fh-library 0.2.17 → 0.2.20

package/dist/index.js

@@ -11,10 +11,10 @@ function dotenvFile(env = process.env) {
   return join(resolve(lib), ".env");
 }
 var loadedFromDotenv = /* @__PURE__ */ new Set();
-function loadDotenv(file2 = dotenvFile()) {
-  if (!existsSync(file2)) return [];
+function loadDotenv(file3 = dotenvFile()) {
+  if (!existsSync(file3)) return [];
   const loaded = [];
-  for (const raw of readFileSync(file2, "utf8").split(/\r?\n/)) {
+  for (const raw of readFileSync(file3, "utf8").split(/\r?\n/)) {
     const line = raw.trim();
     if (!line || line.startsWith("#")) continue;
     const m = /^(?:export\s+)?([A-Za-z_][A-Za-z0-9_]*)\s*=\s*(.*)$/.exec(line);
@@ -1208,1693 +1208,1747 @@ async function unlock(url, postImpl = post) {
   return { bytes: Buffer.from(await res.arrayBuffer()), contentType: res.headers.get("content-type") ?? "" };
 }
 
-// src/tiers/tier.ts
-function oaVersion(v) {
-  if (v === "acceptedVersion") return "accepted_manuscript";
-  if (v === "submittedVersion") return "preprint";
-  return "published";
-}
-var miss = (tier, reason) => ({ ok: false, tier, reason });
-var errMsg = (e) => (e instanceof Error ? e.message : String(e)).replace(/\x1b\[[0-9;]*m/g, "").split("\n")[0].trim().slice(0, 200);
+// src/walls.ts
+import { mkdirSync as mkdirSync3, readFileSync as readFileSync4 } from "fs";
+import { join as join5 } from "path";
 
-// src/download.ts
-var hostOf2 = (url) => {
-  try {
-    return new URL(url).hostname.replace(/^www\./, "");
-  } catch {
-    return url;
-  }
+// src/entry.ts
+import { copyFileSync, existsSync as existsSync4, mkdirSync as mkdirSync2, readdirSync, readFileSync as readFileSync3, renameSync, rmSync, unlinkSync, writeFileSync } from "fs";
+import { join as join4 } from "path";
+import { basename, dirname } from "path";
+
+// src/pubmed.ts
+import { XMLParser as XMLParser2 } from "fast-xml-parser";
+var CORRECTION_TYPES = {
+  ErratumIn: "erratum_in",
+  RetractionIn: "retraction_in",
+  PartialRetractionIn: "partial_retraction_in",
+  RetractedandRepublishedIn: "retracted_and_republished_in",
+  ExpressionOfConcernIn: "expression_of_concern_in",
+  UpdateIn: "update_in"
 };
-var NCBI = /^https?:\/\/([a-z0-9.-]+\.)?(ncbi\.nlm\.nih\.gov|pubmed\.gov)\//i;
-function order(candidates) {
-  const rank = (c) => (c.pdf ? 0 : 2) + (c.host === "publisher" ? 1 : 0);
-  return [...candidates].sort((a, b) => rank(a) - rank(b));
-}
-async function download(tier, p, candidates, ctx) {
-  const usable = candidates.filter((c) => !NCBI.test(c.url));
-  if (!usable.length && candidates.length) return miss(tier, "only pmc copies (covered by the pmc tier)");
-  const ordered = order(usable).filter((c) => !ctx.tried.has(c.url));
-  if (!ordered.length) return miss(tier, "only copies already tried");
-  const reasons = [];
-  for (const c of ordered) {
-    ctx.tried.add(c.url);
-    let r = await tryOne(p, c, tier);
-    const target = !r.ok && r.url && r.url !== c.url ? { ...c, url: r.url } : c;
-    if (target !== c) ctx.onTry?.(`${c.url} \u2192 ${target.url}`);
-    ctx.onTry?.(r.ok ? `${target.url} ok fetch` : `${target.url} failed fetch: ${r.reason}`);
-    const why = (reason) => reason.replace(/^not a pdf \((.*)\)$/, "$1");
-    for (const b of browserRungs()) {
-      if (r.ok || c.headers || c.service || !/^(HTTP 403|HTTP 429|wall page|no methods\/results|unknown content|browser )/.test(why(r.reason))) break;
-      r = await tryOne(p, target, tier, { browser: b.endpoint, label: b.label });
-      ctx.onTry?.(r.ok ? `${target.url} ok browser ${b.label}` : `${target.url} failed browser ${b.label}: ${bare(r.reason)}`);
-      if (!r.ok && bare(r.reason).startsWith(MISMATCH)) ctx.onAlert?.(`browser ${b.label}: ${bare(r.reason)}`);
-    }
-    if (!r.ok && !c.headers && !c.service && unlockerConfigured() && /^(HTTP 403|wall page|browser )/.test(why(r.reason))) {
-      r = await tryOne(p, target, tier, "unlocker");
-      ctx.onTry?.(r.ok ? `${target.url} ok brightdata` : `${target.url} failed brightdata: ${bare(r.reason)}`);
-      if (!r.ok && /^unlocker HTTP/.test(bare(r.reason))) ctx.onAlert?.(`brightdata: ${bare(r.reason)}`);
-    }
-    if (r.ok) return r;
-    reasons.push(`${r.reason}: ${hostOf2(c.url)}`);
-  }
-  return miss(tier, reasons.join("; "));
+var RETRACTION_TYPES = /* @__PURE__ */ new Set(["retraction_in", "partial_retraction_in", "retracted_and_republished_in"]);
+var isRetractionNotice = (c) => RETRACTION_TYPES.has(c.type);
+var EFETCH = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi";
+function efetchUrl(pmid) {
+  const u = new URL(EFETCH);
+  u.searchParams.set("db", "pubmed");
+  u.searchParams.set("id", pmid);
+  u.searchParams.set("retmode", "xml");
+  if (process.env.FH_LIBRARY_NCBI_API_KEY) u.searchParams.set("api_key", process.env.FH_LIBRARY_NCBI_API_KEY);
+  u.searchParams.set("tool", "fh-library");
+  u.searchParams.set("email", contactEmail());
+  return u.toString();
 }
-var bare = (reason) => reason.replace(/^(brightdata|browser [^:]+): /, "");
-async function tryOne(p, c, tier, via = "plain") {
-  let got;
+async function fetchPubmedXml(pmid) {
   try {
-    got = via === "unlocker" ? await unlock(c.url) : typeof via === "object" ? await browserFetch(c.url, via.browser) : await fetchBytes(c.url, { ...DOWNLOAD_HEADERS, ...c.headers ?? {} });
+    return await fetchText(efetchUrl(pmid));
   } catch (e) {
-    const body = e?.body;
-    if (c.service && body?.trim()) return { ok: false, reason: `${errMsg(e)} \u2014 ${serviceMessage(Buffer.from(body), e.contentType)}`, url: e?.url };
-    return { ok: false, reason: (via === "plain" ? "" : via === "unlocker" ? "brightdata: " : `browser ${via.label}: `) + errMsg(e), url: e?.url };
+    throw new Error(`pubmed fetch failed: ${e instanceof Error ? e.message : String(e)}`);
   }
-  const { bytes, contentType } = got;
-  if (c.service && !looksLikePdf(bytes) && !isXmlArticle(bytes.toString("utf8"))) return { ok: false, reason: `service: ${serviceMessage(bytes, contentType)}`, url: got.url };
-  if (via === "unlocker" && !bytes.length) return { ok: false, reason: "brightdata: empty response" };
-  const r = await convertBytes(p, bytes, contentType);
-  if (!r.ok) return { ok: false, reason: c.pdf && r.format !== "pdf" ? `not a pdf (${r.reason})` : r.reason, url: got.url };
-  return { ok: true, tier, format: r.format, raw: r.format === "pdf" ? bytes : bytes.toString("utf8"), md: r.md, version: c.version };
 }
-var isXmlArticle = (text3) => /^\s*<\?xml|<article\b/.test(text3.slice(0, 2e3)) && !/<html/i.test(text3.slice(0, 2e3));
-function serviceMessage(bytes, contentType = "") {
-  const text3 = bytes.toString("utf8").trim();
-  if (!text3) return `empty response (${contentType.split(";")[0] || "no type"})`;
-  try {
-    const j = JSON.parse(text3);
-    const m = j?.message ?? j?.error?.message ?? j?.error ?? j?.detail;
-    if (typeof m === "string" && m.trim()) return m.trim().replace(/\s+/g, " ").slice(0, 160);
-  } catch {
+var parser2 = new XMLParser2({
+  ignoreAttributes: false,
+  attributeNamePrefix: "@_",
+  parseTagValue: false,
+  processEntities: true,
+  stopNodes: ["*.ArticleTitle", "*.AbstractText", "*.CoiStatement"],
+  isArray: (name) => ["PubmedArticle", "PubmedBookArticle", "AbstractText", "Author", "ArticleId", "PublicationType", "CommentsCorrections", "Grant", "AuthorList"].includes(name)
+});
+function text(n) {
+  if (n == null) return "";
+  if (typeof n === "object") return text(n["#text"]);
+  return String(n);
+}
+var plain = (n) => decodeEntities(text(n)).trim();
+function inlineText2(raw) {
+  return decodeEntities(raw.replace(/<[^>]+>/g, "")).replace(/\s+/g, " ").trim();
+}
+function decodeEntities(s) {
+  return s.replace(/&#x([0-9a-f]+);/gi, (_, h) => String.fromCodePoint(parseInt(h, 16))).replace(/&#(\d+);/g, (_, d) => String.fromCodePoint(parseInt(d, 10))).replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&quot;/g, '"').replace(/&apos;/g, "'").replace(/&amp;/g, "&");
+}
+function parsePubmedXml(xml) {
+  const doc = parser2.parse(xml);
+  const article = doc?.PubmedArticleSet?.PubmedArticle?.[0];
+  if (!article) {
+    const book = doc?.PubmedArticleSet?.PubmedBookArticle?.[0];
+    return book ? parseBook(book) : null;
   }
-  const visible = text3.replace(/<script[\s\S]*?<\/script>|<style[\s\S]*?<\/style>/gi, " ").replace(/<title>([\s\S]*?)<\/title>/i, "$1. ").replace(/<[^>]+>/g, " ").replace(/&nbsp;/g, " ").replace(/\s+/g, " ").trim();
-  return (visible || text3).slice(0, 160);
+  const cit = article.MedlineCitation ?? {};
+  const art = cit.Article ?? {};
+  const ids = article.PubmedData?.ArticleIdList?.ArticleId ?? [];
+  const idOf = (t) => {
+    const hit = ids.find((i) => i["@_IdType"] === t);
+    return hit ? text(hit).trim() || null : null;
+  };
+  const authorList = art.AuthorList?.[0]?.Author ?? [];
+  const authors = authorList.map((a) => {
+    if (a.CollectiveName) return plain(a.CollectiveName);
+    const last = plain(a.LastName);
+    const ini = plain(a.Initials);
+    return [last, ini].filter(Boolean).join(" ");
+  }).filter(Boolean);
+  const first = authorList[0];
+  const firstAuthor = first ? plain(first.CollectiveName ?? first.LastName) || null : null;
+  const pub = art.Journal?.JournalIssue?.PubDate ?? {};
+  let year = text(pub.Year).trim() || null;
+  if (!year) year = /\d{4}/.exec(text(pub.MedlineDate))?.[0] ?? null;
+  const abstract = (art.Abstract?.AbstractText ?? []).map((a) => ({
+    label: typeof a === "object" ? a["@_Label"] ?? null : null,
+    text: inlineText2(text(a))
+  })).filter((a) => a.text);
+  const corrections = (cit.CommentsCorrectionsList?.CommentsCorrections ?? []).flatMap((c) => {
+    const type = CORRECTION_TYPES[String(c["@_RefType"] ?? "")];
+    if (!type) return [];
+    return [{ type, pmid: text(c.PMID).trim() || null, source: inlineText2(text(c.RefSource)) }];
+  });
+  const funding = [];
+  for (const g of art.GrantList?.Grant ?? []) {
+    const a = plain(g.Agency);
+    if (a && !funding.includes(a)) funding.push(a);
+  }
+  return {
+    coi: inlineText2(text(cit.CoiStatement)) || null,
+    funding,
+    pmid: text(cit.PMID).trim() || idOf("pubmed") || "",
+    pmcid: idOf("pmc"),
+    doi: idOf("doi"),
+    title: inlineText2(text(art.ArticleTitle)),
+    authors,
+    firstAuthor,
+    journal: plain(art.Journal?.Title) || null,
+    year,
+    publicationTypes: (art.PublicationTypeList?.PublicationType ?? []).map(plain).filter(Boolean),
+    abstract,
+    corrections
+  };
 }
-function describe(reason, format, md) {
-  const words2 = md.split(/\s+/).filter(Boolean).length;
-  const headings2 = md.split("\n").filter((l) => /^#{2,6}\s/.test(l)).length;
-  const seen = `${format}, ${words2} words, ${proseWords(md)} in prose, ${headings2} heading${headings2 === 1 ? "" : "s"}`;
-  if (reason === "no methods/results") return `no article body (${seen})`;
-  if (reason === "title missing") return `neither title nor DOI in text (${seen})`;
-  return reason;
+function parseBook(b) {
+  const d = b.BookDocument ?? {};
+  const book = d.Book ?? {};
+  const ids = [...d.ArticleIdList?.ArticleId ?? [], ...b.PubmedBookData?.ArticleIdList?.ArticleId ?? []];
+  const idOf = (t) => {
+    const hit = ids.find((i) => i["@_IdType"] === t);
+    return hit ? text(hit).trim() || null : null;
+  };
+  const lists = [...d.AuthorList ?? [], ...book.AuthorList ?? []];
+  const pick = lists.find((l) => (l["@_Type"] ?? "authors") === "authors" && (l.Author ?? []).length) ?? lists.find((l) => (l.Author ?? []).length);
+  const people = pick?.Author ?? [];
+  const authors = people.map((a) => a.CollectiveName ? plain(a.CollectiveName) : [plain(a.LastName), plain(a.Initials)].filter(Boolean).join(" ")).filter(Boolean);
+  const first = people[0];
+  const abstract = (d.Abstract?.AbstractText ?? []).map((a) => ({ label: typeof a === "object" ? a["@_Label"] ?? null : null, text: inlineText2(text(a)) })).filter((a) => a.text);
+  const types = ["Book", ...(d.PublicationType ?? []).map(plain).filter(Boolean)];
+  return {
+    pmid: text(d.PMID).trim() || idOf("pubmed") || "",
+    pmcid: null,
+    doi: idOf("doi") ?? (text(book.ELocationID).trim() || null),
+    title: inlineText2(text(d.ArticleTitle)) || inlineText2(text(book.BookTitle)),
+    authors,
+    firstAuthor: first ? plain(first.CollectiveName ?? first.LastName) || null : null,
+    journal: plain(book.Publisher?.PublisherName) || null,
+    year: text(book.PubDate?.Year).trim() || null,
+    publicationTypes: [...new Set(types)],
+    abstract,
+    corrections: [],
+    coi: null,
+    funding: []
+  };
 }
-async function convertBytes(p, bytes, contentType = "") {
-  let format;
-  let md;
-  if (looksLikePdf(bytes)) {
-    format = "pdf";
-    const stats2 = {};
-    try {
-      md = await pdfToMarkdown(bytes, p.title, p.headLine, stats2);
-    } catch (e) {
-      return { ok: false, reason: `pdf: ${errMsg(e)}` };
-    }
-    if ((stats2.pages ?? 0) >= 1 && (stats2.chars ?? 0) < 500 * (stats2.pages ?? 1)) return { ok: false, reason: "scanned pdf", format };
-  } else {
-    const text3 = bytes.toString("utf8");
-    if (isWallPage(text3)) return { ok: false, reason: "wall page" };
-    if (/^\s*<\?xml|<article\b/.test(text3.slice(0, 2e3)) && /<body\b/.test(text3) && !/<html/i.test(text3.slice(0, 2e3))) {
-      format = "xml";
-      const doc = parseJats(/full-text-retrieval-response/.test(text3.slice(0, 3e3)) ? elsevierToJats(text3) : text3);
-      if (!doc.title) doc.title = p.title;
-      if (!doc.hasBody) return { ok: false, reason: "xml without body" };
-      md = renderFulltextMd(doc, p.headLine);
-    } else if (/text\/html|<html/i.test(contentType + text3.slice(0, 500))) {
-      format = "html";
-      const body = /<body[^>]*>([\s\S]*)<\/body>/i.exec(text3)?.[1] ?? text3;
-      md = `# ${p.title}
 
-${p.headLine}
-
-${htmlToMarkdown(body)}`;
-    } else {
-      return { ok: false, reason: `unknown content (${contentType.split(";")[0] || "no type"})` };
-    }
-  }
-  const content = format === "xml" ? md : md.replace(/^# [^\n]*\n/, "");
-  const rawText = format === "html" ? bytes.toString("utf8").replace(/<script[\s\S]*?<\/script>|<style[\s\S]*?<\/style>/gi, " ").replace(/<[^>]+>/g, " ") : void 0;
-  const check = checkFulltext(content, p.title, rawText, p.doi, format === "xml");
-  if (!check.ok) return { ok: false, reason: describe(check.reason, format, content), format };
-  return { ok: true, format, md };
+// src/spl.ts
+import { XMLParser as XMLParser3 } from "fast-xml-parser";
+var DAILYMED_PAGE = (setid) => `https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=${setid}`;
+var SPL_URL = (setid) => `https://dailymed.nlm.nih.gov/dailymed/services/v2/spls/${setid}.xml`;
+async function fetchSpl(setid) {
+  const xml = await fetchText(SPL_URL(setid), { "User-Agent": LIBRARY_UA, Accept: "application/xml, */*;q=0.8" });
+  if (!/<document[\s>]/.test(xml.slice(0, 3e3))) throw new Error("dailymed: not an SPL document");
+  return xml;
 }
-var ELSEVIER_TAGS = [
-  [/ja:article\b/g, "article"],
-  [/ja:body\b/g, "body"],
-  [/ja:head\b/g, "front"],
-  [/ce:sections\b/g, "sec-group"],
-  [/ce:section-title\b/g, "title"],
-  [/ce:section\b/g, "sec"],
-  [/ce:simple-para\b/g, "p"],
-  [/ce:para\b/g, "p"],
-  [/ce:abstract-sec\b/g, "sec"],
-  [/ce:abstract\b/g, "abstract"],
-  [/ce:title\b/g, "article-title"],
-  [/ce:italic\b/g, "italic"],
-  [/ce:bold\b/g, "bold"],
-  [/ce:sup\b/g, "sup"],
-  [/ce:inf\b/g, "sub"],
-  [/ce:cross-refs?\b/g, "xref"],
-  [/ce:list-item\b/g, "list-item"],
-  [/ce:list\b/g, "list"],
-  [/ce:label\b/g, "label"],
-  [/ce:caption\b/g, "caption"],
-  [/ce:table\b/g, "table-wrap"],
-  [/ce:figure\b/g, "fig"],
-  [/ce:bib-reference\b/g, "ref"],
-  [/ce:bibliography\b/g, "ref-list"],
-  [/sb:reference\b/g, "mixed-citation"],
-  [/ce:acknowledgment\b/g, "ack"],
-  [/ce:inter-ref\b/g, "ext-link"]
-];
-function elsevierToJats(xml) {
-  let out = xml;
-  for (const [re, to] of ELSEVIER_TAGS) out = out.replace(re, to);
-  return out.replace(/<\/?full-text-retrieval-response[^>]*>/g, "").replace(/<\/?originalText[^>]*>/g, "").replace(/<\/?xocs:doc[^>]*>/g, "").replace(/<\/?xocs:serial-item[^>]*>/g, "");
-}
-
-// src/tiers/core.ts
-var WINDOW_MS = 6e4;
-var PER_WINDOW = 10;
-var MAX_WAIT_MS = 65e3;
-var TIMEOUT_MS3 = 1e4;
-var calls = [];
-async function pace() {
-  const now = Date.now();
-  while (calls.length && calls[0] <= now - WINDOW_MS) calls.shift();
-  if (calls.length >= PER_WINDOW) await new Promise((r) => setTimeout(r, calls[0] + WINDOW_MS - now + 200));
-  calls.push(Date.now());
+var parser3 = new XMLParser3({ preserveOrder: true, ignoreAttributes: false, attributeNamePrefix: "@_", parseTagValue: false, trimValues: false, processEntities: true, htmlEntities: true });
+var tagOf2 = (n) => Object.keys(n).find((k) => k !== ":@") ?? "";
+var kids2 = (n) => n[tagOf2(n)] ?? [];
+var attr2 = (n, a) => n[":@"]?.[`@_${a}`] ?? "";
+var squash2 = (s) => s.replace(/\s+/g, " ").trim();
+var child = (n, tag) => kids2(n).find((x) => tagOf2(x) === tag);
+function findDeep2(nodes, tag) {
+  for (const n of nodes) {
+    if ("#text" in n) continue;
+    if (tagOf2(n) === tag) return n;
+    const h = findDeep2(kids2(n), tag);
+    if (h) return h;
+  }
+  return void 0;
 }
-async function coreFetch(url, key) {
-  const headers = { "User-Agent": LIBRARY_UA, Authorization: `Bearer ${key}` };
-  for (let attempt = 0; attempt < 2; attempt++) {
-    await pace();
-    const res = await fetch(url, { headers, signal: AbortSignal.timeout(TIMEOUT_MS3), redirect: "follow" });
-    if (res.ok) return await res.text();
-    const body = await res.text().catch(() => "");
-    if (res.status === 429 && attempt === 0) {
-      const until = Date.parse(res.headers.get("x-ratelimit-retry-after") ?? "");
-      const wait = Number.isFinite(until) ? Math.min(Math.max(until - Date.now(), 1e3), MAX_WAIT_MS) : 1e4;
-      await new Promise((r) => setTimeout(r, wait));
+function inline(nodes) {
+  let out = "";
+  for (const n of nodes) {
+    if ("#text" in n) {
+      out += String(n["#text"]);
       continue;
     }
-    throw new Error(res.status < 500 ? httpReason(res.status, body) : `HTTP ${res.status}`);
+    const t = tagOf2(n);
+    const inner = () => inline(kids2(n));
+    switch (t) {
+      case "content": {
+        const style = attr2(n, "styleCode").toLowerCase();
+        const s = inner();
+        out += /bold/.test(style) ? `**${s}**` : /italic/.test(style) ? `*${s}*` : s;
+        break;
+      }
+      case "sup":
+        out += `^${inner()}^`;
+        break;
+      case "sub":
+        out += `~${inner()}~`;
+        break;
+      case "br":
+        out += " ";
+        break;
+      case "footnote":
+        out += ` (${squash2(inner())})`;
+        break;
+      case "footnoteRef":
+      case "renderMultiMedia":
+        break;
+      default:
+        out += inner();
+    }
   }
-  throw new Error("HTTP 429");
+  return out;
 }
-function coreTier(apiKey) {
-  const key = apiKey?.trim();
-  return {
-    name: "core",
-    applies: (p) => p.doi ? true : "no doi",
-    async run(p, ctx) {
-      if (!key) return { ...miss(this.name, "needs api key"), skipped: true };
-      let data;
-      try {
-        const q = encodeURIComponent(`doi:"${p.doi}"`);
-        data = JSON.parse(await coreFetch(`https://api.core.ac.uk/v3/search/works/?q=${q}&limit=5`, key));
-      } catch (e) {
-        return miss(this.name, errMsg(e));
-      }
-      const candidates = coreCandidates(data);
-      if (!candidates.length) return miss(this.name, data.results?.length ? "no full text" : "not in core");
-      ctx.onTry?.(`located ${candidates.length} copies`);
-      return download(this.name, p, candidates, ctx);
+function table(n) {
+  const row3 = (x) => kids2(x).filter((c) => ["td", "th"].includes(tagOf2(c))).map((c) => squash2(inline(kids2(c))).replace(/\|/g, "\\|"));
+  const collect = (nodes, into) => {
+    for (const x of nodes) {
+      if ("#text" in x) continue;
+      if (tagOf2(x) === "tr") into.push(row3(x));
+      else if (!["thead", "tfoot"].includes(tagOf2(x))) collect(kids2(x), into);
     }
   };
+  const head = [], body = [], foot = [];
+  for (const x of kids2(n)) {
+    const t = tagOf2(x);
+    if (t === "thead") collect(kids2(x), head);
+    else if (t === "tfoot") collect(kids2(x), foot);
+    else if (t === "tr") body.push(row3(x));
+    else if (!("#text" in x) && t !== "caption") collect(kids2(x), body);
+  }
+  const rows = [...head, ...body];
+  if (!rows.length) return "";
+  const width = Math.max(...rows.map((r) => r.length));
+  const line = (r) => `| ${[...r, ...Array(width - r.length).fill("")].join(" | ")} |`;
+  const cap = child(n, "caption");
+  const title = cap ? squash2(inline(kids2(cap))) : "";
+  const notes = foot.map((r) => r.filter(Boolean).join(" ")).filter(Boolean);
+  return [title ? `**${title}**` : "", line(rows[0]), `| ${Array(width).fill("---").join(" | ")} |`, ...rows.slice(1).map(line), ...notes.length ? ["", ...notes] : []].filter((l, i) => l || i > 0).join("\n");
 }
-function coreCandidates(d) {
+function blocks(nodes, depth) {
   const out = [];
-  const seen = /* @__PURE__ */ new Set();
-  const add = (url, pdf, host) => {
-    if (!url || seen.has(url)) return;
-    seen.add(url);
-    out.push({ url, version: "published", source: "core", host, pdf });
-  };
-  for (const w of d.results ?? []) {
-    add(w.downloadUrl, true, "core");
-    for (const l of w.links ?? []) if (l.type === "download") add(l.url, /\.pdf($|\?)/i.test(l.url ?? ""), "core");
-    for (const u of w.sourceFulltextUrls ?? []) add(u, /\.pdf($|\?)/i.test(u), "repository");
+  for (const n of nodes) {
+    if ("#text" in n) {
+      const t2 = squash2(String(n["#text"]));
+      if (t2) out.push(t2);
+      continue;
+    }
+    const t = tagOf2(n);
+    const c = kids2(n);
+    switch (t) {
+      case "paragraph": {
+        const s = squash2(inline(c.filter((x) => !["table", "list"].includes(tagOf2(x)))));
+        if (s) out.push(s);
+        out.push(...blocks(c.filter((x) => ["table", "list"].includes(tagOf2(x))), depth));
+        break;
+      }
+      case "list": {
+        const ordered = /^(ordered|numbered|arabic|roman|alpha)/i.test(attr2(n, "listType"));
+        let i = 0;
+        for (const it of c.filter((x) => tagOf2(x) === "item")) {
+          i++;
+          const body = blocks(kids2(it).filter((x) => tagOf2(x) !== "caption"), depth).join(" ");
+          if (body) out.push(`${ordered ? `${i}.` : "-"} ${body}`);
+        }
+        break;
+      }
+      case "table": {
+        const md = table(n);
+        if (md) out.push(md);
+        break;
+      }
+      case "renderMultiMedia":
+      case "observationMedia":
+      case "subject":
+        break;
+      case "section":
+        out.push(...section(n, depth));
+        break;
+      case "component":
+      case "text":
+      case "excerpt":
+      case "highlight":
+      default:
+        out.push(...blocks(c, depth));
+        break;
+    }
   }
   return out;
 }
-
-// src/tiers/europepmc.ts
-var europepmc = {
-  name: "europepmc",
-  applies: (p) => p.pmid ? true : "no pmid",
-  async run(p, ctx) {
-    let data;
-    try {
-      const q = encodeURIComponent(`EXT_ID:${p.pmid} AND SRC:MED`);
-      data = JSON.parse(await fetchText(`https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=${q}&resultType=core&format=json`, { "User-Agent": LIBRARY_UA }));
-    } catch (e) {
-      return miss(this.name, errMsg(e));
-    }
-    const rec = data.resultList?.result?.[0];
-    if (!rec) return miss(this.name, "not in europe pmc");
-    const candidates = europepmcCandidates(rec, p.pmcid);
-    if (!candidates.length) return miss(this.name, "no free copy");
-    ctx.onTry?.(`located ${candidates.length} copies`);
-    return download(this.name, p, candidates, ctx);
-  }
+var SKIP_CODES = /* @__PURE__ */ new Set(["48780-1", "51945-4"]);
+var CODE_NAMES = { "34066-1": "Boxed Warning", "43683-2": "Recent Major Changes" };
+var plain2 = (nodes) => {
+  let out = "";
+  for (const n of nodes) out += "#text" in n ? String(n["#text"]) : tagOf2(n) === "br" ? " " : plain2(kids2(n));
+  return out;
 };
-function europepmcCandidates(r, pmcid) {
+function section(n, depth) {
+  const code = child(n, "code");
+  const c = code ? attr2(code, "code") : "";
+  if (SKIP_CODES.has(c)) return [];
+  const title = child(n, "title");
+  let head = title ? squash2(plain2(kids2(title))) : "";
+  if (c === "34066-1" && head && !/boxed warning/i.test(head)) head = `Boxed Warning: ${head}`;
+  if (!head && CODE_NAMES[c]) head = CODE_NAMES[c];
   const out = [];
-  const seen = /* @__PURE__ */ new Set();
-  const add = (url, pdf, host, version) => {
-    if (!url || seen.has(url)) return;
-    seen.add(url);
-    out.push({ url, version, source: "europepmc", host, pdf });
-  };
-  const id = (r.pmcid ?? pmcid ?? "").toUpperCase() || null;
-  const own = r.isOpenAccess === "Y" ? "published" : "accepted_manuscript";
-  if (id && r.inEPMC === "Y") {
-    if (r.isOpenAccess === "Y") add(`https://www.ebi.ac.uk/europepmc/webservices/rest/${id}/fullTextXML`, false, "europepmc", own);
-    if (r.hasPDF === "Y") add(`https://europepmc.org/articles/${id}?pdf=render`, true, "europepmc", own);
-  }
-  for (const u of r.fullTextUrlList?.fullTextUrl ?? []) {
-    if (!u.url || u.site === "DOI" || !/^(F|OA)$/.test(u.availabilityCode ?? "")) continue;
-    if (u.documentStyle === "abs") continue;
-    if (/^https?:\/\/europepmc\.org\/articles\//i.test(u.url) && u.documentStyle === "html") continue;
-    add(u.url, u.documentStyle === "pdf", u.site?.toLowerCase() ?? "unknown", "published");
-  }
+  if (head) out.push(`${"#".repeat(Math.min(depth, 6))} ${head}`);
+  const rest = kids2(n).filter((x) => !["title", "code", "id", "effectiveTime"].includes(tagOf2(x)));
+  out.push(...blocks(rest, head ? depth + 1 : depth));
   return out;
 }
-
-// src/tiers/openalex.ts
-var openalex = {
-  name: "openalex",
-  applies: (p) => p.pmid || p.doi ? true : "no pmid or doi",
-  async run(p, ctx) {
-    const mail = contactEmail();
-    const key = p.pmid ? `pmid:${p.pmid}` : `doi:${p.doi}`;
-    let data;
-    try {
-      data = JSON.parse(await fetchText(`https://api.openalex.org/works/${encodeURIComponent(key)}?mailto=${encodeURIComponent(mail)}`, { "User-Agent": LIBRARY_UA }));
-    } catch (e) {
-      return miss(this.name, errMsg(e));
-    }
-    const candidates = openalexCandidates(data);
-    if (!candidates.length) return miss(this.name, data.open_access?.is_oa === false ? "closed" : "no oa copy");
-    ctx.onTry?.(`located ${candidates.length} copies`);
-    return download(this.name, p, candidates, ctx);
-  }
-};
-function openalexCandidates(d) {
-  const locs = [...d.best_oa_location ? [d.best_oa_location] : [], ...d.locations ?? []].filter((l) => l.is_oa);
-  const seen = /* @__PURE__ */ new Set();
-  const out = [];
-  for (const l of locs) {
-    for (const [url, pdf] of [[l.pdf_url, true], [l.landing_page_url, false]]) {
-      if (!url || seen.has(url) || /doi\.org\//.test(url)) continue;
-      seen.add(url);
-      out.push({ url, version: oaVersion(l.version), source: "openalex", host: l.source?.display_name ?? "unknown", pdf });
+var isoDate = (v) => /^\d{8}$/.test(v) ? `${v.slice(0, 4)}-${v.slice(4, 6)}-${v.slice(6, 8)}` : v || null;
+function parseSpl(xml) {
+  const doc = parser3.parse(xml);
+  const d = doc.find((x) => tagOf2(x) === "document");
+  if (!d) return null;
+  const k = kids2(d);
+  const setid = attr2(child(d, "setId") ?? {}, "root").toLowerCase();
+  const version = attr2(child(d, "versionNumber") ?? {}, "value") || null;
+  const effective = isoDate(attr2(child(d, "effectiveTime") ?? {}, "value"));
+  const named = (nodes) => {
+    for (const n of nodes) {
+      if ("#text" in n) continue;
+      if (tagOf2(n) === "manufacturedProduct") {
+        const inner2 = child(n, "manufacturedProduct") ?? n;
+        if (child(inner2, "name")) return inner2;
+      }
+      const h = named(kids2(n));
+      if (h) return h;
     }
+    return void 0;
+  };
+  const inner = named(k);
+  const brand = inner ? squash2(inline(kids2(child(inner, "name") ?? { name: [] }))) || null : null;
+  const genericNode = inner ? findDeep2(kids2(inner), "genericMedicine") : void 0;
+  const generic = genericNode ? squash2(inline(kids2(child(genericNode, "name") ?? { name: [] }))) || null : null;
+  const formNode = inner ? child(inner, "formCode") : void 0;
+  const form = formNode ? attr2(formNode, "displayName").toLowerCase() || null : null;
+  const org = findDeep2(k, "representedOrganization");
+  const labeler = org ? squash2(inline(kids2(child(org, "name") ?? { name: [] }))) || null : null;
+  const body = findDeep2(k, "structuredBody");
+  const sections2 = body ? blocks(kids2(body), 2) : [];
+  const docTitle = child(d, "title");
+  const lines = [];
+  let cur = "";
+  for (const x of docTitle ? kids2(docTitle) : []) {
+    if (!("#text" in x) && tagOf2(x) === "br") {
+      lines.push(cur);
+      cur = "";
+    } else cur += "#text" in x ? String(x["#text"]) : inline([x]);
   }
-  return out;
+  lines.push(cur);
+  const highlights = lines.map(squash2).filter((l) => l && !/^these highlights do not include/i.test(l) && !/^see full prescribing information/i.test(l)).map((l) => l.replace(/^see full prescribing information for [^.]+\.\s*/i, "")).filter(Boolean);
+  const approvalYear = /initial u\.?s\.? approval:?\s*(\d{4})/i.exec(highlights.join(" "))?.[1] ?? null;
+  const productLine = highlights.find((l) => !/initial u\.?s\.? approval/i.test(l)) ?? null;
+  const title = [brand, generic ? `(${generic})` : "", form].filter(Boolean).join(" ") || productLine || setid;
+  return { setid, version, effective, brand, generic, form, labeler, title: `${title} \u2014 prescribing information`, highlights, approvalYear, sections: sections2 };
+}
+function headLineSpl(s) {
+  return [s.labeler, `DailyMed set ID ${s.setid}`, s.version ? `version ${s.version}` : "", s.effective ? `effective ${s.effective}` : "", s.approvalYear ? `initial U.S. approval ${s.approvalYear}` : "", DAILYMED_PAGE(s.setid)].filter(Boolean).join(" \xB7 ");
+}
+function renderLabelMd(s) {
+  const parts = [`# ${s.title}`, "", headLineSpl(s), ""];
+  if (s.highlights.length) parts.push("## Highlights", "", ...s.highlights.flatMap((l) => [l, ""]));
+  parts.push(...s.sections.flatMap((b) => [b, ""]));
+  return parts.join("\n").replace(/\n{3,}/g, "\n\n").trimEnd() + "\n";
 }
 
