rsshub
Version:
Make RSS Great Again!
169 lines (167 loc) • 6.63 kB
JavaScript
import "./esm-shims-CzJ_djXG.mjs";
import "./config-C37vj7VH.mjs";
import "./dist-BInvbO1W.mjs";
import "./logger-Czu8UMNd.mjs";
import "./ofetch-BIyrKU3Y.mjs";
import { n as parseRelativeDate } from "./parse-date-BrP7mxXf.mjs";
import { t as cache_default } from "./cache-Bo__VnGm.mjs";
import "./helpers-DxBp0Pty.mjs";
import { t as got_default } from "./got-KxxWdaxq.mjs";
import { load } from "cheerio";
//#region lib/routes/gov/moa/moa.ts
const hostUrl = "http://www.moa.gov.cn/";
const hostUrlObj = new URL(hostUrl);
const route = {
path: "/moa/suburl/:suburl{.+}",
categories: ["government"],
example: "/gov/moa/suburl/gk/zcjd/",
radar: [{
source: ["moa.gov.cn/"],
target: "/moa/suburl/:suburl"
}],
parameters: { suburl: "下级目录，请使用最下级的目录" },
name: "中华人民共和国农业农村部 - 新闻",
maintainers: ["Origami404", "lyqluis"],
handler,
url: "moa.gov.cn/",
description: `更多例子：
- \`农业农村部动态\`的网页链接是\`http://www.moa.gov.cn/xw/zwdt/\`, 对应的\`suburl\`是\`xw/zwdt\`
- \`财务公开\`的网页链接是\`http://www.moa.gov.cn/gk/cwgk_1/\`, 对应的\`suburl\`是\`gk/cwgk_1\`
- 像[政策法规](http://www.moa.gov.cn/gk/zcfg/)这种页面(\`http://www.moa.gov.cn/gk/zcfg/\`), 它**不是**一个合法的分类目录，它是\`法律\`, \`行政法规\`, \`部门规章\`等一堆栏目的集合，这时候请点开对应栏目的\`更多 >>\`进入栏目的最下级目录，再根据上面的规则提取\`suburl\`
- 特别地，\`图片新闻\`对应的\`suburl\`为\`xw/tpxw/\`, \`最新公开\`对应的\`suburl\`为\`govpublic\`, \`数据>最新发布\`对应的\`suburl\`为\`sj/zxfb\``
};
async function handler(ctx) {
const rawSuburl = ctx.req.param("suburl");
const suburl = rawSuburl.slice(-1) === "/" ? rawSuburl : rawSuburl + "/";
if (suburl === "xw/tpxw/") return await dealChannel(suburl, {
channelTitleSelector: ".pub-media2-head",
listSelector: ".tupian_list li",
titleSelector: "a[class=\"block w_fill ellipsis adc ahc\"]",
dateSelector: "span"
});
else if (suburl.startsWith("sj/zxfb")) return await dealLatestDataChannel();
else if (suburl.startsWith("gk")) return await dealChannel(suburl, {
channelTitleSelector: "title",
listSelector: ".commonlist li",
titleSelector: "a",
dateSelector: "span"
});
else if (suburl.startsWith("govpublic")) return await dealChannel("govpublic/1/index.htm", {
channelTitleText: "最新公开",
listSelector: ".commonlist li",
titleSelector: "a",
dateSelector: "span"
});
else return await dealChannel(suburl, {
channelTitleSelector: ".pub-media1-head-title",
listSelector: ".ztlb",
titleSelector: "a",
dateSelector: "span"
});
}
async function dealChannel(suburl, selectors) {
const { channelTitleSelector, listSelector, titleSelector, dateSelector, channelTitleText } = selectors;
const url = suburl.startsWith("http") ? suburl : hostUrl + suburl;
const $ = load((await got_default.get(url)).data);
const channelTitle = channelTitleText ?? $(channelTitleSelector).text();
const pageInfos = $(listSelector).toArray().map((e) => {
const element = $(e);
const titleElement = element.find(titleSelector);
const title = titleElement.text();
const [link, pageType] = dealLink(titleElement, url);
return {
pageType,
title,
link,
pubDate: parseRelativeDate(element.find(dateSelector).text().trim())
};
});
const items = await Promise.all(pageInfos.map(async (item) => {
const link = item.link;
const cacheIn = await cache_default.get(link);
if (cacheIn) return JSON.parse(cacheIn);
if (item.pageType === "normal") item = await dealNormalPage(link, item);
else if (item.pageType === "govpublic") item = await dealGovpublicPage(link, item);
else item.description = `外部链接：${item.link}`;
cache_default.set(link, JSON.stringify(item));
return item;
}));
return {
title: `中华人民共和国农业农村部 - ${channelTitle}`,
link: url,
item: items
};
}
async function dealNormalPage(link, item) {
const $ = load((await got_default.get(link)).data);
if (link.includes("zbft")) {
const pageHeader = $(".nybzb").html() ?? "";
const pics = $(".tpsl").html() ?? "";
const content = $(".wzsl").html() ?? "";
item.description = pageHeader + pics + content;
return item;
}
const metaElements = $(".bjjMAuthorBox span.dc_2").toArray();
item.author = `${$(metaElements[1]).text()} ${$(metaElements[2]).text()}`;
const exactTime = $(metaElements[0]).text();
const dateMatch = /\d{4}-\d{2}-\d{2}/.exec(exactTime);
const timeMatch = /\d{2}:\d{2}/.exec(exactTime);
item.pubDate = parseRelativeDate(`${dateMatch[0]} ${timeMatch[0]}`);
item.description = $(".arc_body").html();
return item;
}
async function dealGovpublicPage(link, item) {
if (item.link.endsWith(".pdf")) return item;
const $ = load((await got_default.get(link)).data);
const body = $(".gsj_htmlcon_bot");
const [, year, month, date] = $(".pubtime").text().match(/：(\d{4})[|年-](\d{1,2})[|月-](\d{1,2})日?/);
const [, author] = $(".pubtime.source")?.text()?.match(/：(.+)/) ?? [null, ""];
if (year && month && date) item.pubDate = `${year}-${month}-${date}`;
item.author = author;
item.description = body.html();
return item;
}
async function dealLatestDataChannel() {
const res = await got_default({
url: "http://zdscxx.moa.gov.cn:8080/nyb/getMessages",
method: "post",
json: {
page: 1,
rows: 20,
type: "最新发布",
isLatestMessage: true
}
});
return {
title: `中华人民共和国农业农村部 - 数据 - 最新发布`,
link: "http://zdscxx.moa.gov.cn:8080/nyb/pc/messageList.jsp",
item: await Promise.all(res.data.result.table.map((item) => {
const { date, id } = item;
item.pubDate = date;
const link = item.link = `http://zdscxx.moa.gov.cn:8080/nyb/pc/messageView.jsp?id=${id}`;
return cache_default.tryGet(link, async () => {
const { content, source } = await getLatestDataArticleDetail(id);
item.description = content;
item.author = source;
return item;
});
}))
};
}
async function getLatestDataArticleDetail(id) {
return (await got_default({
url: "http://zdscxx.moa.gov.cn:8080/nyb/getMessagesById",
method: "post",
form: { id }
})).data.result;
}
function dealLink(element, url) {
const rawLink = element.attr("href");
const { host, href } = new URL(rawLink, url);
let pageType = null;
if (host === hostUrlObj.host) pageType = href.includes("gk") || href.includes("govpublic") ? "govpublic" : "normal";
else pageType = "outside";
return [href, pageType];
}
//#endregion
export { route };