@canonical/pragma-cli 0.34.0 → 0.36.0

package/dist/src/kernel/runtime/graphpack/build.js

@@ -0,0 +1,122 @@
+/**
+ * Build a content-addressed pack from RDF source inputs.
+ *
+ * This is the ONE place the live ke-graphql 7-pass `compile` runs (~210 ms):
+ * building is when we can afford it. The store is created once (parsing the
+ * TTL and writing the `data.nq` n-quads dump via ke's cache), the schema is
+ * extracted and serialized, the entity index is built, and a manifest is
+ * written LAST so the directory is only ever observed complete. Writes go to a
+ * temp directory and are atomically renamed into `packs/<contentHash>/`, so a
+ * crashed build never leaves a half-written pack; a hash whose directory is
+ * already complete is reused untouched.
+ *
+ * Reached only behind a dynamic import (`sources update`), never on the
+ * storeless fast path.
+ */
+import { existsSync, mkdirSync, mkdtempSync, renameSync, rmSync, writeFileSync, } from "node:fs";
+import { join } from "node:path";
+import { createStore } from "@canonical/ke";
+import { compile, createStoreQueryFn, serializeExtraction, } from "@canonical/ke-graphql";
+import { packDir, packsCacheDir } from "../paths.js";
+import { buildIndex } from "./buildIndex.js";
+import { contentHash, hashSources } from "./hash.js";
+import { packIsComplete, readManifest } from "./manifest.js";
+import { DATA_FILE, INDEX_FILE, MANIFEST_FILE, SCHEMA_FILE, STORIES_FILE, } from "./types.js";
+/**
+ * Build (or reuse) the pack for a set of source inputs.
+ *
+ * @param inputs - The RDF sources (path + content).
+ * @param options - Provenance, the prefixes to build with, and the packages'
+ *   carried read stories.
+ * @returns The pack directory, its content hash, and whether it was reused.
+ * @note Impure — creates a store, compiles the schema, writes the pack.
+ */
+export async function buildPack(inputs, options) {
+    const stories = options.stories ?? [];
+    const hash = await contentHash([...inputs, ...stories].map((input) => ({
+        path: input.path,
+        content: input.content,
+    })));
+    const dir = packDir(hash);
+    // Reuse only a COMPLETE pack (manifest + non-empty dump). A corrupt cache
+    // (intact manifest, ruined `data.nq`) must rebuild, not be reused — otherwise
+    // an emptied dump would be permanent.
+    if (packIsComplete(dir)) {
+        const cached = readManifest(dir);
+        if (cached)
+            return { dir, contentHash: hash, manifest: cached, reused: true };
+    }
+    mkdirSync(packsCacheDir(), { recursive: true });
+    const temp = mkdtempSync(join(packsCacheDir(), `.build-${hash.slice(0, 12)}-`));
+    try {
+        const prefixes = options.prefixes ?? {};
+        const store = await createStore({
+            sources: inputs.map((input) => ({
+                content: input.content,
+                path: input.path,
+            })),
+            prefixes,
+            // ke writes the n-quads dump here after loading — our `data.nq`.
+            cache: join(temp, DATA_FILE),
+        });
+        try {
+            const compiled = await compile(createStoreQueryFn(store), store.prefixes);
+            const sourcesHash = hashSources(inputs.map((input) => input.content));
+            writeFileSync(join(temp, SCHEMA_FILE), serializeExtraction(compiled.extraction, sourcesHash));
+            const index = await buildIndex(store, store.prefixes, hash);
+            writeFileSync(join(temp, INDEX_FILE), JSON.stringify(index));
+            // Written UNCONDITIONALLY (`[]` when the packages ship none): the artifact
+            // set is FIVE AND ONLY FIVE, and `packIsComplete` gates on it, so a pack
+            // whose hash covers stories always holds them.
+            writeFileSync(join(temp, STORIES_FILE), JSON.stringify(stories.map((story) => ({
+                source: story.path,
+                content: story.content,
+            }))));
+            const tripleCount = await countTriples(store);
+            // Distinct abox subjects — the same figure `entityTotal` reports (A10),
+            // so `sources status` reads it from the manifest without parsing the
+            // whole index.json. Keep this in sync with `entityTotal` in entitySource.
+            const entityCount = new Set(index.entities
+                .filter((entity) => entity.box === "abox")
+                .map((entity) => entity.uri ?? entity.name)).size;
+            const manifest = {
+                name: options.name,
+                version: options.version,
+                sourceRef: options.sourceRef,
+                contentHash: hash,
+                prefixes: { ...store.prefixes },
+                createdAt: new Date().toISOString(),
+                tripleCount,
+                entityCount,
+            };
+            // Written LAST — the completeness marker for the directory.
+            writeFileSync(join(temp, MANIFEST_FILE), JSON.stringify(manifest));
+            // Publish atomically. A concurrent builder may have won the race — its
+            // pack is content-identical, so keep it and drop ours.
