wiki-graph-core 0.3.1 → 0.4.1

package/dist/retrieval/query/archive-view/search/buckets.js

@@ -0,0 +1,302 @@
+import { listChapters } from "../../../../document/chapter/index.js";
+import {
+  createLexicalQuery,
+  listLexicalQueryCandidateTerms
+} from "../../lexical-search.js";
+import {
+  decodeBucketSearchSessionCursor,
+  encodeBucketSearchSessionCursor,
+  populateSearchSessionObjectCaches,
+  readSearchSessionChunkBucketPage,
+  readSearchSessionMetadataForCursor,
+  readSearchSessionObjectBucketPage
+} from "../../search-cache/index.js";
+import {
+  querySearchIndex,
+  SEARCH_INDEX_FTS_HIT_LIMIT,
+  SEARCH_OBJECT_PROPERTY_OWNER_KIND
+} from "../../../search-index/search/index.js";
+import {
+  BROAD_FIND_LENS_HINT,
+  isDefined,
+  parseFindLens,
+  parseFindMatch,
+  parseFindTypes
+} from "../helpers.js";
+import {
+  assertSearchCursorTypesMatch,
+  createEntitySearchCacheInput
+} from "./cache-input.js";
+import { findEntities, findTriples } from "../find.js";
+import {
+  createFindEvidenceHydrationOptions,
+  hydrateFindHitEvidence
+} from "../evidence.js";
+import { hydrateFindResultBacklinks } from "../backlinks.js";
+import {
+  hydrateSearchObjectHit,
+  hydrateSearchTextHit,
+  parseSearchPropertyIntegerOwnerId,
+  withSearchTerms
+} from "./hydration.js";
+import {
+  hydrateCachedChunkBucketHit,
+  hydrateCachedObjectBucketHit
+} from "./bucket-hydration.js";
+import {
+  compareChapterTitleIndexHits,
+  compareTextIndexHits,
+  getObjectBucketCursorId,
+  isAfterChapterTitleKey,
+  isAfterTextKey
+} from "./bucket-order.js";
+async function readBucketedSearchResultPage(document, cursor, options) {
+  const session = await readSearchSessionMetadataForCursor(
+    cursor.sessionId,
+    options.archiveKey ?? "archive",
+    cursor.createdAt
+  );
+  assertSearchCursorTypesMatch(options.types, session.types);
+  const items = [];
+  let bucketCursor = cursor.cursor;
+  while (bucketCursor !== void 0 && items.length < options.limit) {
+    const remaining = options.limit - items.length;
+    const page = await readBucketPage(
+      document,
+      session,
+      bucketCursor,
+      remaining
+    );
+    items.push(...page.items);
+    bucketCursor = page.nextCursor;
+  }
+  return await hydrateFindResultBacklinks(
+    document,
+    {
+      chapters: session.chapters,
+      items: await hydrateFindHitEvidence(
+        document,
+        items,
+        createFindEvidenceHydrationOptions(options, cursor.sessionId)
+      ),
+      lens: parseFindLens(session.lens),
+      lensHint: session.lens === "broad" ? BROAD_FIND_LENS_HINT : null,
+      limit: options.limit,
+      match: parseFindMatch(session.match),
+      nextCursor: bucketCursor === void 0 ? null : encodeBucketSearchSessionCursor(
+        cursor.sessionId,
+        bucketCursor,
+        session.createdAt
+      ),
+      order: options.order ?? "doc-asc",
+      query: session.query,
+      terms: session.terms,
+      types: parseFindTypes(session.types)
+    },
+    options
+  );
+}
+async function readBucketPage(document, session, cursor, limit) {
+  switch (cursor.bucket) {
+    case 0:
