@cyanheads/zenodo-mcp-server 0.1.1

package/dist/services/zenodo/identifiers.js

@@ -0,0 +1,261 @@
+/**
+ * @fileoverview Pure parsers for the identifier forms Zenodo tools accept: record
+ * references (record ids, Zenodo and external DOIs, zenodo.org and doi.org URLs),
+ * funder, award, ORCID, and community references, plus file-key path encoding.
+ * Nothing here touches the network; URLs are parsed locally and never fetched.
+ * @module services/zenodo/identifiers
+ */
+const ZENODO_HOSTS = new Set(['zenodo.org', 'www.zenodo.org']);
+const DOI_HOSTS = new Set(['doi.org', 'dx.doi.org', 'www.doi.org']);
+const DOI_PATTERN = /^10\.\d{4,9}\/\S+$/;
+const ZENODO_DOI_PATTERN = /^10\.5281\/zenodo\.(\d+)$/i;
+const ZENODO_SUFFIX_PATTERN = /^zenodo\.(\d+)$/i;
+const QUOTE_CHARS = new Set(['"', "'", '“', '”', '‘', '’']);
+const SPACE = /\s/;
+/**
+ * Index just past `s[start, end)` once every trailing character in `chars` is
+ * dropped. A backward scan rather than a `[…]+$` regex, which backtracks
+ * quadratically when a long run of those characters is followed by anything else.
+ */
+function trimEndOf(s, start, end, matches) {
+    let e = end;
+    while (e > start && matches(s[e - 1]))
+        e--;
+    return e;
+}
+const isSpace = (ch) => SPACE.test(ch);
+const isTrailingPunctuation = (ch) => ch === '.' || ch === ',' || ch === ';';
+/** `s` without trailing `.`/`,`/`;`. */
+function stripTrailingPunctuation(s) {
+    return s.slice(0, trimEndOf(s, 0, s.length, isTrailingPunctuation));
+}
+/** `s` without trailing `/`. */
+function stripTrailingSlashes(s) {
+    return s.slice(0, trimEndOf(s, 0, s.length, (ch) => ch === '/'));
+}
+/**
+ * Removes wrapping `<…>`, `"…"`, `'…'`, and surrounding whitespace, repeatedly, in
+ * any nesting order. Trailing `.`/`,`/`;` after a closing wrapper — an id copied
+ * out of a sentence such as `…available at <https://doi.org/…>.` — sits outside the
+ * value and is dropped with it; punctuation inside the wrapper is left to the
+ * per-form rules. Walks two indices inward, so deep nesting stays linear.
+ */
+function stripWrapping(raw) {
+    let start = 0;
+    let end = raw.length;
+    for (;;) {
+        while (start < end && isSpace(raw[start]))
+            start++;
+        end = trimEndOf(raw, start, end, isSpace);
+        const inner = trimEndOf(raw, start, trimEndOf(raw, start, end, isTrailingPunctuation), isSpace);
+        const first = raw[start];
+        const last = raw[inner - 1];
+        const wrapped = inner - start >= 2 &&
+            ((first === '<' && last === '>') || (QUOTE_CHARS.has(first) && QUOTE_CHARS.has(last)));
+        if (!wrapped)
+            return raw.slice(start, end);
+        start++;
+        end = inner - 1;
+    }
+}
+/** Classifies a bare DOI string (no `doi:` prefix, no URL). */
+function classifyDoi(doi, inputKind) {
+    const stripped = stripTrailingPunctuation(doi);
+    const zenodo = ZENODO_DOI_PATTERN.exec(stripped);
+    if (zenodo?.[1]) {
+        return {
+            kind: 'recid',
+            recid: zenodo[1],
+            inputKind: inputKind === 'url' ? 'url' : 'zenodo_doi',
+        };
+    }
+    if (!DOI_PATTERN.test(doi)) {
+        return {
+            kind: 'invalid',
+            message: `"${doi}" is not a well-formed DOI (expected 10.NNNN/suffix).`,
+        };
+    }
+    return {
+        kind: 'external_doi',
+        doi,
+        inputKind: inputKind === 'url' ? 'url' : 'external_doi',
+        ...(stripped !== doi && DOI_PATTERN.test(stripped) ? { strippedDoi: stripped } : {}),
+    };
+}
+function safeDecode(s) {
+    try {
+        return decodeURIComponent(s);
+    }
+    catch {
+        return s;
+    }
+}
+/** Parses a zenodo.org or doi.org URL. Other hosts are rejected. */
+function parseUrl(raw) {
+    let url;
+    try {
+        url = new URL(raw);
+    }
+    catch {
+        return { kind: 'invalid', message: `"${raw}" is not a valid URL.` };
+    }
+    const host = url.hostname.toLowerCase();
+    if (DOI_HOSTS.has(host)) {
+        const doi = safeDecode(url.pathname.replace(/^\/+/, ''));
+        return classifyDoi(doi, 'url');
+    }
+    if (!ZENODO_HOSTS.has(host)) {
+        const note = host.endsWith('zenodo.org')
+            ? `${host} is a different Zenodo instance; only zenodo.org records are served.`
+            : `${host} is not zenodo.org or doi.org.`;
+        return { kind: 'invalid', message: note };
+    }
+    const path = stripTrailingPunctuation(url.pathname);
+    if (/^\/badge\/latestdoi\//i.test(path)) {
+        return {
+            kind: 'invalid',
+            message: 'A zenodo.org/badge/latestdoi URL carries a GitHub repository id, not a Zenodo record id.',
+        };
+    }
+    const record = /^\/(?:api\/)?records?\/(\d+)(?:\/|$)/i.exec(path);
+    if (record?.[1])
+        return { kind: 'recid', recid: record[1], inputKind: 'url' };
+    const doiPath = /^\/doi\/(.+)$/i.exec(path);
+    if (doiPath?.[1])
+        return classifyDoi(safeDecode(doiPath[1]), 'url');
+    const badge = /^\/badge\/doi\/(.+?)(?:\.svg)?$/i.exec(path);
+    if (badge?.[1])
+        return classifyDoi(safeDecode(badge[1]), 'url');
+    return {
+        kind: 'invalid',
+        message: `The zenodo.org path ${url.pathname} does not name a record.`,
+    };
+}
+/**
+ * Classifies a record reference. Accepts a record id, `zenodo.N`, a Zenodo DOI
+ * (any case, optionally `doi:`-prefixed), another DOI, a doi.org URL, or a
+ * zenodo.org record/DOI/badge URL. Wrapping `<>` / quotes (with any punctuation
+ * after them) and trailing punctuation after a record id or Zenodo DOI are
+ * stripped, in whichever order they were combined. Query strings and
+ * fragments are ignored. Never fetches anything.
