rsshub/dist-lib/utils-DUVgrHXS.mjs

Make RSS Great Again!

import { t as rofetch } from "./ofetch-C3ts-Hud.mjs";
import { t as parseDate } from "./parse-date-Cr0wQjV_.mjs";
import { t as cache_default } from "./cache-BkqOokyU.mjs";
import { Fragment, jsx, jsxs } from "hono/jsx/jsx-runtime";
import { load } from "cheerio";
import { renderToString } from "hono/jsx/dom/server";
import { destr } from "destr";
import { raw } from "hono/html";
//#region lib/routes/bloomberg/templates/audio-media.tsx
const renderAudioMedia = ({ img, src, caption, credit }) => renderToString(/* @__PURE__ */ jsxs("figure", { children: [/* @__PURE__ */ jsxs("div", { children: [/* @__PURE__ */ jsx("img", { src: img }), /* @__PURE__ */ jsxs("audio", {
	controls: true,
	children: [/* @__PURE__ */ jsx("source", {
		src,
		type: "audio/mpeg"
	}), "Your browser does not support the audio element."]
})] }), /* @__PURE__ */ jsxs("figcaption", { children: [/* @__PURE__ */ jsx("div", {
	class: "caption",
	children: caption ? raw(caption) : null
}), /* @__PURE__ */ jsx("div", {
	class: "credit",
	children: credit ? raw(credit) : null
})] })] }));
//#endregion
//#region lib/routes/bloomberg/templates/chart-media.tsx
const renderChartMedia = ({ chart }) => renderToString(/* @__PURE__ */ jsxs("figure", { children: [
	chart.title ? /* @__PURE__ */ jsx(Fragment, { children: chart.title }) : null,
	chart.subtitle ? /* @__PURE__ */ jsx("p", { children: chart.subtitle }) : null,
	/* @__PURE__ */ jsx("noscript", { children: /* @__PURE__ */ jsx("img", {
		src: chart.fallback,
		alt: chart.chartAlt,
		loading: "lazy",
		style: "display:block; margin-left:auto; margin-right:auto; width:100%;"
	}) }),
	/* @__PURE__ */ jsx("iframe", {
		id: chart.chartId,
		title: chart.title,
		referrerpolicy: "no-referrer",
		width: "100%",
		height: "150vh",
		frameborder: "0",
		marginheight: "0",
		marginwidth: "0",
		loading: "lazy",
		scrolling: "no",
		style: "border:0; margin:0; padding:0; width:100%; height:150vh;",
		src: chart.url
	}),
	chart.source ? /* @__PURE__ */ jsxs("figcaption", { children: [/* @__PURE__ */ jsx("div", {
		class: "source",
		children: raw(chart.source)
	}), chart.footnote ? /* @__PURE__ */ jsx("p", { children: chart.footnote }) : null] }) : null
] }));
//#endregion
//#region lib/routes/bloomberg/templates/image-figure.tsx
const renderImageFigure = ({ src, alt, caption, credit }) => renderToString(/* @__PURE__ */ jsxs("figure", { children: [/* @__PURE__ */ jsx("img", {
	src,
	alt,
	loading: "lazy",
	style: "display:block; margin-left:auto; margin-right:auto; width:100%;"
}), caption || credit ? /* @__PURE__ */ jsxs("figcaption", { children: [/* @__PURE__ */ jsx("div", {
	class: "caption",
	children: caption ? raw(caption) : null
}), /* @__PURE__ */ jsx("div", {
	class: "credit",
	children: credit ? raw(credit) : null
})] }) : null] }));
//#endregion
//#region lib/routes/bloomberg/templates/video-media.tsx
const renderVideoMedia = ({ stream, mp4, coverUrl, caption }) => renderToString(/* @__PURE__ */ jsxs("figure", { children: [/* @__PURE__ */ jsxs("video", {
	controls: true,
	playsinline: true,
	"webkit-playsinline": "true",
	"x5-playsinline": "true",
	"x5-video-player-type": "h5",
	"x5-video-orientation": "landscape|portrait",
	"x5-video-player-fullscreen": "true",
	"x-webkit-airplay": "allow",
	preload: "metadata",
	poster: coverUrl,
	children: [stream ? /* @__PURE__ */ jsx("source", {
		src: stream,
		type: "application/x-mpegURL"
	}) : null, mp4 ? /* @__PURE__ */ jsx("source", {
		src: mp4,
		type: "video/mp4"
	}) : null]
