[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_154":155,"_156":157},"routes/_app.x.$handle.$name",{"_94":93,"_32":45,"_103":84,"_27":104,"_13":14,"_105":-5,"_11":106,"_107":108,"_109":110,"_111":112,"_113":114,"_20":-5,"_115":-5,"_116":-5,"_117":25,"_118":119,"_120":25,"_121":-5,"_122":-5,"_123":124},"routes/_app.x.$handle.$name.branch.$branch.version.$version.diff",{"_11":12,"_13":14,"_15":16,"_17":-5,"_18":-5},"project",{"_43":44,"_32":45,"_38":46,"_47":-5,"_48":25,"_49":50,"_51":50,"_52":53,"_54":-5,"_18":55,"_56":-5,"_57":25,"_58":-5,"_59":-5,"_60":61,"_62":63,"_64":-5,"_65":-5,"_66":-5,"_67":-5,"_68":69,"_70":69,"_71":72,"_73":72,"_74":72,"_75":69,"_76":69,"_77":78,"_79":80,"_81":82,"_83":25},"version",22,"changes",[19],"aiChatInfo","userId",{"_20":21,"_22":23,"_24":25,"_26":25,"_27":28,"_29":28},"file",{"_30":31,"_32":28,"_33":34,"_35":37,"_38":39,"_40":14,"_41":42,"_24":25},"previous",{"_30":31,"_32":28,"_33":34,"_35":36},"deleted",false,"deletedHere","path","new-file-5265.ts","latestPath","fileId","3dd22cf8-a31b-11f1-b763-1607ee4eb77e","name","fileType","interval","content","/**\n * daily-curiosity.ts — a Val Town cron val\n *\n * Sends Claire one letter each morning: a picture, a word, a finding.\n * Every item is fetched live from a public source and linked back to it.\n * No API keys, no accounts, no cost.\n *\n * SETUP\n * 1. val.town → New val → Cron\n * 2. Paste this file in, and add WORDNIK_API_KEY as an environment variable\n * (free key from developer.wordnik.com; the letter still sends without it,\n * it just drops the word)\n * 3. Set the cron to 07:00, London time\n * 4. Settings → make sure error notifications are on\n *\n * Scheduling is Val Town's job, not this file's. Check once in late October\n * that the letter still lands at 07:00 after the clocks change — if it slips\n * to 06:00, the cron is on a fixed +1 offset rather than Europe/London.\n */\n\nimport { email } from \"https://esm.town/v/std/email\";\nimport { blob } from \"https://esm.town/v/std/blob\";\nimport { XMLParser } from \"npm:fast-xml-parser\";\n\nconst TZ = \"Europe/London\";\n\n// ---------------------------------------------------------------- memory\n\n/** Links already sent, so nothing repeats. Keeps the last SEEN_MEMORY items. */\nconst SEEN_KEY = \"daily-curiosity-seen\";\nconst SEEN_MEMORY = 150;\n\nasync function loadSeen(): Promise {\n try {\n return (await blob.getJSON(SEEN_KEY)) ?? [];\n } catch {\n return []; // first run, or blob unavailable: behave as if nothing is seen\n }\n}\n\nasync function saveSeen(seen: string[], links: string[]) {\n const merged = [...seen, ...links.filter(Boolean)].slice(-SEEN_MEMORY);\n try {\n await blob.setJSON(SEEN_KEY, merged);\n } catch {\n // not worth losing a letter over\n }\n}\n\n// ---------------------------------------------------------------- sources\n\n/**\n * The Metropolitan Museum of Art — public domain works, no API key.\n * (The Art Institute was the obvious choice, but its images sit behind a\n * Cloudflare challenge that blocks mail clients from loading them.)\n *\n * Rotating the department is deliberate: it keeps the letter moving between\n * Egyptian stonework, Edo prints and 20th-century photographs rather than\n * settling into European oil paintings, which dominate on an unfiltered pull.