+ */
+export function parseRecordRef(raw) {
+    const s = stripWrapping(raw);
+    if (!s)
+        return { kind: 'invalid', message: 'The id is empty.' };
+    const withScheme = /^(?:www\.)?(?:zenodo|doi|dx\.doi|sandbox\.zenodo)\.org\//i.test(s)
+        ? `https://${s}`
+        : s;
+    if (/^https?:\/\//i.test(withScheme))
+        return parseUrl(withScheme);
+    const bare = s.replace(/^doi:\s*/i, '');
+    const stripped = stripTrailingPunctuation(bare);
+    if (bare === s && /^\d+$/.test(stripped)) {
+        return { kind: 'recid', recid: stripped, inputKind: 'record_id' };
+    }
+    const suffix = ZENODO_SUFFIX_PATTERN.exec(stripped);
+    if (suffix?.[1])
+        return { kind: 'recid', recid: suffix[1], inputKind: 'zenodo_doi' };
+    if (/^10\./.test(bare))
+        return classifyDoi(bare, 'external_doi');
+    return {
+        kind: 'invalid',
+        message: `"${raw.trim().slice(0, 120)}" is not a Zenodo record id, DOI, or zenodo.org/doi.org URL.`,
+    };
+}
+/** True when `raw` parses as a single DOI or URL record reference (not a bare number). */
+export function isWholeIdentifier(raw) {
+    const s = raw.trim();
+    if (/^[a-z_.]+:/i.test(s) && !/^https?:/i.test(s))
+        return false;
+    const ref = parseRecordRef(s);
+    return ref.kind !== 'invalid' && ref.inputKind !== 'record_id' && !ZENODO_SUFFIX_PATTERN.test(s);
+}
+const ROR_ID = /^0[a-z0-9]{6}\d{2}$/;
+/**
+ * Parses a funder reference: a ROR id (`01cwqze88`), a ROR URL, or a Crossref
+ * Funder DOI (`10.13039/100000002`, bare, `doi:`-prefixed, or as a doi.org URL).
+ * Returns `undefined` for anything else — funder names are never matched.
+ */
+export function parseFunderRef(raw) {
+    const s = stripTrailingSlashes(stripWrapping(raw));
+    const ror = /^(?:https?:\/\/)?(?:www\.)?ror\.org\/(.+)$/i.exec(s)?.[1] ?? s;
+    if (ROR_ID.test(ror.toLowerCase()))
+        return { kind: 'ror', ror: ror.toLowerCase() };
+    const doi = s.replace(/^(?:https?:\/\/)?(?:dx\.)?doi\.org\//i, '').replace(/^doi:\s*/i, '');
+    if (/^10\.13039\/\S+$/i.test(doi))
+        return { kind: 'funder_doi', doi: doi.toLowerCase() };
+    return;
+}
+/** European Commission ROR id — the funder behind CORDIS award DOIs (`10.3030/N`). */
+export const EUROPEAN_COMMISSION_ROR = '00k4n6c32';
+/**
+ * Parses an award reference: `<funder-ror>::<number>`, a CORDIS award DOI
+ * (`10.3030/<n>` → European Commission), or a bare grant number.
+ */
+export function parseAwardRef(raw) {
+    const s = stripWrapping(raw);
+    const id = /^(0[a-z0-9]{6}\d{2})::(.+)$/i.exec(s);
+    if (id?.[1] && id[2])
+        return { field: 'id', value: `${id[1].toLowerCase()}::${id[2]}` };
+    const cordis = /^(?:(?:https?:\/\/)?(?:dx\.)?doi\.org\/|doi:\s*)?10\.3030\/(\S+)$/i.exec(s);
+    if (cordis?.[1])
+        return { field: 'id', value: `${EUROPEAN_COMMISSION_ROR}::${cordis[1]}` };
+    return { field: 'number', value: s };
+}
+/**
+ * Normalizes an ORCID input: strips an `orcid.org/` URL prefix and uppercases a
+ * trailing check character `x`. Validation is separate ({@link isValidOrcid}).
