${cdata(article.title)}

{\n const html = await fetchHtml(link, cookie);\n if (!html) return null;\n\n const $ = cheerio.load(html);\n const article: Article = { title: \"\", link };\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[\"@graph\"] ? data[\"@graph\"] : [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) article.title = String(g.headline).trim();\n const author = Array.isArray(g.author) ? g.author[0] : g.author;\n const authorName = typeof author === \"string\" ? author : author?.name;\n if (authorName) article.author = String(authorName).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) article.image = absolute(String(imageUrl), link);\n\n const date = g.datePublished || g.dateModified;\n if (date && !isNaN(new Date(date).getTime())) article.pubDate = new Date(date);\n }\n } catch {\n // ungültiges JSON-LD ignorieren\n }\n });\n\n if (!article.title) article.title = ($('meta[property=\"og:title\"]').attr(\"content\") ?? \"\").trim();\n if (!article.image) {\n article.image = absolute($('meta[property=\"og:image\"]').attr(\"content\") ?? \"\", link);\n }\n if (!article.title || !article.pubDate) return null;\n\n article.description = $(\"div.article-lead\").first().text().replace(/\\s+/g, \" \").trim() ||\n $('meta[property=\"og:description\"]').attr(\"content\")?.trim();\n article.content = buildContent($, link);\n\n // Viele Artikel haben kein og:image – dann das erste Bild aus dem Text nehmen\n if (!article.image && article.content) {\n article.image = (article.content.match(/]+src=\"([^\"]+)\"/) ?? [])[1] ?? \"\";\n }\n\n return article;\n}\n\n// ---------------------------------------------------------------------------\n// Feed\n// ---------------------------------------------------------------------------\n\nexport default async function (req: Request): Promise {\n /**\n * Zugriffsschutz: Der Feed enthält Abo-Inhalte und darf deshalb nicht\n * offen im Netz stehen. Nur Abrufe mit dem richtigen Schlüssel werden\n * beantwortet.\n */\n const feedKey = Deno.env.get(\"FEED_KEY\");\n const providedKey = new URL(req.url).searchParams.get(\"key\") ??\n req.headers.get(\"x-feed-key\") ?? \"\";\n if (!feedKey) {\n return new Response(\"FEED_KEY ist nicht hinterlegt.\", { status: 500 });\n }\n if (providedKey !== feedKey) {\n return new Response(\"Kein Zugriff.\", { status: 401 });\n }\n\n try {\n const cookie = await login();\n const candidates = await discoverArticles(cookie);\n const articles = (await runWithConcurrency(candidates, CONCURRENCY, (link) => enrichArticle(link, cookie)))\n .filter((a): a is Article => a !== null)\n .sort((a, b) => (b.pubDate?.getTime() ?? 0) - (a.pubDate?.getTime() ?? 0))\n .slice(0, MAX_ITEMS);\n\n const lockedOut = articles.filter((a) => PAYWALL_MARKER.test(a.content ?? \"\")).length;\n const now = new Date();\n\n const itemsXml = articles.map((article) => {\n const date = article.pubDate ?? now;\n // Eigenes Bild für den Text, Logo nur als Vorschaubild in der Liste\n const thumbnail = article.image || FALLBACK_IMAGE;\n const summary = `