\n */\nconst MET = \"https://collectionapi.metmuseum.org/public/collection/v1\";\nconst DEPARTMENTS = [\n 6, // Asian Art\n 9, // Drawings and Prints\n 10, // Egyptian Art\n 11, // European Paintings\n 14, // Islamic Art\n 17, // Medieval Art\n 19, // Photographs\n 21, // Modern and Contemporary Art\n];\n\nasync function fetchArtwork(seen: string[] = []) {\n for (let attempt = 0; attempt < 6; attempt++) {\n const dept = DEPARTMENTS[Math.floor(Math.random() * DEPARTMENTS.length)];\n const search = await fetch(\n `${MET}/search?departmentId=${dept}&isPublicDomain=true&hasImages=true&q=*`,\n );\n if (!search.ok) continue;\n const { objectIDs } = await search.json();\n if (!objectIDs?.length) continue;\n\n const id = objectIDs[Math.floor(Math.random() * objectIDs.length)];\n const res = await fetch(`${MET}/objects/${id}`);\n if (!res.ok) continue;\n const a = await res.json();\n if (!a.primaryImageSmall || !a.title) continue;\n if (seen.includes(a.objectURL)) continue;\n\n return {\n kind: \"picture\" as const,\n title: a.title,\n maker: a.artistDisplayName || a.culture || \"Unknown\",\n year: a.objectDate ?? \"\",\n meta: a.medium ?? \"\",\n image: a.primaryImageSmall,\n link: a.objectURL,\n source: \"The Met\",\n text: \"\",\n };\n }\n throw new Error(\"artwork: no usable result after 6 attempts\");\n}\n\n/**\n * Wordnik — editor-picked word of the day, plus its etymology.\n *\n * Needs a free key (developer.wordnik.com), set as WORDNIK_API_KEY.\n * The etymology field is inconsistent: some words return a clean origin,\n * some return marked-up fragments, some nothing at all. So the definition\n * carries the block and the etymology is a bonus when it exists.\n */\nasync function fetchWord(seen: string[] = []) {\n const key = Deno.env.get(\"WORDNIK_API_KEY\")!;\n\n const res = await fetch(\n `https://api.wordnik.com/v4/words.json/wordOfTheDay?api_key=${key}`,\n );\n if (!res.ok) throw new Error(`word: word-of-the-day returned ${res.status}`);\n const wotd = await res.json();\n\n const word = wotd?.word;\n const def = wotd?.definitions?.[0];\n if (!word || !def?.text) throw new Error(\"word: no usable definition\");\n\n const link = `https://www.wordnik.com/words/${encodeURIComponent(word)}`;\n if (seen.includes(link)) throw new Error(`word: already sent ${word}`);\n\n // Etymology is best-effort; a failure here shouldn't cost you the block.\n let etymology = \"\";\n try {\n const etRes = await fetch(\n `https://api.wordnik.com/v4/word.json/${\n encodeURIComponent(word)\n }/etymologies?api_key=${key}`,\n );\n if (etRes.ok) {\n const [raw] = await etRes.json();\n etymology = String(raw ?? \"\")\n .replace(/<[^>]+>/g, \"\") // strip the markup Wordnik sometimes returns\n .replace(/\\s+/g, \" \")\n .trim();\n }\n } catch {\n // leave it empty\n }\n\n return {\n kind: \"word\" as const,\n title: word,\n maker: def.partOfSpeech ?? \"\",\n year: \"\",\n meta: def.sourceDictionary ?? \"Wordnik\",\n image: null,\n link,\n source: \"Wordnik\",\n text: etymology ? `${def.text}\\n\\n${etymology}` : def.text,\n };\n}\n\n/**\n * The finding.\n *\n * Each source is tried twice: first the WordPress archive, which paginates the\n * entire back catalogue, then the RSS feed, which only carries the latest dozen\n * or so. The archive is the difference between a pool of thirty items and a\n * pool of thousands — an eight-year-old Quanta piece is new to you.