+ */
+export function normalizeOrcid(raw) {
+    return stripTrailingSlashes(stripWrapping(raw).replace(/^(?:https?:\/\/)?(?:www\.)?orcid\.org\//i, '')).replace(/x$/, 'X');
+}
+/** The `0000-0000-0000-000X` shape of an ORCID iD. */
+export const ORCID_PATTERN = /^\d{4}-\d{4}-\d{4}-\d{3}[\dX]$/;
+/** True when `orcid` has the {@link ORCID_PATTERN} shape and a valid ISO 7064 mod 11-2 check digit. */
+export function isValidOrcid(orcid) {
+    if (!ORCID_PATTERN.test(orcid))
+        return false;
+    const digits = orcid.replaceAll('-', '');
+    let total = 0;
+    for (const ch of digits.slice(0, 15))
+        total = (total + Number(ch)) * 2;
+    const result = (12 - (total % 11)) % 11;
+    return digits.at(-1) === (result === 10 ? 'X' : String(result));
+}
+/**
+ * Parses a community reference: a slug, a UUID, or a `zenodo.org/communities/<slug>`
+ * URL (scheme optional). Returns `undefined` when the value cannot be a
+ * community identifier, so it is never interpolated into a request path.
+ */
+export function parseCommunityRef(raw) {
+    const s = stripWrapping(raw);
+    const fromUrl = /^(?:https?:\/\/)?(?:www\.)?zenodo\.org\/communities\/([^/?#]+)/i.exec(s)?.[1] ?? s;
+    const value = stripTrailingSlashes(safeDecode(fromUrl));
+    return /^[A-Za-z0-9][A-Za-z0-9_.-]{0,199}$/.test(value) ? value : undefined;
+}
+/**
+ * True when a `/`-separated key or member path has a `.` or `..` segment.
+ * `encodeURIComponent` leaves those literal and the URL parser resolves them, so
+ * such a path would reach a different zenodo.org endpoint instead of naming a file.
+ */
+export function hasDotSegment(key) {
+    return key.split('/').some((segment) => segment === '.' || segment === '..');
+}
+/**
+ * Encodes a file key (or ZIP member path) for a URL path: each `/`-separated
+ * segment is percent-encoded. Request paths must reject {@link hasDotSegment}
+ * keys first, since `.` and `..` survive encoding.
+ */
+export function encodeKeySegments(key) {
+    return key.split('/').map(encodeURIComponent).join('/');
+}
+/** The public download URL for a top-level file. */
+export function downloadUrl(recid, key) {
+    return `https://zenodo.org/api/records/${recid}/files/${encodeKeySegments(key)}/content`;
+}
+/** The public landing page for a record. */
+export function recordUrl(recid) {
+    return `https://zenodo.org/records/${recid}`;
+}
+//# sourceMappingURL=identifiers.js.map

package/dist/services/zenodo/identifiers.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"identifiers.js","sourceRoot":"","sources":["../../../src/services/zenodo/identifiers.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAgCH,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC,CAAC,YAAY,EAAE,gBAAgB,CAAC,CAAC,CAAC;AAC/D,MAAM,SAAS,GAAG,IAAI,GAAG,CAAC,CAAC,SAAS,EAAE,YAAY,EAAE,aAAa,CAAC,CAAC,CAAC;AACpE,MAAM,WAAW,GAAG,oBAAoB,CAAC;AACzC,MAAM,kBAAkB,GAAG,4BAA4B,CAAC;AACxD,MAAM,qBAAqB,GAAG,kBAAkB,CAAC;AACjD,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAC5D,MAAM,KAAK,GAAG,IAAI,CAAC;AAEnB;;;;GAIG;AACH,SAAS,SAAS,CAAC,CAAS,EAAE,KAAa,EAAE,GAAW,EAAE,OAAgC;IACxF,IAAI,CAAC,GAAG,GAAG,CAAC;IACZ,OAAO,CAAC,GAAG,KAAK,IAAI,OAAO,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAW,CAAC;QAAE,CAAC,EAAE,CAAC;IACrD,OAAO,CAAC,CAAC;AACX,CAAC;AAED,MAAM,OAAO,GAAG,CAAC,EAAU,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AAC/C,MAAM,qBAAqB,GAAG,CAAC,EAAU,EAAE,EAAE,CAAC,EAAE,KAAK,GAAG,IAAI,EAAE,KAAK,GAAG,IAAI,EAAE,KAAK,GAAG,CAAC;AAErF,wCAAwC;AACxC,SAAS,wBAAwB,CAAC,CAAS;IACzC,OAAO,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;AACtE,CAAC;AAED,gCAAgC;AAChC,SAAS,oBAAoB,CAAC,CAAS;IACrC,OAAO,CAAC,CAAC,KAAK,CACZ,CAAC,EACD,SAAS,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,KAAK,GAAG,CAAC,CAC9C,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,SAAS,aAAa,CAAC,GAAW;IAChC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,GAAG,GAAG,GAAG,CAAC,MAAM,CAAC;IACrB,SAAS,CAAC;QACR,OAAO,KAAK,GAAG,GAAG,IAAI,OAAO,CAAC,GAAG,CAAC,KAAK,CAAW,CAAC;YAAE,KAAK,EAAE,CAAC;QAC7D,GAAG,GAAG,SAAS,CAAC,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,OAAO,CAAC,CAAC;QAC1C,MAAM,KAAK,GAAG,SAAS,CAAC,GAAG,EAAE,KAAK,EAAE,SAAS,CAAC,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,qBAAqB,CAAC,EAAE,OAAO,CAAC,CAAC;QAChG,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAW,CAAC;QACnC,MAAM,IAAI,GAAG,GAAG,CAAC,KAAK,GAAG,CAAC,CAAW,CAAC;QACtC,MAAM,OAAO,GACX,KAAK,GAAG,KAAK,IAAI,CAAC;YAClB,CAAC,CAAC,KAAK,KAAK,GAAG,IAAI,IAAI,KAAK,GAAG,CAAC,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QACzF,IAAI,CAAC,OAAO;YAAE,OAAO,GAAG,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;QAC3C,KAAK,EAAE,CAAC;QACR,GAAG,GAAG,KAAK,GAAG,CAAC,CAAC;IAClB,CAAC;AACH,CAAC;AAED,+DAA+D;AAC/D,SAAS,WAAW,CAAC,GAAW,EAAE,SAAoB;IACpD,MAAM,QAAQ,GAAG,wBAAwB,CAAC,GAAG,CAAC,CAAC;IAC/C,MAAM,MAAM,GAAG,kBAAkB,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IACjD,IAAI,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QAChB,OAAO;YACL,IAAI,EAAE,OAAO;YACb,KAAK,EAAE,MAAM,CAAC,CAAC,CAAC;YAChB,SAAS,EAAE,SAAS,KAAK,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,YAAY;SACtD,CAAC;IACJ,CAAC;IACD,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC;QAC3B,OAAO;YACL,IAAI,EAAE,SAAS;YACf,OAAO,EAAE,IAAI,GAAG,uDAAuD;SACxE,CAAC;IACJ,CAAC;IACD,OAAO;QACL,IAAI,EAAE,cAAc;QACpB,GAAG;QACH,SAAS,EAAE,SAAS,KAAK,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,cAAc;QACvD,GAAG,CAAC,QAAQ,KAAK,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,WAAW,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;KACrF,CAAC;AACJ,CAAC;AAED,SAAS,UAAU,CAAC,CAAS;IAC3B,IAAI,CAAC;QACH,OAAO,kBAAkB,CAAC,CAAC,CAAC,CAAC;IAC/B,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,CAAC,CAAC;IACX,CAAC;AACH,CAAC;AAED,oEAAoE;AACpE,SAAS,QAAQ,CAAC,GAAW;IAC3B,IAAI,GAAQ,CAAC;IACb,IAAI,CAAC;QACH,GAAG,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;IACrB,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,IAAI,GAAG,uBAAuB,EAAE,CAAC;IACtE,CAAC;IACD,MAAM,IAAI,GAAG,GAAG,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC;IAExC,IAAI,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;QACxB,MAAM,GAAG,GAAG,UAAU,CAAC,GAAG,CAAC,QAAQ,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,CAAC;QACzD,OAAO,WAAW,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;IACjC,CAAC;IAED,IAAI,CAAC,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;QAC5B,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,YAAY,CAAC;YACtC,CAAC,CAAC,GAAG,IAAI,sEAAsE;YAC/E,CAAC,CAAC,GAAG,IAAI,gCAAgC,CAAC;QAC5C,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC;IAC5C,CAAC;IAED,MAAM,IAAI,GAAG,wBAAwB,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;IACpD,IAAI,wBAAwB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QACxC,OAAO;YACL,IAAI,EAAE,SAAS;YACf,OAAO,EACL,0FAA0F;SAC7F,CAAC;IACJ,CAAC;IACD,MAAM,MAAM,GAAG,uCAAuC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAClE,IAAI,MAAM,EAAE,CAAC,CAAC,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,CAAC,CAAC,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;IAE9E,MAAM,OAAO,GAAG,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC5C,IAAI,OAAO,EAAE,CAAC,CAAC,CAAC;QAAE,OAAO,WAAW,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IAEpE,MAAM,KAAK,GAAG,kCAAkC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC5D,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC;QAAE,OAAO,WAAW,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IAEhE,OAAO;QACL,IAAI,EAAE,SAAS;QACf,OAAO,EAAE,uBAAuB,GAAG,CAAC,QAAQ,0BAA0B;KACvE,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,cAAc,CAAC,GAAW;IACxC,MAAM,CAAC,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC;IAC7B,IAAI,CAAC,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,kBAAkB,EAAE,CAAC;IAEhE,MAAM,UAAU,GAAG,2DAA2D,CAAC,IAAI,CAAC,CAAC,CAAC;QACpF,CAAC,CAAC,WAAW,CAAC,EAAE;QAChB,CAAC,CAAC,CAAC,CAAC;IACN,IAAI,eAAe,CAAC,IAAI,CAAC,UAAU,CAAC;QAAE,OAAO,QAAQ,CAAC,UAAU,CAAC,CAAC;IAElE,MAAM,IAAI,GAAG,CAAC,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC;IACxC,MAAM,QAAQ,GAAG,wBAAwB,CAAC,IAAI,CAAC,CAAC;IAEhD,IAAI,IAAI,KAAK,CAAC,IAAI,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC;QACzC,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,QAAQ,EAAE,SAAS,EAAE,WAAW,EAAE,CAAC;IACpE,CAAC;IACD,MAAM,MAAM,GAAG,qBAAqB,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IACpD,IAAI,MAAM,EAAE,CAAC,CAAC,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,CAAC,CAAC,CAAC,EAAE,SAAS,EAAE,YAAY,EAAE,CAAC;IAErF,IAAI,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,WAAW,CAAC,IAAI,EAAE,cAAc,CAAC,CAAC;IAEjE,OAAO;QACL,IAAI,EAAE,SAAS;QACf,OAAO,EAAE,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,8DAA8D;KACpG,CAAC;AACJ,CAAC;AAED,0FAA0F;AAC1F,MAAM,UAAU,iBAAiB,CAAC,GAAW;IAC3C,MAAM,CAAC,