feat(route): add 国务院政策文件库 (#11301)

* feat(route): add 国务院政策文件库

* fix wrong author

* typo

* refactor: migrate to v2
This commit is contained in:
zxx-457 2022-11-22 01:53:40 +08:00 committed by GitHub
parent 01e01608cc
commit 10cf5a2468
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
12 changed files with 209 additions and 159 deletions

View File

@ -1002,7 +1002,7 @@ pageClass: routes
<Route author="hillerliao" example="/csrc/auditstatus/:apply_id" path="/csrc/auditstatus" :paramsDesc="['事项类别id`https://neris.csrc.gov.cn/alappl/home/xkDetail` 列表中各地址的 appMatrCde 参数']"/>
## 中国政府
## 中国政府
### 数据
@ -1040,7 +1040,7 @@ pageClass: routes
### 最新文件
<Route author="ciaranchen" example="/gov/zhengce/wenjian" path="/gov/zhengce/wenjian/:pcodeJiguan?" :paramsDesc="['文种分类。国令; 国发; 国函; 国发明电; 国办发; 国办函; 国办发明电; 其他']" />
<Route author="ciaranchen" example="/gov/zhengce/wenjian" path="/gov/zhengce/wenjian/:pcodeJiguan?" :paramsDesc="['文种分类。国令、国发、国函、国发明电、国办发、国办函、国办发明电、其他']" />
### 信息稿件
@ -1059,6 +1059,10 @@ pageClass: routes
</Route>
### 国务院政策文件库
<Route author="zxx-457" example="/gov/zhengce/zhengceku/bmwj" path="/gov/zhengce/zhengceku/:department" :paramsDesc="['库名']" />
### 政府新闻
<Route author="EsuRt" example="/gov/news/:uid" path="/gov/news" :paramsDesc="['分类名']">

View File

@ -1174,9 +1174,9 @@ router.get('/cyzone/author/:id', lazyloadRouteHandler('./routes/cyzone/author'))
router.get('/cyzone/label/:name', lazyloadRouteHandler('./routes/cyzone/label'));
// 政府
router.get('/gov/zhengce/zuixin', lazyloadRouteHandler('./routes/gov/zhengce/zuixin'));
router.get('/gov/zhengce/wenjian/:pcodeJiguan?', lazyloadRouteHandler('./routes/gov/zhengce/wenjian'));
router.get('/gov/zhengce/govall/:advance?', lazyloadRouteHandler('./routes/gov/zhengce/govall'));
// router.get('/gov/zhengce/zuixin', lazyloadRouteHandler('./routes/gov/zhengce/zuixin'));
// router.get('/gov/zhengce/wenjian/:pcodeJiguan?', lazyloadRouteHandler('./routes/gov/zhengce/wenjian'));
// router.get('/gov/zhengce/govall/:advance?', lazyloadRouteHandler('./routes/gov/zhengce/govall'));
router.get('/gov/province/:name/:category', lazyloadRouteHandler('./routes/gov/province'));
router.get('/gov/city/:name/:category', lazyloadRouteHandler('./routes/gov/city'));
router.get('/gov/statecouncil/briefing', lazyloadRouteHandler('./routes/gov/statecouncil/briefing'));

View File

@ -1,59 +0,0 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
module.exports = async (ctx) => {
const { advance } = ctx.params;
const link = `http://sousuo.gov.cn/list.htm`;
const params = new URLSearchParams({
n: 20,
t: 'govall',
sort: 'pubtime',
advance: 'true',
});
const query = `${params.toString()}&${advance}`;
const res = await got.get(link, {
searchParams: query.replace(/([\u4e00-\u9fa5])/g, (str) => encodeURIComponent(str)),
});
const $list = cheerio.load(res.data);
ctx.state.data = {
title: `信息稿件 - 中国政府网`,
link: `${link}?${query}`,
item: await Promise.all(
$list('body > div.dataBox > table > tbody > tr')
.slice(1)
.map(async (_, el) => {
const item = $list(el);
const number = item.find('td.info').next();
const title = item.find('td.info > a');
const href = title.attr('href');
const key = `gov_zhengce: ${href}`;
let description;
const value = await ctx.cache.get(key);
if (value) {
description = value;
} else {
try {
const contentData = await got.get(href);
const $content = cheerio.load(contentData.data);
description = $content('#UCAP-CONTENT').html();
} catch (error) {
description = '文章已被删除';
}
ctx.cache.set(key, description);
}
return {
title: number.text() + ' | ' + title.text(),
description,
link: href,
};
})
.get()
),
};
};