\n *\n * Whether a given site exposes wp-json publicly varies, so a failure there is\n * expected rather than alarming; the log line tells you which route each source\n * actually took. A feed that has died contributes nothing instead of breaking\n * the letter, so adding one you're unsure about is safe.\n */\nconst FEEDS = [\n {\n name: \"Quanta Magazine\",\n rss: \"https://api.quantamagazine.org/feed/\",\n wp: \"https://api.quantamagazine.org/wp-json/wp/v2/posts\",\n },\n {\n name: \"Nautilus\",\n rss: \"https://nautil.us/feed/\",\n wp: \"https://nautil.us/wp-json/wp/v2/posts\",\n },\n {\n name: \"Aeon\",\n rss: \"https://aeon.co/feed.rss\",\n },\n {\n name: \"Knowable Magazine\",\n rss: \"https://knowablemagazine.org/rss\",\n },\n {\n // Their Substack rather than worksinprogress.co: it carries most magazine\n // pieces plus Substack-only writing, in a predictable feed shape.\n name: \"Works in Progress\",\n rss: \"https://www.worksinprogress.news/feed\",\n },\n];\n\ntype Feed = { name: string; rss: string; wp?: string };\n\n/** WordPress returns titles with HTML entities in them. */\nfunction decode(str: string) {\n return String(str ?? \"\")\n .replace(/&#(\\d+);/g, (_, n) => String.fromCharCode(Number(n)))\n .replace(\n /&#x([0-9a-f]+);/gi,\n (_, n) => String.fromCharCode(parseInt(n, 16)),\n )\n .replace(/&/g, \"&\")\n .replace(/"/g, '\"')\n .replace(/'|'/g, \"'\")\n .replace(/</g, \"<\")\n .replace(/>/g, \">\")\n .replace(/ /g, \" \")\n .replace(/<[^>]+>/g, \"\")\n .trim();\n}\n\n/** The whole back catalogue, one random page of it. */\nasync function fetchArchive(feed: Feed) {\n const probe = await fetch(`${feed.wp}?per_page=10&_fields=id`);\n if (!probe.ok) {\n throw new Error(`${feed.name}: archive returned ${probe.status}`);\n }\n\n const pages = Number(probe.headers.get(\"x-wp-totalpages\") ?? 0);\n if (!pages) throw new Error(`${feed.name}: archive gave no page count`);\n\n const page = 1 + Math.floor(Math.random() * pages);\n const res = await fetch(`${feed.wp}?per_page=10&page=${page}&_embed=1`);\n if (!res.ok) {\n throw new Error(\n `${feed.name}: archive page ${page} returned ${res.status}`,\n );\n }\n\n const posts = await res.json();\n if (!Array.isArray(posts)) {\n throw new Error(`${feed.name}: archive gave no posts`);\n }\n\n console.log(`${feed.name}: archive, page ${page} of ${pages}`);\n\n return posts.map((p: any) => ({\n kind: \"finding\" as const,\n title: decode(p?.title?.rendered ?? \"\"),\n maker: p?._embedded?.author?.[0]?.name ?? feed.name,\n year: p?.date ? new Date(p.date).getFullYear().toString() : \"\",\n meta: \"\",\n image: p?._embedded?.[\"wp:featuredmedia\"]?.[0]?.source_url ?? null,\n link: String(p?.link ?? \"\"),\n source: feed.name,\n text: \"\",\n })).filter((i: any) => i.title && i.link);\n}\n\n/** The last dozen items, for sources with no reachable archive. */\nasync function fetchRss(feed: Feed) {\n const res = await fetch(feed.rss);\n if (!res.ok) throw new Error(`${feed.name}: feed returned ${res.status}`);\n const parser = new XMLParser({ ignoreAttributes: false });\n const parsed = parser.parse(await res.text());\n const raw = parsed?.rss?.channel?.item ?? [];\n const items = Array.isArray(raw) ? raw : [raw];\n\n console.log(`${feed.name}: rss`);\n\n return items.slice(0, 12).map((item: any) => {\n const body: string = item[\"content:encoded\"] ?? item.description ?? \"\";\n return {\n kind: \"finding\" as const,\n title: decode(String(item.title ?? \"\")),\n maker: String(item[\"dc:creator\"] ?? feed.name),\n year: item.pubDate ? new Date(item.pubDate).getFullYear().toString() : \"\",\n meta: \"\",\n image: item[\"media:content\"]?.