+      return await readChapterTitleBucketPage(
+        document,
+        session,
+        cursor.key,
+        limit
+      );
+    case 1:
+      return await readObjectBucketPage(document, session, cursor.key, limit);
+    case 2:
+      return await readChunkBucketPage(document, session, cursor.key, limit);
+    case 3:
+      return await readTextBucketPage(document, session, cursor.key, limit);
+  }
+}
+async function readChapterTitleBucketPage(document, session, after, limit) {
+  const result = await querySearchIndex(document, session.query, {
+    ...session.chapters === null ? {} : { chapters: session.chapters },
+    match: parseFindMatch(session.match),
+    objectHitLimit: SEARCH_INDEX_FTS_HIT_LIMIT,
+    textHitLimit: 0,
+    types: ["chapter-title"]
+  });
+  const chapters = new Map(
+    (await listChapters(document)).map((chapter) => [
+      chapter.chapterId,
+      chapter
+    ])
+  );
+  const hits = (result?.objectHits ?? []).filter(
+    (hit) => hit.ownerKind === SEARCH_OBJECT_PROPERTY_OWNER_KIND.chapter
+  ).sort(compareChapterTitleIndexHits).filter((hit) => isAfterChapterTitleKey(hit, after));
+  const page = hits.slice(0, limit + 1);
+  const hydrated = (await Promise.all(
+    page.slice(0, limit).map(async (hit) => {
+      const item = await hydrateSearchObjectHit(document, chapters, hit);
+      return item === void 0 ? void 0 : withSearchTerms(item, session.terms);
+    })
+  )).filter(isDefined);
+  const last = page.at(limit - 1);
+  return {
+    items: hydrated,
+    nextCursor: page.length > limit && last !== void 0 ? {
+      bucket: 0,
+      key: {
+        archiveId: last.archiveId,
+        chapterId: parseSearchPropertyIntegerOwnerId(last.ownerId),
+        score: last.score
+      }
+    } : { bucket: 1 }
+  };
+}
+async function readObjectBucketPage(document, session, after, limit) {
+  if (!session.objectCachesPopulated) {
+    await populateObjectBucketCaches(document, session);
+  }
+  const page = await readSearchSessionObjectBucketPage(
+    session.sessionId,
+    1,
+    after,
+    limit
+  );
+  const items = page.slice(0, limit);
+  const hydrated = await Promise.all(
+    items.map(async (hit) => await hydrateCachedObjectBucketHit(document, hit))
+  );
+  const last = items.at(-1);
+  return {
+    items: hydrated,
+    nextCursor: page.length > limit && last !== void 0 ? {
+      bucket: 1,
+      key: {
+        archiveId: getHitArchiveId(last),
+        id: getObjectBucketCursorId(last),
+        kind: last.type === "triple" ? "triple" : "entity",
+        score: last.score ?? 0
+      }
+    } : { bucket: 2 }
+  };
+}
+async function populateObjectBucketCaches(document, session) {
+  const search = createLexicalQuery(session.query);
+  if (search === void 0) {
+    return;
+  }
+  const [allMentions, indexed] = await Promise.all([
+    document.mentions.listBySurfaceTerms(
+      listLexicalQueryCandidateTerms(session.query)
+    ),
+    querySearchIndex(document, session.query, {
+      ...session.chapters === null ? {} : { chapters: session.chapters },
+      match: parseFindMatch(session.match),
+      objectHitLimit: SEARCH_INDEX_FTS_HIT_LIMIT,
+      textHitLimit: 0,