-// src/tiers/pmc.ts
-function pmcEfetchUrl(pmcid) {
-  const u = new URL("https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi");
-  u.searchParams.set("db", "pmc");
-  u.searchParams.set("id", pmcid.replace(/^PMC/i, ""));
-  u.searchParams.set("retmode", "xml");
-  if (process.env.FH_LIBRARY_NCBI_API_KEY) u.searchParams.set("api_key", process.env.FH_LIBRARY_NCBI_API_KEY);
-  u.searchParams.set("tool", "fh-library");
-  u.searchParams.set("email", contactEmail());
-  return u.toString();
+// src/ctgov.ts
+var CTGOV_PAGE = (nctid) => `https://clinicaltrials.gov/study/${nctid}`;
+var API = (nctid) => `https://clinicaltrials.gov/api/v2/studies/${nctid}?format=json`;
+async function fetchStudy(nctid) {
+  const json2 = await fetchText(API(nctid), { "User-Agent": LIBRARY_UA, Accept: "application/json" });
+  if (!/"protocolSection"/.test(json2.slice(0, 2e3))) throw new Error("ctgov: not a study record");
+  return json2;
 }
-var pmc = {
-  name: "pmc",
-  applies: (p) => p.pmcid ? true : "no pmcid",
-  async run(p) {
-    let xml;
-    try {
-      xml = await fetchText(pmcEfetchUrl(p.pmcid));
-    } catch (e) {
-      return miss(this.name, errMsg(e));
-    }
-    const doc = parseJats(xml);
-    if (!doc.hasBody) return miss(this.name, "no body");
-    const md = renderFulltextMd(doc, p.headLine);
-    const check = checkFulltext(md, p.title);
-    if (!check.ok) return miss(this.name, check.reason);
-    return { ok: true, tier: this.name, format: "xml", raw: xml, md, version: "published" };
-  }
+var STATUS = {
+  NOT_YET_RECRUITING: "Not yet recruiting",
+  RECRUITING: "Recruiting",
+  ENROLLING_BY_INVITATION: "Enrolling by invitation",
+  ACTIVE_NOT_RECRUITING: "Active, not recruiting",
+  SUSPENDED: "Suspended",
+  TERMINATED: "Terminated",
+  COMPLETED: "Completed",
+  WITHDRAWN: "Withdrawn",
+  UNKNOWN: "Unknown status",
+  AVAILABLE: "Available",
+  NO_LONGER_AVAILABLE: "No longer available",
+  TEMPORARILY_NOT_AVAILABLE: "Temporarily not available",
+  APPROVED_FOR_MARKETING: "Approved for marketing",
+  WITHHELD: "Withheld"
 };
-
-// src/tiers/tdm.ts
-var CROSSREF = "https://api.crossref.org/works/";
-var TOKENED = [
-  { host: /(^|\.)api\.elsevier\.com$/i, key: "elsevier", env: "ELSEVIER_API_KEY", header: "X-ELS-APIKey" },
-  { host: /(^|\.)api\.wiley\.com$/i, key: "wiley", env: "WILEY_TDM_TOKEN", header: "Wiley-TDM-Client-Token" }
-];
-var hostOf3 = (url) => {
+var PHASE = { EARLY_PHASE1: "Early Phase 1", PHASE1: "Phase 1", PHASE2: "Phase 2", PHASE3: "Phase 3", PHASE4: "Phase 4", NA: "Not applicable" };
+var words = (v) => {
+  const s = String(v ?? "");
+  if (!s) return "";
+  if (STATUS[s]) return STATUS[s];
+  if (PHASE[s]) return PHASE[s];
+  const w = s.toLowerCase().replace(/_/g, " ");
+  return w.charAt(0).toUpperCase() + w.slice(1);
+};
+var dated = (d) => d?.date ? `${d.date}${d.type ? ` (${String(d.type).toLowerCase()})` : ""}` : "";
+var list = (a) => Array.isArray(a) ? a : [];
+var row = (cells) => `| ${cells.map((c) => c.replace(/\|/g, "\\|").replace(/\s+/g, " ").trim()).join(" | ")} |`;
+var table2 = (head, rows) => rows.length ? [row(head), `| ${head.map(() => "---").join(" | ")} |`, ...rows.map(row)].join("\n") : "";
+var para = (s) => String(s ?? "").replace(/\r/g, "").trim();
+function parseStudy(json2) {
+  let d;
   try {
-    return new URL(url).hostname;
+    d = JSON.parse(json2);
   } catch {
-    return "";
-  }
-};
-function tdmCandidates(work, tokens2 = {}, env = process.env) {
-  const links = (work.message?.link ?? []).filter((l) => l["intended-application"] === "text-mining" && l.URL);
-  const seen = /* @__PURE__ */ new Set();
-  const candidates = [];
-  const missingTokens = /* @__PURE__ */ new Set();
-  for (const l of links) {
-    const ct = (l["content-type"] ?? "").toLowerCase();
-    if (seen.has(l.URL)) {
-      const prev = candidates.find((c) => c.url === l.URL);
-      if (prev && /pdf/.test(ct)) prev.pdf = true;
-      continue;
-    }
-    seen.add(l.URL);
-    const rule = TOKENED.find((r) => r.host.test(hostOf3(l.URL)));
-    let headers;
-    if (rule) {
-      const token = tokens2[rule.key]?.trim() || env[rule.env]?.trim();
-      if (!token) {
-        missingTokens.add(rule.key);
-        continue;
-      }
-      headers = { [rule.header]: token };
-    }
-    if (/text\/plain/.test(ct)) continue;
-    candidates.push({ url: l.URL, version: "published", source: "tdm", host: work.message?.publisher ?? hostOf3(l.URL), pdf: /pdf/.test(ct) || /\.pdf(\?|$)/i.test(l.URL), headers });
-  }
-  const rank = (c) => /\/xml|text\/xml/.test(c.url) ? 0 : c.pdf ? 1 : 2;
-  candidates.sort((a, b) => rank(a) - rank(b));
-  return { candidates, missingTokens: [...missingTokens] };
-}
-function tdmTier(tokens2 = {}) {
-  return {
-    name: "tdm",
-    applies: (p) => p.doi ? true : "no doi",
-    async run(p, ctx) {
-      let work;
-      try {
-        work = JSON.parse(await fetchText(CROSSREF + encodeURIComponent(p.doi), { "User-Agent": `${LIBRARY_UA} (mailto:${contactEmail()})` }));
-      } catch (e) {
-        return miss(this.name, errMsg(e));
-      }
-      const { candidates, missingTokens } = tdmCandidates(work, tokens2);
-      if (!candidates.length) return missingTokens.length ? { ...miss(this.name, `needs ${missingTokens.join("/")} token`), skipped: true } : miss(this.name, "no text-mining links");
-      ctx.onTry?.(`${candidates.length} text-mining links${missingTokens.length ? ` (skipped hosts needing ${missingTokens.join("/")} token)` : ""}`);
-      return download(this.name, p, candidates, ctx);
-    }
-  };
-}
-var tdm = tdmTier();
-
-// src/tiers/unpaywall.ts
-var unpaywall = {
-  name: "unpaywall",
-  applies: (p) => p.doi ? true : "no doi",
-  async run(p, ctx) {
-    const mail = contactEmail();
-    let data;
-    try {
-      data = JSON.parse(await fetchText(`https://api.unpaywall.org/v2/${encodeURIComponent(p.doi)}?email=${encodeURIComponent(mail)}`, { "User-Agent": LIBRARY_UA }));
-    } catch (e) {
-      return miss(this.name, errMsg(e));
-    }
-    const candidates = unpaywallCandidates(data);
-    if (!candidates.length) return miss(this.name, data.is_oa === false ? "closed" : "no oa copy");
-    ctx.onTry?.(`located ${candidates.length} copies`);
-    return download(this.name, p, candidates, ctx);
+    return null;
   }
-};
-function unpaywallCandidates(d) {
-  const locs = [...d.best_oa_location ? [d.best_oa_location] : [], ...d.oa_locations ?? []];
-  const seen = /* @__PURE__ */ new Set();
+  const ps = d?.protocolSection;
+  if (!ps) return null;
+  const idm = ps.identificationModule ?? {}, st = ps.statusModule ?? {}, sp = ps.sponsorCollaboratorsModule ?? {}, de = ps.descriptionModule ?? {};
+  const dm = ps.designModule ?? {}, ai = ps.armsInterventionsModule ?? {}, om = ps.outcomesModule ?? {}, el = ps.eligibilityModule ?? {}, cl = ps.contactsLocationsModule ?? {}, rf = ps.referencesModule ?? {};
+  const nctid = String(idm.nctId ?? "").toUpperCase();
   const out = [];
-  for (const l of locs) {
-    for (const [url, pdf] of [[l.url_for_pdf, true], [l.url ?? l.url_for_landing_page, false]]) {
-      if (!url || seen.has(url)) continue;
-      seen.add(url);
-      out.push({ url, version: oaVersion(l.version), source: "unpaywall", host: l.host_type ?? "unknown", pdf });
-    }
-  }
-  return out;
-}
-
-// src/tiers/index.ts
-var BUILTIN = [
-  { name: "pmc", make: () => pmc },
-  { name: "europepmc", make: () => europepmc },
-  { name: "tdm", make: (o) => tdmTier(o.tokens ?? {}) },
-  { name: "unpaywall", make: () => unpaywall },
-  { name: "openalex", make: () => openalex },
-  { name: "core", make: (o) => coreTier(o.api_key) }
-];
-
-// src/tiers/url.ts
-var PLACEHOLDERS = ["doi", "pmid", "pmcid"];
-function urlTier(name, template, headers) {
-  const used = PLACEHOLDERS.filter((k) => template.includes(`{${k}}`));
-  return {
-    name,
-    applies(p) {
-      const missing = used.filter((k) => !p[k]);
-      return missing.length ? `no ${missing.join("/")}` : true;
-    },
-    async run(p, ctx) {
-      if (!used.length) return miss(name, "template has no placeholder");
-      const url = used.reduce((u, k) => u.replaceAll(`{${k}}`, encodeURIComponent(p[k])), template);
-      return download(name, p, [{ url, version: "published", source: name, host: "user", pdf: /\.pdf(\?|$)/i.test(url), headers, service: true }], ctx);
-    }
+  const sec = (title, lines) => {
+    const body = lines.filter((l) => !!l && l.trim() !== "");
+    if (body.length) out.push(`## ${title}`, ...body);
   };
-}
-
-// src/config.ts
-var VARIABLES = [
-  "FH_LIBRARY_DIR",
-  "FH_LIBRARY_ENV",
-  "FH_LIBRARY_TIERS",
-  "FH_LIBRARY_CONTACT_EMAIL",
-  "FH_LIBRARY_ELSEVIER_API_KEY",
-  "FH_LIBRARY_WILEY_TDM_TOKEN",
-  "FH_LIBRARY_CORE_API_KEY",
-  "FH_LIBRARY_PROXY_URL",
-  "FH_LIBRARY_PROXY_HEADERS",
-  "FH_LIBRARY_PLAYWRIGHT_1",
-  "FH_LIBRARY_PLAYWRIGHT_2",
-  "FH_LIBRARY_CHROME_1",
-  "FH_LIBRARY_PLAYWRIGHT_HEADLESS",
-  "FH_LIBRARY_VERIFY_URL",
-  "FH_LIBRARY_VERIFY_API_KEY",
-  "FH_LIBRARY_VERIFY_MODEL",
-  "FH_LIBRARY_VERIFY_THRESHOLD",
-  "FH_LIBRARY_VERIFY_TIMEOUT_SECS",
-  "FH_LIBRARY_RETRY_AFTER_DAYS",
-  "FH_LIBRARY_RETRY_BLOCKED",
-  "FH_LIBRARY_REFRESH_AFTER_DAYS",
-  "FH_LIBRARY_SEARCH_CACHE_DAYS",
-  "FH_LIBRARY_NCBI_API_KEY",
-  "FH_LIBRARY_BRIGHTDATA_API_TOKEN",
-  "FH_LIBRARY_BRIGHTDATA_ZONE"
-];
-var sources = (env = process.env) => Object.fromEntries(VARIABLES.map((v) => [v, sourceOf(v, env)]));
-function settings(env = process.env) {
+  const secondary = list(idm.secondaryIdInfos).map((s) => `${words(s.type)}${s.domain ? ` (${s.domain})` : ""}: ${s.id}`);
+  sec("Identification", [
+    idm.officialTitle ? `Official title: ${para(idm.officialTitle)}` : null,
+    idm.acronym ? `Acronym: ${idm.acronym}` : null,
+    idm.organization?.fullName ? `Organization: ${idm.organization.fullName}` : null,
+    secondary.length ? `Secondary IDs: ${secondary.join("; ")}` : null
+  ]);
+  sec("Status", [
+    `Overall status: ${words(st.overallStatus)}${st.whyStopped ? ` \u2014 why stopped: ${para(st.whyStopped)}` : ""}`,
+    st.startDateStruct ? `Start: ${dated(st.startDateStruct)}` : null,
+    st.primaryCompletionDateStruct ? `Primary completion: ${dated(st.primaryCompletionDateStruct)}` : null,
+    st.completionDateStruct ? `Completion: ${dated(st.completionDateStruct)}` : null,
+    st.studyFirstPostDateStruct ? `First posted: ${dated(st.studyFirstPostDateStruct)}` : null,
+    st.lastUpdatePostDateStruct ? `Last update posted: ${dated(st.lastUpdatePostDateStruct)}` : null,
+    st.statusVerifiedDate ? `Status verified: ${st.statusVerifiedDate}` : null,
+    `Results posted: ${d.hasResults ? "yes" : "no"}`
+  ]);
+  const rp = sp.responsibleParty;
+  sec("Sponsor and collaborators", [
+    sp.leadSponsor?.name ? `Lead sponsor: ${sp.leadSponsor.name}${sp.leadSponsor.class ? ` (${words(sp.leadSponsor.class).toLowerCase()})` : ""}` : null,
+    list(sp.collaborators).length ? `Collaborators: ${list(sp.collaborators).map((c) => c.name).join("; ")}` : null,
+    rp ? `Responsible party: ${[words(rp.type).toLowerCase(), rp.investigatorFullName, rp.investigatorTitle, rp.investigatorAffiliation].filter(Boolean).join(", ")}` : null
+  ]);
+  sec("Description", [de.briefSummary ? para(de.briefSummary) : null, de.detailedDescription ? para(de.detailedDescription) : null]);
+  sec("Conditions", [list(ps.conditionsModule?.conditions).length ? list(ps.conditionsModule.conditions).join("; ") : null, list(ps.conditionsModule?.keywords).length ? `Keywords: ${list(ps.conditionsModule.keywords).join("; ")}` : null]);
+  const di = dm.designInfo ?? {};
+  const mi = di.maskingInfo ?? {};
+  sec("Design", [
+    dm.studyType ? `Study type: ${words(dm.studyType)}` : null,
+    list(dm.phases).length ? `Phase: ${list(dm.phases).map(words).join(", ")}` : null,
+    di.allocation ? `Allocation: ${words(di.allocation)}` : null,
+    di.interventionModel ? `Intervention model: ${words(di.interventionModel)}${di.interventionModelDescription ? ` \u2014 ${para(di.interventionModelDescription)}` : ""}` : null,
+    di.primaryPurpose ? `Primary purpose: ${words(di.primaryPurpose)}` : null,
+    mi.masking ? `Masking: ${words(mi.masking)}${list(mi.whoMasked).length ? ` (${list(mi.whoMasked).map((w) => words(w).toLowerCase()).join(", ")})` : ""}${mi.maskingDescription ? ` \u2014 ${para(mi.maskingDescription)}` : ""}` : null,
+    dm.enrollmentInfo?.count !== void 0 ? `Enrollment: ${dm.enrollmentInfo.count}${dm.enrollmentInfo.type ? ` (${String(dm.enrollmentInfo.type).toLowerCase()})` : ""}` : null
+  ]);
+  const arms = list(ai.armGroups).map((a) => `- **${a.label}**${a.type ? ` (${words(a.type).toLowerCase()})` : ""}${a.description ? `: ${para(a.description)}` : ""}${list(a.interventionNames).length ? ` \u2014 ${list(a.interventionNames).join("; ")}` : ""}`);
+  const ints = list(ai.interventions).map((i) => `- **${words(i.type)}: ${i.name}**${list(i.otherNames).length ? ` (${list(i.otherNames).join(", ")})` : ""}${i.description ? `: ${para(i.description)}` : ""}${list(i.armGroupLabels).length ? ` \u2014 arms: ${list(i.armGroupLabels).join("; ")}` : ""}`);
+  sec("Arms and interventions", [arms.length ? "Arms:" : null, ...arms, ints.length ? "Interventions:" : null, ...ints]);
+  const outcome = (o) => `- **${para(o.measure)}**${o.description ? `: ${para(o.description)}` : ""}${o.timeFrame ? ` Time frame: ${para(o.timeFrame)}.` : ""}`;
+  sec("Outcomes", [
+    list(om.primaryOutcomes).length ? "### Primary outcomes" : null,
+    ...list(om.primaryOutcomes).map(outcome),
+    list(om.secondaryOutcomes).length ? "### Secondary outcomes" : null,
+    ...list(om.secondaryOutcomes).map(outcome),
+    list(om.otherOutcomes).length ? "### Other outcomes" : null,
+    ...list(om.otherOutcomes).map(outcome)
+  ]);
+  sec("Eligibility", [
+    [el.sex ? `Sex: ${words(el.sex).toLowerCase()}` : "", el.minimumAge || el.maximumAge ? `Age: ${el.minimumAge ?? "no minimum"} to ${el.maximumAge ?? "no maximum"}` : "", el.healthyVolunteers !== void 0 ? `Healthy volunteers: ${el.healthyVolunteers ? "yes" : "no"}` : ""].filter(Boolean).join(" \xB7 "),
+    el.eligibilityCriteria ? para(el.eligibilityCriteria) : null
+  ]);
+  const countries = [...new Set(list(cl.locations).map((l) => l.country).filter(Boolean))];
+  sec("Locations", [countries.length ? `${list(cl.locations).length} sites in ${countries.join(", ")}` : null, list(cl.overallOfficials).length ? `Overall officials: ${list(cl.overallOfficials).map((o) => [o.name, o.role ? words(o.role).toLowerCase() : "", o.affiliation].filter(Boolean).join(", ")).join("; ")}` : null]);
+  const refs = list(rf.references);
+  sec("References", refs.map((r) => `- ${para(r.citation)}${r.pmid ? ` pmid:${r.pmid}` : ""}${r.type ? ` (${words(r.type).toLowerCase()})` : ""}`));
+  const publications = refs.filter((r) => r.pmid && ["RESULT", "DERIVED"].includes(String(r.type))).map((r) => `pmid:${r.pmid}`);
+  if (d.hasResults && d.resultsSection) out.push(...results(d.resultsSection));
   return {
-    retry_after_days: num("FH_LIBRARY_RETRY_AFTER_DAYS", 1, env),
-    retry_blocked: bool("FH_LIBRARY_RETRY_BLOCKED", false, env),
-    refresh_after_days: num("FH_LIBRARY_REFRESH_AFTER_DAYS", 30, env),
-    search_cache_days: num("FH_LIBRARY_SEARCH_CACHE_DAYS", 7, env)
+    nctid,
+    title: para(idm.briefTitle) || nctid,
+    acronym: idm.acronym ?? null,
+    sponsor: sp.leadSponsor?.name ?? null,
+    firstPosted: st.studyFirstPostDateStruct?.date ?? null,
+    lastUpdate: st.lastUpdatePostDateStruct?.date ?? null,
+    status: st.overallStatus ? words(st.overallStatus) : null,
+    hasResults: !!d.hasResults,
+    publications,
+    sections: out
   };
 }
-function proxyHeaders(spec) {
-  const out = {};
-  for (const part of spec.split(";")) {
-    const i = part.indexOf(":");
-    if (i < 0) continue;
-    const name = part.slice(0, i).trim(), value = part.slice(i + 1).trim();
-    if (name && value) out[name] = value;
+function results(rs) {
+  const out = ["## Results"];
+  const pf = rs.participantFlowModule;
+  if (pf) {
+    out.push("### Participant flow");
+    const groups = list(pf.groups);
+    out.push(...groups.map((g) => `- **${g.title}**${g.description ? `: ${para(g.description)}` : ""}`));
+    if (pf.recruitmentDetails) out.push(para(pf.recruitmentDetails));
+    for (const p of list(pf.periods)) {
+      const head = ["Period: " + (p.title ?? ""), ...groups.map((g) => g.title)];
+      const rows = list(p.milestones).map((m) => [words(m.type), ...groups.map((g) => list(m.achievements).find((a) => a.groupId === g.id)?.numSubjects ?? "")]);
+      for (const dw of list(p.dropWithdraws)) rows.push([`Not completed: ${dw.type}`, ...groups.map((g) => list(dw.reasons).find((r) => r.groupId === g.id)?.numSubjects ?? "")]);
+      out.push(table2(head, rows));
+    }
   }
-  return out;
-}
-function resolveTiers(env = process.env) {
-  const problems = [];
-  const available = /* @__PURE__ */ new Map();
-  const tokens2 = { elsevier: str("FH_LIBRARY_ELSEVIER_API_KEY", env) || void 0, wiley: str("FH_LIBRARY_WILEY_TDM_TOKEN", env) || void 0 };
-  for (const t of BUILTIN) available.set(t.name, t.make({ tokens: tokens2, api_key: str("FH_LIBRARY_CORE_API_KEY", env) || void 0 }));
-  const proxyUrl = str("FH_LIBRARY_PROXY_URL", env);
-  if (proxyUrl) available.set("proxy", urlTier("proxy", proxyUrl, proxyHeaders(str("FH_LIBRARY_PROXY_HEADERS", env))));
-  const listed = str("FH_LIBRARY_TIERS", env);
-  const order2 = listed ? listed.split(",").map((s) => s.trim()).filter(Boolean) : [...available.keys()];
-  const tiers = [], seen = /* @__PURE__ */ new Set();
-  for (const name of order2) {
-    if (seen.has(name)) continue;
-    seen.add(name);
-    const t = available.get(name);
-    if (t) tiers.push(t);
-    else if (name === "proxy") problems.push('FH_LIBRARY_TIERS: "proxy" listed, but FH_LIBRARY_PROXY_URL is unset');
-    else problems.push(`FH_LIBRARY_TIERS: unknown tier "${name}" (built-in: ${BUILTIN.map((t2) => t2.name).join(", ")}; proxy when FH_LIBRARY_PROXY_URL is set)`);
+  const bc = rs.baselineCharacteristicsModule;
+  if (bc) {
+    out.push("### Baseline characteristics");
+    const groups = list(bc.groups);
+    out.push(...groups.map((g) => `- **${g.title}**${g.description ? `: ${para(g.description)}` : ""}`));
+    for (const m of list(bc.measures)) {
+      const unit = [m.paramType ? words(m.paramType).toLowerCase() : "", m.dispersionType ? words(m.dispersionType).toLowerCase() : "", m.unitOfMeasure].filter(Boolean).join(", ");
+      const rows = [];
+      for (const c of list(m.classes)) for (const cat of list(c.categories)) rows.push([[c.title, cat.title].filter(Boolean).join(" \u2014 ") || m.title, ...groups.map((g) => {
+        const x = list(cat.measurements).find((v) => v.groupId === g.id);
+        return x ? `${x.value ?? ""}${x.spread !== void 0 ? ` \xB1 ${x.spread}` : ""}${x.lowerLimit !== void 0 ? ` (${x.lowerLimit} to ${x.upperLimit})` : ""}` : "";
+      })]);
+      out.push(`**${m.title}**${unit ? ` (${unit})` : ""}`, table2(["", ...groups.map((g) => g.title)], rows));
+    }
   }
-  const all = [...tiers.map((t) => ({ name: t.name, enabled: true })), ...[...available.keys()].filter((n) => !seen.has(n)).map((name) => ({ name, enabled: false }))];
-  return { tiers, all, source: sourceOf("FH_LIBRARY_TIERS", env), problems };
-}
-
-// src/log.ts
-import { appendFileSync, mkdirSync as mkdirSync2, renameSync, statSync } from "fs";
-import { hostname } from "os";
-import { join as join4 } from "path";
-var HOST = hostname();
-var LOG_FILE = join4(FH_LIBRARY_DIR, "library.log");
-function stamp() {
-  const d = /* @__PURE__ */ new Date();
-  const p = (n) => String(n).padStart(2, "0");
-  return `${d.getFullYear()}-${p(d.getMonth() + 1)}-${p(d.getDate())} ${p(d.getHours())}:${p(d.getMinutes())}:${p(d.getSeconds())}`;
-}
-var LOG_MAX_BYTES = 50 * 1024 * 1024;
-function rotateIfLarge(file2, max = LOG_MAX_BYTES) {
-  let size = 0;
-  try {
-    size = statSync(file2).size;
-  } catch {
-    return false;
+  const om = rs.outcomeMeasuresModule;
+  if (om) {
+    out.push("### Outcome measures");
+    for (const m of list(om.outcomeMeasures)) {
+      const groups = list(m.groups);
+      const unit = [m.paramType ? words(m.paramType).toLowerCase() : "", m.dispersionType ? String(m.dispersionType).toLowerCase() : "", m.unitOfMeasure].filter(Boolean).join(", ");
+      out.push(`#### ${m.title}${m.type ? ` (${words(m.type).toLowerCase()})` : ""}`);
+      if (m.description) out.push(para(m.description));
+      out.push([m.timeFrame ? `Time frame: ${para(m.timeFrame)}.` : "", unit ? `Measure: ${unit}.` : "", m.populationDescription ? `Population: ${para(m.populationDescription)}` : ""].filter(Boolean).join(" "));
+      if (m.reportingStatus && m.reportingStatus !== "POSTED") out.push(`Reporting status: ${words(m.reportingStatus)}.`);
+      const rows = [];
+      for (const c of list(m.classes)) for (const cat of list(c.categories)) rows.push([[c.title, cat.title].filter(Boolean).join(" \u2014 ") || m.title, ...groups.map((g) => {
+        const x = list(cat.measurements).find((v) => v.groupId === g.id);
+        return x ? `${x.value ?? ""}${x.spread !== void 0 ? ` \xB1 ${x.spread}` : ""}${x.lowerLimit !== void 0 ? ` (${x.lowerLimit} to ${x.upperLimit})` : ""}` : "";
+      })]);
+      const denoms = list(m.denoms).flatMap((dn) => list(dn.counts)).filter((c) => c.groupId);
+      if (denoms.length) rows.unshift(["Participants analysed", ...groups.map((g) => denoms.find((c) => c.groupId === g.id)?.value ?? "")]);
+      out.push(table2(["", ...groups.map((g) => g.title)], rows));
+      for (const a of list(m.analyses)) out.push(`Analysis${a.groupDescription ? ` (${para(a.groupDescription)})` : ""}: ${[a.statisticalMethod, a.pValue !== void 0 ? `p ${a.pValueComment ?? ""}${a.pValue}`.trim() : "", a.paramType ? `${words(a.paramType).toLowerCase()} ${a.paramValue ?? ""}` : "", a.ciPctValue ? `${a.ciPctValue}% CI ${a.ciLowerLimit ?? ""} to ${a.ciUpperLimit ?? ""}` : ""].filter(Boolean).join("; ")}`);
+    }
   }
-  if (size <= max) return false;
-  try {
-    renameSync(file2, `${file2}.1`);
-    return true;
-  } catch {
-    return false;
+  const ae = rs.adverseEventsModule;
+  if (ae) {
+    out.push("### Adverse events");
+    const groups = list(ae.eventGroups);
+    out.push([ae.timeFrame ? `Time frame: ${para(ae.timeFrame)}.` : "", ae.frequencyThreshold ? `Threshold for other events: ${ae.frequencyThreshold}%.` : "", ae.description ? para(ae.description) : ""].filter(Boolean).join(" "));
+    const summary = [
+      ["Serious events, participants affected / at risk", ...groups.map((g) => `${g.seriousNumAffected ?? ""} / ${g.seriousNumAtRisk ?? ""}`)],
+      ["Other events, participants affected / at risk", ...groups.map((g) => `${g.otherNumAffected ?? ""} / ${g.otherNumAtRisk ?? ""}`)],
+      ["Deaths, participants affected / at risk", ...groups.map((g) => g.deathsNumAffected !== void 0 ? `${g.deathsNumAffected} / ${g.deathsNumAtRisk ?? ""}` : "")]
+    ].filter((r) => r.slice(1).some(Boolean));
+    out.push(table2(["", ...groups.map((g) => g.title)], summary));
+    const events = (title, evs) => {
+      if (!evs.length) return;
+      out.push(`**${title}** (participants affected / at risk)`, table2(["Event", "System", ...groups.map((g) => g.title)], evs.map((e) => [e.term ?? "", e.organSystem ?? "", ...groups.map((g) => {
+        const s = list(e.stats).find((x) => x.groupId === g.id);
+        return s ? `${s.numAffected ?? ""} / ${s.numAtRisk ?? ""}` : "";
+      })])));
+    };
+    events("Serious adverse events", list(ae.seriousEvents));
+    events("Other adverse events", list(ae.otherEvents));
   }
+  return out;
 }
-function log(tool, id, message) {
-  const clean2 = message.replace(/\x1b\[[0-9;]*m/g, "").replace(/\s*\n\s*/g, " | ");
-  const line = `${stamp()} ${HOST} ${tool.replace(/^library_/, "")} ${id ?? "-"} ${clean2}
-`;
-  try {
-    mkdirSync2(FH_LIBRARY_DIR, { recursive: true });
-    rotateIfLarge(LOG_FILE);
-    appendFileSync(LOG_FILE, line);
-  } catch {
-  }
+function headLineStudy(s) {
+  return [s.sponsor, s.nctid, s.status ? s.status.toLowerCase() : "", s.firstPosted ? `first posted ${s.firstPosted}` : "", s.lastUpdate ? `last update posted ${s.lastUpdate}` : "", CTGOV_PAGE(s.nctid)].filter(Boolean).join(" \xB7 ");
+}
+function renderStudyMd(s) {
+  const parts = [`# ${s.title}${s.acronym ? ` (${s.acronym})` : ""} \u2014 ClinicalTrials.gov record`, "", headLineStudy(s), "", ...s.sections.flatMap((b) => [b, ""])];
+  return parts.join("\n").replace(/\n{3,}/g, "\n\n").trimEnd() + "\n";
 }
-
-// src/stats.ts
-import { mkdirSync as mkdirSync4, readdirSync as readdirSync2, readFileSync as readFileSync4, statSync as statSync2, unlinkSync as unlinkSync2 } from "fs";
-import { join as join6 } from "path";
 
