${cdata(article.title)}

stehen, aber nicht zum Artikeltext gehören\n const SKIP_ANCESTORS =\n \".picturebox, .picture-gallery, .c_toolbar_list, ul.social, ul.functions, .lb_content, .linkbox, .box_head, .teaser, .detail_author, .detail_timestamp, .pictures-index\";\n\n const nodes = root\n .find(\"p.detail_lead, div.picturebox, div.picture-gallery, script, h2, h3, p\")\n .toArray();\n\n const parts: string[] = [];\n\n for (const node of nodes) {\n const el = node as unknown as { tagName?: string };\n const $el = $(node as never);\n const cls = $el.attr(\"class\") ?? \"\";\n\n // Medien-Player: das JS-Snippet verrät die mediaInfo.json\n if (el.tagName === \"script\") {\n const code = $el.contents().text();\n const match = code.match(/ardplayer\\.Player\\([^)]*?\"([^\"]*mediaInfo\\.json[^\"]*)\"/);\n if (match) {\n const rendered = await renderMedia(absolute(match[1], pageUrl));\n if (rendered) parts.push(rendered);\n }\n continue;\n }\n\n if (/picture-gallery/.test(cls)) {\n const rendered = await renderGallery($, $el as never, pageUrl);\n if (rendered) parts.push(rendered);\n continue;\n }\n\n if (/picturebox/.test(cls)) {\n const img = $el.find(\"img\").first();\n const src = absolute(img.attr(\"src\") ?? \"\", pageUrl);\n if (src) {\n const caption = $el.find(\"p\").first().text().replace(/\\s+/g, \" \").trim();\n const { credit } = splitCaption(img.attr(\"alt\") ?? \"\");\n const alt = caption\n ? `${caption}${credit ? ` | Bild: ${credit}` : \"\"}`\n : (img.attr(\"alt\") ?? \"\");\n parts.push(figure(src, alt));\n }\n continue;\n }\n\n // Alles Übrige nur, wenn es nicht in einem Fremdbaustein steckt\n if ($el.closest(SKIP_ANCESTORS).length > 0) continue;\n\n if (el.tagName === \"h2\" || el.tagName === \"h3\") {\n const text = $el.text().replace(/\\s+/g, \" \").trim();\n // Überschriften, die nur ein Player-Snippet enthalten, überspringen\n if (!text || /ardplayer|function\\s*\\(/.test(text)) continue;\n parts.push(`

${escapeHtml(text)}

`);\n continue;\n }\n\n if (el.tagName === \"p\") {\n if (/detail_author|detail_timestamp|pictures-index|copyright/.test(cls)) continue;\n const text = $el.text().replace(/\\s+/g, \" \").trim();\n if (!text) continue;\n if (/detail_lead/.test(cls)) {\n parts.push(`

${escapeHtml(text)}

`);\n continue;\n }\n if ($el.closest(\".detail_paragraph\").length === 0) continue;\n parts.push(`

${escapeHtml(text)}

`);\n }\n }\n\n const html = parts.join(\"\\n\");\n return html.length > MAX_CONTENT_CHARS ? html.slice(0, MAX_CONTENT_CHARS) : html;\n}\n\n/** Sammelt die Artikel-Links der Startseite. */\nasync function discoverArticles(): Promise {\n const res = await fetch(OVERVIEW_URL, {\n headers: {\n \"User-Agent\": USER_AGENT,\n Accept: \"text/html,application/xhtml+xml\",\n \"Accept-Language\": \"de-DE,de;q=0.9\",\n \"Cache-Control\": \"no-cache\",\n Pragma: \"no-cache\",\n },\n });\n if (!res.ok) throw new Error(`Startseite nicht erreichbar (${res.status})`);\n\n const $ = cheerio.load(await res.text());\n const seen = new Set();\n const items: Article[] = [];\n\n $(\"a[href]\").each((_, el) => {\n const href = $(el).attr(\"href\");\n if (!href) return;\n let url: URL;\n try {\n url = new URL(href, SOURCE_URL);\n } catch {\n return;\n }\n // Nur ardalpha.de; br.de-Links sind Sendungs- und Podcastseiten\n if (url.hostname.replace(/^www\\./, \"\") !== \"ardalpha.de\") return;\n // Bildseiten und Übersichtsseiten ausschliessen\n if (url.pathname.includes(\"~\") || /\\/index\\.html$/i.test(url.pathname)) return;\n if (!ARTICLE_PATH_RE.test(url.pathname)) return;\n\n const link = `${SOURCE_URL}${url.pathname}`;\n if (seen.has(link)) return;\n seen.add(link);\n items.push({ title: \"\", link });\n });\n\n return items.slice(0, MAX_CANDIDATES);\n}\n\nasync function enrichArticle(article: Article): Promise