}), caption ? /* @__PURE__ */ jsx("figcaption", { children: /* @__PURE__ */ jsx("div", {
	class: "caption",
	children: raw(caption)
}) }) : null] }));
//#endregion
//#region lib/routes/bloomberg/templates/lede-media.tsx
const renderLedeMedia = (media) => {
	if (media?.kind === "video") return renderVideoMedia(media.video ?? {});
	if (media?.kind === "image") return renderToString(/* @__PURE__ */ jsxs("figure", { children: [/* @__PURE__ */ jsx("img", {
		src: media.src,
		alt: media.description,
		loading: "lazy",
		style: "display:block; margin-left:auto; margin-right:auto; width:100%;"
	}), media.caption ? /* @__PURE__ */ jsxs("figcaption", { children: [/* @__PURE__ */ jsx("div", {
		class: "caption",
		children: raw(media.caption)
	}), /* @__PURE__ */ jsx("div", {
		class: "credit",
		children: media.credit ? raw(media.credit) : null
	})] }) : null] }));
	return "";
};
//#endregion
//#region lib/routes/bloomberg/utils.ts
const rootUrl = "https://www.bloomberg.com/feeds";
const idSel = "script[id^=\"article-info\"][type=\"application/json\"], script[class^=\"article-info\"][type=\"application/json\"], script#dvz-config";
const idUrl = "https://www.bloomberg.com/article/api/story/id/";
const headers = {
	accept: "application/json",
	"cache-control": "no-cache",
	referer: "https://www.bloomberg.com"
};
const apiEndpoints = {
	articles: { url: "https://www.bloomberg.com/article/api/story/slug/" },
	features: { url: "https://www.bloomberg.com/article/api/story/slug/" },
	audio: {
		url: "https://www.bloomberg.com/news/audio/",
		sel: "script#__NEXT_DATA__"
	},
	videos: {
		url: "https://www.bloomberg.com/news/videos/",
		sel: "script"
	},
	newsletters: { url: "https://www.bloomberg.com/article/api/story/slug/" },
	"photo-essays": {
		url: "https://www.bloomberg.com/javelin/api/photo-essay_transporter/",
		sel: "script[type = \"application/json\"][data-component-props]"
	},
	"features/": {
		url: "https://www.bloomberg.com/features/",
		sel: idSel,
		prop: "id"
	}
};
const regex = [/\/(?<page>[\w-]*)\/(?<link>\d{4}-\d{2}-\d{2}\/.*)/, /(?<!news|politics)\/(?<page>features\/|graphics\/)(?<link>.*)/];
const capRegex = /<p>|<\/p>/g;
const emptyRegex = /<p\b[^>]*>(&nbsp;|\s)<\/p>/g;
const redirectGot = (url) => rofetch.raw(url, {
	headers,
	parseResponse: (responseText) => ({
		data: destr(responseText),
		body: responseText
	})
});
const parseArticle = (item) => cache_default.tryGet(item.link, async () => {
	const group = regex.map((r) => r.exec(item.link)).filter((e) => e && e.groups).map((a) => a && a.groups)[0];
	if (group) {
		const { page, link } = group;
		if (Object.hasOwn(apiEndpoints, page)) {
			const api = { ...apiEndpoints[page] };
			let res;
			try {
				const apiUrl = `${api.url}${link}`;
				res = await redirectGot(apiUrl);
			} catch (error) {
				const err = error;
				if (err.name && [
					"HTTPError",
					"RequestError",
					"FetchError"
				].includes(err.name)) try {
					res = await redirectGot(item.link);
				} catch {
					return {
						title: item.title,
						link: item.link,
						pubDate: item.pubDate,
						description: item.description
					};
				}
			}
			if (res.redirected && new URL(res.url).pathname === "/tosv2.html" || res.status === 404) return {
				title: item.title,
				link: item.link,
				pubDate: item.pubDate,
				description: item.description
			};
			switch (page) {
				case "audio": return parseAudioPage(res._data, api, item);
				case "videos": return parseVideoPage(res._data, api, item);
				case "photo-essays": return parsePhotoEssaysPage(res._data, api, item);