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC;IACrB,IAAI,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC;QAAE,OAAO,KAAK,CAAC;IAChE,MAAM,GAAG,GAAG,cAAc,CAAC,CAAC,CAAC,CAAC;IAC9B,OAAO,GAAG,CAAC,IAAI,KAAK,SAAS,IAAI,GAAG,CAAC,SAAS,KAAK,WAAW,IAAI,CAAC,qBAAqB,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AACnG,CAAC;AAKD,MAAM,MAAM,GAAG,qBAAqB,CAAC;AAErC;;;;GAIG;AACH,MAAM,UAAU,cAAc,CAAC,GAAW;IACxC,MAAM,CAAC,GAAG,oBAAoB,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC;IACnD,MAAM,GAAG,GAAG,6CAA6C,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;IAC5E,IAAI,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,CAAC,WAAW,EAAE,EAAE,CAAC;IACnF,MAAM,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC,uCAAuC,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC;IAC5F,IAAI,mBAAmB,CAAC,IAAI,CAAC,GAAG,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,EAAE,GAAG,CAAC,WAAW,EAAE,EAAE,CAAC;IACzF,OAAO;AACT,CAAC;AAED,sFAAsF;AACtF,MAAM,CAAC,MAAM,uBAAuB,GAAG,WAAW,CAAC;AAKnD;;;GAGG;AACH,MAAM,UAAU,aAAa,CAAC,GAAW;IACvC,MAAM,CAAC,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC;IAC7B,MAAM,EAAE,GAAG,8BAA8B,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAClD,IAAI,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;QAAE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;IACxF,MAAM,MAAM,GAAG,oEAAoE,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAC5F,IAAI,MAAM,EAAE,CAAC,CAAC,CAAC;QAAE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,uBAAuB,KAAK,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC;IAC3F,OAAO,EAAE,KAAK,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC;AACvC,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,cAAc,CAAC,GAAW;IACxC,OAAO,oBAAoB,CACzB,aAAa,CAAC,GAAG,CAAC,CAAC,OAAO,CAAC,0CAA0C,EAAE,EAAE,CAAC,CAC3E,CAAC,OAAO,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;AACvB,CAAC;AAED,sDAAsD;AACtD,MAAM,CAAC,MAAM,aAAa,GAAG,gCAAgC,CAAC;AAE9D,uGAAuG;AACvG,MAAM,UAAU,YAAY,CAAC,KAAa;IACxC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,KAAK,CAAC;IAC7C,MAAM,MAAM,GAAG,KAAK,CAAC,UAAU,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;IACzC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,EAAE,IAAI,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC;QAAE,KAAK,GAAG,CAAC,KAAK,GAAG,MAAM,CAAC,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC;IACvE,MAAM,MAAM,GAAG,CAAC,EAAE,GAAG,CAAC,KAAK,GAAG,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC;IACxC,OAAO,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,KAAK,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC;AAClE,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,iBAAiB,CAAC,GAAW;IAC3C,MAAM,CAAC,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC;IAC7B,MAAM,OAAO,GACX,iEAAiE,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;IACtF,MAAM,KAAK,GAAG,oBAAoB,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,CAAC;IACxD,OAAO,oCAAoC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAAC;AAC9E,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,aAAa,CAAC,GAAW;IACvC,OAAO,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,KAAK,GAAG,IAAI,OAAO,KAAK,IAAI,CAAC,CAAC;AAC/E,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,iBAAiB,CAAC,GAAW;IAC3C,OAAO,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,kBAAkB,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AAC1D,CAAC;AAED,oDAAoD;AACpD,MAAM,UAAU,WAAW,CAAC,KAAa,EAAE,GAAW;IACpD,OAAO,kCAAkC,KAAK,UAAU,iBAAiB,CAAC,GAAG,CAAC,UAAU,CAAC;AAC3F,CAAC;AAED,4CAA4C;AAC5C,MAAM,UAAU,SAAS,CAAC,KAAa;IACrC,OAAO,8BAA8B,KAAK,EAAE,CAAC;AAC/C,CAAC"}

package/dist/services/zenodo/normalize.d.ts

@@ -0,0 +1,29 @@
+/**
+ * @fileoverview Raw InvenioRDM JSON → domain types. Strips embedded file manifests
+ * from search and version hits, converts HTML descriptions to plain text, and keeps
+ * absent upstream fields absent instead of coercing them to `0`, `''`, or `false`.
+ * @module services/zenodo/normalize
+ */
+import type { AwardEntry, CommunityEntry, ContainerListing, FunderEntry, LicenseEntry, RawAward, RawCommunity, RawContainer, RawFunder, RawLicense, RawRecord, RawSearchResponse, RawTombstone, RecordSummary, SearchFacets, Tombstone, VersionSummary, ZenodoRecord } from './types.js';
+/**
+ * `{ [key]: value }` when the value is present, `{}` otherwise — for building
+ * objects under `exactOptionalPropertyTypes`. `null` and `''` count as absent.