{\n const html = await fetchHtml(article.link);\n if (!html) return article;\n\n const $ = cheerio.load(html);\n const result: Article = {\n ...article,\n title: cleanTitle($('meta[property=\"og:title\"]').attr(\"content\") ?? \"\"),\n description: $('meta[property=\"og:description\"]').attr(\"content\")?.trim(),\n image: $('meta[property=\"og:image\"]').attr(\"content\")?.trim(),\n author: $('meta[name=\"author\"]').attr(\"content\")?.replace(/,\\s*$/, \"\").trim(),\n };\n\n $('script[type=\"application/ld+json\"]').each((_, el) => {\n try {\n const data = JSON.parse($(el).contents().text());\n const nodes = Array.isArray(data) ? data : [data];\n for (const g of nodes) {\n const types = Array.isArray(g?.[\"@type\"]) ? g[\"@type\"] : [g?.[\"@type\"]];\n if (!types.includes(\"NewsArticle\") && !types.includes(\"Article\")) continue;\n\n if (g.headline) result.title = cleanTitle(String(g.headline));\n if (g.description) result.description = String(g.description).trim();\n\n const image = Array.isArray(g.image) ? g.image[0] : g.image;\n const imageUrl = typeof image === \"string\" ? image : image?.url;\n if (imageUrl) result.image = absolute(String(imageUrl), article.link);\n\n const author = Array.isArray(g.author) ? g.author[0] : g.author;\n const authorName = typeof author === \"string\" ? author : author?.name;\n if (authorName) result.author = String(authorName).trim();\n\n const date = g.datePublished || g.dateModified;\n if (date && !isNaN(new Date(date).getTime())) result.pubDate = new Date(date);\n }\n } catch {\n // ungültiges JSON-LD ignorieren\n }\n });\n\n result.content = await buildContent($, article.link);\n return result;\n}\n\nexport default async function (req: Request): Promise {\n try {\n const candidates = await discoverArticles();\n const articles = (await runWithConcurrency(candidates, CONCURRENCY, enrichArticle))\n .filter((a) => a.title)\n .sort((a, b) => (b.pubDate?.getTime() ?? 0) - (a.pubDate?.getTime() ?? 0))\n .slice(0, MAX_ITEMS);\n\n const now = new Date();\n const itemsXml = articles.map((article) => {\n const date = article.pubDate ?? now;\n const summary = `