				case "features/": return parseReactRendererPage(res._data, api, item);
				default: return parseStoryJson(res._data.data, item);
			}
		}
	}
	return item;
});
const parseAudioPage = async (res, api, item) => {
	const audio_json = JSON.parse(load(res.data)(api.sel).html() ?? "").props.pageProps;
	const episode = audio_json.episode;
	return {
		title: episode.title || item.title,
		link: audio_json.pageInfo.canonicalUrl || item.link,
		guid: `bloomberg:${episode.id}`,
		description: (await processBody(episode.articleBody, audio_json)).replaceAll(emptyRegex, ""),
		pubDate: parseDate(episode.publishedAt) || item.pubDate,
		author: audio_json.hero.showTitle,
		media: {
			content: { url: episode.image },
			thumbnails: { url: episode.image }
		},
		enclosure_type: "audio/mpeg",
		enclosure_url: episode.url,
		itunes_item_image: episode.image || audio_json.pageInfo.image.url
	};
};
const parseVideoPage = async (res, api, item) => {
	const $ = load(res.data);
	const json = $(api.sel).filter((i, el) => $(el).text().includes("__PRELOADED_STATE__")).text().trim().match(/window\.__PRELOADED_STATE__ = (.*?);/)?.[1];
	const article_json = JSON.parse(json || "{}");
	const video_story = article_json.video?.videoStory ?? article_json.quicktakeVideo?.videoStory;
	if (video_story) {
		const desc = await processVideo(video_story.video.bmmrId, video_story.summary.html.replaceAll(emptyRegex, ""));
		return {
			title: video_story.headline.text || item.title,
			link: video_story.url || item.link,
			guid: `bloomberg:${video_story.id}`,
			description: renderVideoMedia(desc),
			pubDate: parseDate(video_story.publishedAt) || item.pubDate,
			media: {
				content: { url: video_story.video?.thumbnail.url || "" },
				thumbnails: { url: video_story.video?.thumbnail.url || "" }
			},
			category: desc.keywords ?? []
		};
	}
	return item;
};
const parsePhotoEssaysPage = async (res, api, item) => {
	const $ = load(res.data.html);
	const article_json = {};
	for (const e of $(api.sel).toArray()) {
		const raw = $(e).html();
		if (raw !== null) Object.assign(article_json, JSON.parse(raw));
	}
	return {
		title: article_json.headline || item.title,
		link: article_json.canonical || item.link,
		guid: `bloomberg:${article_json.id}`,
		description: (await processBody(article_json.body, article_json)).replaceAll(emptyRegex, ""),
		pubDate: item.pubDate,
		author: article_json.authors?.map((a) => a.name).join(", ") ?? []
	};
};
const parseReactRendererPage = async (res, api, item) => {
	const json = load(res.data)(api.sel).text().trim();
	const story_id = JSON.parse(json)[api.prop];
	try {
		const res = await redirectGot(`${idUrl}${story_id}`);
		return await parseStoryJson(res._data, item);
	} catch (error) {
		const err = error;
		if (err.name && [
			"HTTPError",
			"RequestError",
			"FetchError"
		].includes(err.name)) return {
			title: item.title,
			link: item.link,
			pubDate: item.pubDate
		};
	}
};
const parseStoryJson = async (story_json, item) => {
	const media_img = story_json.ledeImageUrl || Object.values(story_json.imageAttachments ?? {})[0]?.url;
	return {
		title: story_json.headline || item.title,
		link: story_json.url || item.link,
		guid: `bloomberg:${story_json.id}`,
		description: processHeadline(story_json) + await processLedeMedia(story_json) + await documentToHtmlString(story_json.body || ""),
		pubDate: parseDate(story_json.publishedAt) || item.pubDate,
		author: story_json.authors?.map((a) => a.name).join(", ") ?? [],
		category: story_json.mostRelevantTags ?? [],
		media: {
			content: { url: media_img },