+      types: null
+    })
+  ]);
+  const structuredHits = [
+    ...findEntities(search, { mentions: allMentions }),
+    ...await findTriples(document, search, { mentions: allMentions })
+  ];
+  const entityCacheInput = createEntitySearchCacheInput(
+    structuredHits,
+    indexed
+  );
+  await populateSearchSessionObjectCaches({
+    entityHits: entityCacheInput.entityHits,
+    evidenceEvents: entityCacheInput.evidenceEvents,
+    sessionId: session.sessionId
+  });
+}
+async function readChunkBucketPage(document, session, after, limit) {
+  const page = await readSearchSessionChunkBucketPage(
+    session.sessionId,
+    after,
+    limit
+  );
+  const items = page.slice(0, limit);
+  const hydrated = (await Promise.all(
+    items.map(
+      async (hit) => await hydrateCachedChunkBucketHit(document, hit)
+    )
+  )).filter(isDefined);
+  const last = items.at(-1);
+  return {
+    items: hydrated,
+    nextCursor: page.length > limit && last !== void 0 ? {
+      bucket: 2,
+      key: {
+        archiveId: getHitArchiveId(last),
+        chunkId: parseSearchPropertyIntegerOwnerId(
+          last.id.slice("wikg://chunk/".length)
+        ),
+        score: last.score ?? 0
+      }
+    } : { bucket: 3 }
+  };
+}
+async function readTextBucketPage(document, session, after, limit) {
+  const result = await querySearchIndex(document, session.query, {
+    ...session.chapters === null ? {} : { chapters: session.chapters },
+    match: parseFindMatch(session.match),
+    objectHitLimit: 0,
+    ...after === void 0 ? {} : {
+      textAfter: {
+        archiveId: after.archiveId,
+        chapterId: after.chapterId,
+        kind: after.kind,
+        rank: after.rank,
+        sentenceIndex: after.sentenceIndex
+      }
+    },
+    textHitLimit: createBucketQueryWindow(limit),
+    types: ["source", "summary"]
+  });
+  const chapters = new Map(
+    (await listChapters(document)).map((chapter) => [
+      chapter.chapterId,
+      chapter
+    ])
+  );
+  const hits = [...result?.textHits ?? []].sort(compareTextIndexHits).filter((hit) => isAfterTextKey(hit, after));
+  const page = hits.slice(0, limit + 1);
+  const hydrated = (await Promise.all(
+    page.slice(0, limit).map(async (hit) => {
+      const item = await hydrateSearchTextHit(document, chapters, hit);
+      return item === void 0 ? void 0 : withSearchTerms(item, session.terms);
+    })
+  )).filter(isDefined);
+  const last = page.at(limit - 1);
+  return {
+    items: hydrated,
+    nextCursor: page.length > limit && last !== void 0 ? {
+      bucket: 3,
+      key: {
+        archiveId: last.archiveId,
+        chapterId: last.chapterId,
+        kind: last.kind,
+        rank: last.rank,
+        sentenceIndex: last.sentenceIndex
+      }
+    } : void 0
+  };
+}
+function createBucketQueryWindow(limit) {
+  return Math.max(limit + 1, limit * 3 + 1, 100);
+}
+function getHitArchiveId(hit) {
+  return hit.archiveId ?? 0;
+}
+function tryDecodeBucketSearchSessionCursor(cursor) {
+  try {
+    return decodeBucketSearchSessionCursor(cursor);
+  } catch {
+    return void 0;
+  }
+}
+export {
+  readBucketedSearchResultPage,