${\n article.image ? `

` : \"\"\n }

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

`;\n\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 ` `,\n ` `,\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 ${now.toUTCString()}\n de-CH\n${itemsXml}\n \n`;\n\n return new Response(xml, {\n status: 200,\n headers: {\n \"Content-Type\": \"application/rss+xml; charset=utf-8\",\n // Abo-Inhalte nicht in fremden Zwischenspeichern ablegen\n \"Cache-Control\": \"private, max-age=300\",\n \"X-Robots-Tag\": \"noindex, nofollow\",\n \"X-Ktipp-Login\": lockedOut === 0 ? \"ok\" : `paywall-in-${lockedOut}-artikeln`,\n },\n });\n } catch (err) {\n return new Response(\n `Fehler beim Erzeugen des RSS-Feeds: ${err instanceof Error ? err.message : String(err)}`,\n { status: 500, headers: { \"Content-Type\": \"text/plain; charset=utf-8\" } },\n );\n }\n}\n","import * as cheerio from \"npm:cheerio@1.0.0\";\n\nconst SOURCE_URL = \"https://www.saldo.ch\";\nconst FEED_TITLE = \"Saldo\";\nconst FEED_DESCRIPTION =\n \"Tests, Beratung und Konsumenten-News von Saldo – Volltext für Abonnenten.\";\nconst USER_AGENT =\n \"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36\";\n\nconst SECTION_PATHS = [\"/\", \"/tests/produktetests\", \"/tests/testsieger\", \"/beratung\", \"/service\"];\n\n/**\n * Ersatzbild für Artikel ohne eigenes Bild – sonst bleibt in der Listen-\n * ansicht ein leerer Platz. Es wird nur als Vorschaubild gesetzt, nicht in\n * den Artikeltext eingefügt.\n */\nconst FALLBACK_IMAGE = `${SOURCE_URL}/fileadmin/templates/img/logo_Ktipp.png`;\n\nconst MAX_ITEMS = 30;\nconst MAX_CANDIDATES = 45;\nconst CONCURRENCY = 4;\nconst MAX_CONTENT_CHARS = 300_000;\n\nconst FEED_REVISION = \"v1\";\n\nconst ARTICLE_PATH_RE = /^\\/artikel\\/artikeldetail\\/[a-z0-9-]+\\/?$/i;\n\n/** Text, an dem sich erkennen lässt, dass die Anmeldung nicht gegriffen hat. */\nconst PAYWALL_MARKER = /Melden Sie sich bitte an oder wählen Sie eines unserer/i;\n\n/** Bausteine, die nicht zum Artikel gehören. */\nconst JUNK_SELECTOR = [\n \"script\", \"style\", \"noscript\", \"form\", \"button\", \"svg\",\n \".login-box\", \".login-box-cover\",\n // Buchwerbung am Artikelende (\"Buch bestellen\")\n \".related-interests\", \".interests-slider\", \".interest-item\", \".linkbox-shop\",\n // Zählpixel von ProLitteris – Reader nehmen ihn sonst als Vorschaubild\n 'img[src*=\"prolitteris\"]',\n '[class*=\"social\"]', '[class*=\"share\"]', '[class*=\"teilen\"]',\n '[class*=\"werbung\"]', '[class*=\"advert\"]', '[class*=\"banner\"]',\n '[class*=\"newsletter\"]', '[class*=\"shop\"]', '[class*=\"buchtipp\"]',\n].join(\", \");\n\ninterface Article {\n title: string;\n link: string;\n description?: string;\n image?: string;\n author?: string;\n content?: string;\n pubDate?: Date;\n}\n\nfunction cdata(value: string): string {\n return `/g, \"]]]]>\")}]]>`;\n}\n\nfunction escapeXml(value: string): string {\n return value.replace(/&/g, \"&\").replace(//g, \">\");\n}\n\nfunction absolute(url: string, base = SOURCE_URL): string {\n if (!url || !url.trim() || url.startsWith(\"data:\")) return \"\";\n try {\n if (url.startsWith(\"//\")) return `https:${url}`;\n return new URL(url, base).href;\n } catch {\n return \"\";\n }\n}\n\nfunction imageMime(url: string): string {\n const path = url.split(\"?\")[0].toLowerCase();\n if (path.endsWith(\".png\")) return \"image/png\";\n if (path.endsWith(\".webp\")) return \"image/webp\";\n if (path.endsWith(\".gif\")) return \"image/gif\";\n return \"image/jpeg\";\n}\n\n// ---------------------------------------------------------------------------\n// Anmeldung\n// ---------------------------------------------------------------------------\n\n/**\n * K-Tipp läuft auf TYPO3 und setzt nach der Anmeldung ein fe_typo_user-Cookie.