			thumbnails: { url: media_img }
		}
	};
};
const processHeadline = (story_json) => {
	const dek = story_json.dek || story_json.summary || story_json.headline || "";
	const abs = story_json.abstract?.map((a) => `<li>${a}</li>`).join("");
	return abs ? dek + `<ul>${abs}</ul>` : dek;
};
const processLedeMedia = async (story_json) => {
	if (story_json.ledeMediaKind) {
		const kind = story_json.ledeMediaKind;
		return renderLedeMedia({
			kind: story_json.ledeMediaKind,
			caption: story_json.ledeCaption?.replaceAll(capRegex, "") ?? "",
			description: story_json.ledeDescription?.replaceAll(capRegex, "") ?? "",
			credit: story_json.ledeCredit?.replaceAll(capRegex, "") ?? "",
			src: story_json.ledeImageUrl,
			video: kind === "video" && await processVideo(story_json.ledeAttachment.bmmrId)
		});
	}
	if (story_json.lede) {
		const lede = story_json.lede;
		return renderImageFigure({
			src: lede.url,
			alt: lede.alt || lede.title,
			caption: lede.caption?.replaceAll(capRegex, "") ?? "",
			credit: lede.credit?.replaceAll(capRegex, "") ?? ""
		});
	}
	if (story_json.imageAttachments) {
		const attachment = Object.values(story_json.imageAttachments)[0];
		if (attachment) return renderImageFigure({
			src: attachment.baseUrl || attachment.url,
			alt: attachment.alt || attachment.title,
			caption: attachment.caption?.replaceAll(capRegex, "") ?? "",
			credit: attachment.credit?.replaceAll(capRegex, "") ?? ""
		});
		return "";
	}
	if (story_json.type === "Lede") {
		const props = story_json.props;
		return renderLedeMedia({
			kind: props.media,
			caption: props.caption?.replaceAll(capRegex, "") ?? "",
			description: props.dek?.replaceAll(capRegex, "") ?? "",
			credit: props.credit?.replaceAll(capRegex, "") ?? "",
			src: props.url
		});
	}
};
const processBody = async (body_html, story_json) => {
	const removeSel = [
		"meta",
		"*[class$=\"-footnotes\"]",
		"*[class$=\"for-you\"]",
		"*[class$=\"-newsletter\"]",
		"*[class$=\"page-ad\"]",
		"*[class$=\"-recirc\"]",
		"*[data-ad-placeholder=\"Advertisement\"]"
	];
	const $ = load(body_html);
	for (const sel of removeSel) $(sel).remove();
	$(".paywall").removeAttr("class");
	for await (const e of $("figure")) {
		const imageType = $(e).data("image-type");
		const type = $(e).data("type");
		let new_figure = "";
		if (imageType === "audio") {
			let audio = {};
			if (story_json.audios) {
				const attachment = story_json.audios.find((a) => a.id.toString() === $(e).data("id").toString());
				audio = {
					img: attachment.image?.url || $(e).find("img").attr("src"),
					src: attachment.url || $(e).find("audio source").attr("src"),
					caption: $(e).find("[class$=\"text\"]").html()?.trim() ?? "",
					credit: $(e).find("[class$=\"credit\"]").html()?.trim() ?? ""
				};
			}
			if (story_json.episode) {
				const episode = story_json.episode;
				audio = {
					src: episode.url,
					img: episode.image || story_json.pageInfo.image.url,
					caption: episode.description || ($(e).find("[class$=\"text\"]").html()?.trim() ?? ""),
					credit: (episode.credits.map((c) => c.name).join(", ") ?? []) || ($(e).find("[class$=\"credit\"]").html()?.trim() ?? "")
				};
			}
			new_figure = renderAudioMedia(audio);
		} else if (imageType === "video") {
			if (story_json.videoAttachments) {
				const attachment = story_json.videoAttachments[$(e).data("id")];
				new_figure = renderVideoMedia(await processVideo(attachment.bmmrId));
			}
		} else if (imageType === "photo" || imageType === "image" || type === "image") {
			let src, alt;
			if (story_json.imageAttachments) {