View File

@ -1,53 +0,0 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const queryString = require('query-string');
module.exports = async (ctx) => {
const { pcodeJiguan } = ctx.params;
const link = `http://sousuo.gov.cn/list.htm`;
const res = await got.get(link, {
searchParams: queryString.stringify({
n: 20,
sort: 'pubtime',
t: 'paper',
pcodeJiguan: pcodeJiguan ? pcodeJiguan : '',
}),
});
const $list = cheerio.load(res.data);
ctx.state.data = {
title: `最新文件 - 中国政府网`,
link,
item: await Promise.all(
$list('body > div.dataBox > table > tbody > tr')
.slice(1)
.map(async (_, el) => {
const item = $list(el);
const number = item.find('td.info').next();
const title = item.find('td.info > a');
const href = title.attr('href');
const key = `gov_zhengce: ${href}`;
let description;
const value = await ctx.cache.get(key);
if (value) {
description = value;
} else {
const contentData = await got.get(href);
const $content = cheerio.load(contentData.data);
description = $content('#UCAP-CONTENT').html();
ctx.cache.set(key, description);
}
return {
title: number.text() + ' | ' + title.text(),
description,
link: href,
};
})
.get()
),
};
};

View File

@ -1,42 +0,0 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
module.exports = async (ctx) => {
const link = `http://www.gov.cn/zhengce/zuixin.htm`;
const listData = await got.get(link);
const $list = cheerio.load(listData.data);
ctx.state.data = {
title: `最新政策 - 中国政府网`,
link,
item: await Promise.all(
$list('.news_box .list li:not(.line)')
.map(async (_, el) => {
const $el = $list(el);
const $a = $el.find('h4 a');
const href = $a.attr('href');
const link = href.startsWith('/zhengce') ? `http://www.gov.cn${href}` : href;
const key = `gov_zhengce_zuixin: ${link}`;
let description;
const value = await ctx.cache.get(key);
if (value) {
description = value;
} else {
const contentData = await got.get(link);
const $content = cheerio.load(contentData.data);
description = $content('#UCAP-CONTENT').html();
ctx.cache.set(key, description);
}
return {
title: $a.text(),
description,
guid: href,
link,
pubDate: new Date($el.find('.date').text()).toUTCString(),
};
})
.get()
),
};
};

View File

@ -35,6 +35,10 @@ module.exports = {
'/pbc/tradeAnnouncement': ['nczitzk'],
'/pbc/zcyj': ['Fatpandac'],
'/stats/:path+': ['bigfei'],
'/zhengce/govall/:advance?': ['ciaranchen'],
'/zhengce/wenjian/:pcodeJiguan?': ['ciaranchen'],
'/zhengce/zhengceku/:department': ['zxx-457'],
'/zhengce/zuixin': ['SettingDust'],
// province
'/anhui/kjt/:path?': ['nczitzk'],
'/beijing/bphc/:channel': ['bigfei'],

View File

@ -776,6 +776,35 @@ module.exports = {
},
],
},
'www.gov.cn': {
_name: '中国政府网',
'.': [
{
title: '最新政策',
docs: 'https://docs.rsshub.app/government.html#zhong-guo-zheng-fu-wang',
source: ['/zhengce/zuixin.htm', '/'],
target: '/gov/zhengce/zuixin',
},
{
title: '最新文件',
docs: 'https://docs.rsshub.app/government.html#zhong-guo-zheng-fu-wang',
source: ['/'],
target: '/gov/zhengce/wenjian',
},
{
title: '信息稿件',
docs: 'https://docs.rsshub.app/government.html#zhong-guo-zheng-fu-wang',
source: ['/'],
target: '/gov/zhengce/govall',
},
{
title: '国务院政策文件库',
docs: 'https://docs.rsshub.app/government.html#zhong-guo-zheng-fu-wang',
source: ['/zhengce/zhengceku/:lib'],
target: (params) => `/gov/zhengce/zhengceku/${params.libs}`,
},
],
},
'xz.gov.cn': {
_name: '徐州市人民政府',
hrss: [

View File

@ -27,6 +27,10 @@ module.exports = function (router) {
router.get('/pbc/tradeAnnouncement', require('./pbc/tradeAnnouncement'));
router.get('/pbc/zcyj', require('./pbc/zcyj'));
router.get(/stats(\/[\w/-]+)?/, require('./stats'));
router.get('/zhengce/govall/:advance?', require('./zhengce/govall'));
router.get('/zhengce/wenjian/:pcodeJiguan?', require('./zhengce/wenjian'));
router.get('/zhengce/zhengceku/:department', require('./zhengce/zhengceku'));
router.get('/zhengce/zuixin', require('./zhengce/zuixin'));
// province
router.get(/anhui\/kjt\/([\w\d/-]+)?/, require('./anhui/kjt'));
router.get(/beijing\/bphc(\/[\w/-]+)?/, require('./beijing/bphc'));