\n * Das Cookie wird zwischengespeichert, damit nicht jeder Feed-Abruf eine neue\n * Anmeldung auslöst.\n */\nlet cachedCookie = \"\";\nlet cachedAt = 0;\nconst COOKIE_TTL_MS = 12 * 3600 * 1000;\n\nfunction collectCookies(jar: Map, res: Response): void {\n const headers = res.headers as unknown as { getSetCookie?: () => string[] };\n const list = headers.getSetCookie?.() ?? [];\n const single = res.headers.get(\"set-cookie\");\n for (const raw of list.length > 0 ? list : single ? [single] : []) {\n const pair = String(raw).split(\";\")[0] ?? \"\";\n const idx = pair.indexOf(\"=\");\n if (idx > 0) jar.set(pair.slice(0, idx).trim(), pair.slice(idx + 1).trim());\n }\n}\n\nasync function login(): Promise {\n const user = Deno.env.get(\"KTIPP_USER\");\n const password = Deno.env.get(\"KTIPP_PASSWORD\");\n if (!user || !password) {\n throw new Error(\"KTIPP_USER und KTIPP_PASSWORD sind nicht hinterlegt.\");\n }\n\n if (cachedCookie && Date.now() - cachedAt < COOKIE_TTL_MS) return cachedCookie;\n\n const jar = new Map();\n const cookieHeader = () => [...jar].map(([k, v]) => `${k}=${v}`).join(\"; \");\n\n const page = await fetch(`${SOURCE_URL}/login`, {\n headers: { \"User-Agent\": USER_AGENT, Accept: \"text/html\", \"Accept-Language\": \"de-CH,de;q=0.9\" },\n });\n collectCookies(jar, page);\n\n const $ = cheerio.load(await page.text());\n const form = $(\"#loginForm\");\n if (form.length === 0) throw new Error(\"Anmeldeformular nicht gefunden.\");\n\n const action = absolute(form.attr(\"action\") ?? \"/login\");\n const body = new URLSearchParams();\n\n // Alle versteckten Felder übernehmen – TYPO3 prüft sie beim Absenden\n form.find(\"input\").each((_, el) => {\n const attrs = (el as unknown as { attribs: Record }).attribs ?? {};\n if (!attrs.name || attrs.type === \"text\" || attrs.type === \"password\") return;\n body.set(attrs.name, attrs.value ?? \"\");\n });\n body.set(\"tx_updkonsuminfo_konsuminfofeuserlogin[userid]\", user);\n body.set(\"tx_updkonsuminfo_konsuminfofeuserlogin[password]\", password);\n body.set(\"logIn\", \"senden\");\n\n let res = await fetch(action, {\n method: \"POST\",\n headers: {\n \"User-Agent\": USER_AGENT,\n \"Content-Type\": \"application/x-www-form-urlencoded\",\n Accept: \"text/html\",\n \"Accept-Language\": \"de-CH,de;q=0.9\",\n Referer: `${SOURCE_URL}/login`,\n Origin: SOURCE_URL,\n Cookie: cookieHeader(),\n },\n body,\n redirect: \"manual\",\n });\n collectCookies(jar, res);\n\n // Weiterleitungen von Hand verfolgen, um alle Cookies mitzunehmen\n for (let hop = 0; hop < 5 && [301, 302, 303, 307, 308].includes(res.status); hop++) {\n const location = absolute(res.headers.get(\"location\") ?? \"/\");\n res = await fetch(location, {\n headers: { \"User-Agent\": USER_AGENT, Accept: \"text/html\", Cookie: cookieHeader() },\n redirect: \"manual\",\n });\n collectCookies(jar, res);\n }\n\n cachedCookie = cookieHeader();\n cachedAt = Date.now();\n return cachedCookie;\n}\n\nasync function fetchHtml(url: string, cookie: 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-CH,de;q=0.9\",\n ...(cookie ? { Cookie: cookie } : {}),\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// Inhalt\n// ---------------------------------------------------------------------------\n\nfunction buildContent($: cheerio.CheerioAPI, pageUrl: string): string {\n const body = $(\"section.main-content_article\").first();\n if (body.length === 0) return \"\";\n\n const fragment = cheerio.load(`