+ */
+export declare function opt<K extends string, V>(key: K, value: V | null | undefined): Partial<Record<K, V>>;
+/** Normalizes a full RDM record (from a record GET or a search hit). */
+export declare function normalizeRecord(raw: RawRecord): ZenodoRecord;
+/** Normalizes a 410 tombstone body. */
+export declare function normalizeTombstone(raw: RawTombstone | undefined): Tombstone;
+/** Normalizes a search hit to a summary, dropping its embedded file manifest. */
+export declare function normalizeSearchHit(raw: RawRecord): RecordSummary;
+/** Facet counts from search aggregations. */
+export declare function normalizeFacets(aggs: RawSearchResponse['aggregations']): SearchFacets;
+/** Normalizes a version hit (manifest stripped; this-version usage counts). */
+export declare function normalizeVersionHit(raw: RawRecord): VersionSummary;
+/** Normalizes a `/container` listing. */
+export declare function normalizeContainer(raw: RawContainer): ContainerListing;
+export declare function normalizeCommunity(raw: RawCommunity): CommunityEntry;
+export declare function normalizeFunder(raw: RawFunder): FunderEntry;
+export declare function normalizeAward(raw: RawAward): AwardEntry;
+export declare function normalizeLicense(raw: RawLicense): LicenseEntry;
+//# sourceMappingURL=normalize.d.ts.map

package/dist/services/zenodo/normalize.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"normalize.d.ts","sourceRoot":"","sources":["../../../src/services/zenodo/normalize.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAIH,OAAO,KAAK,EACV,UAAU,EACV,cAAc,EACd,gBAAgB,EAGhB,WAAW,EACX,YAAY,EAGZ,QAAQ,EACR,YAAY,EACZ,YAAY,EAGZ,SAAS,EAGT,UAAU,EACV,SAAS,EACT,iBAAiB,EAEjB,YAAY,EACZ,aAAa,EACb,YAAY,EACZ,SAAS,EAET,cAAc,EACd,YAAY,EACb,MAAM,YAAY,CAAC;AAEpB;;;GAGG;AACH,wBAAgB,GAAG,CAAC,CAAC,SAAS,MAAM,EAAE,CAAC,EACrC,GAAG,EAAE,CAAC,EACN,KAAK,EAAE,CAAC,GAAG,IAAI,GAAG,SAAS,GAC1B,OAAO,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAIvB;AA4DD,wEAAwE;AACxE,wBAAgB,eAAe,CAAC,GAAG,EAAE,SAAS,GAAG,YAAY,CAuF5D;AAED,uCAAuC;AACvC,wBAAgB,kBAAkB,CAAC,GAAG,EAAE,YAAY,GAAG,SAAS,GAAG,SAAS,CAO3E;AAUD,iFAAiF;AACjF,wBAAgB,kBAAkB,CAAC,GAAG,EAAE,SAAS,GAAG,aAAa,CAyChE;AAOD,6CAA6C;AAC7C,wBAAgB,eAAe,CAAC,IAAI,EAAE,iBAAiB,CAAC,cAAc,CAAC,GAAG,YAAY,CAmBrF;AAED,+EAA+E;AAC/E,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,SAAS,GAAG,cAAc,CAelE;AAED,yCAAyC;AACzC,wBAAgB,kBAAkB,CAAC,GAAG,EAAE,YAAY,GAAG,gBAAgB,CActE;AAED,wBAAgB,kBAAkB,CAAC,GAAG,EAAE,YAAY,GAAG,cAAc,CAYpE;AAED,wBAAgB,eAAe,CAAC,GAAG,EAAE,SAAS,GAAG,WAAW,CAS3D;AAED,wBAAgB,cAAc,CAAC,GAAG,EAAE,QAAQ,GAAG,UAAU,CAcxD;AAED,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,UAAU,GAAG,YAAY,CAS9D"}

package/dist/services/zenodo/normalize.js

@@ -0,0 +1,324 @@
+/**
+ * @fileoverview Raw InvenioRDM JSON → domain types. Strips embedded file manifests
+ * from search and version hits, converts HTML descriptions to plain text, and keeps
+ * absent upstream fields absent instead of coercing them to `0`, `''`, or `false`.
+ * @module services/zenodo/normalize
+ */
+import { htmlToText } from './html-to-text.js';
+import { downloadUrl, recordUrl } from './identifiers.js';
+/**
+ * `{ [key]: value }` when the value is present, `{}` otherwise — for building
+ * objects under `exactOptionalPropertyTypes`. `null` and `''` count as absent.
+ */
+export function opt(key, value) {
+    return value === undefined || value === null || value === ''
+        ? {}
+        : { [key]: value };
+}
+/** English value of a multilingual map, else the first non-empty value. */
+function i18n(map) {
+    if (!map)
+        return;
+    return map.en || Object.values(map).find((v) => v) || undefined;
+}
+function identifierOf(ids, scheme) {
+    return ids?.find((i) => i.scheme?.toLowerCase() === scheme)?.identifier || undefined;
+}
+function counts(block) {
+    return { ...opt('views', block?.unique_views), ...opt('downloads', block?.unique_downloads) };
+}
+function stripMd5(checksum) {
+    if (!checksum)
+        return;
+    return checksum.startsWith('md5:') ? checksum.slice(4) : undefined;
+}
+function normalizePerson(raw) {
+    const p = raw.person_or_org ?? {};
+    const name = p.name || [p.given_name, p.family_name].filter(Boolean).join(' ') || p.family_name || '';
+    return {
+        name,
+        ...opt('type', p.type),
+        ...opt('orcid', identifierOf(p.identifiers, 'orcid')),
+        ...opt('role', raw.role ? (i18n(raw.role.title) ?? raw.role.id) : undefined),
+        affiliations: (raw.affiliations ?? [])
+            .filter((a) => a.name || a.id)
+            .map((a) => ({ ...opt('name', a.name), ...opt('ror', a.id) })),
+    };
+}
+/** The complete manifest in upstream order (`files.order` when non-empty, else object order). */
+function normalizeManifest(recid, files) {
+    const entries = files?.entries ?? {};
+    const keys = files?.order?.length
+        ? files.order.filter((k) => k in entries)
+        : Object.keys(entries);
+    return keys.map((mapKey) => {
+        const e = entries[mapKey] ?? {};
+        const key = e.key ?? mapKey;
+        return {
+            key,
+            ...opt('size', e.size),
+            ...opt('mimetype', e.mimetype),
+            ...opt('md5', stripMd5(e.checksum)),
+            download_url: downloadUrl(recid, key),
+        };
+    });
+}
+function embargoUntil(raw) {
+    const embargo = raw.access?.embargo;
+    return embargo?.active ? embargo.until : undefined;