[\"@_url\"] ??\n item[\"media:thumbnail\"]?.[\"@_url\"] ??\n // Substack uses for the lead image\n (String(item.enclosure?.[\"@_type\"] ?? \"\").startsWith(\"image/\")\n ? item.enclosure[\"@_url\"]\n : null) ??\n body.match(/]+src=\"([^\"]+)\"/)?.[1] ??\n null,\n link: String(item.link ?? \"\"),\n source: feed.name,\n text: \"\",\n };\n }).filter((i: any) => i.title && i.link);\n}\n\nasync function fetchFeed(feed: Feed) {\n if (feed.wp) {\n try {\n return await fetchArchive(feed);\n } catch (err) {\n console.warn(\n `${feed.name}: archive unavailable, falling back to rss — ${err}`,\n );\n }\n }\n return await fetchRss(feed);\n}\n\nasync function fetchFinding(seen: string[] = []) {\n const settled = await Promise.allSettled(FEEDS.map(fetchFeed));\n const pool = settled\n .filter((r) => r.status === \"fulfilled\")\n .flatMap((r: any) => r.value)\n .filter((i: any) => !seen.includes(i.link));\n\n if (!pool.length) throw new Error(\"finding: no unseen items across any feed\");\n\n // Prefer something with a picture, since the letter is image-led.\n const withImage = pool.filter((i: any) => i.image);\n const candidates = withImage.length ? withImage : pool;\n return candidates[Math.floor(Math.random() * candidates.length)];\n}\n\n// ---------------------------------------------------------------- rendering\n\nconst INK = \"#1A1A1C\";\nconst META = \"#77777A\";\nconst WALL = \"#F2F2F0\";\nconst RULE = \"#D8D8D4\";\nconst SERIF = \"Georgia, 'Iowan Old Style', 'Times New Roman', serif\";\nconst SANS = \"-apple-system, 'Helvetica Neue', Helvetica, Arial, sans-serif\";\n\nconst esc = (s: string) =>\n String(s ?? \"\").replace(/&/g, \"&\").replace(//g,\n \">\",\n );\n\nconst EYEBROW: Record = {\n picture: \"Picture\",\n word: \"Word\",\n finding: \"Finding\",\n};\n\nfunction label(item: any) {\n const year = item.year\n ? `${\n esc(item.year)\n }`\n : \"\";\n\n return `\n \n \n \n ${year}\n \n

${\n EYEBROW[item.kind]\n }

${\n esc(item.title)\n }

${esc(item.maker)}${\n item.meta ? ` · ${esc(item.meta)}` : \"\"\n }

${\n esc(item.source)\n } ›

`;\n}\n\nfunction block(item: any) {\n const visual = !item.image && !item.text\n ? \"\" // no picture, no text: show the label alone rather than an empty box\n : item.image\n ? ` $\"${\n$ `\n : `

${\n esc(item.text)\n }

`;\n\n return `${visual}

${\n label(item)\n }

`;\n}\n\nfunction renderLetter(items: any[]) {\n const today = new Intl.DateTimeFormat(\"en-GB\", {\n weekday: \"long\",\n day: \"numeric\",\n month: \"long\",\n year: \"numeric\",\n timeZone: TZ,\n }).format(new Date());\n\n return `\n \n \n

\n \n \n \n ${items.map((i) => block(i)).join(\"\")}\n \n

${\n esc(today)\n }

\n ${[...new Set(items.map((i) => i.source))].join(\" · \")}\n

