filter(Boolean)
|
||||
.reduce((prev, curr) => (prev.length && prev[prev.length - 1].link === curr.link ? prev : [...prev, curr]), [])
|
||||
.sort((a, b) => b.pubDate - a.pubDate)
|
||||
.slice(0, ctx.query.limit || 20);
|
||||
|
||||
items = await Promise.all(
|
||||
items.map(
|
||||
async (item) =>
|
||||
await ctx.cache.tryGet(item.link, async () => {
|
||||
const response = await got(item.link);
|
||||
const $ = cheerio.load(response.data);
|
||||
// fix weird format
|
||||
let description = '';
|
||||
const content = $('div.content_all');
|
||||
if (content.length) {
|
||||
content
|
||||
.eq(content.length - 1) // some pages have "summary"
|
||||
.contents()
|
||||
.each((_, child) => {
|
||||
const $child = $(child);
|
||||
let innerHtml;
|
||||
if (child.name === 'div') {
|
||||
innerHtml = $child.html();
|
||||
innerHtml = innerHtml && innerHtml.trim();
|
||||
description += !innerHtml || innerHtml === ' ' ? (description ? '<br>' : '') : innerHtml;
|
||||
} else {
|
||||
// bare text node or something else
|
||||
description += $child.toString().trim();
|
||||
}
|
||||
});
|
||||
} else {
|
||||
// http://rail.ally.net.cn/html/2022/InviteTen_0407/4686.html
|
||||
description = $('div.content div').first().html();
|
||||
}
|
||||
|
||||
description = description.replace(/\s*<br ?\/?>\s*$/, ''); // trim <br> at the end
|
||||
const info = $('.content > em span');
|
||||
return {
|
||||
title: $('.content > h2').text() || item.title,
|
||||
description,
|
||||
// pubDate: timezone(parseDate(info.eq(0).text()), 8),
|
||||
pubDate: item.pubDate,
|
||||
author: info
|
||||
.eq(1)
|
||||
.text()
|
||||
.replace(/^来源：/, ''),
|
||||
link: item.link,
|
||||
};
|
||||
})
|
||||
)
|
||||
);
|
||||
|
||||
ctx.state.data = {
|
||||
title: `世界轨道交通资讯网 - ${title}`,
|
||||
link: pageUrl,
|
||||
item: items,
|
||||
description: $('head > meta[name="description"]').attr('content'),
|
||||
};
|
||||
};
|
||||
|
|
@ -0,0 +1,3 @@
|
|||
module.exports = function (router) {
|
||||
router.get('/rail/:category?/:topic?', require('./rail'));
|
||||
};
|
||||
Loading…
Reference in New Issue