${body.html() ?? \"\"}

`, null, false);\n const root = fragment(\"div\").first();\n\n fragment(JUNK_SELECTOR).remove();\n fragment(\"h1\").remove();\n fragment(\"header.article-header\").remove();\n\n /**\n * Kommentarbereich abschneiden. Er hat keine eigene Klasse, beginnt aber\n * zuverlässig mit dieser Überschrift – ab dort wird alles entfernt,\n * einschliesslich der nachfolgenden Elemente auf jeder Ebene.\n */\n const commentHeading = fragment(\"h2, h3\").filter((_, node) =>\n /^Kommentare zu diesem Artikel/i.test(fragment(node as never).text().replace(/\\s+/g, \" \").trim())\n ).first();\n\n if (commentHeading.length > 0) {\n // Erst alles nach der Überschrift auf ihrer Ebene, dann die Überschrift\n // selbst – und danach auf jeder höheren Ebene nur die nachfolgenden\n // Elemente. Die Container selbst bleiben stehen, sonst verschwindet\n // auch der Artikeltext, der vor ihnen steht.\n commentHeading.nextAll().remove();\n let node = commentHeading.parent();\n commentHeading.remove();\n for (let depth = 0; depth < 12 && node.length > 0 && !node.is(root); depth++) {\n node.nextAll().remove();\n node = node.parent() as never;\n }\n }\n\n fragment(\"*\").each((_, node) => {\n const el = node as unknown as { tagName?: string; attribs?: Record };\n const attrs = el.attribs ?? {};\n const kept: Record = {};\n\n switch (el.tagName) {\n case \"img\": {\n const src = absolute(attrs[\"data-src\"] || attrs.src || \"\", pageUrl);\n if (src) kept.src = src;\n if (attrs.alt) kept.alt = attrs.alt;\n break;\n }\n case \"a\": {\n const href = absolute(attrs.href ?? \"\", pageUrl);\n if (href) kept.href = href;\n break;\n }\n case \"iframe\": {\n const src = absolute(attrs[\"data-src\"] || attrs.src || \"\", pageUrl);\n if (src && /youtube|youtu\\.be|vimeo/i.test(src)) {\n kept.src = src;\n kept.allowfullscreen = \"\";\n kept.frameborder = \"0\";\n }\n break;\n }\n default:\n break;\n }\n\n el.attribs = kept;\n });\n\n fragment(\"img:not([src])\").remove();\n fragment(\"iframe:not([src])\").remove();\n\n for (let pass = 0; pass < 3; pass++) {\n fragment(\"div, p, span, section, li, ul\").each((_, node) => {\n const $el = fragment(node as never);\n if ($el.find(\"img, iframe, table\").length > 0) return;\n if ($el.text().replace(/\\s+/g, \"\").length === 0) $el.remove();\n });\n }\n\n const html = (root.html() ?? \"\").trim();\n return html.length > MAX_CONTENT_CHARS ? html.slice(0, MAX_CONTENT_CHARS) : html;\n}\n\nasync function discoverArticles(cookie: string): Promise {\n const seen = new Set();\n\n const pages = await runWithConcurrency(\n SECTION_PATHS,\n 3,\n (path) => fetchHtml(`${SOURCE_URL}${path}`, cookie, 15000),\n );\n\n for (const html of pages) {\n if (!html) continue;\n const $ = cheerio.load(html);\n $('a[href*=\"/artikel/artikeldetail/\"]').each((_, el) => {\n const href = $(el).attr(\"href\");\n if (!href) return;\n try {\n const url = new URL(href, SOURCE_URL);\n if (url.hostname.replace(/^www\\./, \"\") !== \"ktipp.ch\") return;\n if (!ARTICLE_PATH_RE.test(url.pathname)) return;\n seen.add(`${SOURCE_URL}${url.pathname}`);\n } catch {\n // ungültige URL ignorieren\n }\n });\n }\n\n return [...seen].slice(0, MAX_CANDIDATES);\n}\n\nasync function enrichArticle(link: string, cookie: string): Promise