`;\n}\n\n// ---------------------------------------------------------------- run\n\nexport default async function () {\n const seen = await loadSeen();\n\n // The word block is skipped entirely until WORDNIK_API_KEY exists, so a\n // missing key is silence rather than a daily error mail. Add the variable\n // and it appears on the next run with no code change.\n const tasks = [fetchArtwork(seen), fetchFinding(seen)];\n if (Deno.env.get(\"WORDNIK_API_KEY\")) tasks.splice(1, 0, fetchWord(seen));\n\n const results = await Promise.allSettled(tasks);\n\n const items = results\n .filter((r) => r.status === \"fulfilled\")\n .map((r: any) => r.value);\n const failures = results\n .filter((r) => r.status === \"rejected\")\n .map((r: any) => r.reason?.message ?? String(r.reason));\n\n if (!items.length) {\n throw new Error(`every source failed: ${failures.join(\"; \")}`);\n }\n\n await email({\n subject: `Picture, word, finding — ${\n new Intl.DateTimeFormat(\"en-GB\", {\n day: \"numeric\",\n month: \"long\",\n timeZone: TZ,\n })\n .format(new Date())\n }`,\n html: renderLetter(items),\n });\n\n await saveSeen(seen, items.map((i: any) => i.link));\n\n // Surfaces partial failures in Val Town's run log without losing the letter.\n if (failures.length) {\n throw new Error(`sent, but degraded: ${failures.join(\"; \")}`);\n }\n}","/**\n * daily-curiosity.ts — a Val Town cron val\n *\n * Sends Claire one letter each morning: a picture, a word, a finding.\n * Every item is fetched live from a public source and linked back to it.\n * No API keys, no accounts, no cost.\n *\n * SETUP\n * 1. val.town → New val → Cron\n * 2. Paste this file in, and add WORDNIK_API_KEY as an environment variable\n * (free key from developer.wordnik.com; the letter still sends without it,\n * it just drops the word)\n * 3. Set the cron to 07:00, London time\n * 4. Settings → make sure error notifications are on\n *\n * Scheduling is Val Town's job, not this file's. Check once in late October\n * that the letter still lands at 07:00 after the clocks change — if it slips\n * to 06:00, the cron is on a fixed +1 offset rather than Europe/London.\n */\n\nimport { email } from \"https://esm.town/v/std/email\";\nimport { blob } from \"https://esm.town/v/std/blob\";\nimport { XMLParser } from \"npm:fast-xml-parser\";\n\nconst TZ = \"Europe/London\";\n\n// ---------------------------------------------------------------- memory\n\n/** Links already sent, so nothing repeats. Keeps the last SEEN_MEMORY items. */\nconst SEEN_KEY = \"daily-curiosity-seen\";\nconst SEEN_MEMORY = 150;\n\nasync function loadSeen(): Promise {\n try {\n return (await blob.getJSON(SEEN_KEY)) ?? [];\n } catch {\n return []; // first run, or blob unavailable: behave as if nothing is seen\n }\n}\n\nasync function saveSeen(seen: string[], links: string[]) {\n const merged = [...seen, ...links.filter(Boolean)].slice(-SEEN_MEMORY);\n try {\n await blob.setJSON(SEEN_KEY, merged);\n } catch {\n // not worth losing a letter over\n }\n}\n\n// ---------------------------------------------------------------- sources\n\n/**\n * The Metropolitan Museum of Art — public domain works, no API key.\n * (The Art Institute was the obvious choice, but its images sit behind a\n * Cloudflare challenge that blocks mail clients from loading them.)\n *\n * Rotating the department is deliberate: it keeps the letter moving between\n * Egyptian stonework, Edo prints and 20th-century photographs rather than\n * settling into European oil paintings, which dominate on an unfiltered pull.