View File

@ -0,0 +1,55 @@
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 { advance } = ctx.params;
const link = `http://sousuo.gov.cn/list.htm`;
const params = new URLSearchParams({
n: 20,
t: 'govall',
sort: 'pubtime',
advance: 'true',
});
const query = `${params.toString()}&${advance}`;
const res = await got.get(link, {
searchParams: query.replace(/([\u4e00-\u9fa5])/g, (str) => encodeURIComponent(str)),
});
const $ = cheerio.load(res.data);
const list = $('body > div.dataBox > table > tbody > tr')
.slice(1)
.toArray()
.map((elem) => {
elem = $(elem);
return {
title: elem.find('td:nth-child(2) > a').text(),
link: elem.find('td:nth-child(2) > a').attr('href'),
pubDate: timezone(parseDate(elem.find('td:nth-child(5)').text()), 8),
};
});
const items = await Promise.all(
list.map((item) =>
ctx.cache.tryGet(item.link, async () => {
let description = '';
try {
const contentData = await got(item.link);
const $ = cheerio.load(contentData.data);
description = $('#UCAP-CONTENT').html();
} catch (error) {
description = '文章已被删除';
}
item.description = description;
return item;
})
)
);
ctx.state.data = {
title: '信息稿件 - 中国政府网',
link: `${link}?${query}`,
item: items,
};
};

View File

@ -0,0 +1,47 @@
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 { pcodeJiguan } = ctx.params;
const link = 'http://sousuo.gov.cn/list.htm';
const res = await got(link, {
searchParams: {
n: ctx.query.limit ? Number(ctx.query.limit) : 20,
sort: 'pubtime',
t: 'paper',
pcodeJiguan: pcodeJiguan ? pcodeJiguan : '',
},
});
const $ = cheerio.load(res.data);
const list = $('body > div.dataBox > table > tbody > tr')
.slice(1)
.toArray()
.map((elem) => {
elem = $(elem);
return {
title: elem.find('td:nth-child(2) > a').text(),
link: elem.find('td:nth-child(2) > a').attr('href'),
pubDate: timezone(parseDate(elem.find('td:nth-child(5)').text()), 8),
};
});
const items = await Promise.all(
list.map((item) =>
ctx.cache.tryGet(item.link, async () => {
const contentData = await got(item.link);
const $ = cheerio.load(contentData.data);
item.description = $('#UCAP-CONTENT').html();
return item;
})
)
);
ctx.state.data = {
title: '最新文件 - 中国政府网',
link: res.url,
item: items,
};
};

View File

@ -0,0 +1,22 @@
const buildData = require('@/utils/common-config');
module.exports = async (ctx) => {
const department = ctx.params.department;
const link = `http://www.gov.cn/zhengce/zhengceku/${department}/`;
ctx.state.data = await buildData({
link,
url: link,
title: `%title%`,
description: '政府文件库, 当页的所有列表',
params: {
title: `$('.channel_tab > .noline > a').text().trim() + ' - 政府文件库'`,
},
item: {
item: '.news_box > .list > ul > li:not(.line)',
title: `$('h4 > a').text()`,
link: `$('h4 > a').attr('href')`,
pubDate: `parseDate($('h4 > .date').text().trim())`,
},
});
};

View File

@ -0,0 +1,39 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const { parseDate } = require('@/utils/parse-date');
const timezone = require('@/utils/timezone');
const baseUrl = 'http://www.gov.cn';
module.exports = async (ctx) => {
const link = `${baseUrl}/zhengce/zuixin.htm`;
const listData = await got(link);
const $ = cheerio.load(listData.data);
const list = $('.news_box .list li:not(.line)')
.toArray()
.map((elem) => {
elem = $(elem);
return {
title: elem.find('h4 a').text(),
link: new URL(elem.find('h4 a').attr('href'), baseUrl).href,
pubDate: timezone(parseDate(elem.find('h4 .date').text()), 8),
};
});
const items = await Promise.all(
list.map((item) =>
ctx.cache.tryGet(item.link, async () => {
const contentData = await got(item.link);
const $ = cheerio.load(contentData.data);
item.description = $('#UCAP-CONTENT').html();
return item;
})
)
);
ctx.state.data = {
title: '最新政策 - 中国政府网',
link,
item: items,
};
};