fix(route): 新华网新华社新闻 (#14390)
This commit is contained in:
Ethan Shen
GitHub
parent
64c0c47d44
commit
b552dad8d7
|
|
@ -1,3 +1,3 @@
|
|||
module.exports = {
|
||||
'/whxw': ['nczitzk'],
|
||||
'/xhsxw': ['nczitzk'],
|
||||
};
|
||||
|
|
|
|||
|
|
@ -5,8 +5,8 @@ module.exports = {
|
|||
{
|
||||
title: '新华社新闻',
|
||||
docs: 'https://docs.rsshub.app/routes/new-media#xin-hua-wang-xin-hua-she-xin-wen',
|
||||
source: ['/'],
|
||||
target: '/news/whxw',
|
||||
source: ['/xhsxw.htm'],
|
||||
target: '/news/xhsxw',
|
||||
},
|
||||
],
|
||||
},
|
||||
|
|
|
|||
|
|
@ -1,3 +1,4 @@
|
|||
module.exports = function (router) {
|
||||
router.get('/whxw', require('./whxw'));
|
||||
router.get('/xhsxw', require('./xhsxw'));
|
||||
router.get('/whxw', require('./xhsxw'));
|
||||
};
|
||||
|
|
|
|||
|
|
@ -0,0 +1,21 @@
|
|||
{{ if images }}
|
||||
{{ each images image }}
|
||||
{{ if image?.src }}
|
||||
<figure>
|
||||
<img
|
||||
{{ if image.alt }}
|
||||
alt="{{ image.alt }}"
|
||||
{{ /if }}
|
||||
src="{{ image.src }}">
|
||||
</figure>
|
||||
{{ /if }}
|
||||
{{ /each }}
|
||||
{{ /if }}
|
||||
|
||||
{{ if intro }}
|
||||
<blockquote>{{ intro }}</blockquote>
|
||||
{{ /if }}
|
||||
|
||||
{{ if description }}
|
||||
{{@ description }}
|
||||
{{ /if }}
|
||||
|
|
@ -1,68 +0,0 @@
|
|||
const got = require('@/utils/got');
|
||||
const cheerio = require('cheerio');
|
||||
const { parseDate } = require('@/utils/parse-date');
|
||||
const timezone = require('@/utils/timezone');
|
||||
|
||||
module.exports = async (ctx) => {
|
||||
const rootUrl = 'http://www.news.cn';
|
||||
const currentUrl = `${rootUrl}/whxw.htm`;
|
||||
const response = await got({
|
||||
method: 'get',
|
||||
url: currentUrl,
|
||||
});
|
||||
|
||||
const $ = cheerio.load(response.data);
|
||||
|
||||
let items = $('h3 a')
|
||||
.toArray()
|
||||
.slice(0, ctx.query.limit ? Number.parseInt(ctx.query.limit) : 100)
|
||||
.map((item) => {
|
||||
item = $(item);
|
||||
return {
|
||||
title: item.text(),
|
||||
link: item.attr('href'),
|
||||
};
|
||||
});
|
||||
|
||||
items = await Promise.all(
|
||||
items.map((item) =>
|
||||
ctx.cache.tryGet(item.link, async () => {
|
||||
let detailResponse;
|
||||
|
||||
try {
|
||||
detailResponse = await got(item.link);
|
||||
} catch {
|
||||
item.status = 404;
|
||||
}
|
||||
|
||||
if (item.status !== 404) {
|
||||
const content = cheerio.load(detailResponse.data);
|
||||
const date = content('.header-time.left').text() || content('.h-time').text() || content('.info').text();
|
||||
const author =
|
||||
content('.editor').text() ??