${\n article.image ? `

` : \"\"\n }

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

`;\n\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 ` `,\n ` `,\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 ${now.toUTCString()}\n de-CH\n${itemsXml}\n \n`;\n\n return new Response(xml, {\n status: 200,\n headers: {\n \"Content-Type\": \"application/rss+xml; charset=utf-8\",\n // Abo-Inhalte nicht in fremden Zwischenspeichern ablegen\n \"Cache-Control\": \"private, max-age=300\",\n \"X-Robots-Tag\": \"noindex, nofollow\",\n \"X-Ktipp-Login\": lockedOut === 0 ? \"ok\" : `paywall-in-${lockedOut}-artikeln`,\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","createdAt",["D",1789386040254],"branchVersion","branchId","e9c7f8fc-29f7-4787-93b4-95bf5ae1f56b","id","2b9023ec-da21-48ce-98fe-9f4d4b496c67","ktipp",["D",1787814821425],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","","imageUrl","dbf2ae7a-e596-4450-9e80-f48182352e08","pinnedFileId","c849506e-b02c-11f1-a148-1607ee4eb77e","pinPreview",true,"pinMigrated","screenshotVersion","likeCount",0,"orgId","01a03d75-a2dc-7129-bcfe-464f7359f782","tursoName","tursoHostname","tursoMeasuredAt","pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_97":96,"_105":101},"org",{"_97":96,"_98":99,"_100":101,"_18":55,"_102":103,"_104":25},"branch",{"_13":86,"_32":87,"_43":42,"_88":44,"_89":-5,"_90":-5,"_91":-5,"_92":-5,"_93":-5,"_94":45,"_95":96},"isOwned",61,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","mikefrischknecht","handle","displayName","Mike Frischknecht","avatar","https://lh3.googleusercontent.com/a/ACg8ocJYhZ6TlQC9Px9BhWVC2KxGG5zXrx2pRqdDuxsgIfio5QgUBBEd=s96-c","tier","free","canRestrictVals","profileImageUrl","branchName","adoptOrgId",{"_43":44,"_32":45,"_38":188,"_47":-5,"_48":25,"_49":50,"_51":50,"_52":53,"_54":-5,"_18":55,"_56":57,"_58":59,"_60":-5,"_61":-5,"_62":63,"_64":65,"_66":-5,"_67":-5,"_68":-5,"_69":-5,"_70":71,"_72":71,"_73":74,"_75":74,"_76":74,"_77":71,"_78":71,"_79":189,"_81":190,"_83":191,"_85":25},"tree",[171,172,173,174],"files",[171,172,173,174],"counts",{"_160":63,"_161":135,"_162":63,"_163":164,"_165":166},"contents",[127,128,129,130],"upstreamProject","upstreamBranch","emptyEnvVars","flags",[],"hasHttpFiles","grants","orgs","customDomains",{},{"_43":152,"_32":153,"_30":154,"_132":-5,"_41":42,"_40":155,"_133":25,"_38":156,"_13":63,"_33":157,"_35":158,"_24":25,"_136":-5,"_27":159},{"_43":144,"_32":145,"_30":146,"_132":-5,"_41":42,"_40":147,"_133":59,"_38":148,"_13":149,"_33":34,"_35":150,"_24":25,"_136":-5,"_27":151},{"_43":138,"_32":139,"_30":57,"_132":-5,"_41":42,"_40":140,"_133":59,"_38":141,"_13":142,"_33":34,"_35":36,"_24":25,"_136":-5,"_27":143},{"_43":131,"_32":28,"_30":31,"_132":-5,"_41":42,"_40":14,"_133":25,"_38":134,"_13":135,"_33":34,"_35":37,"_24":25,"_136":-5,"_27":137},"6d7887f0-f342-4664-b16c-e639ba980735","parentId","latest",["D",1789386040254],1,"label",[28],"5dc9db23-342c-477a-b82c-ce084d664592","ktipp.ts",44,["D",1789385217516],6,[139],"17a77f47-a630-4a05-9ab6-6d4ae3161ad4","gesundheitstipp.ts","01a09fb0-b90d-73cf-be74-e1768b4819ee",52,["D",1789385573639],4,"import * as cheerio from \"npm:cheerio@1.0.0\";\n\nconst SOURCE_URL = \"https://www.gesundheitstipp.ch\";\nconst FEED_TITLE = \"Gesundheitstipp\";\nconst FEED_DESCRIPTION =\n \"Tests, Beratung und Gesundheits-News des Gesundheitstipp – Volltext für Abonnenten.