 // src/entry.ts
-import { copyFileSync, existsSync as existsSync4, mkdirSync as mkdirSync3, readdirSync, readFileSync as readFileSync3, renameSync as renameSync2, rmSync, unlinkSync, writeFileSync } from "fs";
-import { join as join5 } from "path";
-import { basename, dirname } from "path";
-
-// src/pubmed.ts
-import { XMLParser as XMLParser2 } from "fast-xml-parser";
-var CORRECTION_TYPES = {
-  ErratumIn: "erratum_in",
-  RetractionIn: "retraction_in",
-  PartialRetractionIn: "partial_retraction_in",
-  RetractedandRepublishedIn: "retracted_and_republished_in",
-  ExpressionOfConcernIn: "expression_of_concern_in",
-  UpdateIn: "update_in"
-};
-var RETRACTION_TYPES = /* @__PURE__ */ new Set(["retraction_in", "partial_retraction_in", "retracted_and_republished_in"]);
-var isRetractionNotice = (c) => RETRACTION_TYPES.has(c.type);
-var EFETCH = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi";
-function efetchUrl(pmid) {
-  const u = new URL(EFETCH);
-  u.searchParams.set("db", "pubmed");
-  u.searchParams.set("id", pmid);
-  u.searchParams.set("retmode", "xml");
-  if (process.env.FH_LIBRARY_NCBI_API_KEY) u.searchParams.set("api_key", process.env.FH_LIBRARY_NCBI_API_KEY);
-  u.searchParams.set("tool", "fh-library");
-  u.searchParams.set("email", contactEmail());
-  return u.toString();
-}
-async function fetchPubmedXml(pmid) {
-  try {
-    return await fetchText(efetchUrl(pmid));
-  } catch (e) {
-    throw new Error(`pubmed fetch failed: ${e instanceof Error ? e.message : String(e)}`);
-  }
-}
-var parser2 = new XMLParser2({
-  ignoreAttributes: false,
-  attributeNamePrefix: "@_",
-  parseTagValue: false,
-  processEntities: true,
-  stopNodes: ["*.ArticleTitle", "*.AbstractText", "*.CoiStatement"],
-  isArray: (name) => ["PubmedArticle", "PubmedBookArticle", "AbstractText", "Author", "ArticleId", "PublicationType", "CommentsCorrections", "Grant", "AuthorList"].includes(name)
-});
-function text(n) {
-  if (n == null) return "";
-  if (typeof n === "object") return text(n["#text"]);
-  return String(n);
-}
-var plain = (n) => decodeEntities(text(n)).trim();
-function inlineText2(raw) {
-  return decodeEntities(raw.replace(/<[^>]+>/g, "")).replace(/\s+/g, " ").trim();
-}
-function decodeEntities(s) {
-  return s.replace(/&#x([0-9a-f]+);/gi, (_, h) => String.fromCodePoint(parseInt(h, 16))).replace(/&#(\d+);/g, (_, d) => String.fromCodePoint(parseInt(d, 10))).replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&quot;/g, '"').replace(/&apos;/g, "'").replace(/&amp;/g, "&");
-}
-function parsePubmedXml(xml) {
-  const doc = parser2.parse(xml);
-  const article = doc?.PubmedArticleSet?.PubmedArticle?.[0];
-  if (!article) {
-    const book = doc?.PubmedArticleSet?.PubmedBookArticle?.[0];
-    return book ? parseBook(book) : null;
-  }
-  const cit = article.MedlineCitation ?? {};
-  const art = cit.Article ?? {};
-  const ids = article.PubmedData?.ArticleIdList?.ArticleId ?? [];
-  const idOf = (t) => {
-    const hit = ids.find((i) => i["@_IdType"] === t);
-    return hit ? text(hit).trim() || null : null;
-  };
-  const authorList = art.AuthorList?.[0]?.Author ?? [];
-  const authors = authorList.map((a) => {
-    if (a.CollectiveName) return plain(a.CollectiveName);
-    const last = plain(a.LastName);
-    const ini = plain(a.Initials);
-    return [last, ini].filter(Boolean).join(" ");
-  }).filter(Boolean);
-  const first = authorList[0];
-  const firstAuthor = first ? plain(first.CollectiveName ?? first.LastName) || null : null;
-  const pub = art.Journal?.JournalIssue?.PubDate ?? {};
-  let year = text(pub.Year).trim() || null;
-  if (!year) year = /\d{4}/.exec(text(pub.MedlineDate))?.[0] ?? null;
-  const abstract = (art.Abstract?.AbstractText ?? []).map((a) => ({
-    label: typeof a === "object" ? a["@_Label"] ?? null : null,
-    text: inlineText2(text(a))
-  })).filter((a) => a.text);
-  const corrections = (cit.CommentsCorrectionsList?.CommentsCorrections ?? []).flatMap((c) => {
-    const type = CORRECTION_TYPES[String(c["@_RefType"] ?? "")];
-    if (!type) return [];
-    return [{ type, pmid: text(c.PMID).trim() || null, source: inlineText2(text(c.RefSource)) }];
-  });
-  const funding = [];
-  for (const g of art.GrantList?.Grant ?? []) {
-    const a = plain(g.Agency);
-    if (a && !funding.includes(a)) funding.push(a);
-  }
-  return {
-    coi: inlineText2(text(cit.CoiStatement)) || null,
-    funding,
-    pmid: text(cit.PMID).trim() || idOf("pubmed") || "",
-    pmcid: idOf("pmc"),
-    doi: idOf("doi"),
-    title: inlineText2(text(art.ArticleTitle)),
-    authors,
-    firstAuthor,
-    journal: plain(art.Journal?.Title) || null,
-    year,
-    publicationTypes: (art.PublicationTypeList?.PublicationType ?? []).map(plain).filter(Boolean),
-    abstract,
-    corrections
-  };
-}
-function parseBook(b) {
-  const d = b.BookDocument ?? {};
-  const book = d.Book ?? {};
-  const ids = [...d.ArticleIdList?.ArticleId ?? [], ...b.PubmedBookData?.ArticleIdList?.ArticleId ?? []];
-  const idOf = (t) => {
-    const hit = ids.find((i) => i["@_IdType"] === t);
-    return hit ? text(hit).trim() || null : null;
-  };
-  const lists = [...d.AuthorList ?? [], ...book.AuthorList ?? []];
-  const pick = lists.find((l) => (l["@_Type"] ?? "authors") === "authors" && (l.Author ?? []).length) ?? lists.find((l) => (l.Author ?? []).length);
-  const people = pick?.Author ?? [];
-  const authors = people.map((a) => a.CollectiveName ? plain(a.CollectiveName) : [plain(a.LastName), plain(a.Initials)].filter(Boolean).join(" ")).filter(Boolean);
-  const first = people[0];
-  const abstract = (d.Abstract?.AbstractText ?? []).map((a) => ({ label: typeof a === "object" ? a["@_Label"] ?? null : null, text: inlineText2(text(a)) })).filter((a) => a.text);
-  const types = ["Book", ...(d.PublicationType ?? []).map(plain).filter(Boolean)];
+function emptyStatus(requested, type, id, libraryDir) {
   return {
-    pmid: text(d.PMID).trim() || idOf("pubmed") || "",
+    requested,
+    type,
+    id,
+    entry: null,
+    entry_dir: null,
+    library_dir: libraryDir,
+    pmid: null,
+    setid: null,
+    nctid: null,
+    publications: [],
     pmcid: null,
-    doi: idOf("doi") ?? (text(book.ELocationID).trim() || null),
-    title: inlineText2(text(d.ArticleTitle)) || inlineText2(text(book.BookTitle)),
-    authors,
-    firstAuthor: first ? plain(first.CollectiveName ?? first.LastName) || null : null,
-    journal: plain(book.Publisher?.PublisherName) || null,
-    year: text(book.PubDate?.Year).trim() || null,
-    publicationTypes: [...new Set(types)],
-    abstract,
+    doi: null,
+    title: null,
+    first_author: null,
+    year: null,
+    publication_types: [],
+    retracted: false,
     corrections: [],
     coi: null,
-    funding: []
+    funding: [],
+    citable: true,
+    comment: null,
+    has_abstract: null,
+    level: null,
+    version: null,
+    retrieved_by: null,
+    retrieved_date: null,
+    pubmed_checked: null,
+    attempts: [],
+    files: [],
+    glyphs_unresolved: null
   };
 }
-
-// src/spl.ts
-import { XMLParser as XMLParser3 } from "fast-xml-parser";
-var DAILYMED_PAGE = (setid) => `https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=${setid}`;
-var SPL_URL = (setid) => `https://dailymed.nlm.nih.gov/dailymed/services/v2/spls/${setid}.xml`;
-async function fetchSpl(setid) {
-  const xml = await fetchText(SPL_URL(setid), { "User-Agent": LIBRARY_UA, Accept: "application/xml, */*;q=0.8" });
-  if (!/<document[\s>]/.test(xml.slice(0, 3e3))) throw new Error("dailymed: not an SPL document");
-  return xml;
+var today = () => (/* @__PURE__ */ new Date()).toISOString().slice(0, 10);
+function headLine(r) {
+  return [
+    r.authors.join(", ") || null,
+    r.journal,
+    r.year,
+    r.pmid ? `PMID ${r.pmid}` : "not in PubMed",
+    r.doi ? `DOI ${r.doi}` : null,
+    r.pmcid ? `PMCID ${r.pmcid}` : null
+  ].filter(Boolean).join(" \xB7 ");
 }
-var parser3 = new XMLParser3({ preserveOrder: true, ignoreAttributes: false, attributeNamePrefix: "@_", parseTagValue: false, trimValues: false, processEntities: true, htmlEntities: true });
-var tagOf2 = (n) => Object.keys(n).find((k) => k !== ":@") ?? "";
-var kids2 = (n) => n[tagOf2(n)] ?? [];
-var attr2 = (n, a) => n[":@"]?.[`@_${a}`] ?? "";
-var squash2 = (s) => s.replace(/\s+/g, " ").trim();
-var child = (n, tag) => kids2(n).find((x) => tagOf2(x) === tag);
-function findDeep2(nodes, tag) {
-  for (const n of nodes) {
-    if ("#text" in n) continue;
-    if (tagOf2(n) === tag) return n;
-    const h = findDeep2(kids2(n), tag);
-    if (h) return h;
-  }
-  return void 0;
+function renderAbstractMd(r) {
+  const body = r.abstract.length ? r.abstract.map((a) => a.label ? `**${a.label}.** ${a.text}` : a.text).join("\n\n") : "_No abstract available._";
+  return `# ${r.title}
+
+${headLine(r)}
+
+## Abstract
+
+${body}
+`;
 }
-function inline(nodes) {
-  let out = "";
-  for (const n of nodes) {
-    if ("#text" in n) {
-      out += String(n["#text"]);
-      continue;
-    }
-    const t = tagOf2(n);
-    const inner = () => inline(kids2(n));
-    switch (t) {
-      case "content": {
-        const style = attr2(n, "styleCode").toLowerCase();
-        const s = inner();
-        out += /bold/.test(style) ? `**${s}**` : /italic/.test(style) ? `*${s}*` : s;
-        break;
-      }
-      case "sup":
-        out += `^${inner()}^`;
-        break;
-      case "sub":
-        out += `~${inner()}~`;
-        break;
-      case "br":
-        out += " ";
-        break;
-      case "footnote":
-        out += ` (${squash2(inner())})`;
-        break;
-      case "footnoteRef":
-      case "renderMultiMedia":
-        break;
-      default:
-        out += inner();
-    }
+function readHeadLine(p) {
+  try {
+    const lines = readFileSync3(p.md, "utf8").split("\n").filter((l) => l.trim());
+    return lines[1] ?? "";
+  } catch {
+    return "";
   }
-  return out;
 }
-function table(n) {
-  const row3 = (x) => kids2(x).filter((c) => ["td", "th"].includes(tagOf2(c))).map((c) => squash2(inline(kids2(c))).replace(/\|/g, "\\|"));
-  const collect = (nodes, into) => {
-    for (const x of nodes) {
-      if ("#text" in x) continue;
-      if (tagOf2(x) === "tr") into.push(row3(x));
-      else if (!["thead", "tfoot"].includes(tagOf2(x))) collect(kids2(x), into);
-    }
+function metaFromRecord(r, files, converter) {
+  return {
+    pmid: r.pmid,
+    setid: null,
+    nctid: null,
+    publications: [],
+    pmcid: r.pmcid,
+    doi: r.doi,
+    title: r.title,
+    first_author: r.firstAuthor,
+    year: r.year,
+    publication_types: r.publicationTypes,
+    retracted: isRetracted(r),
+    corrections: r.corrections,
+    coi: r.coi,
+    funding: r.funding,
+    has_abstract: r.abstract.length > 0,
+    level: "abstract",
+    version: null,
+    files,
+    retrieved_by: "pubmed",
+    retrieved_date: today(),
+    converted_by: converter,
+    pubmed_checked: today(),
+    attempts: []
   };
-  const head = [], body = [], foot = [];
-  for (const x of kids2(n)) {
-    const t = tagOf2(x);
-    if (t === "thead") collect(kids2(x), head);
-    else if (t === "tfoot") collect(kids2(x), foot);
-    else if (t === "tr") body.push(row3(x));
-    else if (!("#text" in x) && t !== "caption") collect(kids2(x), body);
+}
+var isRetracted = (r) => r.publicationTypes.includes("Retracted Publication") || r.corrections.some(isRetractionNotice);
+var renderCorrection = (c) => `${c.type} ${c.pmid ?? "-"} ${JSON.stringify(c.source)}`;
+function parseCorrection(v) {
+  const m = /^(\S+) (\S+) ("[\s\S]*")$/.exec(v.trim());
+  if (!m) return null;
+  let source;
+  try {
+    source = JSON.parse(m[3]);
+  } catch {
+    return null;
   }
-  const rows = [...head, ...body];
-  if (!rows.length) return "";
-  const width = Math.max(...rows.map((r) => r.length));
-  const line = (r) => `| ${[...r, ...Array(width - r.length).fill("")].join(" | ")} |`;
-  const cap = child(n, "caption");
-  const title = cap ? squash2(inline(kids2(cap))) : "";
-  const notes = foot.map((r) => r.filter(Boolean).join(" ")).filter(Boolean);
-  return [title ? `**${title}**` : "", line(rows[0]), `| ${Array(width).fill("---").join(" | ")} |`, ...rows.slice(1).map(line), ...notes.length ? ["", ...notes] : []].filter((l, i) => l || i > 0).join("\n");
+  return { type: m[1], pmid: m[2] === "-" ? null : m[2], source };
 }
-function blocks(nodes, depth) {
-  const out = [];
-  for (const n of nodes) {
-    if ("#text" in n) {
-      const t2 = squash2(String(n["#text"]));
-      if (t2) out.push(t2);
-      continue;
-    }
-    const t = tagOf2(n);
-    const c = kids2(n);
-    switch (t) {
-      case "paragraph": {
-        const s = squash2(inline(c.filter((x) => !["table", "list"].includes(tagOf2(x)))));
-        if (s) out.push(s);
-        out.push(...blocks(c.filter((x) => ["table", "list"].includes(tagOf2(x))), depth));
-        break;
-      }
-      case "list": {
-        const ordered = /^(ordered|numbered|arabic|roman|alpha)/i.test(attr2(n, "listType"));
-        let i = 0;
-        for (const it of c.filter((x) => tagOf2(x) === "item")) {
-          i++;
-          const body = blocks(kids2(it).filter((x) => tagOf2(x) !== "caption"), depth).join(" ");
-          if (body) out.push(`${ordered ? `${i}.` : "-"} ${body}`);
-        }
-        break;
-      }
-      case "table": {
-        const md = table(n);
-        if (md) out.push(md);
-        break;
-      }
-      case "renderMultiMedia":
-      case "observationMedia":
-      case "subject":
-        break;
-      case "section":
-        out.push(...section(n, depth));
-        break;
-      case "component":
-      case "text":
-      case "excerpt":
-      case "highlight":
-      default:
-        out.push(...blocks(c, depth));
-        break;
-    }
+function renderMeta(m) {
+  const lines = [
+    ["pmid", m.pmid],
+    ...m.setid ? [["setid", m.setid]] : [],
+    ...m.nctid ? [["nctid", m.nctid]] : [],
+    ...m.nctid ? [["publications", m.publications.join(", ")]] : [],
+    ["pmcid", m.pmcid ?? ""],
+    ["doi", m.doi ?? ""],
+    ["title", JSON.stringify(m.title)],
+    ["first_author", m.first_author ?? ""],
+    ["year", m.year ?? ""],
+    ["publication_types", m.publication_types.join("; ")],
+    ["retracted", m.retracted ? "yes" : "no"],
+    ...m.corrections.map((c) => ["corrections", renderCorrection(c)]),
+    ["coi", m.coi === null ? "" : JSON.stringify(m.coi)],
+    ["funding", m.funding.length ? JSON.stringify(m.funding) : ""],
+    // a JSON array: a statement may hold semicolons
+    ["has_abstract", m.has_abstract ? "yes" : "no"],
+    ["level", m.level],
+    ["version", m.version ?? ""],
+    ["files", m.files.join(", ")],
+    ["retrieved_by", m.retrieved_by],
+    ["retrieved_date", m.retrieved_date],
+    ["converted_by", m.converted_by],
+    ["pubmed_checked", m.pubmed_checked],
+    ["attempts", m.attempts.join(", ")]
+  ];
+  return "---\n" + lines.map(([k, v]) => v === "" ? `${k}:` : `${k}: ${v}`).join("\n") + "\n---\n";
+}
+function parseFrontmatter(md) {
+  const m = /^---\n([\s\S]*?)\n---/.exec(md);
+  const fm = {}, all = {};
+  for (const line of (m?.[1] ?? "").split("\n")) {
+    const i = line.indexOf(":");
+    if (i <= 0) continue;
+    const k = line.slice(0, i).trim(), v = line.slice(i + 1).trim();
+    fm[k] = v;
+    (all[k] ??= []).push(v);
   }
-  return out;
+  return { fm, all };
 }
-var SKIP_CODES = /* @__PURE__ */ new Set(["48780-1", "51945-4"]);
-var CODE_NAMES = { "34066-1": "Boxed Warning", "43683-2": "Recent Major Changes" };
-var plain2 = (nodes) => {
-  let out = "";
-  for (const n of nodes) out += "#text" in n ? String(n["#text"]) : tagOf2(n) === "br" ? " " : plain2(kids2(n));
-  return out;
-};
-function section(n, depth) {
-  const code = child(n, "code");
-  const c = code ? attr2(code, "code") : "";
-  if (SKIP_CODES.has(c)) return [];
-  const title = child(n, "title");
-  let head = title ? squash2(plain2(kids2(title))) : "";
-  if (c === "34066-1" && head && !/boxed warning/i.test(head)) head = `Boxed Warning: ${head}`;
-  if (!head && CODE_NAMES[c]) head = CODE_NAMES[c];
+var list2 = (v) => v ? v.split(",").map((s) => s.trim()).filter(Boolean) : [];
+function splitAttempts(v) {
+  if (!v) return [];
   const out = [];
-  if (head) out.push(`${"#".repeat(Math.min(depth, 6))} ${head}`);
-  const rest = kids2(n).filter((x) => !["title", "code", "id", "effectiveTime"].includes(tagOf2(x)));
-  out.push(...blocks(rest, head ? depth + 1 : depth));
-  return out;
-}
-var isoDate = (v) => /^\d{8}$/.test(v) ? `${v.slice(0, 4)}-${v.slice(4, 6)}-${v.slice(6, 8)}` : v || null;
-function parseSpl(xml) {
-  const doc = parser3.parse(xml);
-  const d = doc.find((x) => tagOf2(x) === "document");
-  if (!d) return null;
-  const k = kids2(d);
-  const setid = attr2(child(d, "setId") ?? {}, "root").toLowerCase();
-  const version = attr2(child(d, "versionNumber") ?? {}, "value") || null;
-  const effective = isoDate(attr2(child(d, "effectiveTime") ?? {}, "value"));
-  const named = (nodes) => {
-    for (const n of nodes) {
-      if ("#text" in n) continue;
-      if (tagOf2(n) === "manufacturedProduct") {
-        const inner2 = child(n, "manufacturedProduct") ?? n;
-        if (child(inner2, "name")) return inner2;
-      }
-      const h = named(kids2(n));
-      if (h) return h;
-    }
-    return void 0;
-  };
-  const inner = named(k);
-  const brand = inner ? squash2(inline(kids2(child(inner, "name") ?? { name: [] }))) || null : null;
-  const genericNode = inner ? findDeep2(kids2(inner), "genericMedicine") : void 0;
-  const generic = genericNode ? squash2(inline(kids2(child(genericNode, "name") ?? { name: [] }))) || null : null;
-  const formNode = inner ? child(inner, "formCode") : void 0;
-  const form = formNode ? attr2(formNode, "displayName").toLowerCase() || null : null;
-  const org = findDeep2(k, "representedOrganization");
-  const labeler = org ? squash2(inline(kids2(child(org, "name") ?? { name: [] }))) || null : null;
-  const body = findDeep2(k, "structuredBody");
-  const sections2 = body ? blocks(kids2(body), 2) : [];
-  const docTitle = child(d, "title");
-  const lines = [];
-  let cur = "";
-  for (const x of docTitle ? kids2(docTitle) : []) {
-    if (!("#text" in x) && tagOf2(x) === "br") {
-      lines.push(cur);
+  let cur = "", depth = 0;
+  for (const ch of v) {
+    if (ch === "(") depth++;
+    else if (ch === ")") depth = Math.max(0, depth - 1);
+    if (ch === "," && depth === 0) {
+      out.push(cur);
       cur = "";
-    } else cur += "#text" in x ? String(x["#text"]) : inline([x]);
+    } else cur += ch;
   }
-  lines.push(cur);
-  const highlights = lines.map(squash2).filter((l) => l && !/^these highlights do not include/i.test(l) && !/^see full prescribing information/i.test(l)).map((l) => l.replace(/^see full prescribing information for [^.]+\.\s*/i, "")).filter(Boolean);
-  const approvalYear = /initial u\.?s\.? approval:?\s*(\d{4})/i.exec(highlights.join(" "))?.[1] ?? null;
-  const productLine = highlights.find((l) => !/initial u\.?s\.? approval/i.test(l)) ?? null;
-  const title = [brand, generic ? `(${generic})` : "", form].filter(Boolean).join(" ") || productLine || setid;
-  return { setid, version, effective, brand, generic, form, labeler, title: `${title} \u2014 prescribing information`, highlights, approvalYear, sections: sections2 };
+  out.push(cur);
+  const merged = [];
+  for (const piece of out.map((s) => s.trim()).filter(Boolean)) {
+    if (merged.length && !/^\S+ \(/.test(piece)) merged[merged.length - 1] += ", " + piece;
+    else merged.push(piece);
+  }
+  return merged;
 }
-function headLineSpl(s) {
-  return [s.labeler, `DailyMed set ID ${s.setid}`, s.version ? `version ${s.version}` : "", s.effective ? `effective ${s.effective}` : "", s.approvalYear ? `initial U.S. approval ${s.approvalYear}` : "", DAILYMED_PAGE(s.setid)].filter(Boolean).join(" \xB7 ");
+var opt = (v) => v ? v : null;
+var TYPE_CONTINUATION = /* @__PURE__ */ new Set(["N.I.H.", "Extramural", "Intramural", "Non-U.S. Gov't", "U.S. Gov't", "U.S. Government", "P.H.S.", "Non-P.H.S.", "Phase I", "Phase II", "Phase III", "Phase IV", "Veterinary", "NIH", "American Recovery and Reinvestment Act"]);
+function splitPublicationTypes(v) {
+  if (!v) return [];
+  if (v.includes(";")) return v.split(";").map((s) => s.trim()).filter(Boolean);
+  const out = [];
+  for (const piece of list2(v)) {
+    if (out.length && TYPE_CONTINUATION.has(piece)) out[out.length - 1] += ", " + piece;
+    else out.push(piece);
+  }
+  return out;
 }
-function renderLabelMd(s) {
-  const parts = [`# ${s.title}`, "", headLineSpl(s), ""];
-  if (s.highlights.length) parts.push("## Highlights", "", ...s.highlights.flatMap((l) => [l, ""]));
-  parts.push(...s.sections.flatMap((b) => [b, ""]));
-  return parts.join("\n").replace(/\n{3,}/g, "\n\n").trimEnd() + "\n";
+function abstractMissing(p) {
+  try {
+    return readFileSync3(p.md, "utf8").includes("_No abstract available._");
+  } catch {
+    return false;
+  }
 }
-
-// src/ctgov.ts
-var CTGOV_PAGE = (nctid) => `https://clinicaltrials.gov/study/${nctid}`;
-var API = (nctid) => `https://clinicaltrials.gov/api/v2/studies/${nctid}?format=json`;
-async function fetchStudy(nctid) {
-  const json2 = await fetchText(API(nctid), { "User-Agent": LIBRARY_UA, Accept: "application/json" });
-  if (!/"protocolSection"/.test(json2.slice(0, 2e3))) throw new Error("ctgov: not a study record");
-  return json2;
+function correctionsFromXml(p) {
+  try {
+    return parsePubmedXml(readFileSync3(p.xml, "utf8"))?.corrections ?? [];
+  } catch {
+    return [];
+  }
 }
-var STATUS = {
-  NOT_YET_RECRUITING: "Not yet recruiting",
-  RECRUITING: "Recruiting",
-  ENROLLING_BY_INVITATION: "Enrolling by invitation",
-  ACTIVE_NOT_RECRUITING: "Active, not recruiting",
-  SUSPENDED: "Suspended",
-  TERMINATED: "Terminated",
-  COMPLETED: "Completed",
-  WITHDRAWN: "Withdrawn",
-  UNKNOWN: "Unknown status",
-  AVAILABLE: "Available",
-  NO_LONGER_AVAILABLE: "No longer available",
-  TEMPORARILY_NOT_AVAILABLE: "Temporarily not available",
-  APPROVED_FOR_MARKETING: "Approved for marketing",
-  WITHHELD: "Withheld"
-};
-var PHASE = { EARLY_PHASE1: "Early Phase 1", PHASE1: "Phase 1", PHASE2: "Phase 2", PHASE3: "Phase 3", PHASE4: "Phase 4", NA: "Not applicable" };
-var words = (v) => {
-  const s = String(v ?? "");
-  if (!s) return "";
-  if (STATUS[s]) return STATUS[s];
-  if (PHASE[s]) return PHASE[s];
-  const w = s.toLowerCase().replace(/_/g, " ");
-  return w.charAt(0).toUpperCase() + w.slice(1);
-};
-var dated = (d) => d?.date ? `${d.date}${d.type ? ` (${String(d.type).toLowerCase()})` : ""}` : "";
-var list = (a) => Array.isArray(a) ? a : [];
-var row = (cells) => `| ${cells.map((c) => c.replace(/\|/g, "\\|").replace(/\s+/g, " ").trim()).join(" | ")} |`;
-var table2 = (head, rows) => rows.length ? [row(head), `| ${head.map(() => "---").join(" | ")} |`, ...rows.map(row)].join("\n") : "";
-var para = (s) => String(s ?? "").replace(/\r/g, "").trim();
-function parseStudy(json2) {
-  let d;
+function declarationsFromXml(p) {
   try {
-    d = JSON.parse(json2);
+    const r = parsePubmedXml(readFileSync3(p.xml, "utf8"));
+    return { coi: r?.coi ?? null, funding: r?.funding ?? [] };
   } catch {
-    return null;
+    return { coi: null, funding: [] };
   }
-  const ps = d?.protocolSection;
-  if (!ps) return null;
-  const idm = ps.identificationModule ?? {}, st = ps.statusModule ?? {}, sp = ps.sponsorCollaboratorsModule ?? {}, de = ps.descriptionModule ?? {};
-  const dm = ps.designModule ?? {}, ai = ps.armsInterventionsModule ?? {}, om = ps.outcomesModule ?? {}, el = ps.eligibilityModule ?? {}, cl = ps.contactsLocationsModule ?? {}, rf = ps.referencesModule ?? {};
-  const nctid = String(idm.nctId ?? "").toUpperCase();
-  const out = [];
-  const sec = (title, lines) => {
-    const body = lines.filter((l) => !!l && l.trim() !== "");
-    if (body.length) out.push(`## ${title}`, ...body);
+}
+function readMeta(p) {
+  const { fm, all } = parseFrontmatter(readFileSync3(p.meta, "utf8"));
+  let title = fm.title ?? "";
+  if (title.startsWith('"')) {
+    try {
+      title = JSON.parse(title);
+    } catch {
+    }
+  }
+  const legacy = fm.pubmed_checked === void 0;
+  const corrections = legacy ? correctionsFromXml(p) : (all.corrections ?? []).map(parseCorrection).filter((c) => c !== null);
+  let coi = null, funding = [];
+  if (fm.coi === void 0) ({ coi, funding } = declarationsFromXml(p));
+  else {
+    if (fm.coi.startsWith('"')) {
+      try {
+        coi = JSON.parse(fm.coi);
+      } catch {
+        coi = fm.coi;
+      }
+    } else coi = fm.coi || null;