+            if (!packIsComplete(dir)) {
+                rmSync(dir, { recursive: true, force: true });
+                renameSync(temp, dir);
+            }
+            return { dir, contentHash: hash, manifest, reused: false };
+        }
+        finally {
+            store.dispose();
+        }
+    }
+    finally {
+        if (existsSync(temp))
+            rmSync(temp, { recursive: true, force: true });
+    }
+}
+/**
+ * Count a store's triples (a cheap aggregate over the union graph).
+ *
+ * @param store - The booted store.
+ * @returns The number of triples.
+ * @note Impure — runs a COUNT query against the store.
+ */
+async function countTriples(store) {
+    const result = (await store.query("SELECT (COUNT(*) AS ?n) WHERE { ?s ?p ?o }"));
+    return Number(result.bindings.at(0)?.n ?? 0);
+}
+//# sourceMappingURL=build.js.map

package/dist/src/kernel/runtime/graphpack/build.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"build.js","sourceRoot":"","sources":["../../../../../src/kernel/runtime/graphpack/build.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EACL,UAAU,EACV,SAAS,EACT,WAAW,EACX,UAAU,EACV,MAAM,EACN,aAAa,GACd,MAAM,SAAS,CAAC;AACjB,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AACjC,OAAO,EAAE,WAAW,EAAE,MAAM,eAAe,CAAC;AAC5C,OAAO,EACL,OAAO,EACP,kBAAkB,EAClB,mBAAmB,GACpB,MAAM,uBAAuB,CAAC;AAC/B,OAAO,EAAE,OAAO,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AACrD,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAC;AAC7C,OAAO,EAAE,WAAW,EAAE,WAAW,EAAE,MAAM,WAAW,CAAC;AACrD,OAAO,EAAE,cAAc,EAAE,YAAY,EAAE,MAAM,eAAe,CAAC;AAC7D,OAAO,EACL,SAAS,EACT,UAAU,EACV,aAAa,EAEb,WAAW,EACX,YAAY,GACb,MAAM,YAAY,CAAC;AAoCpB;;;;;;;;GAQG;AACH,MAAM,CAAC,KAAK,UAAU,SAAS,CAC7B,MAAiC,EACjC,OAAyB;IAEzB,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC;IACtC,MAAM,IAAI,GAAG,MAAM,WAAW,CAC5B,CAAC,GAAG,MAAM,EAAE,GAAG,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;QACtC,IAAI,EAAE,KAAK,CAAC,IAAI;QAChB,OAAO,EAAE,KAAK,CAAC,OAAO;KACvB,CAAC,CAAC,CACJ,CAAC;IACF,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC;IAE1B,0EAA0E;IAC1E,8EAA8E;IAC9E,sCAAsC;IACtC,IAAI,cAAc,CAAC,GAAG,CAAC,EAAE,CAAC;QACxB,MAAM,MAAM,GAAG,YAAY,CAAC,GAAG,CAAC,CAAC;QACjC,IAAI,MAAM;YACR,OAAO,EAAE,GAAG,EAAE,WAAW,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;IACtE,CAAC;IAED,SAAS,CAAC,aAAa,EAAE,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;IAChD,MAAM,IAAI,GAAG,WAAW,CACtB,IAAI,CAAC,aAAa,EAAE,EAAE,UAAU,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CACtD,CAAC;IACF,IAAI,CAAC;QACH,MAAM,QAAQ,GAAG,OAAO,CAAC,QAAQ,IAAI,EAAE,CAAC;QACxC,MAAM,KAAK,GAAG,MAAM,WAAW,CAAC;YAC9B,OAAO,EAAE,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;gBAC9B,OAAO,EAAE,KAAK,CAAC,OAAO;gBACtB,IAAI,EAAE,KAAK,CAAC,IAAI;aACjB,CAAC,CAAC;YACH,QAAQ;YACR,iEAAiE;YACjE,KAAK,EAAE,IAAI,CAAC,IAAI,EAAE,SAAS,CAAC;SAC7B,CAAC,CAAC;QAEH,IAAI,CAAC;YACH,MAAM,QAAQ,GAAG,MAAM,OAAO,CAAC,kBAAkB,CAAC,KAAK,CAAC,EAAE,KAAK,CAAC,QAAQ,CAAC,CAAC;YAC1E,MAAM,WAAW,GAAG,WAAW,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;YACtE,aAAa,CACX,IAAI,CAAC,IAAI,EAAE,WAAW,CAAC,EACvB,mBAAmB,CAAC,QAAQ,CAAC,UAAU,EAAE,WAAW,CAAC,CACtD,CAAC;YAEF,MAAM,KAAK,GAAG,MAAM,UAAU,CAAC,KAAK,EAAE,KAAK,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;YAC5D,aAAa,CAAC,IAAI,CAAC,IAAI,EAAE,UAAU,CAAC,EAAE,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC;YAE7D,2EAA2E;YAC3E,yEAAyE;YACzE,+CAA+C;YAC/C,aAAa,CACX,IAAI,CAAC,IAAI,EAAE,YAAY,CAAC,EACxB,IAAI,CAAC,SAAS,CACZ,OAAO,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