+}
+/** Normalizes a full RDM record (from a record GET or a search hit). */
+export function normalizeRecord(raw) {
+    const m = raw.metadata ?? {};
+    const recid = String(raw.id);
+    const description = m.description ? htmlToText(m.description) : undefined;
+    const embargo = raw.access?.embargo;
+    const codeRepository = raw.custom_fields?.['code:codeRepository'];
+    return {
+        recid,
+        ...opt('concept_recid', raw.parent?.id),
+        ...opt('doi', raw.pids?.doi?.identifier),
+        ...opt('doi_provider', raw.pids?.doi?.provider),
+        ...opt('concept_doi', raw.parent?.pids?.doi?.identifier),
+        ...opt('oai_id', raw.pids?.oai?.identifier),
+        title: m.title ?? '',
+        ...opt('publication_date', m.publication_date),
+        ...opt('version', m.version),
+        ...opt('publisher', m.publisher),
+        ...(m.resource_type?.id
+            ? { resource_type: { id: m.resource_type.id, ...opt('title', i18n(m.resource_type.title)) } }
+            : {}),
+        ...opt('description', description),
+        additional_descriptions: (m.additional_descriptions ?? [])
+            .filter((d) => d.description)
+            .map((d) => ({
+            ...opt('type', d.type ? (i18n(d.type.title) ?? d.type.id) : undefined),
+            text: htmlToText(d.description ?? ''),
+        })),
+        creators: (m.creators ?? []).map(normalizePerson),
+        contributors: (m.contributors ?? []).map(normalizePerson),
+        keywords: (m.subjects ?? []).map((s) => s.subject).filter((s) => !!s),
+        rights: (m.rights ?? []).map((r) => ({
+            ...opt('id', r.id),
+            title: i18n(r.title) ?? r.id ?? '',
+            ...opt('url', r.props?.url),
+        })),
+        access: {
+            status: raw.access?.status ?? 'unknown',
+            ...opt('record', raw.access?.record),
+            ...opt('files', raw.access?.files),
+            embargo_active: embargo?.active === true,
+            ...opt('embargo_until', embargoUntil(raw)),
+            ...opt('embargo_reason', embargo?.active ? embargo.reason : undefined),
+        },
+        funding: (m.funding ?? []).map((f) => ({
+            ...opt('funder_id', f.funder?.id),
+            ...opt('funder_name', f.funder?.name),
+            ...opt('award_id', f.award?.id),
+            ...opt('award_number', f.award?.number),
+            ...opt('award_acronym', f.award?.acronym),
+            ...opt('award_title', i18n(f.award?.title)),
+            ...opt('award_program', f.award?.program),
+            ...opt('award_doi', identifierOf(f.award?.identifiers, 'doi')),
+            ...opt('award_url', identifierOf(f.award?.identifiers, 'url')),
+        })),
+        related_identifiers: (m.related_identifiers ?? [])
+            .filter((r) => r.identifier)
+            .map((r) => ({
+            identifier: r.identifier ?? '',
+            ...opt('scheme', r.scheme),
+            ...opt('relation', r.relation_type?.id),
+            ...opt('resource_type', r.resource_type?.id),
+        })),
+        ...opt('code_repository', typeof codeRepository === 'string' ? codeRepository : undefined),
+        communities: (raw.parent?.communities?.entries ?? [])
+            .filter((c) => c.slug)
+            .map((c) => ({ slug: c.slug ?? '', ...opt('title', c.metadata?.title) })),
+        versions: {
+            ...opt('index', raw.versions?.index),
+            ...opt('is_latest', raw.versions?.is_latest),
+        },
+        ...(raw.stats
+            ? {
+                stats: {
+                    this_version: counts(raw.stats.this_version),
+                    all_versions: counts(raw.stats.all_versions),
+                },
+            }
+            : {}),
+        files: {
+            enabled: raw.files?.enabled === true,
+            ...opt('count', raw.files?.count),
+            ...opt('total_bytes', raw.files?.total_bytes),
+            entries: normalizeManifest(recid, raw.files),
+        },
+        ...opt('revision', raw.revision_id),
+    };
+}
+/** Normalizes a 410 tombstone body. */
+export function normalizeTombstone(raw) {
+    return {
+        ...opt('removal_date', raw?.removal_date),
+        ...opt('removal_reason', raw?.removal_reason?.id),
+        ...opt('note', raw?.note),
+        ...opt('citation_text', raw?.citation_text),
+    };
+}
+/** First `max` characters at a word boundary, with `…` when cut. */
+function snippet(text, max) {
+    if (text.length <= max)
+        return text;
+    const cut = text.slice(0, max);
+    const space = cut.lastIndexOf(' ');
+    return `${(space > max * 0.6 ? cut.slice(0, space) : cut).trimEnd()}…`;
+}
+/** Normalizes a search hit to a summary, dropping its embedded file manifest. */
+export function normalizeSearchHit(raw) {
+    const m = raw.metadata ?? {};
+    const recid = String(raw.id);
+    const creators = m.creators ?? [];
+    const description = m.description ? htmlToText(m.description) : undefined;
+    return {
+        recid,
+        ...opt('doi', raw.pids?.doi?.identifier),
+        ...opt('doi_provider', raw.pids?.doi?.provider),
+        ...opt('concept_recid', raw.parent?.id),
+        ...opt('concept_doi', raw.parent?.pids?.doi?.identifier),
+        title: m.title ?? '',
+        ...opt('publication_date', m.publication_date),
+        ...(m.resource_type?.id