				const attachment = story_json.imageAttachments[$(e).data("id")];
				alt = attachment?.alt || $(e).find("img").attr("alt")?.trim();
				src = attachment?.baseUrl;
			} else {
				alt = $(e).find("img").attr("alt").trim();
				src = $(e).find("img").data("native-src");
			}
			const caption = $(e).find("[class$=\"text\"], .caption, .photo-essay__text").html()?.trim() ?? "";
			const credit = $(e).find("[class$=\"credit\"], .credit, .photo-essay__source").html()?.trim() ?? "";
			new_figure = renderImageFigure({
				src,
				alt,
				caption,
				credit
			});
		}
		$(new_figure).insertAfter(e);
		$(e).remove();
	}
	return $.html();
};
const processVideo = async (bmmrId, summary) => {
	const api = `https://www.bloomberg.com/multimedia/api/embed?id=${bmmrId}`;
	const res = await redirectGot(api);
	if (res.redirected && new URL(res.url).pathname === "/tosv2.html" || res.status === 404) return {
		stream: "",
		mp4: "",
		coverUrl: "",
		caption: summary
	};
	if (res._data.data) {
		const video_json = res._data.data;
		return {
			stream: video_json.streams ? video_json.streams[0]?.url : "",
			mp4: video_json.downloadURLs ? video_json.downloadURLs["600"] : "",
			coverUrl: video_json.thumbnail?.baseUrl ?? "",
			caption: video_json.description || video_json.title || summary
		};
	}
	return {
		stream: "",
		mp4: "",
		coverUrl: "",
		caption: summary
	};
};
const nodeRenderers = {
	paragraph: async (node, nextNode) => `<p>${await nextNode(node.content)}</p>`,
	text: (node) => {
		const { attributes: attr, value: val } = node;
		if (attr?.emphasis && attr.strong) return `<strong><em>${val}</em></strong>`;
		if (attr?.emphasis) return `<em>${val}</em>`;
		if (attr?.strong) return `<strong>${val}</strong>`;
		return val;
	},
	"inline-newsletter": async (node, nextNode) => `<div>${await nextNode(node.content)}</div>`,
	"inline-recirc": async (node, nextNode) => `<div>${await nextNode(node.content)}</div>`,
	heading: async (node, nextNode) => {
		const nodeData = node.data;
		if (nodeData.level === 2 || nodeData.level === 3) return `<h3>${await nextNode(node.content)}</h3>`;
	},
	link: async (node, nextNode) => {
		const dest = node.data.destination;
		const web = dest.web;
		const bbg = dest.bbg;
		const title = node.data.title;
		if (web) return `<a href="${web}" title="${title}" target="_blank">${await nextNode(node.content)}</a>`;
		if (bbg && bbg.startsWith("bbg://news/stories")) {
			const o = bbg.split("bbg://news/stories/").pop();
			return `<a href="${"https://www.bloomberg.com/news/terminal/".concat(o)}" title="${title}" target="_blank">${await nextNode(node.content)}</a>`;
		}
		return String(await nextNode(node.content));
	},
	entity: async (node, nextNode) => {
		const t = node.subType;
		const web = node.data.link.destination.web;
		if (t === "person") return nextNode(node.content);
		if (t === "story") {
			if (web) return `<a href="${web}" target="_blank">${await nextNode(node.content)}</a>`;
			const a = node.data.story.identifiers.suid;
			return `<a href="${"https://www.bloomberg.com/news/terminal/".concat(a)}" target="_blank">${await nextNode(node.content)}</a>`;
		}
		if (t === "security") {
			const s = node.data.security.identifiers.parsekey;
			if (s) {
				const c = s.split(" ");
				return `<a href="${[..."https://www.bloomberg.com/quote/".concat(c[0], ":"), c[1]]}" target="_blank">${await nextNode(node.content)}</a>`;
			}
		}
		return nextNode(node.content);
	},
	br: () => "<br/>",
	hr: () => "<br/>",
	ad: () => {},
	blockquote: async (node, nextNode) => `<blockquote>${await nextNode(node.content)}</blockquote>`,