;gBACtB,MAAM,EAAE,KAAK,CAAC,IAAI;gBAClB,OAAO,EAAE,KAAK,CAAC,OAAO;aACvB,CAAC,CAAC,CACJ,CACF,CAAC;YAEF,MAAM,WAAW,GAAG,MAAM,YAAY,CAAC,KAAK,CAAC,CAAC;YAC9C,wEAAwE;YACxE,qEAAqE;YACrE,0EAA0E;YAC1E,MAAM,WAAW,GAAG,IAAI,GAAG,CACzB,KAAK,CAAC,QAAQ;iBACX,MAAM,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,GAAG,KAAK,MAAM,CAAC;iBACzC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,CAC9C,CAAC,IAAI,CAAC;YAEP,MAAM,QAAQ,GAAa;gBACzB,IAAI,EAAE,OAAO,CAAC,IAAI;gBAClB,OAAO,EAAE,OAAO,CAAC,OAAO;gBACxB,SAAS,EAAE,OAAO,CAAC,SAAS;gBAC5B,WAAW,EAAE,IAAI;gBACjB,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,QAAQ,EAAE;gBAC/B,SAAS,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;gBACnC,WAAW;gBACX,WAAW;aACZ,CAAC;YACF,4DAA4D;YAC5D,aAAa,CAAC,IAAI,CAAC,IAAI,EAAE,aAAa,CAAC,EAAE,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAC,CAAC;YAEnE,uEAAuE;YACvE,uDAAuD;YACvD,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,EAAE,CAAC;gBACzB,MAAM,CAAC,GAAG,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;gBAC9C,UAAU,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;YACxB,CAAC;YACD,OAAO,EAAE,GAAG,EAAE,WAAW,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC;QAC7D,CAAC;gBAAS,CAAC;YACT,KAAK,CAAC,OAAO,EAAE,CAAC;QAClB,CAAC;IACH,CAAC;YAAS,CAAC;QACT,IAAI,UAAU,CAAC,IAAI,CAAC;YAAE,MAAM,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;IACvE,CAAC;AACH,CAAC;AAED;;;;;;GAMG;AACH,KAAK,UAAU,YAAY,CACzB,KAA8C;IAE9C,MAAM,MAAM,GAAG,CAAC,MAAM,KAAK,CAAC,KAAK,CAC/B,4CAAqD,CACtD,CAAyC,CAAC;IAC3C,OAAO,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC;AAC/C,CAAC"}

package/dist/src/kernel/runtime/graphpack/buildIndex.js

@@ -0,0 +1,303 @@
+/**
+ * Build the storeless entity index (`index.json`) from a populated store.
+ *
+ * A lean port of the v1 `buildGraphIndex` (#856): a few bulk SPARQL queries —
+ * types (which also yield the instance counts), labels, descriptions and
+ * alternative names — joined in memory, rather than a query per subject. Only
+ * the last names a DOMAIN property, and it is read from the distribution's
+ * declaration rather than spelled out here; see {@link resolveAltNamePredicate} for
+ * the pack that binds no such prefix.
+ *
+ * Only typed named-node subjects are indexed (untyped subjects and blank-node
+ * subjects are reachable via `graph query` but never listed). Each entity
+ * carries the FROZEN `{ name, type }` minimum plus the enrichment fields
+ * (`uri`, `prefixed`, `types`, `label`, `box`) the completion tier and reads
+ * use. The output is pure JSON — no store handle survives — so the completion
+ * tier reads it without ever booting oxigraph.
+ *
+ * Real-data correctness (things a single-typed toy graph never exercises):
+ * - Blank-node `rdf:type` objects (SHACL shapes, RDF lists) are ignored — a
+ *   `_:b0` is never a domain class and must not become a `type:"_:b0"` key.
+ * - The abox primary type is the LEXICALLY-SMALLEST domain class, a
+ *   deterministic tie-break: SPARQL returns a subject's types in unspecified
+ *   order, so a multi-domain individual would otherwise get an arbitrary
+ *   completion key that flips across builds/engines.
+ * - Multilingual labels prefer untagged / `@en`, then the lexically smallest —
+ *   never the store's arbitrary first row.
+ * - An OWL-punned subject (a class/property IRI ALSO asserted as a domain
+ *   individual) emits BOTH its tbox and abox facet, so its abox membership
+ *   stays completable instead of vanishing behind the tbox facet.