+    funding = !fm.funding ? [] : fm.funding.startsWith("[") ? JSON.parse(fm.funding) : fm.funding.split(";").map((s) => s.trim()).filter(Boolean);
+  }
+  return {
+    pmid: fm.pmid ?? "",
+    setid: opt(fm.setid),
+    nctid: opt(fm.nctid),
+    publications: list2(fm.publications),
+    pmcid: opt(fm.pmcid),
+    doi: opt(fm.doi),
+    title,
+    first_author: opt(fm.first_author),
+    year: opt(fm.year),
+    publication_types: splitPublicationTypes(fm.publication_types),
+    retracted: fm.retracted === "yes" || corrections.some(isRetractionNotice),
+    corrections,
+    coi,
+    funding,
+    has_abstract: fm.has_abstract ? fm.has_abstract === "yes" : !abstractMissing(p),
+    level: fm.level === "full_text" ? "full_text" : "abstract",
+    version: opt(fm.version),
+    files: list2(fm.files),
+    retrieved_by: fm.retrieved_by ?? "",
+    retrieved_date: fm.retrieved_date ?? "",
+    converted_by: fm.converted_by ?? "",
+    pubmed_checked: legacy ? fm.retrieved_date ?? "" : fm.pubmed_checked,
+    attempts: splitAttempts(fm.attempts)
   };
-  const secondary = list(idm.secondaryIdInfos).map((s) => `${words(s.type)}${s.domain ? ` (${s.domain})` : ""}: ${s.id}`);
-  sec("Identification", [
-    idm.officialTitle ? `Official title: ${para(idm.officialTitle)}` : null,
-    idm.acronym ? `Acronym: ${idm.acronym}` : null,
-    idm.organization?.fullName ? `Organization: ${idm.organization.fullName}` : null,
-    secondary.length ? `Secondary IDs: ${secondary.join("; ")}` : null
-  ]);
-  sec("Status", [
-    `Overall status: ${words(st.overallStatus)}${st.whyStopped ? ` \u2014 why stopped: ${para(st.whyStopped)}` : ""}`,
-    st.startDateStruct ? `Start: ${dated(st.startDateStruct)}` : null,
-    st.primaryCompletionDateStruct ? `Primary completion: ${dated(st.primaryCompletionDateStruct)}` : null,
-    st.completionDateStruct ? `Completion: ${dated(st.completionDateStruct)}` : null,
-    st.studyFirstPostDateStruct ? `First posted: ${dated(st.studyFirstPostDateStruct)}` : null,
-    st.lastUpdatePostDateStruct ? `Last update posted: ${dated(st.lastUpdatePostDateStruct)}` : null,
-    st.statusVerifiedDate ? `Status verified: ${st.statusVerifiedDate}` : null,
-    `Results posted: ${d.hasResults ? "yes" : "no"}`
-  ]);
-  const rp = sp.responsibleParty;
-  sec("Sponsor and collaborators", [
-    sp.leadSponsor?.name ? `Lead sponsor: ${sp.leadSponsor.name}${sp.leadSponsor.class ? ` (${words(sp.leadSponsor.class).toLowerCase()})` : ""}` : null,
-    list(sp.collaborators).length ? `Collaborators: ${list(sp.collaborators).map((c) => c.name).join("; ")}` : null,
-    rp ? `Responsible party: ${[words(rp.type).toLowerCase(), rp.investigatorFullName, rp.investigatorTitle, rp.investigatorAffiliation].filter(Boolean).join(", ")}` : null
-  ]);
-  sec("Description", [de.briefSummary ? para(de.briefSummary) : null, de.detailedDescription ? para(de.detailedDescription) : null]);
-  sec("Conditions", [list(ps.conditionsModule?.conditions).length ? list(ps.conditionsModule.conditions).join("; ") : null, list(ps.conditionsModule?.keywords).length ? `Keywords: ${list(ps.conditionsModule.keywords).join("; ")}` : null]);
-  const di = dm.designInfo ?? {};
-  const mi = di.maskingInfo ?? {};
-  sec("Design", [
-    dm.studyType ? `Study type: ${words(dm.studyType)}` : null,
-    list(dm.phases).length ? `Phase: ${list(dm.phases).map(words).join(", ")}` : null,
-    di.allocation ? `Allocation: ${words(di.allocation)}` : null,
-    di.interventionModel ? `Intervention model: ${words(di.interventionModel)}${di.interventionModelDescription ? ` \u2014 ${para(di.interventionModelDescription)}` : ""}` : null,
-    di.primaryPurpose ? `Primary purpose: ${words(di.primaryPurpose)}` : null,
-    mi.masking ? `Masking: ${words(mi.masking)}${list(mi.whoMasked).length ? ` (${list(mi.whoMasked).map((w) => words(w).toLowerCase()).join(", ")})` : ""}${mi.maskingDescription ? ` \u2014 ${para(mi.maskingDescription)}` : ""}` : null,
-    dm.enrollmentInfo?.count !== void 0 ? `Enrollment: ${dm.enrollmentInfo.count}${dm.enrollmentInfo.type ? ` (${String(dm.enrollmentInfo.type).toLowerCase()})` : ""}` : null
-  ]);
-  const arms = list(ai.armGroups).map((a) => `- **${a.label}**${a.type ? ` (${words(a.type).toLowerCase()})` : ""}${a.description ? `: ${para(a.description)}` : ""}${list(a.interventionNames).length ? ` \u2014 ${list(a.interventionNames).join("; ")}` : ""}`);
-  const ints = list(ai.interventions).map((i) => `- **${words(i.type)}: ${i.name}**${list(i.otherNames).length ? ` (${list(i.otherNames).join(", ")})` : ""}${i.description ? `: ${para(i.description)}` : ""}${list(i.armGroupLabels).length ? ` \u2014 arms: ${list(i.armGroupLabels).join("; ")}` : ""}`);
-  sec("Arms and interventions", [arms.length ? "Arms:" : null, ...arms, ints.length ? "Interventions:" : null, ...ints]);
-  const outcome = (o) => `- **${para(o.measure)}**${o.description ? `: ${para(o.description)}` : ""}${o.timeFrame ? ` Time frame: ${para(o.timeFrame)}.` : ""}`;
-  sec("Outcomes", [
-    list(om.primaryOutcomes).length ? "### Primary outcomes" : null,
-    ...list(om.primaryOutcomes).map(outcome),
-    list(om.secondaryOutcomes).length ? "### Secondary outcomes" : null,
-    ...list(om.secondaryOutcomes).map(outcome),
-    list(om.otherOutcomes).length ? "### Other outcomes" : null,
-    ...list(om.otherOutcomes).map(outcome)
-  ]);
-  sec("Eligibility", [
-    [el.sex ? `Sex: ${words(el.sex).toLowerCase()}` : "", el.minimumAge || el.maximumAge ? `Age: ${el.minimumAge ?? "no minimum"} to ${el.maximumAge ?? "no maximum"}` : "", el.healthyVolunteers !== void 0 ? `Healthy volunteers: ${el.healthyVolunteers ? "yes" : "no"}` : ""].filter(Boolean).join(" \xB7 "),
-    el.eligibilityCriteria ? para(el.eligibilityCriteria) : null
-  ]);
-  const countries = [...new Set(list(cl.locations).map((l) => l.country).filter(Boolean))];
-  sec("Locations", [countries.length ? `${list(cl.locations).length} sites in ${countries.join(", ")}` : null, list(cl.overallOfficials).length ? `Overall officials: ${list(cl.overallOfficials).map((o) => [o.name, o.role ? words(o.role).toLowerCase() : "", o.affiliation].filter(Boolean).join(", ")).join("; ")}` : null]);
-  const refs = list(rf.references);
-  sec("References", refs.map((r) => `- ${para(r.citation)}${r.pmid ? ` pmid:${r.pmid}` : ""}${r.type ? ` (${words(r.type).toLowerCase()})` : ""}`));
-  const publications = refs.filter((r) => r.pmid && ["RESULT", "DERIVED"].includes(String(r.type))).map((r) => `pmid:${r.pmid}`);
-  if (d.hasResults && d.resultsSection) out.push(...results(d.resultsSection));
+}
+function writeAtomic(file3, data) {
+  const tmp = `${file3}.${process.pid}.${Math.random().toString(36).slice(2, 8)}.tmp`;
+  writeFileSync(tmp, data);
+  renameSync(tmp, file3);
+}
+function writeMeta(p, m) {
+  writeAtomic(p.meta, renderMeta(m));
+  try {
+    unlinkSync(join4(p.entryDir, `${p.entry}-missing.md`));
+  } catch {
+  }
+}
+function copyEntry(from, to) {
+  const tmp = `${to.entryDir}.${process.pid}.tmp`, old = `${to.entryDir}.${process.pid}.old`;
+  rmSync(tmp, { recursive: true, force: true });
+  mkdirSync2(tmp, { recursive: true });
+  const rename = (n) => n.startsWith(from.entry) ? to.entry + n.slice(from.entry.length) : n;
+  for (const n of readdirSync(from.entryDir)) if (join4(from.entryDir, n) !== from.meta && !n.endsWith(".tmp")) copyFileSync(join4(from.entryDir, n), join4(tmp, rename(n)));
+  const m = readMeta(from);
+  const copy = { ...m, files: m.files.map(rename) };
+  writeFileSync(join4(tmp, basename(to.meta)), renderMeta(copy));
+  mkdirSync2(dirname(to.entryDir), { recursive: true });
+  if (existsSync4(to.entryDir)) renameSync(to.entryDir, old);
+  renameSync(tmp, to.entryDir);
+  rmSync(old, { recursive: true, force: true });
+  return copy;
+}
+var NOTHING_TO_VERIFY = "no abstract and no full text: nothing to verify";
+function citability(m) {
+  const of = (t) => m.corrections.filter((c) => c.type === t);
+  const cite = (c) => `${c.source}${c.pmid ? ` (pmid:${c.pmid})` : ""}`;
+  if (m.retracted) {
+    const r = m.corrections.filter(isRetractionNotice);
+    return { citable: false, comment: r.length ? `retracted: ${r.map(cite).join("; ")}` : "retracted (PubMed publication type)" };
+  }
+  if (m.level !== "full_text" && !m.has_abstract) return { citable: false, comment: NOTHING_TO_VERIFY };
+  const notes = [
+    ...of("erratum_in").map((c) => `erratum, not assessed: ${cite(c)}`),
+    ...of("expression_of_concern_in").map((c) => `expression of concern: ${cite(c)}`),
+    ...of("update_in").map((c) => `superseded by a newer version: ${cite(c)}`)
+  ];
+  return { citable: true, comment: notes.length ? notes.join("; ") : null };
+}
+function statusFromMeta(p, m, base) {
+  const full = existsSync4(p.fulltextMd) ? readFileSync3(p.fulltextMd, "utf8") : null;
+  const cit = citability(m);
+  if (full?.includes(`
+${CORRECTED_HEADING}
+`) && cit.comment) cit.comment = cit.comment.replace(/erratum, not assessed: /g, "erratum, corrected figures stored (Corrected in PubMed abstract): ");
   return {
-    nctid,
-    title: para(idm.briefTitle) || nctid,
-    acronym: idm.acronym ?? null,
-    sponsor: sp.leadSponsor?.name ?? null,
-    firstPosted: st.studyFirstPostDateStruct?.date ?? null,
-    lastUpdate: st.lastUpdatePostDateStruct?.date ?? null,
-    status: st.overallStatus ? words(st.overallStatus) : null,
-    hasResults: !!d.hasResults,
-    publications,
-    sections: out
+    ...base,
+    entry: p.entry,
+    entry_dir: p.entryDir,
+    pmid: m.pmid || null,
+    setid: m.setid,
+    nctid: m.nctid,
+    publications: m.publications,
+    pmcid: m.pmcid,
+    doi: m.doi,
+    title: m.title || null,
+    first_author: m.first_author,
+    year: m.year,
+    publication_types: m.publication_types,
+    retracted: m.retracted,
+    corrections: m.corrections,
+    coi: m.coi,
+    funding: m.funding,
+    ...cit,
+    // What the entry holds: full_text, abstract, or only the title (the meta's
+    // level is "abstract" until a full text arrives; has_abstract says whether
+    // there is abstract text).
+    has_abstract: m.has_abstract,
+    level: m.level === "abstract" && !m.has_abstract ? "title" : m.level,
+    version: m.version,
+    retrieved_by: m.retrieved_by || null,
+    retrieved_date: m.retrieved_date || null,
+    pubmed_checked: m.pubmed_checked || null,
+    attempts: m.attempts,
+    files: m.files,
+    glyphs_unresolved: full === null ? null : countUnresolved(full)
   };
 }
-function results(rs) {
-  const out = ["## Results"];
-  const pf = rs.participantFlowModule;
-  if (pf) {
-    out.push("### Participant flow");
-    const groups = list(pf.groups);
-    out.push(...groups.map((g) => `- **${g.title}**${g.description ? `: ${para(g.description)}` : ""}`));
-    if (pf.recruitmentDetails) out.push(para(pf.recruitmentDetails));
-    for (const p of list(pf.periods)) {
-      const head = ["Period: " + (p.title ?? ""), ...groups.map((g) => g.title)];
-      const rows = list(p.milestones).map((m) => [words(m.type), ...groups.map((g) => list(m.achievements).find((a) => a.groupId === g.id)?.numSubjects ?? "")]);
-      for (const dw of list(p.dropWithdraws)) rows.push([`Not completed: ${dw.type}`, ...groups.map((g) => list(dw.reasons).find((r) => r.groupId === g.id)?.numSubjects ?? "")]);
-      out.push(table2(head, rows));
-    }
-  }
-  const bc = rs.baselineCharacteristicsModule;
-  if (bc) {
-    out.push("### Baseline characteristics");
-    const groups = list(bc.groups);
-    out.push(...groups.map((g) => `- **${g.title}**${g.description ? `: ${para(g.description)}` : ""}`));
-    for (const m of list(bc.measures)) {
-      const unit = [m.paramType ? words(m.paramType).toLowerCase() : "", m.dispersionType ? words(m.dispersionType).toLowerCase() : "", m.unitOfMeasure].filter(Boolean).join(", ");
-      const rows = [];
-      for (const c of list(m.classes)) for (const cat of list(c.categories)) rows.push([[c.title, cat.title].filter(Boolean).join(" \u2014 ") || m.title, ...groups.map((g) => {
-        const x = list(cat.measurements).find((v) => v.groupId === g.id);
-        return x ? `${x.value ?? ""}${x.spread !== void 0 ? ` \xB1 ${x.spread}` : ""}${x.lowerLimit !== void 0 ? ` (${x.lowerLimit} to ${x.upperLimit})` : ""}` : "";
-      })]);
-      out.push(`**${m.title}**${unit ? ` (${unit})` : ""}`, table2(["", ...groups.map((g) => g.title)], rows));
+function readStatus(p, base) {
+  return statusFromMeta(p, readMeta(p), base);
+}
+function entryExists(p) {
+  return existsSync4(p.meta);
+}
+function writeEntry(p, r, xml, converter) {
+  mkdirSync2(p.entryDir, { recursive: true });
+  writeAtomic(p.xml, xml);
+  writeAtomic(p.md, renderAbstractMd(r));
+  const meta = metaFromRecord(r, [basename(p.xml), basename(p.md)], converter);
+  writeMeta(p, meta);
+  return meta;
+}
+function refreshRecord(p, m, r, xml) {
+  writeAtomic(p.xml, xml);
+  syncCorrectedBlock(p);
+  const next = {
+    ...m,
+    pmcid: r.pmcid ?? m.pmcid,
+    doi: r.doi ?? m.doi,
+    publication_types: r.publicationTypes,
+    retracted: isRetracted(r),
+    corrections: r.corrections,
+    coi: r.coi ?? m.coi,
+    funding: r.funding.length ? r.funding : m.funding,
+    // the paper's own statements stay when PubMed has none
+    pubmed_checked: today()
+  };
+  writeMeta(p, next);
+  return next;
+}
+function daysSince(date, now = /* @__PURE__ */ new Date()) {
+  const t = new Date(date).getTime();
+  return Number.isFinite(t) ? (now.getTime() - t) / 864e5 : Infinity;
+}
+var FUNDING_RE = /^[*_\s]*(funding|financial support|sources? of (financial )?support|support(ed by)?|grant support|this (work|study|research) was (supported|funded))\b[^:]*:?[*_\s]*(.*)$/i;
+var COI_RE = /^[*_\s]*(conflicts? of interest|competing interests?|disclosures?|financial disclosures?|declarations? of (competing |conflicting )?interests?|conflict[- ]of[- ]interest statement)\b[^:]*:?[*_\s]*(.*)$/i;
+function declarationsFromMd(md) {
+  let coi = null;
+  const funding = [];
+  const lines = md.split("\n");
+  let section2 = "";
+  for (let i = 0; i < lines.length; i++) {
+    const l = lines[i].trim();
+    const h = /^#{2,6}\s+(.*)$/.exec(l);
+    if (h) {
+      section2 = h[1];
+      const body = lines.slice(i + 1).find((x) => x.trim() && !/^#/.test(x))?.trim() ?? "";
+      if (body && /^(funding|financial support|sources? of funding|role of the funding source)/i.test(section2) && !funding.length) funding.push(body);
+      if (body && /^(conflicts? of interest|competing interests?|disclosures?|financial disclosures?|declarations? of (competing |conflicting )?interests?)/i.test(section2) && coi === null) coi = body;
+      continue;
     }
+    if (section2 !== "Declarations" || !l) continue;
+    const f = FUNDING_RE.exec(l);
+    const c = COI_RE.exec(l);
+    if (c && c[3]?.trim() && coi === null) coi = c[3].trim().replace(/[*_]+$/, "");
+    else if (f && f[6]?.trim() && !funding.length) funding.push(f[6].trim().replace(/[*_]+$/, ""));
   }
-  const om = rs.outcomeMeasuresModule;
-  if (om) {
-    out.push("### Outcome measures");
-    for (const m of list(om.outcomeMeasures)) {
-      const groups = list(m.groups);
-      const unit = [m.paramType ? words(m.paramType).toLowerCase() : "", m.dispersionType ? String(m.dispersionType).toLowerCase() : "", m.unitOfMeasure].filter(Boolean).join(", ");
-      out.push(`#### ${m.title}${m.type ? ` (${words(m.type).toLowerCase()})` : ""}`);
-      if (m.description) out.push(para(m.description));
-      out.push([m.timeFrame ? `Time frame: ${para(m.timeFrame)}.` : "", unit ? `Measure: ${unit}.` : "", m.populationDescription ? `Population: ${para(m.populationDescription)}` : ""].filter(Boolean).join(" "));
-      if (m.reportingStatus && m.reportingStatus !== "POSTED") out.push(`Reporting status: ${words(m.reportingStatus)}.`);
-      const rows = [];
-      for (const c of list(m.classes)) for (const cat of list(c.categories)) rows.push([[c.title, cat.title].filter(Boolean).join(" \u2014 ") || m.title, ...groups.map((g) => {
-        const x = list(cat.measurements).find((v) => v.groupId === g.id);
-        return x ? `${x.value ?? ""}${x.spread !== void 0 ? ` \xB1 ${x.spread}` : ""}${x.lowerLimit !== void 0 ? ` (${x.lowerLimit} to ${x.upperLimit})` : ""}` : "";
-      })]);
-      const denoms = list(m.denoms).flatMap((dn) => list(dn.counts)).filter((c) => c.groupId);
-      if (denoms.length) rows.unshift(["Participants analysed", ...groups.map((g) => denoms.find((c) => c.groupId === g.id)?.value ?? "")]);
-      out.push(table2(["", ...groups.map((g) => g.title)], rows));
-      for (const a of list(m.analyses)) out.push(`Analysis${a.groupDescription ? ` (${para(a.groupDescription)})` : ""}: ${[a.statisticalMethod, a.pValue !== void 0 ? `p ${a.pValueComment ?? ""}${a.pValue}`.trim() : "", a.paramType ? `${words(a.paramType).toLowerCase()} ${a.paramValue ?? ""}` : "", a.ciPctValue ? `${a.ciPctValue}% CI ${a.ciLowerLimit ?? ""} to ${a.ciUpperLimit ?? ""}` : ""].filter(Boolean).join("; ")}`);
+  return { coi, funding };
+}
+var CORRECTED_HEADING = "## Corrected in PubMed abstract";
+function correctedSentences(r) {
+  const out = [];
+  for (const a of r.abstract) {
+    for (const s of a.text.split(/(?<=[.!?\]])\s+(?=[A-Z(\u2212-])/)) {
+      const i = s.indexOf("[corrected]");
+      if (i >= 0) out.push(s.slice(0, i).trim().replace(/\s+([.,;])$/, "$1"));
     }
   }
-  const ae = rs.adverseEventsModule;
-  if (ae) {
-    out.push("### Adverse events");
-    const groups = list(ae.eventGroups);
-    out.push([ae.timeFrame ? `Time frame: ${para(ae.timeFrame)}.` : "", ae.frequencyThreshold ? `Threshold for other events: ${ae.frequencyThreshold}%.` : "", ae.description ? para(ae.description) : ""].filter(Boolean).join(" "));
-    const summary = [
-      ["Serious events, participants affected / at risk", ...groups.map((g) => `${g.seriousNumAffected ?? ""} / ${g.seriousNumAtRisk ?? ""}`)],
-      ["Other events, participants affected / at risk", ...groups.map((g) => `${g.otherNumAffected ?? ""} / ${g.otherNumAtRisk ?? ""}`)],
-      ["Deaths, participants affected / at risk", ...groups.map((g) => g.deathsNumAffected !== void 0 ? `${g.deathsNumAffected} / ${g.deathsNumAtRisk ?? ""}` : "")]
-    ].filter((r) => r.slice(1).some(Boolean));
-    out.push(table2(["", ...groups.map((g) => g.title)], summary));
-    const events = (title, evs) => {
-      if (!evs.length) return;
-      out.push(`**${title}** (participants affected / at risk)`, table2(["Event", "System", ...groups.map((g) => g.title)], evs.map((e) => [e.term ?? "", e.organSystem ?? "", ...groups.map((g) => {
-        const s = list(e.stats).find((x) => x.groupId === g.id);
-        return s ? `${s.numAffected ?? ""} / ${s.numAtRisk ?? ""}` : "";
-      })])));
-    };
-    events("Serious adverse events", list(ae.seriousEvents));
-    events("Other adverse events", list(ae.otherEvents));
+  return out.filter(Boolean);
+}
+function withCorrectedBlock(md, sentences) {
+  const base = md.split(`
+${CORRECTED_HEADING}
+`)[0].trimEnd();
+  if (!sentences.length) return base + "\n";
+  return `${base}
+
+${CORRECTED_HEADING}
+
+NLM corrected these sentences of the abstract after an erratum; the text above may predate the correction.
+
+${sentences.join("\n\n")}
+`;
+}
+function syncCorrectedBlock(p) {
+  if (!existsSync4(p.fulltextMd) || !existsSync4(p.xml)) return;
+  let r;
+  try {
+    r = parsePubmedXml(readFileSync3(p.xml, "utf8"));
+  } catch {
+    return;
   }
-  return out;
+  if (!r) return;
+  const md = readFileSync3(p.fulltextMd, "utf8");
+  const next = withCorrectedBlock(md, correctedSentences(r));
+  if (next !== md) writeAtomic(p.fulltextMd, next);
 }
-function headLineStudy(s) {
-  return [s.sponsor, s.nctid, s.status ? s.status.toLowerCase() : "", s.firstPosted ? `first posted ${s.firstPosted}` : "", s.lastUpdate ? `last update posted ${s.lastUpdate}` : "", CTGOV_PAGE(s.nctid)].filter(Boolean).join(" \xB7 ");
+function applyFulltext(p, m, hit, converter) {
+  const raw = p.fulltextRaw(hit.format);
+  writeAtomic(raw, hit.raw);
+  writeAtomic(p.fulltextMd, hit.md);
+  syncCorrectedBlock(p);
+  const files = [...m.files.filter((f) => !f.includes("-fulltext.")), basename(raw), basename(p.fulltextMd)];
+  const own = declarationsFromMd(hit.md);
+  const next = {
+    ...m,
+    level: "full_text",
+    version: hit.version,
+    files,
+    retrieved_by: hit.tier,
+    retrieved_date: today(),
+    converted_by: converter,
+    coi: m.coi ?? own.coi,
+    funding: m.funding.length ? m.funding : own.funding
+  };
+  writeMeta(p, next);
+  return next;
 }
-function renderStudyMd(s) {
-  const parts = [`# ${s.title}${s.acronym ? ` (${s.acronym})` : ""} \u2014 ClinicalTrials.gov record`, "", headLineStudy(s), "", ...s.sections.flatMap((b) => [b, ""])];
-  return parts.join("\n").replace(/\n{3,}/g, "\n\n").trimEnd() + "\n";
+function recordAttempts(p, m, labels) {
+  if (!labels.length) return m;
+  const tierOf = (a) => /^(\S+) \(/.exec(a.trim())?.[1] ?? a;
+  const replaced = new Set(labels.map(tierOf));
+  const next = { ...m, attempts: [...m.attempts.filter((a) => !replaced.has(tierOf(a))), ...labels] };
+  writeMeta(p, next);
+  return next;
 }
-
-// src/entry.ts
-function emptyStatus(requested, type, id, libraryDir) {
-  return {
-    requested,
-    type,
-    id,
-    entry: null,
-    entry_dir: null,
-    library_dir: libraryDir,
-    pmid: null,
+function writeLabelEntry(p, s, xml, converter) {
+  mkdirSync2(p.entryDir, { recursive: true });
+  const raw = p.fulltextRaw("xml");
+  writeAtomic(raw, xml);
+  writeAtomic(p.fulltextMd, renderLabelMd(s));
+  const meta = {
+    pmid: "",
+    setid: s.setid,
+    nctid: null,
+    publications: [],
+    pmcid: null,
+    doi: null,
+    title: s.title,
+    first_author: s.labeler,
+    year: s.effective?.slice(0, 4) ?? null,
+    publication_types: ["Drug Label"],
+    retracted: false,
+    corrections: [],
+    coi: null,
+    funding: [],
+    has_abstract: false,
+    level: "full_text",
+    version: s.version ? `spl ${s.version}` : null,
+    files: [basename(raw), basename(p.fulltextMd)],
+    retrieved_by: "dailymed",
+    retrieved_date: today(),
+    converted_by: converter,
+    pubmed_checked: today(),
+    attempts: []
+  };
+  writeMeta(p, meta);
+  return meta;
+}
+function writeStudyEntry(p, s, json2, converter) {
+  mkdirSync2(p.entryDir, { recursive: true });
+  const raw = p.fulltextRaw("json");
+  writeAtomic(raw, json2);
+  writeAtomic(p.fulltextMd, renderStudyMd(s));
+  const meta = {
+    pmid: "",
     setid: null,
-    nctid: null,
-    publications: [],
+    nctid: s.nctid,
+    publications: s.publications,
     pmcid: null,
     doi: null,
-    title: null,
-    first_author: null,
-    year: null,
-    publication_types: [],
+    title: s.title,
+    first_author: s.sponsor,
+    year: s.firstPosted?.slice(0, 4) ?? null,
+    publication_types: ["Trial Registration"],
     retracted: false,
     corrections: [],
     coi: null,
     funding: [],
-    citable: true,
-    comment: null,
-    has_abstract: null,
-    level: null,
-    version: null,
-    retrieved_by: null,
-    retrieved_date: null,
-    pubmed_checked: null,
-    attempts: [],
-    files: [],
-    glyphs_unresolved: null
+    has_abstract: false,
+    level: "full_text",
+    version: s.lastUpdate,
+    files: [basename(raw), basename(p.fulltextMd)],
+    retrieved_by: "ctgov",
+    retrieved_date: today(),
+    converted_by: converter,
+    pubmed_checked: today(),
+    attempts: []
   };
+  writeMeta(p, meta);
+  return meta;
 }
-var today = () => (/* @__PURE__ */ new Date()).toISOString().slice(0, 10);
-function headLine(r) {
-  return [
-    r.authors.join(", ") || null,
-    r.journal,
-    r.year,
-    r.pmid ? `PMID ${r.pmid}` : "not in PubMed",
-    r.doi ? `DOI ${r.doi}` : null,
-    r.pmcid ? `PMCID ${r.pmcid}` : null
-  ].filter(Boolean).join(" \xB7 ");
+function writeDoiEntry(p, r, json2, converter, abstractFrom = "crossref", recordFrom = "crossref") {
+  mkdirSync2(p.entryDir, { recursive: true });
+  const raw = p.fulltextRaw("json").replace(/-fulltext\.json$/, recordFrom === "crossref" ? "-crossref.json" : "-csl.json");
+  writeAtomic(raw, json2);
+  writeAtomic(p.md, renderAbstractMd(r));
+  const meta = { ...metaFromRecord(r, [basename(raw), basename(p.md)], converter), retrieved_by: r.abstract.length ? abstractFrom : recordFrom };
+  writeMeta(p, meta);
+  return meta;
 }
-function renderAbstractMd(r) {
-  const body = r.abstract.length ? r.abstract.map((a) => a.label ? `**${a.label}.** ${a.text}` : a.text).join("\n\n") : "_No abstract available._";
-  return `# ${r.title}
 