+  tryDecodeBucketSearchSessionCursor
+};
+//# sourceMappingURL=buckets.js.map

package/dist/retrieval/query/archive-view/search/buckets.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../src/retrieval/query/archive-view/search/buckets.ts"],"sourcesContent":["import type { ReadonlyDocument } from \"../../../../document/index.js\";\nimport { listChapters } from \"../../../../document/chapter/index.js\";\nimport {\n  createLexicalQuery,\n  listLexicalQueryCandidateTerms,\n} from \"../../lexical-search.js\";\nimport {\n  decodeBucketSearchSessionCursor,\n  encodeBucketSearchSessionCursor,\n  populateSearchSessionObjectCaches,\n  readSearchSessionChunkBucketPage,\n  readSearchSessionMetadataForCursor,\n  readSearchSessionObjectBucketPage,\n} from \"../../search-cache/index.js\";\nimport type {\n  BucketSearchCursor,\n  SearchChapterTitleCursorKey,\n  SearchChunkCursorKey,\n  SearchObjectCursorKey,\n  SearchSessionDescriptor,\n  SearchTextCursorKey,\n} from \"../../search-cache/types.js\";\nimport {\n  querySearchIndex,\n  SEARCH_INDEX_FTS_HIT_LIMIT,\n  SEARCH_OBJECT_PROPERTY_OWNER_KIND,\n  type SearchIndexTextHit,\n} from \"../../../search-index/search/index.js\";\n\nimport {\n  BROAD_FIND_LENS_HINT,\n  isDefined,\n  parseFindLens,\n  parseFindMatch,\n  parseFindTypes,\n} from \"../helpers.js\";\nimport {\n  assertSearchCursorTypesMatch,\n  createEntitySearchCacheInput,\n} from \"./cache-input.js\";\nimport { findEntities, findTriples } from \"../find.js\";\nimport {\n  createFindEvidenceHydrationOptions,\n  hydrateFindHitEvidence,\n} from \"../evidence.js\";\nimport { hydrateFindResultBacklinks } from \"../backlinks.js\";\nimport {\n  hydrateSearchObjectHit,\n  hydrateSearchTextHit,\n  parseSearchPropertyIntegerOwnerId,\n  withSearchTerms,\n} from \"./hydration.js\";\nimport {\n  hydrateCachedChunkBucketHit,\n  hydrateCachedObjectBucketHit,\n} from \"./bucket-hydration.js\";\nimport {\n  compareChapterTitleIndexHits,\n  compareTextIndexHits,\n  getObjectBucketCursorId,\n  isAfterChapterTitleKey,\n  isAfterTextKey,\n} from \"./bucket-order.js\";\nimport type {\n  ArchiveFindHit,\n  ArchiveFindOptions,\n  ArchiveFindResult,\n} from \"../types.js\";\n\nexport async function readBucketedSearchResultPage(\n  document: ReadonlyDocument,\n  cursor: {\n    readonly createdAt: number;\n    readonly cursor: BucketSearchCursor;\n    readonly sessionId: string;\n  },\n  options: ArchiveFindOptions & { readonly limit: number },\n): Promise<ArchiveFindResult> {\n  const session = await readSearchSessionMetadataForCursor(\n    cursor.sessionId,\n    options.archiveKey ?? \"archive\",\n    cursor.createdAt,\n  );\n\n  assertSearchCursorTypesMatch(options.types, session.types);\n\n  const items: ArchiveFindHit[] = [];\n  let bucketCursor: BucketSearchCursor | undefined = cursor.cursor;\n\n  while (bucketCursor !