+            ? { resource_type: { id: m.resource_type.id, ...opt('title', i18n(m.resource_type.title)) } }
+            : {}),
+        ...opt('version', m.version),
+        ...opt('is_latest', raw.versions?.is_latest),
+        ...opt('version_index', raw.versions?.index),
+        creators: creators.slice(0, 5).map((c) => {
+            const p = normalizePerson(c);
+            return { name: p.name, ...opt('orcid', p.orcid) };
+        }),
+        creator_count: creators.length,
+        license_ids: (m.rights ?? []).map((r) => r.id).filter((id) => !!id),
+        access: {
+            status: raw.access?.status ?? 'unknown',
+            ...opt('files', raw.access?.files),
+            ...opt('embargo_until', embargoUntil(raw)),
+        },
+        communities: (raw.parent?.communities?.entries ?? [])
+            .map((c) => c.slug)
+            .filter((s) => !!s)
+            .slice(0, 5),
+        ...opt('file_count', raw.files?.count),
+        ...opt('total_bytes', raw.files?.total_bytes),
+        ...opt('views', raw.stats?.all_versions?.unique_views),
+        ...opt('downloads', raw.stats?.all_versions?.unique_downloads),
+        ...opt('description_snippet', description ? snippet(description, 280) : undefined),
+        zenodo_url: recordUrl(recid),
+    };
+}
+function bucket(b) {
+    const id = String(b.key ?? '');
+    return { id, label: b.label ?? id, count: b.doc_count ?? 0 };
+}
+/** Facet counts from search aggregations. */
+export function normalizeFacets(aggs) {
+    const buckets = (name) => aggs?.[name]?.buckets ?? [];
+    return {
+        resource_type: buckets('resource_type')
+            .slice(0, 10)
+            .map((b) => {
+            const subtypes = (b.inner?.buckets ?? []).slice(0, 5).map(bucket);
+            return { ...bucket(b), ...(subtypes.length ? { subtypes } : {}) };
+        }),
+        access_status: buckets('access_status').map(bucket),
+        file_type: buckets('file_type').slice(0, 10).map(bucket),
+        subject: buckets('subject')
+            .slice(0, 10)
+            .map((b) => ({ label: b.label ?? String(b.key ?? ''), count: b.doc_count ?? 0 })),
+        publication_year: buckets('publication_date')
+            .map((b) => ({ year: String(b.key ?? ''), count: b.doc_count ?? 0 }))
+            .sort((a, b) => b.year.localeCompare(a.year))
+            .slice(0, 10),
+    };
+}
+/** Normalizes a version hit (manifest stripped; this-version usage counts). */
+export function normalizeVersionHit(raw) {
+    const m = raw.metadata ?? {};
+    return {
+        recid: String(raw.id),
+        ...opt('doi', raw.pids?.doi?.identifier),
+        ...opt('version', m.version),
+        title: m.title ?? '',
+        ...opt('publication_date', m.publication_date),
+        ...opt('index', raw.versions?.index),
+        ...opt('is_latest', raw.versions?.is_latest),
+        ...opt('file_count', raw.files?.count),
+        ...opt('total_bytes', raw.files?.total_bytes),
+        ...opt('views', raw.stats?.this_version?.unique_views),
+        ...opt('downloads', raw.stats?.this_version?.unique_downloads),
+    };
+}
+/** Normalizes a `/container` listing. */
+export function normalizeContainer(raw) {
+    return {
+        members: (raw.entries ?? [])
+            .filter((e) => e.key)
+            .map((e) => ({
+            path: e.key ?? '',
+            ...opt('size', e.size),
+            ...opt('compressed_size', e.compressed_size),
+            ...opt('mimetype', e.mimetype),
+        })),
+        ...opt('total', raw.total),
+        upstream_truncated: raw.truncated === true,
+        directory_count: raw.directories?.length ?? 0,
+    };
+}
+export function normalizeCommunity(raw) {
+    const m = raw.metadata ?? {};
+    return {
+        slug: raw.slug ?? '',
+        uuid: raw.id ?? '',
+        ...opt('title', m.title),
+        ...opt('community_type', m.type ? (i18n(m.type.title) ?? m.type.id) : undefined),
+        ...opt('website', m.website),
+        organizations: (m.organizations ?? [])
+            .map((o) => o.name || (o.id ? `ROR ${o.id}` : ''))
+            .filter(Boolean),
+    };
+}
+export function normalizeFunder(raw) {
+    return {
+        ror_id: raw.id ?? '',
+        ...opt('name', raw.name ?? i18n(raw.title)),
+        ...opt('funder_doi', identifierOf(raw.identifiers, 'doi')),
+        ...opt('acronym', raw.acronym),
+        ...opt('country', raw.country),
+        ...opt('country_name', raw.country_name),
+    };
+}
+export function normalizeAward(raw) {
+    return {
+        id: raw.id ?? '',
+        ...opt('number', raw.number),
+        ...opt('acronym', raw.acronym),
+        ...opt('title', i18n(raw.title)),
+        ...opt('program', raw.program),
+        ...opt('funder_id', raw.funder?.id),
+        ...opt('funder_name', raw.funder?.name),
+        ...opt('award_doi', identifierOf(raw.identifiers, 'doi')),
+        ...opt('award_url', identifierOf(raw.identifiers, 'url')),
+        ...opt('start_date', raw.start_date),
+        ...opt('end_date', raw.end_date),
+    };
+}
+export function normalizeLicense(raw) {
+    const osi = raw.props?.osi_approved;
+    return {
+        id: raw.id ?? '',
+        ...opt('title', i18n(raw.title)),
+        ...opt('url', raw.props?.url),
+        ...(osi === 'y' || osi === 'n' ? { osi_approved: osi === 'y' } : {}),
+        tags: raw.tags ?? [],
+    };
+}
+//# sourceMappingURL=normalize.js.map