-${headLine(r)}
+// src/walls.ts
+var WALL2 = /^(challenge not cleared|empty response)/i;
+var WALL_STREAK = 2;
+var host = (url) => {
+  try {
+    return new URL(url).hostname.replace(/^www\./, "").toLowerCase();
+  } catch {
+    return "";
+  }
+};
+var file = (dir, h) => join5(dir, "walls", `${h.replace(/[^a-z0-9.-]/g, "_")}.json`);
+function read(dir, h) {
+  try {
+    return JSON.parse(readFileSync4(file(dir, h), "utf8"));
+  } catch {
+    return {};
+  }
+}
+function walled(url, rung2, dir = FH_LIBRARY_DIR) {
+  const h = host(url);
+  if (!h) return null;
+  const r = read(dir, h)[rung2];
+  return r && r.streak >= WALL_STREAK && daysSince(r.date) < settings().retry_after_days ? r : null;
+}
+function recordRung(url, rung2, ok, reason = "", dir = FH_LIBRARY_DIR) {
+  const h = host(url);
+  if (!h) return;
+  const why = reason.replace(/^(browser [^:]+|brightdata): /, "");
+  if (!ok && !WALL2.test(why)) return;
+  const all = read(dir, h);
+  if (ok) {
+    if (!all[rung2]) return;
+    delete all[rung2];
+  } else all[rung2] = { streak: (all[rung2]?.streak ?? 0) + 1, date: today(), reason: why };
+  try {
+    mkdirSync3(join5(dir, "walls"), { recursive: true });
+    writeAtomic(file(dir, h), JSON.stringify(all) + "\n");
+  } catch {
+  }
+}
+var wallNote = (r) => `${r.reason} ${r.streak}\xD7 on this host, last ${r.date}`;
 