== undefined && items.length < options.limit) {\n    const remaining = options.limit - items.length;\n    const page = await readBucketPage(\n      document,\n      session,\n      bucketCursor,\n      remaining,\n    );\n\n    items.push(...page.items);\n    bucketCursor = page.nextCursor;\n  }\n\n  return await hydrateFindResultBacklinks(\n    document,\n    {\n      chapters: session.chapters,\n      items: await hydrateFindHitEvidence(\n        document,\n        items,\n        createFindEvidenceHydrationOptions(options, cursor.sessionId),\n      ),\n      lens: parseFindLens(session.lens),\n      lensHint: session.lens === \"broad\" ? BROAD_FIND_LENS_HINT : null,\n      limit: options.limit,\n      match: parseFindMatch(session.match),\n      nextCursor:\n        bucketCursor === undefined\n          ? null\n          : encodeBucketSearchSessionCursor(\n              cursor.sessionId,\n              bucketCursor,\n              session.createdAt,\n            ),\n      order: options.order ?? \"doc-asc\",\n      query: session.query,\n      terms: session.terms,\n      types: parseFindTypes(session.types),\n    },\n    options,\n  );\n}\n\nasync function readBucketPage(\n  document: ReadonlyDocument,\n  session: SearchSessionDescriptor,\n  cursor: BucketSearchCursor,\n  limit: number,\n): Promise<{\n  readonly items: readonly ArchiveFindHit[];\n  readonly nextCursor: BucketSearchCursor | undefined;\n}> {\n  switch (cursor.bucket) {\n    case 0:\n      return await readChapterTitleBucketPage(\n        document,\n        session,\n        cursor.key,\n        limit,\n      );\n    case 1:\n      return await readObjectBucketPage(document, session, cursor.key, limit);\n    case 2:\n      return await readChunkBucketPage(document, session, cursor.key, limit);\n    case 3:\n      return await readTextBucketPage(document, session, cursor.key, limit);\n  }\n}\n\nasync function readChapterTitleBucketPage(\n  document: ReadonlyDocument,\n  session: SearchSessionDescriptor,\n  after: SearchChapterTitleCursorKey | undefined,\n  limit: number,\n): Promise<{\n  readonly items: readonly ArchiveFindHit[];\n  readonly nextCursor: BucketSearchCursor | undefined;\n}> {\n  const result = await querySearchIndex(document, session.query, {\n    ...(session.chapters === null ? {} : { chapters: session.chapters }),\n    match: parseFindMatch(session.match),\n    objectHitLimit: SEARCH_INDEX_FTS_HIT_LIMIT,\n    textHitLimit: 0,\n    types: [\"chapter-title\"],\n  });\n  const chapters = new Map(\n    (await listChapters(document)).map((chapter) => [\n      chapter.chapterId,\n      chapter,\n    ]),\n  );\n  const hits = (result?.objectHits ?? [])\n    .filter(\n      (hit) => hit.ownerKind === SEARCH_OBJECT_PROPERTY_OWNER_KIND.chapter,\n    )\n    .sort(compareChapterTitleIndexHits)\n    .filter((hit) => isAfterChapterTitleKey(hit, after));\n  const page = hits.slice(0, limit + 1);\n  const hydrated = (\n    await Promise.all(\n      page.slice(0, limit).map(async (hit) => {\n        const item = await hydrateSearchObjectHit(document, chapters, hit);\n\n        return item === undefined\n          ? undefined\n          : withSearchTerms(item, session.terms);\n      }),\n    )\n  ).filter(isDefined);\n  const last = page.at(limit - 1);\n\n  return {\n    items: hydrated,\n    nextCursor:\n      page.length > limit && last !