+ */
+import { compactUri } from "../../render/index.js";
+import { VOCABULARY } from "../../vocabulary.js";
+/** `rdf:type`. */
+const RDF_TYPE = "http://www.w3.org/1999/02/22-rdf-syntax-ns#type";
+/** Meta-types that mark a subject as TBox schema (a class). */
+const CLASS_METATYPES = new Set([
+    "http://www.w3.org/2002/07/owl#Class",
+    "http://www.w3.org/2000/01/rdf-schema#Class",
+]);
+/** Meta-types that mark a subject as TBox schema (a property). */
+const PROPERTY_METATYPES = new Set([
+    "http://www.w3.org/2002/07/owl#ObjectProperty",
+    "http://www.w3.org/2002/07/owl#DatatypeProperty",
+    "http://www.w3.org/2002/07/owl#AnnotationProperty",
+    "http://www.w3.org/1999/02/22-rdf-syntax-ns#Property",
+]);
+/** `owl:NamedIndividual` — a typing artifact, never a subject's domain class. */
+const NAMED_INDIVIDUAL = "http://www.w3.org/2002/07/owl#NamedIndividual";
+/** Namespaces whose types are structural, not a subject's domain class. */
+const STD_VOCAB_PREFIXES = [
+    "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
+    "http://www.w3.org/2000/01/rdf-schema#",
+    "http://www.w3.org/2002/07/owl#",
+    "http://www.w3.org/2001/XMLSchema#",
+    "http://www.w3.org/2004/02/skos/core#",
+    "http://www.w3.org/ns/shacl#",
+];
+/** Label predicates, in preference order. */
+const LABEL_PREDICATES = [
+    "http://www.w3.org/2000/01/rdf-schema#label",
+    "http://www.w3.org/2004/02/skos/core#prefLabel",
+    "http://purl.org/dc/terms/title",
+    "http://schema.org/name",
+];
+/** Description predicates, in preference order. */
+const DESCRIPTION_PREDICATES = [
+    "http://www.w3.org/2000/01/rdf-schema#comment",
+    "http://purl.org/dc/terms/description",
+    "http://www.w3.org/2004/02/skos/core#definition",
+    "http://schema.org/description",
+];
+const isStdVocab = (uri) => STD_VOCAB_PREFIXES.some((ns) => uri.startsWith(ns));
+/**
+ * Classify a subject's full types into its facets — one per box it occupies.
+ *
+ * A plain individual or a plain class yields ONE facet. An OWL-punned subject
+ * (a class/property IRI ALSO asserted as a domain individual) yields BOTH its
+ * tbox and abox facet (A8), so its abox membership stays completable rather than
+ * disappearing behind the tbox one. The abox primary is the lexically-smallest
+ * domain class — a deterministic tie-break (A3), not the store's first row.
+ *
+ * @param fullTypes - The subject's asserted `rdf:type` values (blank nodes
+ *   already dropped), in store order.
+ * @returns The subject's facets (empty when it has no usable type).
+ */
+function classify(fullTypes) {
+    if (fullTypes.length === 0)
+        return [];
+    const facets = [];
+    const classMeta = fullTypes.find((t) => CLASS_METATYPES.has(t));
+    const propMeta = fullTypes.find((t) => PROPERTY_METATYPES.has(t));
+    if (classMeta !== undefined)
+        facets.push({ box: "tbox", primary: classMeta });
+    else if (propMeta !== undefined)
+        facets.push({ box: "tbox", primary: propMeta });
+    // A subject's domain classes: its own types minus the structural std vocab
+    // and the `owl:NamedIndividual` typing artifact. Sorted, so the abox primary
+    // (the smallest) is stable across builds/engines regardless of store order.
+    const domainClasses = fullTypes
+        .filter((t) => t !== NAMED_INDIVIDUAL && !isStdVocab(t))
+        .sort();
+    const primaryDomain = domainClasses.at(0);
+    if (primaryDomain !== undefined) {
+        facets.push({ box: "abox", primary: primaryDomain });
+    }
+    else if (facets.length === 0) {
+        // No schema facet and no domain class — a bare individual (only
+        // `owl:NamedIndividual`, or only std-vocab types). Keep it indexed under a
+        // deterministic primary rather than dropping it silently.
+        const fallback = fullTypes
+            .filter((t) => t !== NAMED_INDIVIDUAL)
+            .sort()
+            .at(0) ?? [...fullTypes].sort().at(0);
+        if (fallback !== undefined)
+            facets.push({ box: "abox", primary: fallback });
+    }
+    return facets;
+}
+/** A `VALUES ?p { … }` body from a predicate list. */
+const valuesList = (uris) => uris.map((uri) => `<${uri}>`).join(" ");
+/**
+ * Expand the declared alternative-name property against a pack's prefix map.
+ *
+ * That property is what a family is addressed by, so the storeless completion
+ * sources can offer those tokens exactly and a bespoke lookup can match them,
+ * without a store boot. Distinct from the display `label`: a subject may carry
+ * both.
+ *
+ * The distribution DECLARES it as a prefixed name; this resolves it against the
+ * PACK's own prefixes, which are not the distribution's. A pack that binds no
+ * such prefix simply carries no alternative names — the normal case for a
+ * third-party pack outside the distribution's namespace. Expanding through
+ * `packs/iri.ts` instead would THROW on exactly that pack and take its whole
+ * build down, so this degrades and the caller skips the query.