\n */\nconst MET = \"https://collectionapi.metmuseum.org/public/collection/v1\";\nconst DEPARTMENTS = [\n 6, // Asian Art\n 9, // Drawings and Prints\n 10, // Egyptian Art\n 11, // European Paintings\n 14, // Islamic Art\n 17, // Medieval Art\n 19, // Photographs\n 21, // Modern and Contemporary Art\n];\n\nasync function fetchArtwork(seen: string[] = []) {\n for (let attempt = 0; attempt < 6; attempt++) {\n const dept = DEPARTMENTS[Math.floor(Math.random() * DEPARTMENTS.length)];\n const search = await fetch(\n `${MET}/search?departmentId=${dept}&isPublicDomain=true&hasImages=true&q=*`,\n );\n if (!search.ok) continue;\n const { objectIDs } = await search.json();\n if (!objectIDs?.length) continue;\n\n const id = objectIDs[Math.floor(Math.random() * objectIDs.length)];\n const res = await fetch(`${MET}/objects/${id}`);\n if (!res.ok) continue;\n const a = await res.json();\n if (!a.primaryImageSmall || !a.title) continue;\n if (seen.includes(a.objectURL)) continue;\n\n return {\n kind: \"picture\" as const,\n title: a.title,\n maker: a.artistDisplayName || a.culture || \"Unknown\",\n year: a.objectDate ?? \"\",\n meta: a.medium ?? \"\",\n image: a.primaryImageSmall,\n link: a.objectURL,\n source: \"The Met\",\n text: \"\",\n };\n }\n throw new Error(\"artwork: no usable result after 6 attempts\");\n}\n\n/**\n * Wordnik — the editor-picked word of the day, plus its etymology.\n *\n * Needs a free key (developer.wordnik.com), set as WORDNIK_API_KEY.\n *\n * The word of the day is the same word all day, so a second run in one day\n * would otherwise repeat it. When that happens it falls back to a random word\n * with a definition, and if even that fails it returns null and the letter\n * simply arrives without a word — a repeat run is not worth an error mail.\n *\n * The etymology field is inconsistent: some words return a clean origin, some\n * marked-up fragments, some nothing. So the definition carries the block.\n */\nasync function wordnik(path: string, key: string) {\n const join = path.includes(\"?\") ? \"&\" : \"?\";\n const res = await fetch(`https://api.wordnik.com/v4/${path}${join}api_key=${key}`);\n if (!res.ok) throw new Error(`word: ${path} returned ${res.status}`);\n return await res.json();\n}\n\nasync function buildWord(word: string, def: any, key: string, seen: string[]) {\n if (!word || !def?.text) return null;\n\n const link = `https://www.wordnik.com/words/${encodeURIComponent(word)}`;\n if (seen.includes(link)) return null;\n\n let etymology = \"\";\n try {\n const [raw] = await wordnik(`word.json/${encodeURIComponent(word)}/etymologies`, key);\n etymology = String(raw ?? \"\")\n .replace(/<[^>]+>/g, \"\") // strip the markup Wordnik sometimes returns\n .replace(/\\s+/g, \" \")\n .trim();\n } catch {\n // best-effort; a missing etymology shouldn't cost you the block\n }\n\n return {\n kind: \"word\" as const,\n title: word,\n maker: def.partOfSpeech ?? \"\",\n year: \"\",\n meta: def.sourceDictionary ?? \"Wordnik\",\n image: null,\n link,\n source: \"Wordnik\",\n text: etymology ? `${def.text}\\n\\n${etymology}` : def.text,\n };\n}\n\nasync function fetchWord(seen: string[] = []) {\n const key = Deno.env.get(\"WORDNIK_API_KEY\")!;\n\n const wotd = await wordnik(\"words.json/wordOfTheDay\", key);\n const first = await buildWord(wotd?.word, wotd?.definitions?.