== undefined\n        ? {\n            bucket: 0,\n            key: {\n              archiveId: last.archiveId,\n              chapterId: parseSearchPropertyIntegerOwnerId(last.ownerId),\n              score: last.score,\n            },\n          }\n        : { bucket: 1 },\n  };\n}\n\nasync function readObjectBucketPage(\n  document: ReadonlyDocument,\n  session: SearchSessionDescriptor,\n  after: SearchObjectCursorKey | undefined,\n  limit: number,\n): Promise<{\n  readonly items: readonly ArchiveFindHit[];\n  readonly nextCursor: BucketSearchCursor | undefined;\n}> {\n  if (!session.objectCachesPopulated) {\n    await populateObjectBucketCaches(document, session);\n  }\n  const page = await readSearchSessionObjectBucketPage(\n    session.sessionId,\n    1,\n    after,\n    limit,\n  );\n  const items = page.slice(0, limit);\n  const hydrated = await Promise.all(\n    items.map(async (hit) => await hydrateCachedObjectBucketHit(document, hit)),\n  );\n  const last = items.at(-1);\n\n  return {\n    items: hydrated,\n    nextCursor:\n      page.length > limit && last !== undefined\n        ? {\n            bucket: 1,\n            key: {\n              archiveId: getHitArchiveId(last),\n              id: getObjectBucketCursorId(last),\n              kind: last.type === \"triple\" ? \"triple\" : \"entity\",\n              score: last.score ?? 0,\n            },\n          }\n        : { bucket: 2 },\n  };\n}\n\nasync function populateObjectBucketCaches(\n  document: ReadonlyDocument,\n  session: SearchSessionDescriptor,\n): Promise<void> {\n  const search = createLexicalQuery(session.query);\n\n  if (search === undefined) {\n    return;\n  }\n\n  const [allMentions, indexed] = await Promise.all([\n    document.mentions.listBySurfaceTerms(\n      listLexicalQueryCandidateTerms(session.query),\n    ),\n    querySearchIndex(document, session.query, {\n      ...(session.chapters === null ? {} : { chapters: session.chapters }),\n      match: parseFindMatch(session.match),\n      objectHitLimit: SEARCH_INDEX_FTS_HIT_LIMIT,\n      textHitLimit: 0,\n      types: null,\n    }),\n  ]);\n  const structuredHits = [\n    ...findEntities(search, { mentions: allMentions }),\n    ...(await findTriples(document, search, { mentions: allMentions })),\n  ];\n  const entityCacheInput = createEntitySearchCacheInput(\n    structuredHits,\n    indexed,\n  );\n\n  await populateSearchSessionObjectCaches({\n    entityHits: entityCacheInput.entityHits,\n    evidenceEvents: entityCacheInput.evidenceEvents,\n    sessionId: session.sessionId,\n  });\n}\n\nasync function readChunkBucketPage(\n  document: ReadonlyDocument,\n  session: SearchSessionDescriptor,\n  after: SearchChunkCursorKey | undefined,\n  limit: number,\n): Promise<{\n  readonly items: readonly ArchiveFindHit[];\n  readonly nextCursor: BucketSearchCursor | undefined;\n}> {\n  const page = await readSearchSessionChunkBucketPage(\n    session.sessionId,\n    after,\n    limit,\n  );\n  const items = page.slice(0, limit);\n  const hydrated = (\n    await Promise.all(\n      items.map(\n        async (hit) => await hydrateCachedChunkBucketHit(document, hit),\n      ),\n    )\n  ).filter(isDefined);\n  const last = items.at(-1);\n\n  return {\n    items: hydrated,\n    nextCursor:\n      page.length > limit && last !