\";\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 SECTION_PATHS = [\"/\", \"/tests\", \"/beratung\", \"/service\"];\n\n/**\n * Ersatzbild für Artikel ohne eigenes Bild – sonst bleibt in der Listen-\n * ansicht ein leerer Platz. Es wird nur als Vorschaubild gesetzt, nicht in\n * den Artikeltext eingefügt.\n */\nconst FALLBACK_IMAGE = `${SOURCE_URL}/fileadmin/templates/img/logo_gesundheitstipp.svg`;\n\nconst MAX_ITEMS = 30;\nconst MAX_CANDIDATES = 45;\nconst CONCURRENCY = 4;\nconst MAX_CONTENT_CHARS = 300_000;\n\nconst FEED_REVISION = \"v1\";\n\nconst ARTICLE_PATH_RE = /^\\/artikel\\/artikeldetail\\/[a-z0-9-]+\\/?$/i;\n\n/** Text, an dem sich erkennen lässt, dass die Anmeldung nicht gegriffen hat. */\nconst PAYWALL_MARKER = /Melden Sie sich bitte an oder wählen Sie eines unserer/i;\n\n/** Bausteine, die nicht zum Artikel gehören. */\nconst JUNK_SELECTOR = [\n \"script\", \"style\", \"noscript\", \"form\", \"button\", \"svg\",\n \".login-box\", \".login-box-cover\",\n // Buchwerbung am Artikelende (\"Buch bestellen\")\n \".related-interests\", \".interests-slider\", \".interest-item\", \".linkbox-shop\",\n // Zählpixel von ProLitteris – Reader nehmen ihn sonst als Vorschaubild\n 'img[src*=\"prolitteris\"]',\n '[class*=\"social\"]', '[class*=\"share\"]', '[class*=\"teilen\"]',\n '[class*=\"werbung\"]', '[class*=\"advert\"]', '[class*=\"banner\"]',\n '[class*=\"newsletter\"]', '[class*=\"shop\"]', '[class*=\"buchtipp\"]',\n].join(\", \");\n\ninterface Article {\n title: string;\n link: string;\n description?: string;\n image?: string;\n author?: string;\n content?: string;\n pubDate?: Date;\n}\n\nfunction cdata(value: string): string {\n return `/g, \"]]]]>\")}]]>`;\n}\n\nfunction escapeXml(value: string): string {\n return value.replace(/&/g, \"&\").replace(//g, \">\");\n}\n\nfunction absolute(url: string, base = SOURCE_URL): string {\n if (!url || !url.trim() || url.startsWith(\"data:\")) return \"\";\n try {\n if (url.startsWith(\"//\")) return `https:${url}`;\n return new URL(url, base).href;\n } catch {\n return \"\";\n }\n}\n\nfunction imageMime(url: string): string {\n const path = url.split(\"?\")[0].toLowerCase();\n if (path.endsWith(\".png\")) return \"image/png\";\n if (path.endsWith(\".webp\")) return \"image/webp\";\n if (path.endsWith(\".gif\")) return \"image/gif\";\n return \"image/jpeg\";\n}\n\n// ---------------------------------------------------------------------------\n// Anmeldung\n// ---------------------------------------------------------------------------\n\n/**\n * K-Tipp läuft auf TYPO3 und setzt nach der Anmeldung ein fe_typo_user-Cookie.\n * Das Cookie wird zwischengespeichert, damit nicht jeder Feed-Abruf eine neue\n * Anmeldung auslöst.\n */\nlet cachedCookie = \"\";\nlet cachedAt = 0;\nconst COOKIE_TTL_MS = 12 * 3600 * 1000;\n\nfunction collectCookies(jar: Map, res: Response): void {\n const headers = res.headers as unknown as { getSetCookie?: () => string[] };\n const list = headers.getSetCookie?.