[0], key, seen);\n if (first) return first;\n\n console.log(\"word: today's word already sent, trying a random one\");\n\n for (let attempt = 0; attempt < 5; attempt++) {\n try {\n const rand = await wordnik(\n \"words.json/randomWord?hasDictionaryDef=true&minLength=4\",\n key,\n );\n const [def] = await wordnik(\n `word.json/${encodeURIComponent(rand?.word)}/definitions?limit=1`,\n key,\n );\n const item = await buildWord(rand?.word, def, key, seen);\n if (item) return item;\n } catch {\n // try again\n }\n }\n\n console.warn(\"word: nothing unseen available, skipping the block\");\n return null;\n}\n\n/**\n * The finding.\n *\n * Each source is tried twice: first the WordPress archive, which paginates the\n * entire back catalogue, then the RSS feed, which only carries the latest dozen\n * or so. The archive is the difference between a pool of thirty items and a\n * pool of thousands — an eight-year-old Quanta piece is new to you.\n *\n * Whether a given site exposes wp-json publicly varies, so a failure there is\n * expected rather than alarming; the log line tells you which route each source\n * actually took. A feed that has died contributes nothing instead of breaking\n * the letter, so adding one you're unsure about is safe.\n */\nconst FEEDS = [\n {\n name: \"Quanta Magazine\",\n rss: \"https://api.quantamagazine.org/feed/\",\n wp: \"https://api.quantamagazine.org/wp-json/wp/v2/posts\",\n },\n {\n // No wp-json here: it 404s, so this is feed-only.\n name: \"Nautilus\",\n rss: \"https://nautil.us/feed/\",\n },\n {\n name: \"Aeon\",\n rss: \"https://aeon.co/feed.rss\",\n },\n {\n name: \"Knowable Magazine\",\n rss: \"https://knowablemagazine.org/rss\",\n },\n {\n // Their Substack rather than worksinprogress.co: it carries most magazine\n // pieces plus Substack-only writing, in a predictable feed shape.\n name: \"Works in Progress\",\n rss: \"https://www.worksinprogress.news/feed\",\n },\n];\n\ntype Feed = { name: string; rss: string; wp?: string };\n\n/** WordPress returns titles with HTML entities in them. */\nfunction decode(str: string) {\n return String(str ?? \"\")\n .replace(/&#(\\d+);/g, (_, n) => String.fromCharCode(Number(n)))\n .replace(/&#x([0-9a-f]+);/gi, (_, n) => String.fromCharCode(parseInt(n, 16)))\n .replace(/&/g, \"&\")\n .replace(/"/g, '\"')\n .replace(/'|'/g, \"'\")\n .replace(/</g, \"<\")\n .replace(/>/g, \">\")\n .replace(/ /g, \" \")\n .replace(/<[^>]+>/g, \"\")\n .trim();\n}\n\n/** The whole back catalogue, one random page of it. */\nasync function fetchArchive(feed: Feed) {\n const probe = await fetch(`${feed.wp}?per_page=10&_fields=id`);\n if (!probe.ok) throw new Error(`${feed.name}: archive returned ${probe.status}`);\n\n const pages = Number(probe.headers.get(\"x-wp-totalpages\") ?? 0);\n if (!pages) throw new Error(`${feed.name}: archive gave no page count`);\n\n const page = 1 + Math.floor(Math.random() * pages);\n const res = await fetch(`${feed.wp}?per_page=10&page=${page}&_embed=1`);\n if (!res.ok) throw new Error(`${feed.name}: archive page ${page} returned ${res.status}`);\n\n const posts = await res.json();\n if (!Array.isArray(posts)) throw new Error(`${feed.name}: archive gave no posts`);\n\n console.log(`${feed.name}: archive, page ${page} of ${pages}`);\n\n return posts.map((p: any) => ({\n kind: \"finding\" as const,\n title: decode(p?.title?.rendered ?? \"\"),\n maker: p?._embedded?.author?.