${\n article.image ? `

` : \"\"\n }

${article.description ?? \"\"}

`;\n return [\n \" \",\n ` ${cdata(article.title)}`,\n ` ${cdata(summary)}`,\n article.content ? ` ${cdata(article.content)}` : \"\",\n ` ${escapeXml(article.link)}`,\n ` ${escapeXml(article.link)}#${FEED_REVISION}`,\n article.author ? ` ${cdata(article.author)}` : \"\",\n ` ${date.toUTCString()}`,\n \" \",\n ].filter(Boolean).join(\"\\n\");\n }).join(\"\\n\");\n\n const xml = `\n\n \n ${cdata(FEED_TITLE)}\n ${cdata(FEED_DESCRIPTION)}\n ${SOURCE_URL}/\n Val.town RSS Scraper (cheerio)\n ${now.toUTCString()}\n \n de\n${itemsXml}\n \n`;\n\n return new Response(xml, {\n status: 200,\n headers: {\n \"Content-Type\": \"application/rss+xml; charset=utf-8\",\n \"Cache-Control\": \"public, max-age=600\",\n },\n });\n } catch (err) {\n return new Response(\n `Fehler beim Erzeugen des RSS-Feeds: ${err instanceof Error ? err.message : String(err)}`,\n { status: 500, headers: { \"Content-Type\": \"text/plain; charset=utf-8\" } },\n );\n }\n}\n","import * as cheerio from \"npm:cheerio@1.0.0\";\n\nconst SOURCE_URL = \"https://www.ardalpha.de\";\nconst OVERVIEW_URL = `${SOURCE_URL}/index.html`;\nconst FEED_TITLE = \"ARD alpha\";\nconst FEED_DESCRIPTION =\n \"ARD alpha macht euch fit für Schule, Ausbildung, Uni, Beruf und euren Alltag. Hier gibt es Lernangebote, Hintergrundwissen und Tipps, die euch weiterbringen.\";\nconst USER_AGENT =\n \"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36\";\n\nconst MAX_ITEMS = 25;\nconst MAX_CANDIDATES = 40;\nconst CONCURRENCY = 6;\nconst MAX_CONTENT_CHARS = 300_000;\n// Obergrenze pro Bildergalerie. Jedes Galeriebild kostet einen Seitenabruf,\n// weil die grosse Fassung nur auf der jeweiligen Bildseite steht.\nconst MAX_GALLERY_IMAGES = 24;\n\nconst FEED_REVISION = \"v1\";\n\n// Artikel-URLs von ARD alpha enden auf -.html\nconst ARTICLE_PATH_RE = /^\\/[a-z0-9/_-]+-\\d+\\.html$/i;\n\ninterface Article {\n title: string;\n link: string;\n description?: string;\n image?: string;\n author?: string;\n content?: string;\n pubDate?: Date;\n}\n\nfunction cdata(value: string): string {\n return `/g, \"]]]]>\")}]]>`;\n}\n\nfunction escapeXml(value: string): string {\n return value.replace(/&/g, \"&\").replace(//g, \">\");\n}\n\nfunction escapeHtml(value: string): string {\n return value\n .replace(/&/g, \"&\")\n .replace(//g, \">\")\n .replace(/\"/g, \""\");\n}\n\nfunction absolute(url: string, base = SOURCE_URL): string {\n try {\n if (url.startsWith(\"//\")) return `https:${url}`;\n return new URL(url, base).href;\n } catch {\n return \"\";\n }\n}\n\n/** BR hängt an jeden Seitentitel \" | BR.de\" an – im Feed stört das. */\nfunction cleanTitle(value: string): string {\n return value.replace(/\\s*\\|\\s*BR\\.de\\s*$/i, \"\").trim();\n}\n\n/**\n * Bildunterschriften stecken bei BR im alt-Attribut, inklusive Quelle nach\n * einem senkrechten Strich: \"Beschreibung | Bild: colourbox.com\".\n */\nfunction splitCaption(alt: string): { text: string; credit: string } {\n const parts = alt.split(/\\s*\\|\\s*Bild:\\s*/i);\n return { text: (parts[0] ?? \"\").trim(), credit: (parts[1] ?? \"\").trim() };\n}\n\nfunction figure(src: string, alt: string, extra = \"\"): string {\n const { text, credit } = splitCaption(alt);\n const caption = [\n extra ? escapeHtml(extra) : \"\",\n text ? escapeHtml(text) : \"\",\n credit ? `bild: ${escapeHtml(credit.toLowerCase())}` : \"\",\n ].filter(Boolean).join(\" – \");\n return `

`;\n}\n\nasync function fetchHtml(url: string, timeoutMs = 12000): Promise {\n const controller = new AbortController();\n const timeout = setTimeout(() => controller.abort(), timeoutMs);\n try {\n const res = await fetch(url, {\n headers: {\n \"User-Agent\": USER_AGENT,\n Accept: \"text/html,application/xhtml+xml\",\n \"Accept-Language\": \"de-DE,de;q=0.9\",\n },\n signal: controller.signal,\n });\n if (!res.ok) return null;\n return await res.text();\n } catch {\n return null;\n } finally {\n clearTimeout(timeout);\n }\n}\n\nasync function runWithConcurrency(\n items: T[],\n limit: number,\n worker: (item: T) => Promise,\n): Promise {\n const results: R[] = new Array(items.length);\n let index = 0;\n\n async function next(): Promise {\n const current = index++;\n if (current >= items.length) return;\n results[current] = await worker(items[current]);\n return next();\n }\n\n await Promise.all(Array.from({ length: Math.min(limit, items.length) }, () => next()));\n return results;\n}\n\n/**\n * Löst einen ARD-Player auf. Audio und Video werden per JS eingebunden; die\n * eigentliche Mediendatei steht in der zugehörigen mediaInfo.json.\n */\nasync function renderMedia(mediaInfoUrl: string): Promise {\n const controller = new AbortController();\n const timeout = setTimeout(() => controller.abort(), 10000);\n try {\n const res = await fetch(mediaInfoUrl, {\n headers: { \"User-Agent\": USER_AGENT, Accept: \"application/json\" },\n signal: controller.signal,\n });\n if (!res.ok) return \"\";\n const info = await res.json();\n\n const stream = info?.streams?.[0];\n const media: Record[] = Array.isArray(stream?.media) ? stream.media : [];\n if (media.length === 0) return \"\";\n\n // Beste Qualität anhand der Bitrate im Label wählen\n let best = media[0];\n let bestRate = -1;\n for (const m of media) {\n const rate = Number(String(m.forcedLabel ?? \"\").match(/(\\d+)\\s*Kbit/i)?.[1] ?? 0);\n if (rate > bestRate) {\n bestRate = rate;\n best = m;\n }\n }\n\n const src = absolute(String(best.url ?? \"\"));\n if (!src) return \"\";\n\n const isAudio = stream?.isAudioOnly === true;\n const comment = String(info?.comment ?? \"\").trim();\n const player = isAudio\n ? `