+ *
+ * The split needs no guard: `kernel/vocabulary.ts` has already held the term to
+ * `prefix:local`, so the colon is there and both sides are non-empty.
+ *
+ * @param prefixes - The store's merged prefix map.
+ * @returns The full IRI to collect, or `undefined` when the prefix is unbound.
+ */
+function resolveAltNamePredicate(prefixes) {
+    const colon = VOCABULARY.altName.indexOf(":");
+    const namespace = prefixes[VOCABULARY.altName.slice(0, colon)];
+    return namespace === undefined
+        ? undefined
+        : `${namespace}${VOCABULARY.altName.slice(colon + 1)}`;
+}
+/** Language rank for label selection: untagged (0) < `@en` (1) < any other (2). */
+function rankLanguage(language) {
+    if (language === "")
+        return 0;
+    const lower = language.toLowerCase();
+    return lower === "en" || lower.startsWith("en-") ? 1 : 2;
+}
+/**
+ * Per subject, keep the object whose predicate ranks HIGHEST (lowest index) in
+ * `preference`. The `?p` binding is projected so the winner is chosen by declared
+ * preference order: SPARQL returns solutions in an unspecified order, so keeping
+ * the first row per subject would be store-order-dependent (an entity carrying
+ * both `rdfs:label` and `skos:prefLabel` could index either), not ranked.
+ *
+ * Ties within a predicate (a subject with several `rdfs:label` values — the
+ * multilingual case) are broken deterministically: untagged / `@en` first, then
+ * the lexically smallest value. So `"Button"@en`/`"Bouton"@fr` always resolves
+ * to `Button`, never whichever row the store happened to return first (A7).
+ */
+function preferredBySubject(result, valueVar, preference) {
+    const best = new Map();
+    for (const term of result.termBindings) {
+        if (term.s?.termType !== "NamedNode")
+            continue;
+        const valueTerm = term[valueVar];
+        const predicate = term.p?.value;
+        if (valueTerm === undefined || predicate === undefined)
+            continue;
+        const value = valueTerm.value;
+        if (!value)
+            continue;
+        const rank = preference.indexOf(predicate);
+        if (rank === -1)
+            continue;
+        const language = valueTerm.termType === "Literal" ? (valueTerm.language ?? "") : "";
+        const langRank = rankLanguage(language);
+        const current = best.get(term.s.value);
+        const better = current === undefined ||
+            rank < current.rank ||
+            (rank === current.rank &&
+                (langRank < current.langRank ||
+                    (langRank === current.langRank && value < current.value)));
+        if (better)
+            best.set(term.s.value, { value, rank, langRank });
+    }
+    const picked = new Map();
+    for (const [subject, entry] of best)
+        picked.set(subject, entry.value);
+    return picked;
+}
+/**
+ * Collect ALL matching object values per subject (deduped, insertion order) —
+ * for multi-valued enrichment like `altNames`, where every value is a
+ * completable token rather than a single preferred one.
+ */
+function allBySubject(result, valueVar) {
+    const collected = new Map();
+    for (const term of result.termBindings) {
+        if (term.s?.termType !== "NamedNode")
+            continue;
+        const value = term[valueVar]?.value;
+        if (!value)
+            continue;
+        const bucket = collected.get(term.s.value) ?? [];
+        if (!bucket.includes(value))
+            bucket.push(value);
+        collected.set(term.s.value, bucket);
+    }
+    return collected;
+}
+/**
+ * Build the entity index from a populated store.
+ *
+ * @param store - The ke store to query.
+ * @param prefixes - The store's merged prefix map.
+ * @param contentHash - The pack's content hash, embedded in the index.
+ * @returns The storeless {@link PackIndex}.
+ * @note Impure — runs bulk SPARQL queries against the store.
+ */
+export async function buildIndex(store, prefixes, contentHash) {
+    const altNameUri = resolveAltNamePredicate(prefixes);
+    const [typesResult, labelResult, descResult, altNameResult] = await Promise.all([
+        store.query(`SELECT ?s ?type WHERE { ?s <${RDF_TYPE}> ?type }`),
+        store.query(`SELECT ?s ?p ?label WHERE { ?s ?p ?label . VALUES ?p { ${valuesList(LABEL_PREDICATES)} } }`),
+        store.query(`SELECT ?s ?p ?desc WHERE { ?s ?p ?desc . VALUES ?p { ${valuesList(DESCRIPTION_PREDICATES)} } }`),
+        altNameUri === undefined
+            ? undefined
+            : store.query(`SELECT ?s ?p ?alt WHERE { ?s ?p ?alt . VALUES ?p { <${altNameUri}> } }`),
+    ]);
+    const typesBySubject = new Map();
+    const instanceCountByType = {};
+    for (const term of typesResult.termBindings) {
+        if (term.s?.termType !== "NamedNode")
+            continue;
+        const typeTerm = term.type;
+        // Ignore blank-node `rdf:type` objects (SHACL shapes / RDF lists): a `_:b0`
+        // is never a domain class, so it must not become a garbage primary type or
+        // inflate the per-type instance counts (A6).