-## Abstract
+// src/tiers/tier.ts
+var looked = (t, p, ctx) => ctx.lookups?.get(t.name) ?? t.lookup(p);
+function oaVersion(v) {
+  if (v === "acceptedVersion") return "accepted_manuscript";
+  if (v === "submittedVersion") return "preprint";
+  return "published";
+}
+var miss = (tier, reason) => ({ ok: false, tier, reason });
+var errMsg = (e) => (e instanceof Error ? e.message : String(e)).replace(/\x1b\[[0-9;]*m/g, "").split("\n")[0].trim().slice(0, 200);
 
-${body}
-`;
+// src/download.ts
+var hostOf2 = (url) => {
+  try {
+    return new URL(url).hostname.replace(/^www\./, "");
+  } catch {
+    return url;
+  }
+};
+var NCBI = /^https?:\/\/([a-z0-9.-]+\.)?(ncbi\.nlm\.nih\.gov|pubmed\.gov)\//i;
+function order(candidates) {
+  const rank = (c) => (c.pdf ? 0 : 2) + (c.host === "publisher" ? 1 : 0);
+  return [...candidates].sort((a, b) => rank(a) - rank(b));
+}
+async function download(tier, p, candidates, ctx) {
+  const usable = candidates.filter((c) => !NCBI.test(c.url));
+  if (!usable.length && candidates.length) return miss(tier, "only pmc copies (covered by the pmc tier)");
+  const ordered = order(usable).filter((c) => !ctx.tried.has(c.url));
+  if (!ordered.length) return miss(tier, "only copies already tried");
+  const reasons = [];
+  for (const c of ordered) {
+    ctx.tried.add(c.url);
+    let r = await tryOne(p, c, tier);
+    const target = !r.ok && r.url && r.url !== c.url ? { ...c, url: r.url } : c;
+    if (target !== c) ctx.onTry?.(`${c.url} \u2192 ${target.url}`);
+    ctx.onTry?.(r.ok ? `${target.url} ok fetch` : `${target.url} failed fetch: ${r.reason}`);
+    const why = (reason) => reason.replace(/^not a pdf \((.*)\)$/, "$1");
+    for (const b of browserRungs()) {
+      if (r.ok || c.headers || c.service || !/^(HTTP 403|HTTP 429|wall page|no methods\/results|unknown content|browser )/.test(why(r.reason))) break;
+      const w = walled(target.url, `browser ${b.label}`);
+      if (w) {
+        ctx.onTry?.(`${target.url} skipped browser ${b.label}: ${wallNote(w)}`);
+        continue;
+      }
+      r = await tryOne(p, target, tier, { browser: b.endpoint, label: b.label });
+      recordRung(target.url, `browser ${b.label}`, r.ok, r.ok ? "" : r.reason);
+      ctx.onTry?.(r.ok ? `${target.url} ok browser ${b.label}` : `${target.url} failed browser ${b.label}: ${bare(r.reason)}`);
+      if (!r.ok && bare(r.reason).startsWith(MISMATCH)) ctx.onAlert?.(`browser ${b.label}: ${bare(r.reason)}`);
+    }
+    const uw = walled(target.url, "brightdata");
+    if (uw && !r.ok && !c.headers && !c.service && unlockerConfigured()) ctx.onTry?.(`${target.url} skipped brightdata: ${wallNote(uw)}`);
+    else if (!r.ok && !c.headers && !c.service && unlockerConfigured() && /^(HTTP 403|wall page|browser )/.test(why(r.reason))) {
+      r = await tryOne(p, target, tier, "unlocker");
+      recordRung(target.url, "brightdata", r.ok, r.ok ? "" : r.reason);
+      ctx.onTry?.(r.ok ? `${target.url} ok brightdata` : `${target.url} failed brightdata: ${bare(r.reason)}`);
+      if (!r.ok && /^unlocker HTTP/.test(bare(r.reason))) ctx.onAlert?.(`brightdata: ${bare(r.reason)}`);
+    }
+    if (r.ok) return r;
+    reasons.push(`${r.reason}: ${hostOf2(c.url)}`);
+  }
+  return miss(tier, reasons.join("; "));
+}
+var bare = (reason) => reason.replace(/^(brightdata|browser [^:]+): /, "");
+async function tryOne(p, c, tier, via = "plain") {
+  let got;
+  try {
+    got = via === "unlocker" ? await unlock(c.url) : typeof via === "object" ? await browserFetch(c.url, via.browser) : await fetchBytes(c.url, { ...DOWNLOAD_HEADERS, ...c.headers ?? {} });
+  } catch (e) {
+    const body = e?.body;
+    if (c.service && body?.trim()) return { ok: false, reason: `${errMsg(e)} \u2014 ${serviceMessage(Buffer.from(body), e.contentType)}`, url: e?.url };
+    return { ok: false, reason: (via === "plain" ? "" : via === "unlocker" ? "brightdata: " : `browser ${via.label}: `) + errMsg(e), url: e?.url };
+  }
+  const { bytes, contentType } = got;
+  if (c.service && !looksLikePdf(bytes) && !isXmlArticle(bytes.toString("utf8"))) return { ok: false, reason: `service: ${serviceMessage(bytes, contentType)}`, url: got.url };
+  if (via === "unlocker" && !bytes.length) return { ok: false, reason: "brightdata: empty response" };
+  const r = await convertBytes(p, bytes, contentType);
+  if (!r.ok) return { ok: false, reason: c.pdf && r.format !== "pdf" ? `not a pdf (${r.reason})` : r.reason, url: got.url };
+  return { ok: true, tier, format: r.format, raw: r.format === "pdf" ? bytes : bytes.toString("utf8"), md: r.md, version: c.version };
+}
+var isXmlArticle = (text3) => /^\s*<\?xml|<article\b/.test(text3.slice(0, 2e3)) && !/<html/i.test(text3.slice(0, 2e3));
+function serviceMessage(bytes, contentType = "") {
+  const text3 = bytes.toString("utf8").trim();
+  if (!text3) return `empty response (${contentType.split(";")[0] || "no type"})`;
+  try {
+    const j = JSON.parse(text3);
+    const m = j?.message ?? j?.error?.message ?? j?.error ?? j?.detail;
+    if (typeof m === "string" && m.trim()) return m.trim().replace(/\s+/g, " ").slice(0, 160);
+  } catch {
+  }
+  const visible = text3.replace(/<script[\s\S]*?<\/script>|<style[\s\S]*?<\/style>/gi, " ").replace(/<title>([\s\S]*?)<\/title>/i, "$1. ").replace(/<[^>]+>/g, " ").replace(/&nbsp;/g, " ").replace(/\s+/g, " ").trim();
+  return (visible || text3).slice(0, 160);
+}
+function describe(reason, format, md) {
+  const words2 = md.split(/\s+/).filter(Boolean).length;
+  const headings2 = md.split("\n").filter((l) => /^#{2,6}\s/.test(l)).length;
+  const seen = `${format}, ${words2} words, ${proseWords(md)} in prose, ${headings2} heading${headings2 === 1 ? "" : "s"}`;
+  if (reason === "no methods/results") return `no article body (${seen})`;
+  if (reason === "title missing") return `neither title nor DOI in text (${seen})`;
+  return reason;
+}
+async function convertBytes(p, bytes, contentType = "") {
+  let format;
+  let md;
+  if (looksLikePdf(bytes)) {
+    format = "pdf";
+    const stats2 = {};
+    try {
+      md = await pdfToMarkdown(bytes, p.title, p.headLine, stats2);
+    } catch (e) {
+      return { ok: false, reason: `pdf: ${errMsg(e)}` };
+    }
+    if ((stats2.pages ?? 0) >= 1 && (stats2.chars ?? 0) < 500 * (stats2.pages ?? 1)) return { ok: false, reason: "scanned pdf", format };
+  } else {
+    const text3 = bytes.toString("utf8");
+    if (isWallPage(text3)) return { ok: false, reason: "wall page" };
+    if (/^\s*<\?xml|<article\b/.test(text3.slice(0, 2e3)) && /<body\b/.test(text3) && !/<html/i.test(text3.slice(0, 2e3))) {
+      format = "xml";
+      const doc = parseJats(/full-text-retrieval-response/.test(text3.slice(0, 3e3)) ? elsevierToJats(text3) : text3);
+      if (!doc.title) doc.title = p.title;
+      if (!doc.hasBody) return { ok: false, reason: "xml without body" };
+      md = renderFulltextMd(doc, p.headLine);
+    } else if (/text\/html|<html/i.test(contentType + text3.slice(0, 500))) {
+      format = "html";
+      const body = /<body[^>]*>([\s\S]*)<\/body>/i.exec(text3)?.[1] ?? text3;
+      md = `# ${p.title}
+
+${p.headLine}
+
+${htmlToMarkdown(body)}`;
+    } else {
+      return { ok: false, reason: `unknown content (${contentType.split(";")[0] || "no type"})` };
+    }
+  }
+  const content = format === "xml" ? md : md.replace(/^# [^\n]*\n/, "");
+  const rawText = format === "html" ? bytes.toString("utf8").replace(/<script[\s\S]*?<\/script>|<style[\s\S]*?<\/style>/gi, " ").replace(/<[^>]+>/g, " ") : void 0;
+  const check = checkFulltext(content, p.title, rawText, p.doi, format === "xml");
+  if (!check.ok) return { ok: false, reason: describe(check.reason, format, content), format };
+  return { ok: true, format, md };
+}
+var ELSEVIER_TAGS = [
+  [/ja:article\b/g, "article"],
+  [/ja:body\b/g, "body"],
+  [/ja:head\b/g, "front"],
+  [/ce:sections\b/g, "sec-group"],
+  [/ce:section-title\b/g, "title"],
+  [/ce:section\b/g, "sec"],
+  [/ce:simple-para\b/g, "p"],
+  [/ce:para\b/g, "p"],
+  [/ce:abstract-sec\b/g, "sec"],
+  [/ce:abstract\b/g, "abstract"],
+  [/ce:title\b/g, "article-title"],
+  [/ce:italic\b/g, "italic"],
+  [/ce:bold\b/g, "bold"],
+  [/ce:sup\b/g, "sup"],
+  [/ce:inf\b/g, "sub"],
+  [/ce:cross-refs?\b/g, "xref"],
+  [/ce:list-item\b/g, "list-item"],
+  [/ce:list\b/g, "list"],
+  [/ce:label\b/g, "label"],
+  [/ce:caption\b/g, "caption"],
+  [/ce:table\b/g, "table-wrap"],
+  [/ce:figure\b/g, "fig"],
+  [/ce:bib-reference\b/g, "ref"],
+  [/ce:bibliography\b/g, "ref-list"],
+  [/sb:reference\b/g, "mixed-citation"],
+  [/ce:acknowledgment\b/g, "ack"],
+  [/ce:inter-ref\b/g, "ext-link"]
+];
+function elsevierToJats(xml) {
+  let out = xml;
+  for (const [re, to] of ELSEVIER_TAGS) out = out.replace(re, to);
+  return out.replace(/<\/?full-text-retrieval-response[^>]*>/g, "").replace(/<\/?originalText[^>]*>/g, "").replace(/<\/?xocs:doc[^>]*>/g, "").replace(/<\/?xocs:serial-item[^>]*>/g, "");
+}
+
+// src/tiers/core.ts
+var WINDOW_MS = 6e4;
+var PER_WINDOW = 10;
+var MAX_WAIT_MS = 65e3;
+var TIMEOUT_MS3 = 1e4;
+var calls = [];
+async function pace() {
+  const now = Date.now();
+  while (calls.length && calls[0] <= now - WINDOW_MS) calls.shift();
+  if (calls.length >= PER_WINDOW) await new Promise((r) => setTimeout(r, calls[0] + WINDOW_MS - now + 200));
+  calls.push(Date.now());
 }
-function readHeadLine(p) {
-  try {
-    const lines = readFileSync3(p.md, "utf8").split("\n").filter((l) => l.trim());
-    return lines[1] ?? "";
-  } catch {
-    return "";
+async function coreFetch(url, key) {
+  const headers = { "User-Agent": LIBRARY_UA, Authorization: `Bearer ${key}` };
+  for (let attempt = 0; attempt < 2; attempt++) {
+    await pace();
+    const res = await fetch(url, { headers, signal: AbortSignal.timeout(TIMEOUT_MS3), redirect: "follow" });
+    if (res.ok) return await res.text();
+    const body = await res.text().catch(() => "");
+    if (res.status === 429 && attempt === 0) {
+      const until = Date.parse(res.headers.get("x-ratelimit-retry-after") ?? "");
+      const wait = Number.isFinite(until) ? Math.min(Math.max(until - Date.now(), 1e3), MAX_WAIT_MS) : 1e4;
+      await new Promise((r) => setTimeout(r, wait));
+      continue;
+    }
+    throw new Error(res.status < 500 ? httpReason(res.status, body) : `HTTP ${res.status}`);
   }
+  throw new Error("HTTP 429");
 }
-function metaFromRecord(r, files, converter) {
+function coreTier(apiKey) {
+  const key = apiKey?.trim();
   return {
-    pmid: r.pmid,
-    setid: null,
-    nctid: null,
-    publications: [],
-    pmcid: r.pmcid,
-    doi: r.doi,
-    title: r.title,
-    first_author: r.firstAuthor,
-    year: r.year,
-    publication_types: r.publicationTypes,
-    retracted: isRetracted(r),
-    corrections: r.corrections,
-    coi: r.coi,
-    funding: r.funding,
-    has_abstract: r.abstract.length > 0,
-    level: "abstract",
-    version: null,
-    files,
-    retrieved_by: "pubmed",
-    retrieved_date: today(),
-    converted_by: converter,
-    pubmed_checked: today(),
-    attempts: []
+    name: "core",
+    applies: (p) => p.doi ? true : "no doi",
+    async run(p, ctx) {
+      if (!key) return { ...miss(this.name, "needs api key"), skipped: true };
+      let data;
+      try {
+        const q = encodeURIComponent(`doi:"${p.doi}"`);
+        data = JSON.parse(await coreFetch(`https://api.core.ac.uk/v3/search/works/?q=${q}&limit=5`, key));
+      } catch (e) {
+        return miss(this.name, errMsg(e));
+      }
+      const candidates = coreCandidates(data);
+      if (!candidates.length) return miss(this.name, data.results?.length ? "no full text" : "not in core");
+      ctx.onTry?.(`located ${candidates.length} copies`);
+      return download(this.name, p, candidates, ctx);
+    }
   };
 }
-var isRetracted = (r) => r.publicationTypes.includes("Retracted Publication") || r.corrections.some(isRetractionNotice);
-var renderCorrection = (c) => `${c.type} ${c.pmid ?? "-"} ${JSON.stringify(c.source)}`;
-function parseCorrection(v) {
-  const m = /^(\S+) (\S+) ("[\s\S]*")$/.exec(v.trim());
-  if (!m) return null;
-  let source;
-  try {
-    source = JSON.parse(m[3]);
-  } catch {
-    return null;
+function coreCandidates(d) {
+  const out = [];
+  const seen = /* @__PURE__ */ new Set();
+  const add = (url, pdf, host2) => {
+    if (!url || seen.has(url)) return;
+    seen.add(url);
+    out.push({ url, version: "published", source: "core", host: host2, pdf });
+  };
+  for (const w of d.results ?? []) {
+    add(w.downloadUrl, true, "core");
+    for (const l of w.links ?? []) if (l.type === "download") add(l.url, /\.pdf($|\?)/i.test(l.url ?? ""), "core");
+    for (const u of w.sourceFulltextUrls ?? []) add(u, /\.pdf($|\?)/i.test(u), "repository");
   }
-  return { type: m[1], pmid: m[2] === "-" ? null : m[2], source };
-}
-function renderMeta(m) {
-  const lines = [
-    ["pmid", m.pmid],
-    ...m.setid ? [["setid", m.setid]] : [],
-    ...m.nctid ? [["nctid", m.nctid]] : [],
-    ...m.nctid ? [["publications", m.publications.join(", ")]] : [],
-    ["pmcid", m.pmcid ?? ""],
-    ["doi", m.doi ?? ""],
-    ["title", JSON.stringify(m.title)],
-    ["first_author", m.first_author ?? ""],
-    ["year", m.year ?? ""],
-    ["publication_types", m.publication_types.join("; ")],
-    ["retracted", m.retracted ? "yes" : "no"],
-    ...m.corrections.map((c) => ["corrections", renderCorrection(c)]),
-    ["coi", m.coi === null ? "" : JSON.stringify(m.coi)],
-    ["funding", m.funding.length ? JSON.stringify(m.funding) : ""],
-    // a JSON array: a statement may hold semicolons
-    ["has_abstract", m.has_abstract ? "yes" : "no"],
-    ["level", m.level],
-    ["version", m.version ?? ""],
-    ["files", m.files.join(", ")],
-    ["retrieved_by", m.retrieved_by],
-    ["retrieved_date", m.retrieved_date],
-    ["converted_by", m.converted_by],
-    ["pubmed_checked", m.pubmed_checked],
-    ["attempts", m.attempts.join(", ")]
-  ];
-  return "---\n" + lines.map(([k, v]) => v === "" ? `${k}:` : `${k}: ${v}`).join("\n") + "\n---\n";
+  return out;
 }
-function parseFrontmatter(md) {
-  const m = /^---\n([\s\S]*?)\n---/.exec(md);
-  const fm = {}, all = {};
-  for (const line of (m?.[1] ?? "").split("\n")) {
-    const i = line.indexOf(":");
-    if (i <= 0) continue;
-    const k = line.slice(0, i).trim(), v = line.slice(i + 1).trim();
-    fm[k] = v;
-    (all[k] ??= []).push(v);
+
+// src/tiers/europepmc.ts
+var europepmc = {
+  name: "europepmc",
+  applies: (p) => p.pmid ? true : "no pmid",
+  lookup: (p) => fetchText(`https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=${encodeURIComponent(`EXT_ID:${p.pmid} AND SRC:MED`)}&resultType=core&format=json`, { "User-Agent": LIBRARY_UA }),
+  async run(p, ctx) {
+    let data;
+    try {
+      data = JSON.parse(await looked(this, p, ctx));
+    } catch (e) {
+      return miss(this.name, errMsg(e));
+    }
+    const rec = data.resultList?.result?.[0];
+    if (!rec) return miss(this.name, "not in europe pmc");
+    const candidates = europepmcCandidates(rec, p.pmcid);
+    if (!candidates.length) return miss(this.name, "no free copy");
+    ctx.onTry?.(`located ${candidates.length} copies`);
+    return download(this.name, p, candidates, ctx);
   }
-  return { fm, all };
-}
-var list2 = (v) => v ? v.split(",").map((s) => s.trim()).filter(Boolean) : [];
-function splitAttempts(v) {
-  if (!v) return [];
+};
+function europepmcCandidates(r, pmcid) {
   const out = [];
-  let cur = "", depth = 0;
-  for (const ch of v) {
-    if (ch === "(") depth++;
-    else if (ch === ")") depth = Math.max(0, depth - 1);
-    if (ch === "," && depth === 0) {
-      out.push(cur);
-      cur = "";
-    } else cur += ch;
+  const seen = /* @__PURE__ */ new Set();
+  const add = (url, pdf, host2, version) => {
+    if (!url || seen.has(url)) return;
+    seen.add(url);
+    out.push({ url, version, source: "europepmc", host: host2, pdf });
+  };
+  const id = (r.pmcid ?? pmcid ?? "").toUpperCase() || null;
+  const own = r.isOpenAccess === "Y" ? "published" : "accepted_manuscript";
+  if (id && r.inEPMC === "Y") {
+    if (r.isOpenAccess === "Y") add(`https://www.ebi.ac.uk/europepmc/webservices/rest/${id}/fullTextXML`, false, "europepmc", own);
+    if (r.hasPDF === "Y") add(`https://europepmc.org/articles/${id}?pdf=render`, true, "europepmc", own);
   }
-  out.push(cur);
-  const merged = [];
-  for (const piece of out.map((s) => s.trim()).filter(Boolean)) {
-    if (merged.length && !/^\S+ \(/.test(piece)) merged[merged.length - 1] += ", " + piece;
-    else merged.push(piece);
+  for (const u of r.fullTextUrlList?.fullTextUrl ?? []) {
+    if (!u.url || u.site === "DOI" || !/^(F|OA)$/.test(u.availabilityCode ?? "")) continue;
+    if (u.documentStyle === "abs") continue;
+    if (/^https?:\/\/europepmc\.org\/articles\//i.test(u.url) && u.documentStyle === "html") continue;
+    add(u.url, u.documentStyle === "pdf", u.site?.toLowerCase() ?? "unknown", "published");
   }
-  return merged;
+  return out;
 }
-var opt = (v) => v ? v : null;
-var TYPE_CONTINUATION = /* @__PURE__ */ new Set(["N.I.H.", "Extramural", "Intramural", "Non-U.S. Gov't", "U.S. Gov't", "U.S. Government", "P.H.S.", "Non-P.H.S.", "Phase I", "Phase II", "Phase III", "Phase IV", "Veterinary", "NIH", "American Recovery and Reinvestment Act"]);
-function splitPublicationTypes(v) {
-  if (!v) return [];
-  if (v.includes(";")) return v.split(";").map((s) => s.trim()).filter(Boolean);
+
+// src/tiers/openalex.ts
+var openalex = {
+  name: "openalex",
+  applies: (p) => p.pmid || p.doi ? true : "no pmid or doi",
+  lookup: (p) => fetchText(`https://api.openalex.org/works/${encodeURIComponent(p.pmid ? `pmid:${p.pmid}` : `doi:${p.doi}`)}?mailto=${encodeURIComponent(contactEmail())}`, { "User-Agent": LIBRARY_UA }),
+  async run(p, ctx) {
+    let data;
+    try {
+      data = JSON.parse(await looked(this, p, ctx));
+    } catch (e) {
+      return miss(this.name, errMsg(e));
+    }
+    const candidates = openalexCandidates(data);
+    if (!candidates.length) return miss(this.name, data.open_access?.is_oa === false ? "closed" : "no oa copy");
+    ctx.onTry?.(`located ${candidates.length} copies`);
+    return download(this.name, p, candidates, ctx);
+  }
+};
+function openalexCandidates(d) {
+  const locs = [...d.best_oa_location ? [d.best_oa_location] : [], ...d.locations ?? []].filter((l) => l.is_oa);
+  const seen = /* @__PURE__ */ new Set();
   const out = [];
-  for (const piece of list2(v)) {
-    if (out.length && TYPE_CONTINUATION.has(piece)) out[out.length - 1] += ", " + piece;
-    else out.push(piece);
+  for (const l of locs) {
+    for (const [url, pdf] of [[l.pdf_url, true], [l.landing_page_url, false]]) {
+      if (!url || seen.has(url) || /doi\.org\//.test(url)) continue;
+      seen.add(url);
+      out.push({ url, version: oaVersion(l.version), source: "openalex", host: l.source?.display_name ?? "unknown", pdf });
+    }
   }
   return out;
 }
-function abstractMissing(p) {
-  try {
-    return readFileSync3(p.md, "utf8").includes("_No abstract available._");
-  } catch {
-    return false;
-  }
+
+// src/tiers/pmc.ts
+function pmcEfetchUrl(pmcid) {
+  const u = new URL("https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi");
+  u.searchParams.set("db", "pmc");
+  u.searchParams.set("id", pmcid.replace(/^PMC/i, ""));
+  u.searchParams.set("retmode", "xml");
+  if (process.env.FH_LIBRARY_NCBI_API_KEY) u.searchParams.set("api_key", process.env.FH_LIBRARY_NCBI_API_KEY);
+  u.searchParams.set("tool", "fh-library");
+  u.searchParams.set("email", contactEmail());
+  return u.toString();
 }
-function correctionsFromXml(p) {
-  try {
-    return parsePubmedXml(readFileSync3(p.xml, "utf8"))?.corrections ?? [];
-  } catch {
-    return [];
+var pmc = {
+  name: "pmc",
+  applies: (p) => p.pmcid ? true : "no pmcid",
+  async run(p) {
+    let xml;
+    try {
+      xml = await fetchText(pmcEfetchUrl(p.pmcid));
+    } catch (e) {
+      return miss(this.name, errMsg(e));
+    }
+    const doc = parseJats(xml);
+    if (!doc.hasBody) return miss(this.name, "no body");
+    const md = renderFulltextMd(doc, p.headLine);
+    const check = checkFulltext(md, p.title);
+    if (!check.ok) return miss(this.name, check.reason);
+    return { ok: true, tier: this.name, format: "xml", raw: xml, md, version: "published" };
   }
-}
-function declarationsFromXml(p) {
+};
+
+// src/tiers/tdm.ts
+var CROSSREF = "https://api.crossref.org/works/";
+var TOKENED = [
+  { host: /(^|\.)api\.elsevier\.com$/i, key: "elsevier", env: "ELSEVIER_API_KEY", header: "X-ELS-APIKey" },
+  { host: /(^|\.)api\.wiley\.com$/i, key: "wiley", env: "WILEY_TDM_TOKEN", header: "Wiley-TDM-Client-Token" }
+];
+var hostOf3 = (url) => {
   try {
-    const r = parsePubmedXml(readFileSync3(p.xml, "utf8"));
-    return { coi: r?.coi ?? null, funding: r?.funding ?? [] };
+    return new URL(url).hostname;
   } catch {
-    return { coi: null, funding: [] };
+    return "";
   }
-}
-function readMeta(p) {
-  const { fm, all } = parseFrontmatter(readFileSync3(p.meta, "utf8"));
-  let title = fm.title ?? "";
-  if (title.startsWith('"')) {
-    try {
-      title = JSON.parse(title);
-    } catch {
+};
+function tdmCandidates(work, tokens2 = {}, env = process.env) {
+  const links = (work.message?.link ?? []).filter((l) => l["intended-application"] === "text-mining" && l.URL);
+  const seen = /* @__PURE__ */ new Set();
+  const candidates = [];
+  const missingTokens = /* @__PURE__ */ new Set();
+  for (const l of links) {
+    const ct = (l["content-type"] ?? "").toLowerCase();
+    if (seen.has(l.URL)) {
+      const prev = candidates.find((c) => c.url === l.URL);
+      if (prev && /pdf/.test(ct)) prev.pdf = true;
+      continue;
     }
-  }
-  const legacy = fm.pubmed_checked === void 0;
-  const corrections = legacy ? correctionsFromXml(p) : (all.corrections ?? []).map(parseCorrection).filter((c) => c !== null);
-  let coi = null, funding = [];
-  if (fm.coi === void 0) ({ coi, funding } = declarationsFromXml(p));
-  else {
-    if (fm.coi.startsWith('"')) {
-      try {
-        coi = JSON.parse(fm.coi);
-      } catch {
-        coi = fm.coi;
+    seen.add(l.URL);
+    const rule = TOKENED.find((r) => r.host.test(hostOf3(l.URL)));
+    let headers;
+    if (rule) {
+      const token = tokens2[rule.key]?.trim() || env[rule.env]?.trim();
+      if (!token) {
+        missingTokens.add(rule.key);
+        continue;
       }
-    } else coi = fm.coi || null;
-    funding = !fm.funding ? [] : fm.funding.startsWith("[") ? JSON.parse(fm.funding) : fm.funding.split(";").map((s) => s.trim()).filter(Boolean);
+      headers = { [rule.header]: token };
+    }
+    if (/text\/plain/.test(ct)) continue;
+    candidates.push({ url: l.URL, version: "published", source: "tdm", host: work.message?.publisher ?? hostOf3(l.URL), pdf: /pdf/.test(ct) || /\.pdf(\?|$)/i.test(l.URL), headers });
   }
+  const rank = (c) => /\/xml|text\/xml/.test(c.url) ? 0 : c.pdf ? 1 : 2;
+  candidates.sort((a, b) => rank(a) - rank(b));
+  return { candidates, missingTokens: [...missingTokens] };
+}
+function tdmTier(tokens2 = {}) {
   return {
-    pmid: fm.pmid ?? "",
-    setid: opt(fm.setid),
-    nctid: opt(fm.nctid),
-    publications: list2(fm.publications),
-    pmcid: opt(fm.pmcid),
-    doi: opt(fm.doi),
-    title,
-    first_author: opt(fm.first_author),
-    year: opt(fm.year),
-    publication_types: splitPublicationTypes(fm.publication_types),
-    retracted: fm.retracted === "yes" || corrections.some(isRetractionNotice),
-    corrections,
-    coi,
-    funding,
-    has_abstract: fm.has_abstract ? fm.has_abstract === "yes" : !abstractMissing(p),
-    level: fm.level === "full_text" ? "full_text" : "abstract",
-    version: opt(fm.version),
-    files: list2(fm.files),
-    retrieved_by: fm.retrieved_by ?? "",
-    retrieved_date: fm.retrieved_date ?? "",
-    converted_by: fm.converted_by ?? "",
-    pubmed_checked: legacy ? fm.retrieved_date ?? "" : fm.pubmed_checked,
-    attempts: splitAttempts(fm.attempts)
+    name: "tdm",
+    applies: (p) => p.doi ? true : "no doi",
+    lookup: (p) => fetchText(CROSSREF + encodeURIComponent(p.doi), { "User-Agent": `${LIBRARY_UA} (mailto:${contactEmail()})` }),
+    async run(p, ctx) {
+      let work;
+      try {
+        work = JSON.parse(await looked(this, p, ctx));
+      } catch (e) {
+        return miss(this.name, errMsg(e));
+      }
+      const { candidates, missingTokens } = tdmCandidates(work, tokens2);
+      if (!candidates.length) return missingTokens.length ? { ...miss(this.name, `needs ${missingTokens.join("/")} token`), skipped: true } : miss(this.name, "no text-mining links");
+      ctx.onTry?.(`${candidates.length} text-mining links${missingTokens.length ? ` (skipped hosts needing ${missingTokens.join("/")} token)` : ""}`);
+      return download(this.name, p, candidates, ctx);
+    }
   };
 }
-function writeAtomic(file2, data) {
-  const tmp = `${file2}.${process.pid}.${Math.random().toString(36).slice(2, 8)}.tmp`;
-  writeFileSync(tmp, data);
-  renameSync2(tmp, file2);
-}
-function writeMeta(p, m) {
-  writeAtomic(p.meta, renderMeta(m));
-  try {
-    unlinkSync(join5(p.entryDir, `${p.entry}-missing.md`));
-  } catch {
+var tdm = tdmTier();
+
+// src/tiers/unpaywall.ts
+var unpaywall = {
+  name: "unpaywall",
+  applies: (p) => p.doi ? true : "no doi",
+  lookup: (p) => fetchText(`https://api.unpaywall.org/v2/${encodeURIComponent(p.doi)}?email=${encodeURIComponent(contactEmail())}`, { "User-Agent": LIBRARY_UA }),
+  async run(p, ctx) {
+    let data;
+    try {
+      data = JSON.parse(await looked(this, p, ctx));
+    } catch (e) {
+      return miss(this.name, errMsg(e));
+    }
+    const candidates = unpaywallCandidates(data);
+    if (!candidates.length) return miss(this.name, data.is_oa === false ? "closed" : "no oa copy");
+    ctx.onTry?.(`located ${candidates.length} copies`);
+    return download(this.name, p, candidates, ctx);
   }
-}
-function copyEntry(from, to) {
-  const tmp = `${to.entryDir}.${process.pid}.tmp`, old = `${to.entryDir}.${process.pid}.old`;
-  rmSync(tmp, { recursive: true, force: true });
-  mkdirSync3(tmp, { recursive: true });
-  const rename = (n) => n.startsWith(from.entry) ? to.entry + n.slice(from.entry.length) : n;
-  for (const n of readdirSync(from.entryDir)) if (join5(from.entryDir, n) !== from.meta && !n.endsWith(".tmp")) copyFileSync(join5(from.entryDir, n), join5(tmp, rename(n)));
-  const m = readMeta(from);
-  const copy = { ...m, files: m.files.map(rename) };
-  writeFileSync(join5(tmp, basename(to.meta)), renderMeta(copy));
-  mkdirSync3(dirname(to.entryDir), { recursive: true });
-  if (existsSync4(to.entryDir)) renameSync2(to.entryDir, old);
-  renameSync2(tmp, to.entryDir);
-  rmSync(old, { recursive: true, force: true });
-  return copy;
-}
-var NOTHING_TO_VERIFY = "no abstract and no full text: nothing to verify";
-function citability(m) {
-  const of = (t) => m.corrections.filter((c) => c.type === t);
-  const cite = (c) => `${c.source}${c.pmid ? ` (pmid:${c.pmid})` : ""}`;
-  if (m.retracted) {
-    const r = m.corrections.filter(isRetractionNotice);
-    return { citable: false, comment: r.length ? `retracted: ${r.map(cite).join("; ")}` : "retracted (PubMed publication type)" };
+};
+function unpaywallCandidates(d) {
+  const locs = [...d.best_oa_location ? [d.best_oa_location] : [], ...d.oa_locations ?? []];
+  const seen = /* @__PURE__ */ new Set();
+  const out = [];
+  for (const l of locs) {
+    for (const [url, pdf] of [[l.url_for_pdf, true], [l.url ?? l.url_for_landing_page, false]]) {
+      if (!url || seen.has(url)) continue;
+      seen.add(url);
+      out.push({ url, version: oaVersion(l.version), source: "unpaywall", host: l.host_type ?? "unknown", pdf });
+    }
   }
-  if (m.level !== "full_text" && !m.has_abstract) return { citable: false, comment: NOTHING_TO_VERIFY };
-  const notes = [
-    ...of("erratum_in").map((c) => `erratum, not assessed: ${cite(c)}`),
-    ...of("expression_of_concern_in").map((c) => `expression of concern: ${cite(c)}`),
-    ...of("update_in").map((c) => `superseded by a newer version: ${cite(c)}`)
-  ];
-  return { citable: true, comment: notes.length ? notes.join("; ") : null };
+  return out;
 }
-function statusFromMeta(p, m, base) {
-  const full = existsSync4(p.fulltextMd) ? readFileSync3(p.fulltextMd, "utf8") : null;
-  const cit = citability(m);
-  if (full?.includes(`
-${CORRECTED_HEADING}
-`) && cit.comment) cit.comment = cit.comment.replace(/erratum, not assessed: /g, "erratum, corrected figures stored (Corrected in PubMed abstract): ");
+
+// src/tiers/index.ts
+var BUILTIN = [
+  { name: "pmc", make: () => pmc },
+  { name: "europepmc", make: () => europepmc },
+  { name: "tdm", make: (o) => tdmTier(o.tokens ?? {}) },
+  { name: "unpaywall", make: () => unpaywall },
+  { name: "openalex", make: () => openalex },
+  { name: "core", make: (o) => coreTier(o.api_key) }
+];
+
+// src/tiers/url.ts
+var PLACEHOLDERS = ["doi", "pmid", "pmcid"];
+function urlTier(name, template, headers) {
+  const used = PLACEHOLDERS.filter((k) => template.includes(`{${k}}`));
   return {
-    ...base,
-    entry: p.entry,
-    entry_dir: p.entryDir,
-    pmid: m.pmid || null,
-    setid: m.setid,
-    nctid: m.nctid,
-    publications: m.publications,
-    pmcid: m.pmcid,
-    doi: m.doi,
-    title: m.title || null,
-    first_author: m.first_author,
-    year: m.year,
-    publication_types: m.publication_types,
-    retracted: m.retracted,
-    corrections: m.corrections,
-    coi: m.coi,
-    funding: m.funding,
-    ...cit,
-    // What the entry holds: full_text, abstract, or only the title (the meta's
-    // level is "abstract" until a full text arrives; has_abstract says whether
-    // there is abstract text).
-    has_abstract: m.has_abstract,
-    level: m.level === "abstract" && !m.has_abstract ? "title" : m.level,
-    version: m.version,
-    retrieved_by: m.retrieved_by || null,
-    retrieved_date: m.retrieved_date || null,
-    pubmed_checked: m.pubmed_checked || null,
-    attempts: m.attempts,
-    files: m.files,
-    glyphs_unresolved: full === null ? null : countUnresolved(full)
+    name,
+    applies(p) {
+      const missing = used.filter((k) => !p[k]);
+      return missing.length ? `no ${missing.join("/")}` : true;
+    },
+    async run(p, ctx) {
+      if (!used.length) return miss(name, "template has no placeholder");
+      const url = used.reduce((u, k) => u.replaceAll(`{${k}}`, encodeURIComponent(p[k])), template);
+      return download(name, p, [{ url, version: "published", source: name, host: "user", pdf: /\.pdf(\?|$)/i.test(url), headers, service: true }], ctx);
+    }
   };
 }
-function readStatus(p, base) {
-  return statusFromMeta(p, readMeta(p), base);
-}
-function entryExists(p) {
-  return existsSync4(p.meta);
-}
-function writeEntry(p, r, xml, converter) {
-  mkdirSync3(p.entryDir, { recursive: true });
-  writeAtomic(p.xml, xml);
-  writeAtomic(p.md, renderAbstractMd(r));
-  const meta = metaFromRecord(r, [basename(p.xml), basename(p.md)], converter);
-  writeMeta(p, meta);
-  return meta;
-}
-function refreshRecord(p, m, r, xml) {
-  writeAtomic(p.xml, xml);
-  syncCorrectedBlock(p);
-  const next = {
-    ...m,
-    pmcid: r.pmcid ?? m.pmcid,
-    doi: r.doi ?? m.doi,
-    publication_types: r.publicationTypes,
-    retracted: isRetracted(r),
-    corrections: r.corrections,
-    coi: r.coi ?? m.coi,
-    funding: r.funding.length ? r.funding : m.funding,
-    // the paper's own statements stay when PubMed has none
-    pubmed_checked: today()
+
+// src/config.ts
+var VARIABLES = [
+  "FH_LIBRARY_DIR",
+  "FH_LIBRARY_ENV",
+  "FH_LIBRARY_TIERS",
+  "FH_LIBRARY_CONTACT_EMAIL",
+  "FH_LIBRARY_ELSEVIER_API_KEY",
+  "FH_LIBRARY_WILEY_TDM_TOKEN",
+  "FH_LIBRARY_CORE_API_KEY",
+  "FH_LIBRARY_PROXY_URL",
+  "FH_LIBRARY_PROXY_HEADERS",
+  "FH_LIBRARY_PLAYWRIGHT_1",
+  "FH_LIBRARY_PLAYWRIGHT_2",
+  "FH_LIBRARY_CHROME_1",
+  "FH_LIBRARY_PLAYWRIGHT_HEADLESS",
+  "FH_LIBRARY_VERIFY_URL",
+  "FH_LIBRARY_VERIFY_API_KEY",
+  "FH_LIBRARY_VERIFY_MODEL",
+  "FH_LIBRARY_VERIFY_THRESHOLD",
+  "FH_LIBRARY_VERIFY_TIMEOUT_SECS",
+  "FH_LIBRARY_RETRY_AFTER_DAYS",
+  "FH_LIBRARY_RETRY_BLOCKED",
+  "FH_LIBRARY_REFRESH_AFTER_DAYS",
+  "FH_LIBRARY_SEARCH_CACHE_DAYS",
+  "FH_LIBRARY_NCBI_API_KEY",
+  "FH_LIBRARY_BRIGHTDATA_API_TOKEN",
+  "FH_LIBRARY_BRIGHTDATA_ZONE"
+];
+var sources = (env = process.env) => Object.fromEntries(VARIABLES.map((v) => [v, sourceOf(v, env)]));
+function settings(env = process.env) {
+  return {
+    retry_after_days: num("FH_LIBRARY_RETRY_AFTER_DAYS", 1, env),
+    retry_blocked: bool("FH_LIBRARY_RETRY_BLOCKED", false, env),
+    refresh_after_days: num("FH_LIBRARY_REFRESH_AFTER_DAYS", 30, env),
+    search_cache_days: num("FH_LIBRARY_SEARCH_CACHE_DAYS", 7, env)
   };
-  writeMeta(p, next);
-  return next;
-}
-function daysSince(date, now = /* @__PURE__ */ new Date()) {
-  const t = new Date(date).getTime();
-  return Number.isFinite(t) ? (now.getTime() - t) / 864e5 : Infinity;
 }
-var FUNDING_RE = /^[*_\s]*(funding|financial support|sources? of (financial )?support|support(ed by)?|grant support|this (work|study|research) was (supported|funded))\b[^:]*:?[*_\s]*(.*)$/i;
-var COI_RE = /^[*_\s]*(conflicts? of interest|competing interests?|disclosures?|financial disclosures?|declarations? of (competing |conflicting )?interests?|conflict[- ]of[- ]interest statement)\b[^:]*:?[*_\s]*(.*)$/i;
-function declarationsFromMd(md) {
-  let coi = null;
-  const funding = [];
-  const lines = md.split("\n");
-  let section2 = "";
-  for (let i = 0; i < lines.length; i++) {
-    const l = lines[i].trim();
-    const h = /^#{2,6}\s+(.*)$/.exec(l);
-    if (h) {
-      section2 = h[1];
-      const body = lines.slice(i + 1).find((x) => x.trim() && !/^#/.test(x))?.trim() ?? "";
-      if (body && /^(funding|financial support|sources? of funding|role of the funding source)/i.test(section2) && !funding.length) funding.push(body);
-      if (body && /^(conflicts? of interest|competing interests?|disclosures?|financial disclosures?|declarations? of (competing |conflicting )?interests?)/i.test(section2) && coi === null) coi = body;
-      continue;
-    }
-    if (section2 !== "Declarations" || !l) continue;
-    const f = FUNDING_RE.exec(l);
-    const c = COI_RE.exec(l);
-    if (c && c[3]?.trim() && coi === null) coi = c[3].trim().replace(/[*_]+$/, "");
-    else if (f && f[6]?.trim() && !funding.length) funding.push(f[6].trim().replace(/[*_]+$/, ""));
+function proxyHeaders(spec) {
+  const out = {};
+  for (const part of spec.split(";")) {
+    const i = part.indexOf(":");
+    if (i < 0) continue;
+    const name = part.slice(0, i).trim(), value = part.slice(i + 1).trim();
+    if (name && value) out[name] = value;
   }
-  return { coi, funding };
+  return out;
 }
-var CORRECTED_HEADING = "## Corrected in PubMed abstract";
-function correctedSentences(r) {
-  const out = [];
-  for (const a of r.abstract) {
-    for (const s of a.text.split(/(?<=[.!?\]])\s+(?=[A-Z(\u2212-])/)) {
-      const i = s.indexOf("[corrected]");
-      if (i >= 0) out.push(s.slice(0, i).trim().replace(/\s+([.,;])$/, "$1"));
-    }
+function resolveTiers(env = process.env) {
+  const problems = [];
+  const available = /* @__PURE__ */ new Map();
+  const tokens2 = { elsevier: str("FH_LIBRARY_ELSEVIER_API_KEY", env) || void 0, wiley: str("FH_LIBRARY_WILEY_TDM_TOKEN", env) || void 0 };
+  for (const t of BUILTIN) available.set(t.name, t.make({ tokens: tokens2, api_key: str("FH_LIBRARY_CORE_API_KEY", env) || void 0 }));
+  const proxyUrl = str("FH_LIBRARY_PROXY_URL", env);
+  if (proxyUrl) available.set("proxy", urlTier("proxy", proxyUrl, proxyHeaders(str("FH_LIBRARY_PROXY_HEADERS", env))));
+  const listed = str("FH_LIBRARY_TIERS", env);
+  const order2 = listed ? listed.split(",").map((s) => s.trim()).filter(Boolean) : [...available.keys()];
+  const tiers = [], seen = /* @__PURE__ */ new Set();
+  for (const name of order2) {
+    if (seen.has(name)) continue;
+    seen.add(name);
+    const t = available.get(name);
+    if (t) tiers.push(t);
+    else if (name === "proxy") problems.push('FH_LIBRARY_TIERS: "proxy" listed, but FH_LIBRARY_PROXY_URL is unset');
+    else problems.push(`FH_LIBRARY_TIERS: unknown tier "${name}" (built-in: ${BUILTIN.map((t2) => t2.name).join(", ")}; proxy when FH_LIBRARY_PROXY_URL is set)`);
   }
-  return out.filter(Boolean);
+  const all = [...tiers.map((t) => ({ name: t.name, enabled: true })), ...[...available.keys()].filter((n) => !seen.has(n)).map((name) => ({ name, enabled: false }))];
+  return { tiers, all, source: sourceOf("FH_LIBRARY_TIERS", env), problems };
 }
-function withCorrectedBlock(md, sentences) {
-  const base = md.split(`
-${CORRECTED_HEADING}
-`)[0].trimEnd();
-  if (!sentences.length) return base + "\n";
-  return `${base}
-
-${CORRECTED_HEADING}
-
-NLM corrected these sentences of the abstract after an erratum; the text above may predate the correction.
 