`\n : `

`;\n\n return [\n player,\n comment ? `

${escapeHtml(comment)}

` : \"\",\n `

${isAudio ? \"Audio\" : \"Video\"} direkt öffnen

`,\n ].filter(Boolean).join(\"\\n\");\n } catch {\n return \"\";\n } finally {\n clearTimeout(timeout);\n }\n}\n\n/**\n * Baut eine Bildergalerie auf. Im Artikel-HTML stehen nur kleine Vorschauen\n * und ein Teil der Bilder. Die grosse Fassung und die restlichen Bilder\n * ergeben sich erst aus den einzelnen Bildseiten, die hier nachgeladen und\n * dabei nach weiteren Bildern durchsucht werden.\n */\nasync function renderGallery(\n $: cheerio.CheerioAPI,\n gallery: cheerio.Cheerio,\n pageUrl: string,\n): Promise {\n const known = new Map();\n\n function collect(scope$: cheerio.CheerioAPI, scope: cheerio.Cheerio) {\n scope.find('a[href*=\"~_image-\"]').each((_, el) => {\n const href = scope$(el).attr(\"href\") ?? \"\";\n const idx = Number(href.match(/~_image-(\\d+)_/)?.[1] ?? 0);\n if (idx > 0 && !known.has(idx)) known.set(idx, absolute(href, pageUrl));\n });\n }\n\n collect($, gallery);\n if (known.size === 0) return \"\";\n\n const done = new Map();\n\n while (done.size < MAX_GALLERY_IMAGES) {\n const pending = [...known.entries()]\n .filter(([idx]) => !done.has(idx))\n .sort((a, b) => a[0] - b[0])\n .slice(0, CONCURRENCY);\n if (pending.length === 0) break;\n\n await runWithConcurrency(pending, CONCURRENCY, async ([idx, url]) => {\n const html = await fetchHtml(url, 8000);\n if (!html) {\n done.set(idx, { src: \"\", alt: \"\" });\n return;\n }\n const page$ = cheerio.load(html);\n const src = page$('meta[property=\"og:image\"]').attr(\"content\")?.trim() ?? \"\";\n const alt = page$(\".picture-gallery li.active img\").first().attr(\"alt\") ??\n page$('meta[property=\"og:description\"]').attr(\"content\") ?? \"\";\n done.set(idx, { src, alt });\n collect(page$, page$(\".picture-gallery\") as never);\n });\n }\n\n const total = Math.max(known.size, done.size);\n return [...done.entries()]\n .sort((a, b) => a[0] - b[0])\n .filter(([, v]) => v.src)\n .map(([idx, v]) => figure(v.src, v.alt, `${idx} / ${total}`))\n .join(\"\\n\");\n}\n\n/** Baut den Volltext: Lead, Zwischentitel, Absätze, Bilder, Galerien, Medien. */\nasync function buildContent(\n $: cheerio.CheerioAPI,\n pageUrl: string,\n leadImage?: string,\n): Promise {\n const root = $(\"article\").first();\n if (root.length === 0) return \"\";\n\n // Bausteine, die im

${escapeHtml(text)}

${escapeHtml(text)}

`);\n continue;\n }\n if ($el.closest(\".detail_paragraph\").length === 0) continue;\n parts.push(`

${escapeHtml(text)}

`);\n }\n }\n\n // Aufmacherbild ergänzen, wenn der Artikelkörper gar kein Bild enthält\n if (leadImage && !parts.some((p) => p.startsWith(\"

Diff · v14