[0]?.name ?? feed.name,\n year: p?.date ? new Date(p.date).getFullYear().toString() : \"\",\n meta: \"\",\n image: p?._embedded?.[\"wp:featuredmedia\"]?.[0]?.source_url ?? null,\n link: String(p?.link ?? \"\"),\n source: feed.name,\n text: \"\",\n })).filter((i: any) => i.title && i.link);\n}\n\n/** The last dozen items, for sources with no reachable archive. */\nasync function fetchRss(feed: Feed) {\n const res = await fetch(feed.rss);\n if (!res.ok) throw new Error(`${feed.name}: feed returned ${res.status}`);\n const parser = new XMLParser({ ignoreAttributes: false });\n const parsed = parser.parse(await res.text());\n const raw = parsed?.rss?.channel?.item ?? [];\n const items = Array.isArray(raw) ? raw : [raw];\n\n console.log(`${feed.name}: rss`);\n\n return items.slice(0, 12).map((item: any) => {\n const body: string = item[\"content:encoded\"] ?? item.description ?? \"\";\n return {\n kind: \"finding\" as const,\n title: decode(String(item.title ?? \"\")),\n maker: String(item[\"dc:creator\"] ?? feed.name),\n year: item.pubDate ? new Date(item.pubDate).getFullYear().toString() : \"\",\n meta: \"\",\n image: item[\"media:content\"]?.[\"@_url\"] ??\n item[\"media:thumbnail\"]?.[\"@_url\"] ??\n // Substack uses for the lead image\n (String(item.enclosure?.[\"@_type\"] ?? \"\").startsWith(\"image/\")\n ? item.enclosure[\"@_url\"]\n : null) ??\n body.match(/]+src=\"([^\"]+)\"/)?.[1] ??\n null,\n link: String(item.link ?? \"\"),\n source: feed.name,\n text: \"\",\n };\n }).filter((i: any) => i.title && i.link);\n}\n\nasync function fetchFeed(feed: Feed) {\n if (feed.wp) {\n try {\n return await fetchArchive(feed);\n } catch (err) {\n console.warn(`${feed.name}: archive unavailable, falling back to rss — ${err}`);\n }\n }\n return await fetchRss(feed);\n}\n\nasync function fetchFinding(seen: string[] = []) {\n const settled = await Promise.allSettled(FEEDS.map(fetchFeed));\n const pool = settled\n .filter((r) => r.status === \"fulfilled\")\n .flatMap((r: any) => r.value)\n .filter((i: any) => !seen.includes(i.link));\n\n if (!pool.length) throw new Error(\"finding: no unseen items across any feed\");\n\n // Prefer something with a picture, since the letter is image-led.\n const withImage = pool.filter((i: any) => i.image);\n const candidates = withImage.length ? withImage : pool;\n return candidates[Math.floor(Math.random() * candidates.length)];\n}\n\n// ---------------------------------------------------------------- rendering\n\nconst INK = \"#1A1A1C\";\nconst META = \"#77777A\";\nconst WALL = \"#F2F2F0\";\nconst RULE = \"#D8D8D4\";\nconst SERIF = \"Georgia, 'Iowan Old Style', 'Times New Roman', serif\";\nconst SANS = \"-apple-system, 'Helvetica Neue', Helvetica, Arial, sans-serif\";\n\nconst esc = (s: string) =>\n String(s ?? \"\").replace(/&/g, \"&\").replace(//g,\n \">\",\n );\n\nconst EYEBROW: Record = {\n picture: \"Picture\",\n word: \"Word\",\n finding: \"Finding\",\n};\n\nfunction label(item: any) {\n const year = item.year\n ? `${\n esc(item.year)\n }`\n : \"\";\n\n return `\n \n \n \n ${year}\n \n

Diff · v22