|
||||
content('.p-jc')
|
||||
.text()
|
||||
.replaceAll(/[\n\r]/g, '')
|
||||
.replace('： ', ':')
|
||||
.trim();
|
||||
|
||||
item.author = author.split(':').pop().replace('】', '');
|
||||
|
||||
content('#articleEdit').remove();
|
||||
|
||||
item.description = content('#detail').html();
|
||||
item.pubDate = timezone(parseDate(date, date.includes('/') ? 'YYYYMM/DD HH:mm:ss' : 'YYYY-MM-DD HH:mm:ss'), +8);
|
||||
}
|
||||
|
||||
delete item.status;
|
||||
return item;
|
||||
})
|
||||
)
|
||||
);
|
||||
|
||||
ctx.state.data = {
|
||||
title: $('title').text(),
|
||||
link: currentUrl,
|
||||
item: items,
|
||||
};
|
||||
};
|
||||
|
|
@ -0,0 +1,81 @@
|
|||
const got = require('@/utils/got');
|
||||
const cheerio = require('cheerio');
|
||||
const timezone = require('@/utils/timezone');
|
||||
const { parseDate } = require('@/utils/parse-date');
|
||||
const { art } = require('@/utils/render');
|
||||
const path = require('path');
|
||||
|
||||
module.exports = async (ctx) => {
|
||||
const limit = ctx.query.limit ? Number.parseInt(ctx.query.limit, 10) : 100;
|
||||
|
||||
const rootUrl = 'http://www.news.cn';
|
||||
const currentUrl = new URL('xhsxw.htm', rootUrl).href;
|
||||
|
||||
const { data: currentResponse } = await got(currentUrl);
|
||||
|
||||
const $ = cheerio.load(currentResponse);
|
||||
|
||||
const id = $('ul.wz-list')
|
||||
.prop('data')
|
||||
.replace(/datasource:/, '');
|
||||
|
||||
const apiUrl = new URL(`ds_${id}.json`, rootUrl).href;
|
||||
|
||||
const {
|
||||
data: { datasource: response },
|
||||
} = await got(apiUrl);
|
||||
|
||||
let items = response.slice(0, limit).map((item) => ({
|
||||
title: item.title,
|
||||
link: new URL(item.publishUrl, rootUrl).href,
|
||||
description: art(path.join(__dirname, 'templates/description.art'), {
|
||||
images:
|
||||
item.shareImages?.map((i) => ({
|
||||
src: i.imageUrl,
|
||||
alt: item.title,
|
||||
})) ?? undefined,
|
||||
intro: item.summary,
|
||||
}),
|
||||
author: item.author,
|
||||
category: item.keywords.split(/-|,/),
|
||||
guid: `news-${item.contentId}`,
|
||||
pubDate: timezone(parseDate(item.publishTime), +8),
|
||||
}));
|
||||
|
||||
items = await Promise.all(
|
||||
items.map((item) =>
|
||||
ctx.cache.tryGet(item.link, async () => {
|
||||
try {
|
||||
const { data: detailResponse } = await got(item.link);
|
||||
|
||||
const content = cheerio.load(detailResponse);
|
||||
|
||||
item.description += art(path.join(__dirname, 'templates/description.art'), {
|
||||
description: content('#detailContent').html(),
|
||||
});
|
||||
} catch {
|
||||
// no-empty
|
||||
}
|
||||
|
||||
return item;
|
||||
})
|
||||
)
|
||||
);
|
||||
|
||||
const title = $('title').text();
|
||||
const image = new URL('20141223_xhsxw_logo_v1.png', rootUrl).href;
|
||||
const icon = new URL('favicon.ico', rootUrl).href;
|
||||
|
||||
ctx.state.data = {
|
||||
item: items,
|
||||
title,
|
||||
link: currentUrl,
|
||||
description: title.split(/_/)[0],
|
||||
language: 'zh',
|
||||
image,
|
||||
icon,
|
||||
logo: icon,
|
||||
author: title.split(/_/).pop(),
|
||||
allowEmpty: true,
|
||||
};
|
||||
};
|
||||
|
|
@ -4774,7 +4774,7 @@
|
|||
|
||||
### 新华社新闻 {#xin-hua-wang-xin-hua-she-xin-wen}
|
||||
|
||||
<Route author="nczitzk" example="/news/whxw" path="/news/whxw" />
|
||||
<Route author="nczitzk" example="/news/xhsxw" path="/news/xhsxw" radar="1" />
|
||||
|
||||
## 新浪 {#xin-lang}
|
||||
|
||||
|
|
|
|||
Loading…
Reference in New Issue