-${sentences.join("\n\n")}
-`;
+// src/log.ts
+import { appendFileSync, mkdirSync as mkdirSync4, renameSync as renameSync2, statSync } from "fs";
+import { hostname } from "os";
+import { join as join6 } from "path";
+var HOST = hostname();
+var LOG_FILE = join6(FH_LIBRARY_DIR, "library.log");
+function stamp() {
+  const d = /* @__PURE__ */ new Date();
+  const p = (n) => String(n).padStart(2, "0");
+  return `${d.getFullYear()}-${p(d.getMonth() + 1)}-${p(d.getDate())} ${p(d.getHours())}:${p(d.getMinutes())}:${p(d.getSeconds())}`;
 }
-function syncCorrectedBlock(p) {
-  if (!existsSync4(p.fulltextMd) || !existsSync4(p.xml)) return;
-  let r;
+var LOG_MAX_BYTES = 50 * 1024 * 1024;
+function rotateIfLarge(file3, max = LOG_MAX_BYTES) {
+  let size = 0;
   try {
-    r = parsePubmedXml(readFileSync3(p.xml, "utf8"));
+    size = statSync(file3).size;
   } catch {
-    return;
+    return false;
+  }
+  if (size <= max) return false;
+  try {
+    renameSync2(file3, `${file3}.1`);
+    return true;
+  } catch {
+    return false;
   }
-  if (!r) return;
-  const md = readFileSync3(p.fulltextMd, "utf8");
-  const next = withCorrectedBlock(md, correctedSentences(r));
-  if (next !== md) writeAtomic(p.fulltextMd, next);
-}
-function applyFulltext(p, m, hit, converter) {
-  const raw = p.fulltextRaw(hit.format);
-  writeAtomic(raw, hit.raw);
-  writeAtomic(p.fulltextMd, hit.md);
-  syncCorrectedBlock(p);
-  const files = [...m.files.filter((f) => !f.includes("-fulltext.")), basename(raw), basename(p.fulltextMd)];
-  const own = declarationsFromMd(hit.md);
-  const next = {
-    ...m,
-    level: "full_text",
-    version: hit.version,
-    files,
-    retrieved_by: hit.tier,
-    retrieved_date: today(),
-    converted_by: converter,
-    coi: m.coi ?? own.coi,
-    funding: m.funding.length ? m.funding : own.funding
-  };
-  writeMeta(p, next);
-  return next;
-}
-function recordAttempts(p, m, labels) {
-  if (!labels.length) return m;
-  const tierOf = (a) => /^(\S+) \(/.exec(a.trim())?.[1] ?? a;
-  const replaced = new Set(labels.map(tierOf));
-  const next = { ...m, attempts: [...m.attempts.filter((a) => !replaced.has(tierOf(a))), ...labels] };
-  writeMeta(p, next);
-  return next;
-}
-function writeLabelEntry(p, s, xml, converter) {
-  mkdirSync3(p.entryDir, { recursive: true });
-  const raw = p.fulltextRaw("xml");
-  writeAtomic(raw, xml);
-  writeAtomic(p.fulltextMd, renderLabelMd(s));
-  const meta = {
-    pmid: "",
-    setid: s.setid,
-    nctid: null,
-    publications: [],
-    pmcid: null,
-    doi: null,
-    title: s.title,
-    first_author: s.labeler,
-    year: s.effective?.slice(0, 4) ?? null,
-    publication_types: ["Drug Label"],
-    retracted: false,
-    corrections: [],
-    coi: null,
-    funding: [],
-    has_abstract: false,
-    level: "full_text",
-    version: s.version ? `spl ${s.version}` : null,
-    files: [basename(raw), basename(p.fulltextMd)],
-    retrieved_by: "dailymed",
-    retrieved_date: today(),
-    converted_by: converter,
-    pubmed_checked: today(),
-    attempts: []
-  };
-  writeMeta(p, meta);
-  return meta;
-}
-function writeStudyEntry(p, s, json2, converter) {
-  mkdirSync3(p.entryDir, { recursive: true });
-  const raw = p.fulltextRaw("json");
-  writeAtomic(raw, json2);
-  writeAtomic(p.fulltextMd, renderStudyMd(s));
-  const meta = {
-    pmid: "",
-    setid: null,
-    nctid: s.nctid,
-    publications: s.publications,
-    pmcid: null,
-    doi: null,
-    title: s.title,
-    first_author: s.sponsor,
-    year: s.firstPosted?.slice(0, 4) ?? null,
-    publication_types: ["Trial Registration"],
-    retracted: false,
-    corrections: [],
-    coi: null,
-    funding: [],
-    has_abstract: false,
-    level: "full_text",
-    version: s.lastUpdate,
-    files: [basename(raw), basename(p.fulltextMd)],
-    retrieved_by: "ctgov",
-    retrieved_date: today(),
-    converted_by: converter,
-    pubmed_checked: today(),
-    attempts: []
-  };
-  writeMeta(p, meta);
-  return meta;
 }
-function writeDoiEntry(p, r, json2, converter, abstractFrom = "crossref", recordFrom = "crossref") {
-  mkdirSync3(p.entryDir, { recursive: true });
-  const raw = p.fulltextRaw("json").replace(/-fulltext\.json$/, recordFrom === "crossref" ? "-crossref.json" : "-csl.json");
-  writeAtomic(raw, json2);
-  writeAtomic(p.md, renderAbstractMd(r));
-  const meta = { ...metaFromRecord(r, [basename(raw), basename(p.md)], converter), retrieved_by: r.abstract.length ? abstractFrom : recordFrom };
-  writeMeta(p, meta);
-  return meta;
+function log(tool, id, message) {
+  const clean2 = message.replace(/\x1b\[[0-9;]*m/g, "").replace(/\s*\n\s*/g, " | ");
+  const line = `${stamp()} ${HOST} ${tool.replace(/^library_/, "")} ${id ?? "-"} ${clean2}
+`;
+  try {
+    mkdirSync4(FH_LIBRARY_DIR, { recursive: true });
+    rotateIfLarge(LOG_FILE);
+    appendFileSync(LOG_FILE, line);
+  } catch {
+  }
 }
 
 // src/stats.ts
+import { mkdirSync as mkdirSync5, readdirSync as readdirSync2, readFileSync as readFileSync5, statSync as statSync2, unlinkSync as unlinkSync2 } from "fs";
+import { join as join7 } from "path";
 var HEARTBEAT_MS = 5 * 6e4;
 var STALE_MS = 20 * 6e4;
 var stats = null;
-var file = null;
+var file2 = null;
 function startStats(pkg2, libraryDir = FH_LIBRARY_DIR) {
   stats = { host: HOST, pid: process.pid, package: pkg2.name, version: pkg2.version, started: (/* @__PURE__ */ new Date()).toISOString(), verify: { hits: 0, misses: 0 }, fetch: { hits: 0, misses: 0 }, search: { hits: 0, misses: 0 }, pubmed: { hits: 0, misses: 0 } };
-  file = join6(libraryDir, ".servers", `${HOST}-${process.pid}.json`);
-  sweep(join6(libraryDir, ".servers"));
+  file2 = join7(libraryDir, ".servers", `${HOST}-${process.pid}.json`);
+  sweep(join7(libraryDir, ".servers"));
   publish();
   setInterval(publish, HEARTBEAT_MS).unref();
   const stop = () => {
     try {
-      if (file) unlinkSync2(file);
+      if (file2) unlinkSync2(file2);
     } catch {
     }
   };
@@ -2912,7 +2966,7 @@ function sweep(dir, staleMs = STALE_MS) {
   }
   for (const name of names) {
     if (!name.endsWith(".json")) continue;
-    const p = join6(dir, name);
+    const p = join7(dir, name);
     try {
       if (Date.now() - statSync2(p).mtimeMs > staleMs) {
         unlinkSync2(p);
@@ -2930,16 +2984,16 @@ function bump(kind, hit, n = 1) {
 }
 var currentStats = () => stats;
 function publish() {
-  if (!stats || !file) return;
+  if (!stats || !file2) return;
   try {
-    mkdirSync4(join6(file, ".."), { recursive: true });
-    writeAtomic(file, JSON.stringify(stats, null, 2) + "\n");
+    mkdirSync5(join7(file2, ".."), { recursive: true });
+    writeAtomic(file2, JSON.stringify(stats, null, 2) + "\n");
   } catch {
   }
 }
 
 // src/tools.ts
-import { existsSync as existsSync12, readFileSync as readFileSync12 } from "fs";
+import { existsSync as existsSync12, readFileSync as readFileSync13 } from "fs";
 import { z } from "zod";
 
 // src/cascade.ts
@@ -2971,10 +3025,18 @@ function shouldSkip(attempts, tier, retryAfterDays, now = /* @__PURE__ */ new Da
 async function fulltextCascade(p, attempts, onEvent, tiers = resolveTiers().tiers, retryAfterDays = settings().retry_after_days, retryBlocked = settings().retry_blocked) {
   const misses = [];
   const tried = /* @__PURE__ */ new Set();
+  const lookups = /* @__PURE__ */ new Map();
+  for (const tier of tiers) {
+    if (!tier.lookup || shouldSkip(attempts, tier.name, retryAfterDays, /* @__PURE__ */ new Date(), retryBlocked) || tier.applies(p) !== true) continue;
+    const q = tier.lookup(p);
+    q.catch(() => {
+    });
+    lookups.set(tier.name, q);
+  }
   for (const tier of tiers) {
     if (shouldSkip(attempts, tier.name, retryAfterDays, /* @__PURE__ */ new Date(), retryBlocked)) continue;
     const applies = tier.applies(p);
-    const ctx = { tried, onTry: (message) => onEvent?.({ ok: "try", tier: tier.name, message }), onAlert: (message) => onEvent?.({ ok: "alert", tier: tier.name, message }) };
+    const ctx = { tried, lookups, onTry: (message) => onEvent?.({ ok: "try", tier: tier.name, message }), onAlert: (message) => onEvent?.({ ok: "alert", tier: tier.name, message }) };
     const r = applies === true ? await tier.run(p, ctx) : { ...miss(tier.name, applies), skipped: true };
     onEvent?.(r);
     if (r.ok) return { hit: r, misses };
@@ -3014,11 +3076,11 @@ function parseId(raw) {
 }
 
 // src/info.ts
-import { existsSync as existsSync6, readdirSync as readdirSync3, readFileSync as readFileSync6, statSync as statSync3 } from "fs";
-import { dirname as dirname2, join as join7 } from "path";
+import { existsSync as existsSync6, readdirSync as readdirSync3, readFileSync as readFileSync7, statSync as statSync3 } from "fs";
+import { dirname as dirname2, join as join8 } from "path";
 
 // src/verify.ts
-import { existsSync as existsSync5, readFileSync as readFileSync5 } from "fs";
+import { existsSync as existsSync5, readFileSync as readFileSync6 } from "fs";
 var DEFAULT_MODEL = "pplx-decider-flash-v1.2";
 var QUESTION = "Does the paper state this, in substance? Ignore section labels and definitions of terms in parentheses. Every fact and number of the statement must be stated in the paper or follow from its numbers; a rounded number counts. If any part of the statement is absent from the paper or contradicted by it, answer false.";
 function verifyConfig(env = process.env) {
@@ -3051,7 +3113,26 @@ function isLocal(baseUrl) {
   }
 }
 function cleanMarkdown(md) {
-  return md.replace(/!\[[^\]]*\]\([^)]*\)/g, "").replace(/\]\([^)]*\)/g, "]").replace(/^\s*#+\s*(see more about|more for you|back to top)\b.*$/gim, "").replace(/\n{3,}/g, "\n\n");
+  return paperYearOnly(dropReferences(md)).replace(/!\[[^\]]*\]\([^)]*\)/g, "").replace(/\]\([^)]*\)/g, "]").replace(/^\s*#+\s*(see more about|more for you|back to top)\b.*$/gim, "").replace(/\n{3,}/g, "\n\n");
+}
+var REF_HEAD = /^(#{1,6})\s*\**\s*(references|bibliography|literature cited|reference list|works cited|cited literature)\b[^\n]*$|^\**(references|bibliography|literature cited)\**:?[ \t]*$/gim;
+function dropReferences(md) {
+  let out = md;
+  for (; ; ) {
+    REF_HEAD.lastIndex = 0;
+    const m = REF_HEAD.exec(out);
+    if (!m) return out;
+    const level = m[1]?.length ?? 6;
+    const rest = out.slice(m.index + m[0].length);
+    const next = new RegExp(`^#{1,${level}}\\s`, "m").exec(rest);
+    out = out.slice(0, m.index) + (next ? rest.slice(next.index) : "");
+  }
+}
+function paperYearOnly(md) {
+  return md.replace(/^(#[^\n]*\n\n)([^\n]*· (?:PMID \d+|DOI 10\.)[^\n]*)$/m, (_, title, line) => {
+    const year = / · (\d{4})(?: · |$)/.exec(line)?.[1];
+    return year ? `${title}${year}` : title.trimEnd();
+  });
 }
 function windows(md, max) {
   const out = [];
@@ -3121,11 +3202,11 @@ async function readPaper(raw, libraryDir = FH_LIBRARY_DIR) {
   const paths = entryPaths(libraryDir, p.type, p.id);
   if (!entryExists(paths)) return { id: p.requested, error: "not in library" };
   const full = existsSync5(paths.fulltextMd);
-  if (full) return { id: p.requested, basis: "full_text", text: readFileSync5(paths.fulltextMd, "utf8") };
+  if (full) return { id: p.requested, basis: "full_text", text: readFileSync6(paths.fulltextMd, "utf8") };
   const meta = readMeta(paths);
-  return { id: p.requested, basis: meta.has_abstract ? "abstract" : "title", text: readFileSync5(paths.md, "utf8") };
+  return { id: p.requested, basis: meta.has_abstract ? "abstract" : "title", text: readFileSync6(paths.md, "utf8") };
 }
-async function verifyClaim(claim, id, cfg, decide, read = readPaper) {
+async function verifyClaim(claim, id, cfg, decide, read2 = readPaper) {
   const out = { claim, id, verdict: "error", reason: null, score: null, basis: null, windows: 0, glyphs_unresolved: 0, superseded: null, model: cfg?.model ?? null };
   const words2 = claim.trim().split(/\s+/).filter(Boolean).length;
   if (!words2) return { ...out, error: "claim empty" };
@@ -3134,7 +3215,7 @@ async function verifyClaim(claim, id, cfg, decide, read = readPaper) {
   if (frag) return { ...out, error: frag };
   const groups = citationGroups(claim);
   if (groups.length) return { ...out, error: groups.length > 1 ? `compound claim: ${groups.length} citation groups ${groups.join(" ")} \u2014 one finding per call, without its sources` : `claim names a source ${groups[0]} \u2014 send the finding without its sources; the paper is the id` };
-  const paper = await read(id);
+  const paper = await read2(id);
   out.id = paper.id;
   if ("error" in paper) return { ...out, error: paper.error };
   if (!cfg || !decide) return { ...out, error: "no verify model configured (FH_LIBRARY_VERIFY_URL)" };
@@ -3166,7 +3247,7 @@ async function verifyClaim(claim, id, cfg, decide, read = readPaper) {
 function* metaFiles(dir) {
   if (!existsSync6(dir)) return;
   for (const name of readdirSync3(dir)) {
-    const p = join7(dir, name);
+    const p = join8(dir, name);
     if (statSync3(p).isDirectory()) yield* metaFiles(p);
     else if (name.endsWith("-meta.md")) yield p;
   }
@@ -3184,10 +3265,10 @@ function field(md, key) {
   const m = new RegExp(`^${key}:\\s*(.*)$`, "m").exec(md);
   return m ? m[1].trim() : "";
 }
-function libraryInfo(libraryDir, host, withIds, tiers = resolveTiers()) {
+function libraryInfo(libraryDir, host2, withIds, tiers = resolveTiers()) {
   const info = {
     library_dir: libraryDir,
-    host,
+    host: host2,
     tiers: tiers.all,
     config: { ...settings(), problems: tiers.problems, sources: sources() },
     download: downloadInfo(),
@@ -3206,10 +3287,10 @@ function libraryInfo(libraryDir, host, withIds, tiers = resolveTiers()) {
   };
   const ids = [];
   for (const type of existsSync6(libraryDir) ? readdirSync3(libraryDir) : []) {
-    const typeDir = join7(libraryDir, type);
+    const typeDir = join8(libraryDir, type);
     if (!statSync3(typeDir).isDirectory()) continue;
     for (const meta of metaFiles(typeDir)) {
-      const md = readFileSync6(meta, "utf8");
+      const md = readFileSync7(meta, "utf8");
       const raw = field(md, "level") || "unknown";
       const level = raw === "abstract" && field(md, "has_abstract") === "no" ? "title" : raw;
       info.entries++;
@@ -3221,7 +3302,7 @@ function libraryInfo(libraryDir, host, withIds, tiers = resolveTiers()) {
       const dir = dirname2(meta);
       for (const f of readdirSync3(dir)) {
         try {
-          const st = statSync3(join7(dir, f));
+          const st = statSync3(join8(dir, f));
           if (st.isFile()) info.size_bytes += st.size;
         } catch {
         }
@@ -3250,7 +3331,7 @@ function noticeTypes(md, metaPath) {
     return out;
   }
   try {
-    const xml = readFileSync6(metaPath.replace(/-meta\.md$/, "-abstract.xml"), "utf8");
+    const xml = readFileSync7(metaPath.replace(/-meta\.md$/, "-abstract.xml"), "utf8");
     for (const m of xml.matchAll(/<CommentsCorrections RefType="([A-Za-z]+)"/g)) if (XML_TYPES[m[1]]) out.add(XML_TYPES[m[1]]);
   } catch {
   }
@@ -3376,15 +3457,18 @@ function search(md, query, maxHits, opts = {}) {
   const df = /* @__PURE__ */ new Map();
   for (const x of scored) for (const s of x.matched) df.set(s, (df.get(s) ?? 0) + 1);
   const rarity = (x) => x.matched.reduce((n, s) => n + 1 / df.get(s), 0);
-  const pick = (partial ? scored : full).sort((a, b) => b.matched.length - a.matched.length || (partial ? rarity(b) - rarity(a) : 0) || b.hits - a.hits || a.p.text.length - b.p.text.length).slice(0, maxHits);
-  const passages = pick.map(({ p, matched, hits }) => {
+  const pool = partial ? scored : full;
+  const pick = pool.sort((a, b) => b.matched.length - a.matched.length || (partial ? rarity(b) - rarity(a) : 0) || b.hits - a.hits || a.p.text.length - b.p.text.length).slice(0, maxHits);
+  const passages = pick.map(({ p, matched }) => {
     const w = clip(p, matched, window);
-    const out = { section: p.section, text: w.text, hits };
+    const out = { section: p.section, text: w.text };
     if (partial) out.matched = terms.filter((t) => matched.includes(stem(t)));
     if (w.clipped) out.clipped = true;
     return out;
   });
-  return { passages, partial, missing_terms };
+  const r = { passages, partial, missing_terms };
+  if (pool.length > passages.length) r.more = pool.length - passages.length;
+  return r;
 }
 function paragraphs(md, scope) {
   const lines = md.split("\n");
@@ -3629,8 +3713,8 @@ function parseCsl(json2) {
 
 // src/urlfetch.ts
 import { createHash as createHash2 } from "crypto";
-import { existsSync as existsSync7, mkdirSync as mkdirSync5, readFileSync as readFileSync7 } from "fs";
-import { join as join8 } from "path";
+import { existsSync as existsSync7, mkdirSync as mkdirSync6, readFileSync as readFileSync8 } from "fs";
+import { join as join9 } from "path";
 var TRACKING = /^(utm_[a-z]+|fbclid|gclid|dclid|msclkid|mc_cid|mc_eid|igshid|ref_src)$/i;
 function normalizeUrl(raw) {
   let u;
@@ -3652,8 +3736,8 @@ var urlHash = (normalized2) => createHash2("sha1").update(normalized2).digest("h
 function urlPaths(libraryDir, normalized2) {
   const hash = urlHash(normalized2);
   const entry = `url-${hash}`;
-  const entryDir = join8(libraryDir, "url", hash.slice(0, 2), hash.slice(0, 4), hash);
-  return { hash, entry, entryDir, meta: join8(entryDir, `${entry}-meta.md`), md: join8(entryDir, `${entry}-page.md`), raw: (ext) => join8(entryDir, `${entry}-page.${ext}`) };
+  const entryDir = join9(libraryDir, "url", hash.slice(0, 2), hash.slice(0, 4), hash);
+  return { hash, entry, entryDir, meta: join9(entryDir, `${entry}-meta.md`), md: join9(entryDir, `${entry}-page.md`), raw: (ext) => join9(entryDir, `${entry}-page.${ext}`) };
 }
 function renderUrlMeta(m) {
   const lines = [
@@ -3673,7 +3757,7 @@ function renderUrlMeta(m) {
 }
 function readUrlMeta(p) {
   if (!existsSync7(p.meta)) return null;
-  const md = readFileSync7(p.meta, "utf8");
+  const md = readFileSync8(p.meta, "utf8");
   const f = (k) => {
     const m = new RegExp(`^${k}:[ \\t]*(.*)$`, "m").exec(md);
     return m ? m[1].trim() : "";
@@ -3699,7 +3783,7 @@ function readUrlMeta(p) {
   };
 }
 function writeUrlMeta(p, m) {
-  mkdirSync5(p.entryDir, { recursive: true });
+  mkdirSync6(p.entryDir, { recursive: true });
   writeAtomic(p.meta, renderUrlMeta(m));
 }
 var SHELL_WORDS = 100;
@@ -3746,13 +3830,22 @@ var climb = async (url, onTry, onAlert) => {
   let shell = !r.ok ? r.shell : void 0;
   for (const b of browserRungs()) {
     if (!escalates(r)) break;
+    const w = walled(target, `browser ${b.label}`);
+    if (w) {
+      onTry?.(`skipped browser ${b.label}: ${wallNote(w)}`);
+      continue;
+    }
     r = await rung(target, { browser: b.endpoint, label: b.label });
+    recordRung(target, `browser ${b.label}`, r.ok, r.ok ? "" : r.reason);
     if (!r.ok && r.shell && !shell) shell = r.shell;
     onTry?.(r.ok ? `ok browser ${b.label}` : `failed browser ${b.label}: ${r.reason}`);
     if (!r.ok && r.reason.startsWith(MISMATCH)) onAlert?.(`browser ${b.label}: ${r.reason}`);
   }
-  if (!r.ok && unlockerConfigured() && (r.via !== "fetch" || /^(HTTP 403|wall page|script shell)/.test(r.reason))) {
+  const uw = !r.ok && unlockerConfigured() ? walled(target, "brightdata") : null;
+  if (uw) onTry?.(`skipped brightdata: ${wallNote(uw)}`);
+  else if (!r.ok && unlockerConfigured() && (r.via !== "fetch" || /^(HTTP 403|wall page|script shell)/.test(r.reason))) {
     r = await rung(target, "brightdata");
+    recordRung(target, "brightdata", r.ok, r.ok ? "" : r.reason);
     onTry?.(r.ok ? "ok brightdata" : `failed brightdata: ${r.reason}`);
     if (!r.ok && /^unlocker HTTP/.test(r.reason)) onAlert?.(`brightdata: ${r.reason}`);
   }
@@ -3857,16 +3950,16 @@ function libraryIdForUrl(raw) {
   } catch {
     return null;
   }
-  const host = u.hostname.toLowerCase().replace(/^www\./, ""), path = u.pathname;
+  const host2 = u.hostname.toLowerCase().replace(/^www\./, ""), path = u.pathname;
   let m;
-  if (host === "pubmed.ncbi.nlm.nih.gov" && (m = /^\/(\d{1,8})\/?$/.exec(path))) return `pmid:${m[1]}`;
-  if (host === "ncbi.nlm.nih.gov" && (m = /^\/pubmed\/(\d{1,8})\/?$/.exec(path))) return `pmid:${m[1]}`;
-  if (host === "clinicaltrials.gov" && (m = /^\/(?:study|ct2\/show)\/(NCT\d{8})\/?$/i.exec(path))) return `nct:${m[1].slice(3)}`;
-  if (host === "dailymed.nlm.nih.gov" && /\/(drugInfo|lookup)\.cfm$/i.test(path)) {
+  if (host2 === "pubmed.ncbi.nlm.nih.gov" && (m = /^\/(\d{1,8})\/?$/.exec(path))) return `pmid:${m[1]}`;
+  if (host2 === "ncbi.nlm.nih.gov" && (m = /^\/pubmed\/(\d{1,8})\/?$/.exec(path))) return `pmid:${m[1]}`;
+  if (host2 === "clinicaltrials.gov" && (m = /^\/(?:study|ct2\/show)\/(NCT\d{8})\/?$/i.exec(path))) return `nct:${m[1].slice(3)}`;
+  if (host2 === "dailymed.nlm.nih.gov" && /\/(drugInfo|lookup)\.cfm$/i.test(path)) {
     const id = u.searchParams.get("setid");
     if (id && /^[0-9a-f-]{36}$/i.test(id)) return `dailymed:${id.toLowerCase()}`;
   }
-  if ((host === "doi.org" || host === "dx.doi.org") && (m = /^\/(10\.\S+)$/.exec(decodeURIComponent(path)))) return `doi:${m[1]}`;
+  if ((host2 === "doi.org" || host2 === "dx.doi.org") && (m = /^\/(10\.\S+)$/.exec(decodeURIComponent(path)))) return `doi:${m[1]}`;
   return null;
 }
 var failureText = (url, reason) => {
@@ -3881,17 +3974,17 @@ function titleLine(m, md) {
 }
 function serve(requested, p, m, args, cached) {
   const maxLength = args.max_length ?? 5e3, startIndex = args.start_index ?? 0;
-  const read = (file2) => existsSync7(file2) ? readFileSync7(file2, "utf8") : "";
+  const read2 = (file3) => existsSync7(file3) ? readFileSync8(file3, "utf8") : "";
   if (args.title_only) return { ok: true, text: `Contents of ${requested}:
-${titleLine(m, read(p.md))}`, cached, meta: m };
+${titleLine(m, read2(p.md))}`, cached, meta: m };
   const rawFile = m.files.find((f) => !f.endsWith("-meta.md") && !f.endsWith("-page.md"));
   let body, prefix = "";
-  if (m.format === "markdown") body = read(p.md);
+  if (m.format === "markdown") body = read2(p.md);
   else if (m.format === "text" || args.raw) {
-    body = rawFile ? read(join8(p.entryDir, rawFile)) : read(p.md);
+    body = rawFile ? read2(join9(p.entryDir, rawFile)) : read2(p.md);
     prefix = `Content type ${m.content_type || "(not declared)"} cannot be simplified to markdown, but here is the raw content:
 `;
-  } else body = read(p.md);
+  } else body = read2(p.md);
   return { ok: true, text: fetchAnswer(requested, body, prefix, maxLength, startIndex), cached, meta: m };
 }
 var inflight = /* @__PURE__ */ new Map();
@@ -3915,7 +4008,7 @@ async function libraryFetch(libraryDir, requested, args, converter, logs = {}, f
         return;
       }
       const page = await convert(r.bytes, r.contentType);
-      mkdirSync5(p.entryDir, { recursive: true });
+      mkdirSync6(p.entryDir, { recursive: true });
       const files = [`${p.entry}-meta.md`];
       if (page.md !== null) {
         writeAtomic(p.md, page.md);
@@ -3937,8 +4030,8 @@ async function libraryFetch(libraryDir, requested, args, converter, logs = {}, f
 
 // src/verifycache.ts
 import { createHash as createHash3 } from "crypto";
-import { existsSync as existsSync8, mkdirSync as mkdirSync6, readFileSync as readFileSync8 } from "fs";
-import { dirname as dirname3, join as join9 } from "path";
+import { existsSync as existsSync8, mkdirSync as mkdirSync7, readFileSync as readFileSync9 } from "fs";
+import { dirname as dirname3, join as join10 } from "path";
 var sha = (s) => createHash3("sha1").update(s).digest("hex");
 var normalizeClaim = (claim) => claim.replace(/\s+/g, " ").trim();
 function foldClaim(claim) {
@@ -3950,36 +4043,36 @@ ${id}
 ${model}
 ${threshold}
 ${text3}`).slice(0, 16);
-var verdictFile = (libraryDir, key) => join9(libraryDir, "verify", key.slice(0, 2), key.slice(0, 4), `${key}.json`);
-function readVerdict(file2) {
-  if (!existsSync8(file2)) return null;
+var verdictFile = (libraryDir, key) => join10(libraryDir, "verify", key.slice(0, 2), key.slice(0, 4), `${key}.json`);
+function readVerdict(file3) {
+  if (!existsSync8(file3)) return null;
   try {
-    return JSON.parse(readFileSync8(file2, "utf8"));
+    return JSON.parse(readFileSync9(file3, "utf8"));
   } catch {
     return null;
   }
 }
-function lookup(file2, maxDays) {
-  const c = readVerdict(file2);
+function lookup(file3, maxDays) {
+  const c = readVerdict(file3);
   return c && daysSince(c.date) < maxDays ? c : null;
 }
-function remember(file2, c) {
-  mkdirSync6(dirname3(file2), { recursive: true });
-  writeAtomic(file2, JSON.stringify(c, null, 2) + "\n");
+function remember(file3, c) {
+  mkdirSync7(dirname3(file3), { recursive: true });
+  writeAtomic(file3, JSON.stringify(c, null, 2) + "\n");
 }
-async function verifyCached(claim, id, cfg, decide, libraryDir = FH_LIBRARY_DIR, read = readPaper) {
-  const paper = await read(id, libraryDir);
+async function verifyCached(claim, id, cfg, decide, libraryDir = FH_LIBRARY_DIR, read2 = readPaper) {
+  const paper = await read2(id, libraryDir);
   const once = () => paper;
   if ("error" in paper || !cfg) return { ...await verifyClaim(claim, id, cfg, decide, once), cached: false };
   const text3 = textHash(paper.text);
-  const file2 = verdictFile(libraryDir, cacheKey(claim, paper.id, cfg.model, cfg.threshold, text3));
-  const hit = lookup(file2, settings().refresh_after_days);
+  const file3 = verdictFile(libraryDir, cacheKey(claim, paper.id, cfg.model, cfg.threshold, text3));
+  const hit = lookup(file3, settings().refresh_after_days);
   if (hit) {
     return { claim, id: paper.id, verdict: hit.verdict, reason: hit.reason, score: hit.score, basis: hit.basis, windows: hit.windows, glyphs_unresolved: hit.glyphs_unresolved, superseded: hit.superseded, model: hit.model, cached: true };
   }
   const r = await verifyClaim(claim, id, cfg, decide, once);
   if (!r.error && r.score !== null) {
-    remember(file2, { key: file2.slice(-21, -5), id: paper.id, claim: normalizeClaim(claim), model: cfg.model, threshold: cfg.threshold, text: text3, date: today(), verdict: r.verdict, reason: r.reason, score: r.score, basis: r.basis, windows: r.windows, glyphs_unresolved: r.glyphs_unresolved, superseded: r.superseded });
+    remember(file3, { key: file3.slice(-21, -5), id: paper.id, claim: normalizeClaim(claim), model: cfg.model, threshold: cfg.threshold, text: text3, date: today(), verdict: r.verdict, reason: r.reason, score: r.score, basis: r.basis, windows: r.windows, glyphs_unresolved: r.glyphs_unresolved, superseded: r.superseded });
   }
   return { ...r, cached: false };
 }
@@ -4001,11 +4094,35 @@ function leanObject(o) {
   return out;
 }
 var leanJson = (o) => compact2(lean(o));
+var STATUS_DROP = /* @__PURE__ */ new Set(["requested", "type", "id", "entry", "retrieved_date", "pubmed_checked", "attempts", "coi", "funding", "title", "first_author", "doi"]);
+function statusAnswer(s) {
+  const o = {};
+  for (const [k, v] of Object.entries(s)) if (!STATUS_DROP.has(k)) o[k] = v;
+  if (Array.isArray(o.publication_types)) o.publication_types = reviewTypes(o.publication_types);
+  return leanJson(o);
+}
+function verdictAnswer(r) {
+  if (r.error) return compact2({ verdict: "error", error: r.error });
+  const o = { verdict: r.verdict, score: r.score === null ? null : Math.round(r.score * 100) / 100, basis: r.basis ?? null };
+  if (r.reason) o.reason = r.reason;
+  if (r.superseded) o.superseded = true;
+  if (r.glyphs_unresolved) o.glyphs_unresolved = r.glyphs_unresolved;
+  return compact2(o);
+}
+function readAnswer(o) {
+  const { requested: _r, entry: _e, ...rest } = o;
+  if (rest.partial === false) delete rest.partial;
+  return leanJson(rest);
+}
+function searchAnswer(r) {
+  return compact2({ total: r.total, columns: r.columns, rows: r.rows });
+}
+var reviewTypes = (types) => types.filter((t) => t !== "Journal Article" && !/^Research Support\b/.test(t));
 