+        if (typeTerm === undefined || typeTerm.termType === "BlankNode")
+            continue;
+        const subject = term.s.value;
+        const typeUri = typeTerm.value;
+        const bucket = typesBySubject.get(subject) ?? [];
+        bucket.push(typeUri);
+        typesBySubject.set(subject, bucket);
+        instanceCountByType[typeUri] = (instanceCountByType[typeUri] ?? 0) + 1;
+    }
+    const labelBySubject = preferredBySubject(labelResult, "label", LABEL_PREDICATES);
+    const descBySubject = preferredBySubject(descResult, "desc", DESCRIPTION_PREDICATES);
+    const altNamesBySubject = altNameResult === undefined
+        ? new Map()
+        : allBySubject(altNameResult, "alt");
+    const entities = [];
+    for (const [subject, fullTypes] of typesBySubject) {
+        const facets = classify(fullTypes);
+        if (facets.length === 0)
+            continue;
+        const prefixed = compactUri(subject, prefixes);
+        const altNames = altNamesBySubject.get(subject);
+        const typesCompact = fullTypes.map((t) => compactUri(t, prefixes));
+        const label = labelBySubject.get(subject) ?? null;
+        const description = descBySubject.get(subject) ?? null;
+        // A punned subject emits one entity per facet (tbox class + abox individual).
+        for (const facet of facets) {
+            entities.push({
+                name: prefixed,
+                type: compactUri(facet.primary, prefixes),
+                uri: subject,
+                prefixed,
+                types: typesCompact,
+                label,
+                ...(altNames && altNames.length > 0 ? { altNames } : {}),
+                box: facet.box,
+                description,
+            });
+        }
+    }
+    // Sort by name, then a deterministic tie-break (box, then type) so a punned
+    // subject's two facets keep a stable order across builds.
+    entities.sort((a, b) => {
+        if (a.name !== b.name)
+            return a.name < b.name ? -1 : 1;
+        if (a.box !== b.box)
+            return a.box === "tbox" ? -1 : 1;
+        return a.type < b.type ? -1 : a.type > b.type ? 1 : 0;
+    });
+    return {
+        version: 2,
+        contentHash,
+        prefixes,
+        entities,
+        instanceCountByType,
+    };
+}
+//# sourceMappingURL=buildIndex.js.map

package/dist/src/kernel/runtime/graphpack/buildIndex.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"buildIndex.js","sourceRoot":"","sources":["../../../../../src/kernel/runtime/graphpack/buildIndex.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AAEH,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAC;AACnD,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAKjD,kBAAkB;AAClB,MAAM,QAAQ,GAAG,iDAAiD,CAAC;AAEnE,+DAA+D;AAC/D,MAAM,eAAe,GAAG,IAAI,GAAG,CAAC;IAC9B,qCAAqC;IACrC,4CAA4C;CAC7C,CAAC,CAAC;AAEH,kEAAkE;AAClE,MAAM,kBAAkB,GAAG,IAAI,GAAG,CAAC;IACjC,8CAA8C;IAC9C,gDAAgD;IAChD,kDAAkD;IAClD,qDAAqD;CACtD,CAAC,CAAC;AAEH,iFAAiF;AACjF,MAAM,gBAAgB,GAAG,+CAA+C,CAAC;AAEzE,2EAA2E;AAC3E,MAAM,kBAAkB,GAAG;IACzB,6CAA6C;IAC7C,uCAAuC;IACvC,gCAAgC;IAChC,mCAAmC;IACnC,sCAAsC;IACtC,6BAA6B;CAC9B,CAAC;AAEF,6CAA6C;AAC7C,MAAM,gBAAgB,GAAG;IACvB,4CAA4C;IAC5C,+CAA+C;IAC/C,gCAAgC;IAChC,wBAAwB;CACzB,CAAC;AAEF,mDAAmD;AACnD,MAAM,sBAAsB,GAAG;IAC7B,8CAA8C;IAC9C,sCAAsC;IACtC,gDAAgD;IAChD,+BAA+B;CAChC,CAAC;AAEF,MAAM,UAAU,GAAG,CAAC,GAAW,EAAW,EAAE,CAC1C,kBAAkB,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,GAAG,CAAC,UAAU,CAAC,EAAE,CAAC,CAAC,CAAC;AAQtD;;;;;;;;;;;;GAYG;AACH,SAAS,QAAQ,CAAC,SAA4B;IAC5C,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IACtC,MAAM,MAAM,GAAY,EAAE,CAAC;IAC3B,MAAM,SAAS,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAChE,MAAM,QAAQ,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,kBAAkB,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAClE,IAAI,SAAS,KAAK,SAAS;QAAE,MAAM,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,OAAO,EAAE,SAAS,EAAE,CAAC,CAAC;SACzE,IAAI,QAAQ,KAAK,SAAS;QAC7B,MAAM,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,CAAC,CAAC;IAElD,2EAA2E;IAC3E,6EAA6E;IAC7E,4EAA4E;IAC5E,MAAM,aAAa,GAAG,SAAS;SAC5B,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,KAAK,gBAAgB,IAAI,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;SACvD,IAAI,EAAE,CAAC;IACV,MAAM,aAAa,GAAG,aAAa,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IAC1C,IAAI,aAAa,KAAK,SAAS,EAAE,CAAC;QAChC,MAAM,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,OAAO,EAAE,aAAa,EAAE,CAAC,CAAC;IACvD,CAAC;SAAM,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC/B,gEAAgE;QAChE,2EAA2E;QAC3E,0DAA0D;QAC1D,MAAM,QAAQ,GACZ,SAAS;aACN,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,KAAK,gBAAgB,CAAC;aACrC,IAAI,EAAE;aACN,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,SAAS,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;QAC1C,IAAI,QAAQ,KAAK,SAAS;YAAE,MAAM,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,CAAC,CAAC;IAC9E,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,sDAAsD;AACtD,MAAM,UAAU,GAAG,CAAC,IAAuB,EAAU,EAAE,CACrD,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,IAAI,GAAG,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AAE1C;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,SAAS,uBAAuB,CAC9B,QAA0C;IAE1C,MAAM,KAAK,GAAG,UAAU,CAAC,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IAC9C,MAAM,SAAS,GAAG,QAAQ,CAAC,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC;IAC/D,OAAO,SAAS,KAAK,SAAS;QAC5B,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,GAAG,SAAS,GAAG,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,CAAC;AAC3D,CAAC;AAED,mFAAmF;AACnF,SAAS,YAAY,CAAC,QAAgB;IACpC,IAAI,QAAQ,KAAK,EAAE;QAAE,OAAO,CAAC,CAAC;IAC9B,MAAM,KAAK,GAAG,QAAQ,CAAC,WAAW,EAAE,CAAC;IACrC,OAAO,KAAK,KAAK,IAAI,IAAI,KAAK,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;AAC3D,CAAC;AAED;;;;;;;;;;;GAWG;AACH,SAAS,kBAAkB,CACzB,MAA4C,EAC5C,QAAgB,EAChB,UAA6B;IAE7B,MAAM,IAAI,GAAG,IAAI,GAAG,EAGjB,CAAC;IACJ,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,YAAY,EAAE,CAAC;QACvC,IAAI,IAAI,CAAC,CAAC,EAAE,QAAQ,KAAK,WAAW;YAAE,SAAS;QAC/C,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAC;QACjC,MAAM,SAAS,GAAG,IAAI,CAAC,CAAC,EAAE,KAAK,CAAC;QAChC,IAAI,SAAS,KAAK,SAAS,IAAI,SAAS,KAAK,SAAS;YAAE,SAAS;QACjE,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC;QAC9B,IAAI,CAAC,KAAK;YAAE,SAAS;QACrB,MAAM,IAAI,GAAG,UAAU,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC;QAC3C,IAAI,IAAI,KAAK,CAAC,CAAC;YAAE,SAAS;QAC1B,MAAM,QAAQ,GACZ,SAAS,CAAC,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QACrE,MAAM,QAAQ,GAAG,YAAY,CAAC,QAAQ,CAAC,CAAC;QACxC,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC;QACvC,MAAM,MAAM,GACV,OAAO,KAAK,SAAS;YACrB,IAAI,GAAG,OAAO,CAAC,IAAI;YACnB,CAAC,IAAI,KAAK,OAAO,CAAC,IAAI;gBACpB,CAAC,QAAQ,GAAG,OAAO,CAAC,QAAQ;oBAC1B,CAAC,QAAQ,KAAK,OAAO,CAAC,QAAQ,IAAI,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QACjE,IAAI,MAAM;YAAE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC;IAChE,CAAC;IACD,MAAM,MAAM,GAAG,IAAI,GAAG,EAAkB,CAAC;IACzC,KAAK,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,IAAI,IAAI;QAAE,MAAM,CAAC,GAAG,CAAC,OAAO,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC;IACtE,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;;;GAIG;AACH,SAAS,YAAY,CACnB,MAA4C,EAC5C,QAAgB;IAEhB,MAAM,SAAS,GAAG,IAAI,GAAG,EAAoB,CAAC;IAC9C,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,YAAY,EAAE,CAAC;QACvC,IAAI,IAAI