() ?? [];\n const single = res.headers.get(\"set-cookie\");\n for (const raw of list.length > 0 ? list : single ? [single] : []) {\n const pair = String(raw).split(\";\")[0] ?? \"\";\n const idx = pair.indexOf(\"=\");\n if (idx > 0) jar.set(pair.slice(0, idx).trim(), pair.slice(idx + 1).trim());\n }\n}\n\nasync function login(): Promise {\n const user = Deno.env.get(\"KTIPP_USER\");\n const password = Deno.env.get(\"KTIPP_PASSWORD\");\n if (!user || !password) {\n throw new Error(\"KTIPP_USER und KTIPP_PASSWORD sind nicht hinterlegt.\");\n }\n\n if (cachedCookie && Date.now() - cachedAt < COOKIE_TTL_MS) return cachedCookie;\n\n const jar = new Map();\n const cookieHeader = () => [...jar].map(([k, v]) => `${k}=${v}`).join(\"; \");\n\n const page = await fetch(`${SOURCE_URL}/login`, {\n headers: { \"User-Agent\": USER_AGENT, Accept: \"text/html\", \"Accept-Language\": \"de-CH,de;q=0.9\" },\n });\n collectCookies(jar, page);\n\n const $ = cheerio.load(await page.text());\n const form = $(\"#loginForm\");\n if (form.length === 0) throw new Error(\"Anmeldeformular nicht gefunden.\");\n\n const action = absolute(form.attr(\"action\") ?? \"/login\");\n const body = new URLSearchParams();\n\n // Alle versteckten Felder übernehmen – TYPO3 prüft sie beim Absenden\n form.find(\"input\").each((_, el) => {\n const attrs = (el as unknown as { attribs: Record }).attribs ?? {};\n if (!attrs.name || attrs.type === \"text\" || attrs.type === \"password\") return;\n body.set(attrs.name, attrs.value ?? \"\");\n });\n body.set(\"tx_updkonsuminfo_konsuminfofeuserlogin[userid]\", user);\n body.set(\"tx_updkonsuminfo_konsuminfofeuserlogin[password]\", password);\n body.set(\"logIn\", \"senden\");\n\n let res = await fetch(action, {\n method: \"POST\",\n headers: {\n \"User-Agent\": USER_AGENT,\n \"Content-Type\": \"application/x-www-form-urlencoded\",\n Accept: \"text/html\",\n \"Accept-Language\": \"de-CH,de;q=0.9\",\n Referer: `${SOURCE_URL}/login`,\n Origin: SOURCE_URL,\n Cookie: cookieHeader(),\n },\n body,\n redirect: \"manual\",\n });\n collectCookies(jar, res);\n\n // Weiterleitungen von Hand verfolgen, um alle Cookies mitzunehmen\n for (let hop = 0; hop < 5 && [301, 302, 303, 307, 308].includes(res.status); hop++) {\n const location = absolute(res.headers.get(\"location\") ?? \"/\");\n res = await fetch(location, {\n headers: { \"User-Agent\": USER_AGENT, Accept: \"text/html\", Cookie: cookieHeader() },\n redirect: \"manual\",\n });\n collectCookies(jar, res);\n }\n\n cachedCookie = cookieHeader();\n cachedAt = Date.now();\n return cachedCookie;\n}\n\nasync function fetchHtml(url: string, cookie: 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-CH,de;q=0.9\",\n ...(cookie ? { Cookie: cookie } : {}),\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// Inhalt\n// ---------------------------------------------------------------------------\n\nfunction buildContent($: cheerio.CheerioAPI, pageUrl: string): string {\n const body = $(\"section.main-content_article\").first();\n if (body.length === 0) return \"\";\n\n const fragment = cheerio.load(`

