rsshub
Version:
Make RSS Great Again!
60 lines (59 loc) • 2.43 kB
JavaScript
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 { load } from "cheerio";
//#region lib/routes/apnews/utils.ts
function removeDuplicateByKey(items, key) {
return new Map(items.map((x) => [x[key], x])).values().toArray();
}
function fetchArticle(item) {
return cache_default.tryGet(item.link, async () => {
const $ = load(await rofetch(item.link));
if ($("#link-ld-json").length === 0) {
const gtmRaw = $("meta[name=\"gtm-dataLayer\"]").attr("content");
if (gtmRaw) {
const gtmParsed = JSON.parse(gtmRaw);
return {
title: gtmParsed.headline,
pubDate: parseDate(gtmParsed.publication_date),
description: $("div.RichTextStoryBody").html() || $(":is(.VideoLead, .VideoPage-pageSubHeading)").html(),
category: gtmParsed.tag_array.split(","),
guid: $("meta[name='brightspot.contentId']").attr("content"),
author: gtmParsed.author,
...item
};
}
return item;
}
const rawLdjson = JSON.parse($("#link-ld-json").text());
let ldjson;
if (rawLdjson["@type"] === "NewsArticle" || Array.isArray(rawLdjson) && rawLdjson.some((e) => e["@type"] === "NewsArticle")) {
ldjson = Array.isArray(rawLdjson) ? rawLdjson.find((e) => e["@type"] === "NewsArticle") : rawLdjson;
$("div.Enhancement").remove();
const section = $("meta[property='article:section']").attr("content");
return {
...item,
title: ldjson.headline,
pubDate: parseDate(ldjson.datePublished),
updated: parseDate(ldjson.dateModified),
description: $("div.RichTextStoryBody").html() || $(":is(.VideoLead, .VideoPage-pageSubHeading)").html(),
category: [...section ? [section] : [], ...ldjson.keywords ?? []],
guid: $("meta[name='brightspot.contentId']").attr("content"),
author: ldjson.author
};
}
ldjson = rawLdjson;
const url = new URL(item.link);
const description = url.hash ? $(url.hash).parent().find(".LiveBlogPost-body").html() : ldjson.description;
const pubDate = url.hash ? parseDate(Number($(url.hash).parent().attr("data-posted-date-timestamp"))) : parseDate(ldjson.coverageStartTime);
return {
...item,
category: ldjson.keywords,
pubDate,
description,
guid: $("meta[name='brightspot.contentId']").attr("content")
};
});
}
//#endregion
export { removeDuplicateByKey as n, fetchArticle as t };