,CAAC,CAAC,EAAE,QAAQ,KAAK,WAAW;YAAE,SAAS;QAC/C,MAAM,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,EAAE,KAAK,CAAC;QACpC,IAAI,CAAC,KAAK;YAAE,SAAS;QACrB,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;QACjD,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC;YAAE,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QAChD,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;IACtC,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,CAAC,KAAK,UAAU,UAAU,CAC9B,KAAY,EACZ,QAA0C,EAC1C,WAAmB;IAEnB,MAAM,UAAU,GAAG,uBAAuB,CAAC,QAAQ,CAAC,CAAC;IACrD,MAAM,CAAC,WAAW,EAAE,WAAW,EAAE,UAAU,EAAE,aAAa,CAAC,GACzD,MAAM,OAAO,CAAC,GAAG,CAAC;QAChB,KAAK,CAAC,KAAK,CACT,+BAA+B,QAAQ,WAAoB,CACX;QAClD,KAAK,CAAC,KAAK,CACT,0DAA0D,UAAU,CAAC,gBAAgB,CAAC,MAAe,CACrD;QAClD,KAAK,CAAC,KAAK,CACT,wDAAwD,UAAU,CAAC,sBAAsB,CAAC,MAAe,CACzD;QAClD,UAAU,KAAK,SAAS;YACtB,CAAC,CAAC,SAAS;YACX,CAAC,CAAE,KAAK,CAAC,KAAK,CACV,uDAAuD,UAAU,OAAgB,CAChC;KACxD,CAAC,CAAC;IAEL,MAAM,cAAc,GAAG,IAAI,GAAG,EAAoB,CAAC;IACnD,MAAM,mBAAmB,GAA2B,EAAE,CAAC;IACvD,KAAK,MAAM,IAAI,IAAI,WAAW,CAAC,YAAY,EAAE,CAAC;QAC5C,IAAI,IAAI,CAAC,CAAC,EAAE,QAAQ,KAAK,WAAW;YAAE,SAAS;QAC/C,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC;QAC3B,4EAA4E;QAC5E,2EAA2E;QAC3E,6CAA6C;QAC7C,IAAI,QAAQ,KAAK,SAAS,IAAI,QAAQ,CAAC,QAAQ,KAAK,WAAW;YAAE,SAAS;QAC1E,MAAM,OAAO,GAAG,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC;QAC7B,MAAM,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC;QAC/B,MAAM,MAAM,GAAG,cAAc,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;QACjD,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACrB,cAAc,CAAC,GAAG,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;QACpC,mBAAmB,CAAC,OAAO,CAAC,GAAG,CAAC,mBAAmB,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC;IACzE,CAAC;IAED,MAAM,cAAc,GAAG,kBAAkB,CACvC,WAAW,EACX,OAAO,EACP,gBAAgB,CACjB,CAAC;IACF,MAAM,aAAa,GAAG,kBAAkB,CACtC,UAAU,EACV,MAAM,EACN,sBAAsB,CACvB,CAAC;IACF,MAAM,iBAAiB,GACrB,aAAa,KAAK,SAAS;QACzB,CAAC,CAAC,IAAI,GAAG,EAAoB;QAC7B,CAAC,CAAC,YAAY,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;IAEzC,MAAM,QAAQ,GAAsB,EAAE,CAAC;IACvC,KAAK,MAAM,CAAC,OAAO,EAAE,SAAS,CAAC,IAAI,cAAc,EAAE,CAAC;QAClD,MAAM,MAAM,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAC;QACnC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;YAAE,SAAS;QAClC,MAAM,QAAQ,GAAG,UAAU,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;QAC/C,MAAM,QAAQ,GAAG,iBAAiB,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QAChD,MAAM,YAAY,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,UAAU,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC,CAAC;QACnE,MAAM,KAAK,GAAG,cAAc,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC;QAClD,MAAM,WAAW,GAAG,aAAa,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC;QACvD,8EAA8E;QAC9E,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,QAAQ,CAAC,IAAI,CAAC;gBACZ,IAAI,EAAE,QAAQ;gBACd,IAAI,EAAE,UAAU,CAAC,KAAK,CAAC,OAAO,EAAE,QAAQ,CAAC;gBACzC,GAAG,EAAE,OAAO;gBACZ,QAAQ;gBACR,KAAK,EAAE,YAAY;gBACnB,KAAK;gBACL,GAAG,CAAC,QAAQ,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACxD,GAAG,EAAE,KAAK,CAAC,GAAG;gBACd,WAAW;aACZ,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IACD,4EAA4E;IAC5E,0DAA0D;IAC1D,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;QACrB,IAAI,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,IAAI;YAAE,OAAO,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACvD,IAAI,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,GAAG;YAAE,OAAO,CAAC,CAAC,GAAG,KAAK,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACtD,OAAO,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACxD,CAAC,CAAC,CAAC;IAEH,OAAO;QACL,OAAO,EAAE,CAAC;QACV,WAAW;QACX,QAAQ;QACR,QAAQ;QACR,mBAAmB;KACpB,CAAC;AACJ,CAAC"}