rsshub
Version:
Make RSS Great Again!
158 lines (153 loc) • 5.67 kB
JavaScript
import { t as parseDate } from "./parse-date-Cr0wQjV_.mjs";
import { t as PRESETS } from "./header-generator-DACPoxdp.mjs";
import { t as cache_default } from "./cache-BkqOokyU.mjs";
import { t as got_default } from "./got-BTowW50G.mjs";
import { jsx, jsxs } from "hono/jsx/jsx-runtime";
import { load } from "cheerio";
import { renderToString } from "hono/jsx/dom/server";
import { raw } from "hono/html";
import pMap from "p-map";
//#region lib/routes/wsj/utils.tsx
const parseArticle = (item) => cache_default.tryGet(item.link, async () => {
const html = (await got_default({
url: item.link.replace(/(?<=^https:\/\/\w+\.wsj\.com)/, "/amp"),
method: "get",
headerGeneratorOptions: PRESETS.MODERN_ANDROID
})).data;
const $ = load(html);
const summary = $("head > meta[name=\"description\"]").attr("content");
const updatedAt = $("meta[itemprop=\"dateModified\"]").attr("content");
const publishedAt = $("meta[itemprop=\"datePublished\"]").attr("content");
const author = $(".author > a[rel=\"author\"]").text();
const categories = $("meta[name=\"keywords\"]").attr("content").split(",").map((c) => c.trim());
const article = $("article");
item.subTitle = $("h2.sub-head").html();
article.find(".media-object-podcast").remove();
article.find(".bylineWrap").each((i, e) => {
$(e).find("p").each((_, el) => {
$(el).replaceWith($(el).html() ?? "");
});
});
article.find(".bigTop-hero").each((i, e) => {
const imgSrc = $(e).find("amp-img").attr("src");
const imgAlt = $(e).find("amp-img").attr("alt");
const figCaption = $(e).find(".imageCaption").text().trim();
const figCredit = $(e).find(".imageCredit").text().trim();
const fig = $(`<figure><img src="${imgSrc}" alt="${imgAlt}"><figcaption>${figCaption} <span>${figCredit}</span></figcaption></figure>`);
$(fig).insertBefore(e);
$(e).remove();
});
article.find("amp-img").each((i, e) => {
const img = $(`<img width="${e.attribs.width}" height="${e.attribs.height}" src="${e.attribs.src}" alt="${e.attribs.alt}">`);
$(img).insertBefore(e);
$(e).remove();
});
article.find("amp-iframe").each((i, e) => {
const iframe = $(`<iframe width="${e.attribs.width}" height="${e.attribs.height}" src="${e.attribs.src}">`);
$(iframe).insertBefore(e);
$(e).remove();
});
for (const selector of [
"amp-ad",
".wsj-ad",
"h1",
"h2",
".zonedModule",
".snippet",
".newsletter-inset"
]) article.find(selector).each((i, e) => {
$(e).remove();
});
article.find(".paywall").each((i, e) => {
$(e.childNodes).insertBefore(e);
$(e).remove();
});
item.article = article.html();
item.description = renderToString(/* @__PURE__ */ jsx(WsjDescription, {
subTitle: item.subTitle,
article: item.article
}));
return {
title: item.title,
pubDate: parseDate(publishedAt),
updated: parseDate(updatedAt),
author,
link: item.link,
summary,
description: item.description,
category: categories,
icon: "https://s.wsj.net/media/wsj_launcher-icon-4x.png",
logo: "https://vir.wsj.net/fp/assets/webpack4/img/wsj-logo-big-black.165e51cc.svg"
};
});
const WsjDescription = ({ subTitle, article }) => /* @__PURE__ */ jsxs("div", { children: [subTitle ? raw(subTitle) : null, article ? raw(article) : null] });
//#endregion
//#region lib/routes/wsj/news.ts
const hostMap = {
"en-us": "https://www.wsj.com",
"zh-cn": "https://cn.wsj.com/zh-hans",
"zh-tw": "https://cn.wsj.com/zh-hant"
};
const route = {
path: "/:lang/:category?",
categories: ["traditional-media"],
example: "/wsj/en-us/opinion",
parameters: {
lang: "Language, `en-us`, `zh-cn`, `zh-tw`",
category: "Category. See below"
},
features: {
requireConfig: false,
requirePuppeteer: false,
antiCrawler: false,
supportBT: false,
supportPodcast: false,
supportScihub: false
},
name: "News",
maintainers: ["oppilate"],
handler,
description: `en\\_us
| World | U.S. | Politics | Economy | Business | Tech | Markets | Opinion | Books & Arts | Real Estate | Life & Work | Sytle | Sports |
| ----- | ---- | -------- | ------- | -------- | ---------- | ------- | ------- | ------------ | ----------- | ----------- | ------------------- | ------ |
| world | us | politics | economy | business | technology | markets | opinion | books-arts | realestate | life-work | style-entertainment | sports |
zh-cn / zh-tw
| 国际 | 中国 | 金融市场 | 经济 | 商业 | 科技 | 派 | 专栏与观点 |
| ----- | ----- | -------- | ------- | -------- | ---------- | --------- | ---------- |
| world | china | markets | economy | business | technology | life-arts | opinion |
Provide full article RSS for WSJ topics.`
};
async function handler(ctx) {
const lang = ctx.req.param("lang");
const category = ctx.req.param("category") || "";
const host = hostMap[lang];
let subtitle = ` - ${lang.toUpperCase()}`;
let url = host;
if (category.length > 0) {
url = `${host}/news/${category}`;
subtitle += ` - ${category}`;
}
const contents = load((await got_default({
method: "get",
url
})).data)("script:contains(\"window.__STATE__\")").text();
const data = JSON.parse(contents.match(/\{.*\}/)[0]).data;
const items = await pMap(Object.entries(data).filter(([key, value]) => {
if (!key.startsWith("article")) return false;
return value.data.data.url.includes("wsj.com/articles/");
}).map(([key]) => key).map((key) => {
return {
title: data[key].data.data.headline,
link: data[key].data.data.url,
test: key
};
}), (item) => parseArticle(item), { concurrency: 10 });
return {
title: `WSJ${subtitle}`,
link: url,
description: `WSJ${subtitle}`,
item: items
};
}
//#endregion
export { route };