 // src/trials.ts
 import { createHash as createHash4 } from "crypto";
-import { existsSync as existsSync9, mkdirSync as mkdirSync7, readFileSync as readFileSync9 } from "fs";
-import { dirname as dirname4, join as join10 } from "path";
+import { existsSync as existsSync9, mkdirSync as mkdirSync8, readFileSync as readFileSync10 } from "fs";
+import { dirname as dirname4, join as join11 } from "path";
 var searchCacheDays = () => settings().search_cache_days;
 var API2 = "https://clinicaltrials.gov/api/v2/studies";
 var FIELDS = ["NCTId", "BriefTitle", "OverallStatus", "WhyStopped", "Phase", "StudyType", "EnrollmentCount", "EnrollmentType", "StartDate", "PrimaryCompletionDate", "CompletionDate", "LastUpdatePostDate", "PrimaryOutcomeMeasure", "PrimaryOutcomeTimeFrame", "Condition", "InterventionName", "LeadSponsorName", "HasResults"];
@@ -4020,7 +4137,7 @@ var COLUMNS = [
   // "Terminated — slow accrual": a trial past its completion date is not "ongoing" (ER.md 28.7)
   "phase",
   "enrollment",
-  // "100 (estimated)" / "48 (actual)" — participant count (28.5)
+  // "48" actual, "~100" estimated — participant count (28.5)
   "start",
   "primary_completion",
   "primary_outcome",
@@ -4043,7 +4160,7 @@ function normalizeQuery(q) {
   return out;
 }
 var searchKey = (q) => createHash4("sha1").update(JSON.stringify(["clinicaltrials", q.query ?? "", q.intervention ?? "", q.condition ?? "", q.status, q.sort, q.max_results])).digest("hex").slice(0, 16);
-var searchFile = (libraryDir, key) => join10(libraryDir, "search", key.slice(0, 2), key.slice(0, 4), `${key}.json`);
+var searchFile = (libraryDir, key) => join11(libraryDir, "search", key.slice(0, 2), key.slice(0, 4), `${key}.json`);
 function searchUrl(q) {
   const p = new URLSearchParams();
   if (q.query) p.set("query.term", q.query);
@@ -4079,7 +4196,7 @@ function parseRows(json2) {
       String(id.briefTitle ?? ""),
       status,
       phases.length ? phases.map((x) => PHASE2[x] ?? x).join(" / ") : null,
-      typeof en.count === "number" ? `${en.count}${en.type ? ` (${String(en.type).toLowerCase()})` : ""}` : null,
+      typeof en.count === "number" ? `${String(en.type).toUpperCase() === "ESTIMATED" ? "~" : ""}${en.count}` : null,
       st.startDateStruct?.date ?? null,
       st.primaryCompletionDateStruct?.date ?? null,
       primary,
@@ -4094,10 +4211,10 @@ var liveFetch = (url) => fetchText(url, { "User-Agent": LIBRARY_UA, Accept: "app
 async function searchTrials(raw, libraryDir = FH_LIBRARY_DIR, fetchImpl = liveFetch) {
   const q = normalizeQuery(raw);
   if ("error" in q) return { ok: false, error: q.error };
-  const file2 = searchFile(libraryDir, searchKey(q));
-  if (existsSync9(file2)) {
+  const file3 = searchFile(libraryDir, searchKey(q));
+  if (existsSync9(file3)) {
     try {
-      const r = JSON.parse(readFileSync9(file2, "utf8"));
+      const r = JSON.parse(readFileSync10(file3, "utf8"));
       if (daysSince(r.searched) < searchCacheDays()) return { ok: true, result: r, cached: true };
     } catch {
     }
@@ -4116,8 +4233,8 @@ async function searchTrials(raw, libraryDir = FH_LIBRARY_DIR, fetchImpl = liveFe
   }
   const result = { source: "clinicaltrials", query: q, total: parsed.total, columns: COLUMNS, rows: parsed.rows, searched: today() };
   try {
-    mkdirSync7(dirname4(file2), { recursive: true });
-    writeAtomic(file2, JSON.stringify(result, null, 2) + "\n");
+    mkdirSync8(dirname4(file3), { recursive: true });
+    writeAtomic(file3, JSON.stringify(result, null, 2) + "\n");
   } catch {
   }
   return { ok: true, result, cached: false };
@@ -4129,7 +4246,7 @@ function trialsLogId(q) {
 
 // src/pubmedsearch.ts
 import { createHash as createHash5 } from "crypto";
-import { existsSync as existsSync10, mkdirSync as mkdirSync8, readFileSync as readFileSync10 } from "fs";
+import { existsSync as existsSync10, mkdirSync as mkdirSync9, readFileSync as readFileSync11 } from "fs";
 import { dirname as dirname5 } from "path";
 var MAX_RESULTS2 = 100;
 var DEFAULT_RESULTS2 = 20;
@@ -4198,7 +4315,7 @@ function parseEsummary(json2, ids, libraryDir) {
     const year = /\b(\d{4})\b/.exec(String(s.pubdate ?? ""))?.[1] ?? null;
     const first = String(s.sortfirstauthor ?? (Array.isArray(s.authors) && s.authors[0]?.name) ?? "").replace(/\s+[A-Z]{1,3}$/, "").trim() || null;
     const pmc2 = (Array.isArray(s.articleids) ? s.articleids : []).find((a) => a.idtype === "pmc")?.value ?? null;
-    const types = Array.isArray(s.pubtype) ? s.pubtype.filter((t) => t !== "Journal Article").join("; ") : "";
+    const types = Array.isArray(s.pubtype) ? reviewTypes(s.pubtype.map(String)).join("; ") : "";
     const paths = entryPaths(libraryDir, "pmid", id);
     const level = entryExists(paths) ? String(readStatus(paths, emptyStatus(`pmid:${id}`, "pmid", id, libraryDir)).level) : null;
     rows.push([id, String(s.title ?? "").replace(/\.$/, "") || null, first, year, s.source ? String(s.source) : null, types || null, pmc2 ? String(pmc2) : null, level]);
@@ -4209,10 +4326,10 @@ var live = (url) => fetchText(url, { "User-Agent": "fh-library", Accept: "applic
 async function searchArticles(raw, libraryDir = FH_LIBRARY_DIR, fetchImpl = live) {
   const q = normalizeQuery2(raw);
   if ("error" in q) return { ok: false, error: q.error };
-  const file2 = searchFile(libraryDir, searchKey2(q));
-  if (existsSync10(file2)) {
+  const file3 = searchFile(libraryDir, searchKey2(q));
+  if (existsSync10(file3)) {
     try {
-      const r = JSON.parse(readFileSync10(file2, "utf8"));
+      const r = JSON.parse(readFileSync11(file3, "utf8"));
       if (daysSince(r.searched) < searchCacheDays()) return { ok: true, result: r, cached: true };
     } catch {
     }
@@ -4233,8 +4350,8 @@ async function searchArticles(raw, libraryDir = FH_LIBRARY_DIR, fetchImpl = live
   }
   const result = { source: "pubmed", query: q, total, columns: COLUMNS2, rows, searched: today() };
   try {
-    mkdirSync8(dirname5(file2), { recursive: true });
-    writeAtomic(file2, JSON.stringify(result, null, 2) + "\n");
+    mkdirSync9(dirname5(file3), { recursive: true });
+    writeAtomic(file3, JSON.stringify(result, null, 2) + "\n");
   } catch {
   }
   return { ok: true, result, cached: false };
@@ -4290,11 +4407,10 @@ function recordSummary(json2) {
     start: dated2(st.startDateStruct),
     primary_completion: dated2(st.primaryCompletionDateStruct),
     completion: dated2(st.completionDateStruct),
-    first_posted: st.studyFirstPostDateStruct?.date ?? null,
     last_update: st.lastUpdatePostDateStruct?.date ?? null,
     results_first_posted: st.resultsFirstPostDateStruct?.date ?? null,
     sponsor: para2(sp.leadSponsor?.name),
-    collaborators: joined2(list3(sp.collaborators).map((c) => para2(c.name))),
+    collaborators: collaboratorsLine(list3(sp.collaborators).map((c) => para2(c.name)).filter(Boolean)),
     conditions: joined2(list3(ps.conditionsModule?.conditions).map((c) => para2(c))),
     interventions: joined2(list3(ai.interventions).map((i) => para2(`${i.type ? `${words(i.type)}: ` : ""}${i.name ?? ""}`))),
     primary_outcomes: list3(om.primaryOutcomes).map(outcome).filter(Boolean),
@@ -4306,23 +4422,69 @@ function recordSummary(json2) {
     summary: para2(de.briefSummary, 400)
   };
 }
+function collaboratorsLine(names) {
+  if (!names.length) return null;
+  return names.slice(0, 3).join("; ") + (names.length > 3 ? ` +${names.length - 3}` : "");
+}
+var ORDER = ["Outcome measures", "Adverse events", "Groups", "Participant flow", "Baseline characteristics"];
+function reorderResults(part) {
+  const chunks = part.split(/^(?=### )/m).map((c) => c.trim()).filter(Boolean);
+  const groups = /* @__PURE__ */ new Map();
+  const sections2 = [];
+  for (const c of chunks) {
+    const name = /^### (.+)$/m.exec(c)?.[1].trim() ?? "";
+    const lines = c.split("\n").slice(1);
+    let a = lines.findIndex((l) => /^- \*\*[^*]+\*\*/.test(l));
+    let b = a < 0 ? -1 : lines.findIndex((l, i) => i > a && (/^\|/.test(l) || /^\*\*/.test(l) || /^#### /.test(l)));
+    if (a >= 0 && b < 0) b = lines.length;
+    if (a < 0) {
+      sections2.push({ name, body: lines.join("\n").trim() });
+      continue;
+    }
+    const kept = [];
+    let cur = [];
+    const flush = () => {
+      if (!cur.length) return;
+      const block = cur.join("\n").trim();
+      const title = /^- \*\*([^*]+)\*\*/.exec(block)[1];
+      if (!groups.has(title)) groups.set(title, block);
+      else if (groups.get(title) !== block) kept.push(block);
+      cur = [];
+    };
+    for (const l of lines.slice(a, b)) {
+      if (/^- \*\*[^*]+\*\*/.test(l)) flush();
+      cur.push(l);
+    }
+    flush();
+    const body = [...lines.slice(0, a), ...kept, ...lines.slice(b)].join("\n").replace(/\n{3,}/g, "\n\n").trim();
+    sections2.push({ name, body });
+  }
+  if (groups.size) sections2.push({ name: "Groups", body: [...groups.values()].join("\n\n") });
+  const rank = (n) => {
+    const i = ORDER.indexOf(n);
+    return i < 0 ? ORDER.length : i;
+  };
+  return sections2.map((x, i) => ({ ...x, i })).sort((x, y) => rank(x.name) - rank(y.name) || x.i - y.i).filter((x) => x.body).map((x) => `### ${x.name}
+
+${x.body}`).join("\n\n");
+}
 function resultsText(nct, md, maxChars) {
   const i = md.search(/^## Results\s*$/m);
   if (i < 0) return { nct, has_results: false, text: null, truncated: false, outcomes: [], note: "no posted results in the registry record" };
-  const part = md.slice(i).replace(/^## Results\s*\n/, "").trim();
+  const part = reorderResults(md.slice(i).replace(/^## Results\s*\n/, "").replace(/\n## [\s\S]*$/, "").trim());
   const outcomes = [...part.matchAll(/^#### (.+)$/gm)].map((m) => m[1].trim());
   const truncated = part.length > maxChars;
   return { nct, has_results: true, text: truncated ? part.slice(0, maxChars).trimEnd() + "\n\n[truncated]" : part, truncated, outcomes, note: truncated ? "cut at max_chars; library_read(id, section: <an outcome title>) returns one measure whole" : null };
 }
 
 // src/missing.ts
-import { existsSync as existsSync11, mkdirSync as mkdirSync9, readFileSync as readFileSync11 } from "fs";
-import { join as join11 } from "path";
-var missingFile = (p) => join11(p.entryDir, `${p.entry}-missing.md`);
+import { existsSync as existsSync11, mkdirSync as mkdirSync10, readFileSync as readFileSync12 } from "fs";
+import { join as join12 } from "path";
+var missingFile = (p) => join12(p.entryDir, `${p.entry}-missing.md`);
 function readMissing(p) {
   const f = missingFile(p);
   if (!existsSync11(f)) return null;
-  const md = readFileSync11(f, "utf8");
+  const md = readFileSync12(f, "utf8");
   const get = (k) => /^$/.test(k) ? "" : (new RegExp(`^${k}:[ \\t]*(.*)$`, "m").exec(md)?.[1] ?? "").trim();
   const checked = get("checked"), reason = get("reason");
   return checked ? { checked, reason: reason || "not found" } : null;
@@ -4333,7 +4495,7 @@ function freshMissing(p, retryDays) {
 }
 function rememberMissing(p, reason) {
   const m = { checked: today(), reason };
-  mkdirSync9(p.entryDir, { recursive: true });
+  mkdirSync10(p.entryDir, { recursive: true });
   writeAtomic(missingFile(p), `checked: ${m.checked}
 reason: ${m.reason}
 `);
@@ -4365,7 +4527,7 @@ function splitArticles(xml) {
   }
   return out;
 }
-var row2 = (pmid, s, level) => s ? [pmid, s.title, s.first_author, s.year, s.publication_types.length ? s.publication_types.join("; ") : null, s.retracted, level] : [pmid, null, null, null, null, null, level];
+var row2 = (pmid, s, level) => s ? [pmid, s.title ? s.title.replace(/\.$/, "") : s.title, s.first_author, s.year, reviewTypes(s.publication_types).join("; ") || null, s.retracted, level] : [pmid, null, null, null, null, null, level];
 async function fetchArticles(ids, libraryDir = FH_LIBRARY_DIR, converter = "fh-library", fetchImpl = fetchText) {
   const result = { columns: COLUMNS3, rows: [], new: 0, cached: 0, not_found: 0, invalid: 0 };
   const order2 = [];
@@ -4428,7 +4590,7 @@ async function fetchArticles(ids, libraryDir = FH_LIBRARY_DIR, converter = "fh-l
 // src/tools.ts
 var pkg = (() => {
   try {
-    return JSON.parse(readFileSync12(new URL("../package.json", import.meta.url), "utf8"));
+    return JSON.parse(readFileSync13(new URL("../package.json", import.meta.url), "utf8"));
   } catch {
     return { name: "fh-library", version: "0.0.0" };
   }
@@ -4458,7 +4620,7 @@ Tools
 Hints
 - library_read and library_verify store a missing record themselves; the library_read answer then carries added: true. No library_add necessary first.
 - citable false \u2014 the paper is retracted (a retraction notice, or PubMed's publication type), or holds nothing to read: no abstract and no full text. comment says which; corrections lists the notices as {type, pmid, source}.
-- level \u2014 what is stored: "full_text", "abstract" (no full text could be found; attempts names each source tried and why it failed) or "title" (not even an abstract; citable false).
+- level \u2014 what is stored: "full_text", "abstract" (no full text could be found; the log names each source tried and why it failed) or "title" (not even an abstract; citable false).
 - error "pmid not found", "doi not found", "ctgov: not a study record", "dailymed: not an SPL document" \u2014 no service knows the ID. Remembered for a day (FH_LIBRARY_RETRY_AFTER_DAYS): the same call answers from disk with the date it was checked; after that the service is asked again. "invalid id" \u2014 the ID itself is malformed; the same call gives the same answer.
 - Calls may run in parallel: library_add, library_read and library_verify. Two calls on the same missing record share one add.
 - library_fetch takes any URL that is not a paper, label or trial record; those have IDs, and library_add, library_read and library_verify work on the stored document. A second pass over the same review reads every page from disk; a failed fetch is remembered for a day.
@@ -4483,14 +4645,14 @@ Reading a document
 - partial true \u2014 no passage holds every term; missing_terms are the words the document never uses. The same query without them finds the passage if there is one.
 - Numbers are often in tables; a matching table row comes with its header row.
 - section and query together search inside that section only.
-- section "Abstract" on a full text without one returns the stored abstract. section "Declarations" returns funding, conflicts of interest, acknowledgements and ethics statements, assembled when the document has no section of that name.
+- section "Abstract" on a full text without one returns the stored abstract. section "Declarations" returns funding, conflicts of interest, acknowledgements and ethics statements, assembled when the document has no section of that name, else the statements of the PubMed or Crossref record.
 
-Status fields: requested, type, id, entry, pmid, pmcid, doi, setid, nctid, publications, title, first_author, year, publication_types, retracted, corrections, coi (conflict-of-interest statement), funding, citable, comment, has_abstract, level, version, retrieved_by, retrieved_date, pubmed_checked, attempts, glyphs_unresolved, error. Answers are compact JSON; a field that is null or an empty list is left out (library_verify keeps its nulls: score and reason null mean something).
+Status fields: pmid, pmcid, setid, nctid, publications, year, publication_types, retracted, corrections, citable, comment, has_abstract, level, version, retrieved_by, glyphs_unresolved, error. Funding and conflicts of interest: library_read section "Declarations". Answers carry only what the call did not send: no echo of the ID, claim or query; a field that is null or an empty list is left out. library_verify always has verdict, score (null: no judgement) and basis.
 
 Where records come from: PubMed for the record and abstract; the full text from PubMed Central, Europe PMC, publishers' text-mining links, Unpaywall, OpenAlex, CORE, and a proxy URL template from the configuration. Every action is appended as one line to <library>/library.log.`;
 var idSchema = z.string().describe('ID as "type:value": a paper "pmid:30955454" (bare digits count as pmid) or "doi:10.5812/asjsm.110753" (a paper PubMed does not index; a DOI PubMed indexes is taken as its PMID), a drug label "dailymed:<set ID>" (the set ID from its DailyMed page URL), or a trial record "nct:05173896" (the eight digits; nct:NCT05173896 and a bare NCT05173896 are accepted).');
 var levelWord = (m) => m.level === "abstract" && !m.has_abstract ? "title" : m.level;
-var json = (s) => ({ content: [{ type: "text", text: leanJson(s) }] });
+var json = (s) => ({ content: [{ type: "text", text: statusAnswer(s) }] });
 function noticeTag(m) {
   const n = (t) => m.corrections.filter((c) => c.type === t).length;
   const parts = [
@@ -4798,7 +4960,7 @@ function registerTools(server2) {
   );
   server2.tool(
     "library_read",
-    `One piece of a stored document's Markdown, never the whole. No section/query: the outline (headings with word counts; full text or abstract only). section: that heading's own text (case-insensitive substring; a repeated heading as "Participants [3/12]"), capped at max_chars, its subsections listed. query: passages holding every term, each a window of at most 1000 chars with its section, best first, at most max_hits; partial: true with missing_terms when no passage holds them all. section and query together: inside that section. A missing paper is added first (added: true).`,
+    `One piece of a stored document's Markdown, never the whole. No section/query: the outline (headings with word counts; full text or abstract only). section: that heading's own text (case-insensitive substring; a repeated heading as "Participants [3/12]"), capped at max_chars, its subsections listed. query: passages holding every term, each a window of at most 1000 chars with its section, best first, at most max_hits (more: how many further passages match); partial: true with missing_terms when no passage holds them all. section and query together: inside that section. A missing paper is added first (added: true).`,
     {
       id: idSchema,
       section: z.string().optional().describe("Heading from the outline (case-insensitive substring)."),
@@ -4808,7 +4970,7 @@ function registerTools(server2) {
     },
     async ({ id, section: section2, query, max_chars, max_hits }) => {
       const parsed = parseId(id);
-      const out = (o2) => ({ content: [{ type: "text", text: leanJson(o2) }] });
+      const out = (o2) => ({ content: [{ type: "text", text: readAnswer(o2) }] });
       if (parsed.error) {
         log("library_read", parsed.requested, `error ${parsed.error}`);
         return out({ requested: parsed.requested, error: parsed.error });
@@ -4825,7 +4987,7 @@ function registerTools(server2) {
         added = true;
       }
       const full = existsSync12(paths.fulltextMd);
-      const md = readFileSync12(full ? paths.fulltextMd : paths.md, "utf8");
+      const md = readFileSync13(full ? paths.fulltextMd : paths.md, "utf8");
       const text3 = full ? "full_text" : "abstract";
       const base = added ? { requested: parsed.requested, entry: paths.entry, text: text3, added: true } : { requested: parsed.requested, entry: paths.entry, text: text3 };
       const headings2 = () => outline(md, text3).sections;
@@ -4846,13 +5008,25 @@ function registerTools(server2) {
         let sec = readSection(md, section2, cap);
         let via = "";
         if (!sec && /abstract/i.test(section2) && full && existsSync12(paths.md)) {
-          sec = readSection(readFileSync12(paths.md, "utf8"), "abstract", cap);
+          sec = readSection(readFileSync13(paths.md, "utf8"), "abstract", cap);
           via = " (from the abstract file)";
         }
         if (!sec && /declaration/i.test(section2)) {
           sec = assembleDeclarations(md, cap);
           via = " (assembled)";
         }
+        if (!sec && /declaration/i.test(section2)) {
+          const st = readStatus(paths, emptyStatus(parsed.requested, parsed.type, parsed.id, FH_LIBRARY_DIR));
+          const body = [st.coi ? `### Conflicts of interest
+
+${st.coi}` : "", st.funding?.length ? `### Funding
+
+${st.funding.map((f) => `- ${f}`).join("\n")}` : ""].filter(Boolean).join("\n\n");
+          if (body) {
+            sec = { heading: "Declarations (from the record)", words: body.split(/\s+/).length, text: body, returned_words: body.split(/\s+/).length, truncated: false };
+            via = " (from the record)";
+          }
+        }
         if (!sec) {
           log("library_read", parsed.requested, `section "${section2}" not found`);
           return out({ ...base, error: `no section matching "${section2}"`, outline: headings2() });
@@ -4894,7 +5068,7 @@ function registerTools(server2) {
       if (!r.error && r.score !== null) bump("verify", cached);
       const secs = `${((Date.now() - t0) / 1e3).toFixed(1).padStart(4, "0")}s`;
       log("library_verify", r.id, r.error ? `${modelFailure(r.error, cfg) ? "!! " : ""}${secs} verdict=${"error".padEnd(12)} ${r.error} | ${claim}` : `${secs} verdict=${r.verdict.padEnd(12)} basis=${r.basis.padEnd(9)} score=${r.score === null ? "-    " : r.score.toFixed(3)}${r.windows > 1 ? ` windows=${r.windows}` : ""}${cached ? " cached" : ""}${r.superseded ? " superseded" : ""}${r.glyphs_unresolved ? ` glyphs=${r.glyphs_unresolved}` : ""} | ${claim}`);
-      return { content: [{ type: "text", text: compact2(r) }] };
+      return { content: [{ type: "text", text: verdictAnswer(r) }] };
     }
   );
   server2.tool(
@@ -4957,8 +5131,8 @@ function registerTools(server2) {
             return { content: [{ type: "text", text: `Failed to fetch ${url}: ${added.error}` }], isError: true };
           }
         }
-        const file2 = pid.type === "pmid" ? paths.md : existsSync12(paths.fulltextMd) ? paths.fulltextMd : paths.md;
-        const md = existsSync12(file2) ? readFileSync12(file2, "utf8") : "";
+        const file3 = pid.type === "pmid" ? paths.md : existsSync12(paths.fulltextMd) ? paths.fulltextMd : paths.md;
+        const md = existsSync12(file3) ? readFileSync13(file3, "utf8") : "";
         const text3 = title_only ? `Contents of ${url}:
 ${(md.split("\n").find((l) => l.trim()) ?? "").trim()}` : fetchAnswer(url, md, "", max_length, start_index);
         bump("fetch", had);
@@ -4990,7 +5164,7 @@ ${(md.split("\n").find((l) => l.trim()) ?? "").trim()}` : fetchAnswer(url, md, "
       if (r.new) bump("pubmed", false, r.new);
       const secs = ((Date.now() - t0) / 1e3).toFixed(1).padStart(4, "0") + "s";
       log("library_pubmed", `pmids:${r.rows.length}`, `articles ${secs} cached=${r.cached} new=${r.new}${r.not_found ? ` not_found=${r.not_found}` : ""}${r.invalid ? ` invalid=${r.invalid}` : ""}`);
-      return { content: [{ type: "text", text: leanJson(r) }] };
+      return { content: [{ type: "text", text: compact2({ columns: r.columns, rows: r.rows }) }] };
     }
   );
   server2.tool(
@@ -5013,12 +5187,12 @@ ${(md.split("\n").find((l) => l.trim()) ?? "").trim()}` : fetchAnswer(url, md, "
       bump("search", r.cached);
       const q = r.result.query;
       log("library_search", `pm:${logId2(q)}`, `${r.cached ? "cached" : `searched ${secs}`} rows=${r.result.rows.length} total=${r.result.total}${q.sort !== "relevance" ? ` sort=${q.sort}` : ""}${q.years ? ` years=${q.years}` : ""} | ${q.query}`);
-      return { content: [{ type: "text", text: leanJson(r.result) }] };
+      return { content: [{ type: "text", text: searchAnswer(r.result) }] };
     }
   );
   server2.tool(
     "library_clinicaltrials_search_studies",
-    `Search ClinicalTrials.gov by free text, intervention or condition; status ongoing | completed | all; sort relevance | newest | largest | completion. Returns the total and up to ${MAX_RESULTS} rows as arrays in columns order: NCT ID, title, status, phase, enrollment, start, primary completion, primary outcome, conditions, interventions, has_results. Link: https://clinicaltrials.gov/study/<NCT ID>. Cached ${searchCacheDays()} days. One trial in detail: library_clinicaltrials_get_study_record.`,
+    `Search ClinicalTrials.gov by free text, intervention or condition; status ongoing | completed | all; sort relevance | newest | largest | completion. Returns the total and up to ${MAX_RESULTS} rows as arrays in columns order: NCT ID, title, status, phase, enrollment ("~100" estimated), start, primary completion, primary outcome, conditions, interventions, has_results. Link: https://clinicaltrials.gov/study/<NCT ID>. Cached ${searchCacheDays()} days. One trial in detail: library_clinicaltrials_get_study_record.`,
     {
       query: z.string().optional().describe("Free-text search across all fields (title, conditions, interventions, summary, \u2026); AND / OR / NOT allowed."),
       intervention: z.string().optional().describe('Intervention or treatment, e.g. "tadalafil", "cognitive behavioral therapy".'),
@@ -5038,7 +5212,7 @@ ${(md.split("\n").find((l) => l.trim()) ?? "").trim()}` : fetchAnswer(url, md, "
       bump("search", r.cached);
       const q = r.result.query, parts = [q.intervention && `intervention=${q.intervention}`, q.condition && `condition=${q.condition}`, q.query && `query=${q.query}`].filter(Boolean).join(" ");
       log("library_search", `ct:${trialsLogId(q)}`, `${r.cached ? "cached" : `searched ${secs}`} rows=${r.result.rows.length} total=${r.result.total} status=${q.status}${q.sort !== "relevance" ? ` sort=${q.sort}` : ""} | ${parts}`);
-      return { content: [{ type: "text", text: leanJson(r.result) }] };
+      return { content: [{ type: "text", text: searchAnswer(r.result) }] };
     }
   );
   const nctEntry = async (id) => {
@@ -5065,7 +5239,7 @@ ${(md.split("\n").find((l) => l.trim()) ?? "").trim()}` : fetchAnswer(url, md, "
         return { content: [{ type: "text", text: compact2({ error: e.error }) }], isError: true };
       }
       const raw = e.paths.fulltextRaw("json");
-      const r = existsSync12(raw) ? recordSummary(readFileSync12(raw, "utf8")) : null;
+      const r = existsSync12(raw) ? recordSummary(readFileSync13(raw, "utf8")) : null;
       if (!r) {
         log("library_trials", e.requested, "record error: no stored record");
         return { content: [{ type: "text", text: compact2({ error: "no stored record" }) }], isError: true };
@@ -5084,7 +5258,7 @@ ${(md.split("\n").find((l) => l.trim()) ?? "").trim()}` : fetchAnswer(url, md, "
         log("library_trials", nct, `results error: ${e.error}`);
         return { content: [{ type: "text", text: compact2({ error: e.error }) }], isError: true };
       }
-      const md = existsSync12(e.paths.fulltextMd) ? readFileSync12(e.paths.fulltextMd, "utf8") : "";
+      const md = existsSync12(e.paths.fulltextMd) ? readFileSync13(e.paths.fulltextMd, "utf8") : "";
       const r = resultsText(e.requested.replace(/^nct:/, "NCT"), md, max_chars);
       log("library_trials", e.requested, `results ${r.has_results ? `${r.outcomes.length} outcomes${r.truncated ? " truncated" : ""}` : "none"}`);
       return { content: [{ type: "text", text: leanJson(r) }] };