	quote: async (node, nextNode) => `<blockquote>${await nextNode(node.content)}</blockquote>`,
	aside: async (node, nextNode) => `<aside>${await nextNode(node.content)}</aside>`,
	list: async (node, nextNode) => {
		const t = node.subType;
		if (t === "unordered") return `<ul>${await nextNode(node.content)}</ul>`;
		if (t === "ordered") return `<ol>${await nextNode(node.content)}</ol>`;
	},
	listItem: async (node, nextNode) => `<li>${await nextNode(node.content)}</li>`,
	media: async (node) => {
		const t = node.subType;
		if (t === "chart" && node.data.attachment) {
			if (node.data.attachment.creator === "TOASTER") {
				const c = node.data.chart;
				const e = {
					src: c && c.fallback || "",
					chart: node.data.attachment,
					id: c && c.id || "",
					alt: c && c.alt || ""
				};
				const w = e.chart;
				return renderChartMedia({ chart: {
					source: w.source,
					footnote: w.footnote,
					url: w.url,
					title: w.title,
					subtitle: w.subtitle,
					chartId: "toaster-chart-".concat(e.id),
					chartAlt: e.alt,
					fallback: e.src
				} });
			}
			return renderImageFigure({
				alt: node.data.attachment?.footnote || "",
				caption: node.data.attachment?.title + node.data.attachment.subtitle || "",
				credit: node.data.attachment?.source || "",
				src: node.data.chart?.fallback || ""
			});
		}
		if (t === "photo") {
			const h = node.data;
			let img = "";
			if (h.attachment) img = renderImageFigure({
				src: h.photo?.src,
				alt: h.photo?.alt,
				caption: h.photo?.caption,
				credit: h.photo?.credit
			});
			if (h.link && h.link.destination && h.link.destination.web) return `<a href="${h.link.destination.web}" target="_blank">${img}</a>`;
			return img;
		}
		if (t === "video") {
			const h = node.data;
			const id = h.attachment?.id;
			if (id) return renderVideoMedia(await processVideo(id, h.attachment?.title));
		}
		if (t === "audio" && node.data.attachment) {
			const B = node.data.attachment;
			const P = B.title;
			const D = B.url;
			const M = B.image;
			if (P && D) return renderAudioMedia({
				src: D,
				img: M.url,
				caption: P,
				credit: ""
			});
		}
		return "";
	},
	tabularData: async (node, nextNode) => `<table>${await nextNode(node.content)}</table>`,
	columns: (node) => {
		return `<tr>${node.data.definitions.map((e) => ({
			title: e.title,
			span: e.colSpan || 1,
			type: e.dataType
		})).map((e) => `<th colspan=${e.span}>${e.title}</th>`)}</tr>`;
	},
	row: async (node, nextNode) => `<tr>${await nextNode(node.content)}</tr>`,
	cell: async (node, nextNode) => {
		return `<td data-coltype=${{
			"news-rsf-table-number": "number",
			"news-rsf-table-string": "text"
		}[node.data.class] || "text"} colspan=${node.data.colspan}>${await nextNode(node.content)}</td>`;
	}
};
const nextNode = async (nodes) => {
	return await nodeListToHtmlString(nodes);
};
const nodeToHtmlString = async (node, obj) => {
	if (!node.type || !Object.hasOwn(nodeRenderers, node.type)) return `<node>${node.type}</node>`;
	return await nodeRenderers[node.type](node, nextNode, obj);
};
const nodeListToHtmlString = async (nodes) => {
	return (await Promise.all(nodes.map(async (node, index) => {
		return await nodeToHtmlString(node, {
			index,
			prev: nodes[index - 1]?.type,
			next: nodes[index + 1]?.type
		});
	}))).join("");
};
const documentToHtmlString = async (document) => {
	if (!document || !document.content) return "";
	return await nodeListToHtmlString(document.content);
};
//#endregion
export { rootUrl as n, parseArticle as t };