${body.html() ?? \"\"}

`, null, false);\n const root = fragment(\"div\").first();\n\n fragment(JUNK_SELECTOR).remove();\n fragment(\"h1\").remove();\n fragment(\"header.article-header\").remove();\n\n /**\n * Kommentarbereich abschneiden. Er hat keine eigene Klasse, beginnt aber\n * zuverlässig mit dieser Überschrift – ab dort wird alles entfernt,\n * einschliesslich der nachfolgenden Elemente auf jeder Ebene.\n */\n const commentHeading = fragment(\"h2, h3\").filter((_, node) =>\n /^Kommentare zu diesem Artikel/i.test(fragment(node as never).text().replace(/\\s+/g, \" \").trim())\n ).first();\n\n if (commentHeading.length > 0) {\n // Erst alles nach der Überschrift auf ihrer Ebene, dann die Überschrift\n // selbst – und danach auf jeder höheren Ebene nur die nachfolgenden\n // Elemente. Die Container selbst bleiben stehen, sonst verschwindet\n // auch der Artikeltext, der vor ihnen steht.\n commentHeading.nextAll().remove();\n let node = commentHeading.parent();\n commentHeading.remove();\n for (let depth = 0; depth < 12 && node.length > 0 && !node.is(root); depth++) {\n node.nextAll().remove();\n node = node.parent() as never;\n }\n }\n\n fragment(\"*\").each((_, node) => {\n const el = node as unknown as { tagName?: string; attribs?: Record };\n const attrs = el.attribs ?? {};\n const kept: Record = {};\n\n switch (el.tagName) {\n case \"img\": {\n const src = absolute(attrs[\"data-src\"] || attrs.src || \"\", pageUrl);\n if (src) kept.src = src;\n if (attrs.alt) kept.alt = attrs.alt;\n break;\n }\n case \"a\": {\n const href = absolute(attrs.href ?? \"\", pageUrl);\n if (href) kept.href = href;\n break;\n }\n case \"iframe\": {\n const src = absolute(attrs[\"data-src\"] || attrs.src || \"\", pageUrl);\n if (src && /youtube|youtu\\.be|vimeo/i.test(src)) {\n kept.src = src;\n kept.allowfullscreen = \"\";\n kept.frameborder = \"0\";\n }\n break;\n }\n default:\n break;\n }\n\n el.attribs = kept;\n });\n\n fragment(\"img:not([src])\").remove();\n fragment(\"iframe:not([src])\").remove();\n\n for (let pass = 0; pass < 3; pass++) {\n fragment(\"div, p, span, section, li, ul\").each((_, node) => {\n const $el = fragment(node as never);\n if ($el.find(\"img, iframe, table\").length > 0) return;\n if ($el.text().replace(/\\s+/g, \"\").length === 0) $el.remove();\n });\n }\n\n const html = (root.html() ?? \"\").trim();\n return html.length > MAX_CONTENT_CHARS ? html.slice(0, MAX_CONTENT_CHARS) : html;\n}\n\nasync function discoverArticles(cookie: string): Promise {\n const seen = new Set();\n\n const pages = await runWithConcurrency(\n SECTION_PATHS,\n 3,\n (path) => fetchHtml(`${SOURCE_URL}${path}`, cookie, 15000),\n );\n\n for (const html of pages) {\n if (!html) continue;\n const $ = cheerio.load(html);\n $('a[href*=\"/artikel/artikeldetail/\"]').each((_, el) => {\n const href = $(el).attr(\"href\");\n if (!href) return;\n try {\n const url = new URL(href, SOURCE_URL);\n if (url.hostname.replace(/^www\\./, \"\") !== \"gesundheitstipp.ch\") return;\n if (!ARTICLE_PATH_RE.test(url.pathname)) return;\n seen.add(`${SOURCE_URL}${url.pathname}`);\n } catch {\n // ungültige URL ignorieren\n }\n });\n }\n\n return [...seen].slice(0, MAX_CANDIDATES);\n}\n\nasync function enrichArticle(link: string, cookie: string): Promise

${\n article.image ? `

` : \"\"\n }

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

Diff · v57