== undefined\n        ? {\n            bucket: 2,\n            key: {\n              archiveId: getHitArchiveId(last),\n              chunkId: parseSearchPropertyIntegerOwnerId(\n                last.id.slice(\"wikg://chunk/\".length),\n              ),\n              score: last.score ?? 0,\n            },\n          }\n        : { bucket: 3 },\n  };\n}\n\nasync function readTextBucketPage(\n  document: ReadonlyDocument,\n  session: SearchSessionDescriptor,\n  after: SearchTextCursorKey | undefined,\n  limit: number,\n): Promise<{\n  readonly items: readonly ArchiveFindHit[];\n  readonly nextCursor: BucketSearchCursor | undefined;\n}> {\n  const result = await querySearchIndex(document, session.query, {\n    ...(session.chapters === null ? {} : { chapters: session.chapters }),\n    match: parseFindMatch(session.match),\n    objectHitLimit: 0,\n    ...(after === undefined\n      ? {}\n      : {\n          textAfter: {\n            archiveId: after.archiveId,\n            chapterId: after.chapterId,\n            kind: after.kind as SearchIndexTextHit[\"kind\"],\n            rank: after.rank,\n            sentenceIndex: after.sentenceIndex,\n          },\n        }),\n    textHitLimit: createBucketQueryWindow(limit),\n    types: [\"source\", \"summary\"],\n  });\n  const chapters = new Map(\n    (await listChapters(document)).map((chapter) => [\n      chapter.chapterId,\n      chapter,\n    ]),\n  );\n  const hits = [...(result?.textHits ?? [])]\n    .sort(compareTextIndexHits)\n    .filter((hit) => isAfterTextKey(hit, after));\n  const page = hits.slice(0, limit + 1);\n  const hydrated = (\n    await Promise.all(\n      page.slice(0, limit).map(async (hit) => {\n        const item = await hydrateSearchTextHit(document, chapters, hit);\n\n        return item === undefined\n          ? undefined\n          : withSearchTerms(item, session.terms);\n      }),\n    )\n  ).filter(isDefined);\n  const last = page.at(limit - 1);\n\n  return {\n    items: hydrated,\n    nextCursor:\n      page.length > limit && last !== undefined\n        ? {\n            bucket: 3,\n            key: {\n              archiveId: last.archiveId,\n              chapterId: last.chapterId,\n              kind: last.kind,\n              rank: last.rank,\n              sentenceIndex: last.sentenceIndex,\n            },\n          }\n        : undefined,\n  };\n}\n\nfunction createBucketQueryWindow(limit: number): number {\n  return Math.max(limit + 1, limit * 3 + 1, 100);\n}\n\nfunction getHitArchiveId(hit: ArchiveFindHit): number {\n  return hit.archiveId ?? 0;\n}\n\nexport function tryDecodeBucketSearchSessionCursor(cursor: string):\n  | {\n      readonly createdAt: number;\n      readonly cursor: BucketSearchCursor;\n      readonly sessionId: string;\n    }\n  | undefined {\n  try {\n    return decodeBucketSearchSessionCursor(cursor);\n  } catch {\n    return undefined;\n  }\n}\n"],"mappings":"AACA,SAAS,oBAAoB;AAC7B;AAAA,EACE;AAAA,EACA;AAAA,OACK;AACP;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OACK;AASP;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,OAEK;AAEP;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OACK;AACP;AAAA,EACE;AAAA,EACA;AAAA,OACK;AACP,SAAS,cAAc,mBAAmB;AAC1C;AAAA,EACE;AAAA,EACA;AAAA,OACK;AACP,SAAS,kCAAkC;AAC3C;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OACK;AACP;AAAA,EACE;AAAA,EACA;AAAA,OACK;AACP;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OACK;AAOP,eAAsB,6BACpB,UACA,QAKA,SAC4B;AAC5B,QAAM,UAAU,MAAM;AAAA,IACpB,OAAO;AAAA,IACP,QAAQ,cAAc;AAAA,IACtB,OAAO;AAAA,EACT;AAEA,+BAA6B,QAAQ,OAAO,QAAQ,KAAK;AAEzD,QAAM,QAA0B,CAAC;AACjC,MAAI,eAA+C,OAAO;AAE1D,SAAO,iBAAiB,UAAa,MAAM,SAAS,QAAQ,OAAO;AACjE,UAAM,YAAY,QAAQ,QAAQ,MAAM;AACxC,UAAM,OAAO,MAAM;AAAA,MACjB;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAEA,UAAM,KAAK,GAAG,KAAK,KAAK;AACxB,mBAAe,KAAK;AAAA,EACtB;AAEA,SAAO,MAAM;AAAA,IACX;AAAA,IACA;AAAA,MACE,UAAU,QAAQ;AAAA,MAClB,OAAO,MAAM;AAAA,QACX;AAAA,QACA;AAAA,QACA,mCAAmC,SAAS,OAAO,SAAS;AAAA,MAC9D;AAAA,MACA,MAAM,cAAc,QAAQ,IAAI;AAAA,MAChC,UAAU,QAAQ,SAAS,UAAU,uBAAuB;AAAA,MAC5D,OAAO,QAAQ;AAAA,MACf,OAAO,eAAe,QAAQ,KAAK;AAAA,MACnC,YACE,iBAAiB,SACb,OACA;AAAA,QACE,OAAO;AAAA,QACP;AAAA,QACA,QAAQ;AAAA,MACV;AAAA,MACN,OAAO,QAAQ,SAAS;AAAA,MACxB,OAAO,QAAQ;AAAA,MACf,OAAO,QAAQ;AAAA,MACf,OAAO,eAAe,QAAQ,KAAK;AAAA,IACrC;AAAA,IACA;AAAA,EACF;AACF;AAEA,eAAe,eACb,UACA,SACA,QACA,OAIC;AACD,UAAQ,OAAO,QAAQ;AAAA,IACrB,KAAK;AACH,aAAO,MAAM;AAAA,QACX;AAAA,QACA;AAAA,QACA,OAAO;AAAA,QACP;AAAA,MACF;AAAA,IACF,KAAK;AACH,aAAO,MAAM,qBAAqB,UAAU,SAAS,OAAO,KAAK,KAAK;AAAA,IACxE,KAAK;AACH,aAAO,MAAM,oBAAoB,UAAU,SAAS,OAAO,KAAK,KAAK;AAAA,IACvE,KAAK;AACH,aAAO,MAAM,mBAAmB,UAAU,SAAS,OAAO,KAAK,KAAK;AAAA,EACxE;AACF;AAEA,eAAe,2BACb,UACA,SACA,OACA,OAIC;AACD,QAAM,SAAS,MAAM,iBAAiB,UAAU,QAAQ,OAAO;AAAA,IAC7D,GAAI,QAAQ,aAAa,OAAO,CAAC,IAAI,EAAE,UAAU,QAAQ,SAAS;AAAA,IAClE,OAAO,eAAe,QAAQ,KAAK;AAAA,IACnC,gBAAgB;AAAA,IAChB,cAAc;AAAA,IACd,OAAO,CAAC,eAAe;AAAA,EACzB,CAAC;AACD,QAAM,WAAW,IAAI;AAAA,KAClB,MAAM,aAAa,QAAQ,GAAG,IAAI,CAAC,YAAY;AAAA,MAC9C,QAAQ;AAAA,MACR;AAAA,IACF,CAAC;AAAA,EACH;AACA,QAAM,QAAQ,QAAQ,cAAc,CAAC,GAClC;AAAA,IACC,CAAC,QAAQ,IAAI,cAAc,kCAAkC;AAAA,EAC/D,EACC,KAAK,4BAA4B,EACjC,OAAO,CAAC,QAAQ,uBAAuB,KAAK,KAAK,CAAC;AACrD,QAAM,OAAO,KAAK,MAAM,GAAG,QAAQ,CAAC;AACpC,QAAM,YACJ,MAAM,QAAQ;AAAA,IACZ,KAAK,MAAM,GAAG,KAAK,EAAE,IAAI,OAAO,QAAQ;AACtC,YAAM,OAAO,MAAM,uBAAuB,UAAU,UAAU,GAAG;AAEjE,aAAO,SAAS,SACZ,SACA,gBAAgB,MAAM,QAAQ,KAAK;AAAA,IACzC,CAAC;AAAA,EACH,GACA,OAAO,SAAS;AAClB,QAAM,OAAO,KAAK,GAAG,QAAQ,CAAC;AAE9B,SAAO;AAAA,IACL,OAAO;AAAA,IACP,YACE,KAAK,SAAS,SAAS,SAAS,SAC5B;AAAA,MACE,QAAQ;AAAA,MACR,KAAK;AAAA,QACH,WAAW,KAAK;AAAA,QAChB,WAAW,kCAAkC,KAAK,OAAO;AAAA,QACzD,OAAO,KAAK;AAAA,MACd;AAAA,IACF,IACA,EAAE,QAAQ,EAAE;AAAA,EACpB;AACF;AAEA,eAAe,qBACb,UACA,SACA,OACA,OAIC;AACD,MAAI,CAAC,QAAQ,uBAAuB;AAClC,UAAM,2BAA2B,UAAU,OAAO;AAAA,EACpD;AACA,QAAM,OAAO,MAAM;AAAA,IACjB,QAAQ;AAAA,IACR;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACA,QAAM,QAAQ,KAAK,MAAM,GAAG,KAAK;AACjC,QAAM,WAAW,MAAM,QAAQ;AAAA,IAC7B,MAAM,IAAI,OAAO,QAAQ,MAAM,6BAA6B,UAAU,GAAG,CAAC;AAAA,EAC5E;AACA,QAAM,OAAO,MAAM,GAAG,EAAE;AAExB,SAAO;AAAA,IACL,OAAO;AAAA,IACP,YACE,KAAK,SAAS,SAAS,SAAS,SAC5B;AAAA,MACE,QAAQ;AAAA,MACR,KAAK;AAAA,QACH,WAAW,gBAAgB,IAAI;AAAA,QAC/B,IAAI,wBAAwB,IAAI;AAAA,QAChC,MAAM,KAAK,SAAS,WAAW,WAAW;AAAA,QAC1C,OAAO,KAAK,SAAS;AAAA,MACvB;AAAA,IACF,IACA,EAAE,QAAQ,EAAE;AAAA,EACpB;AACF;AAEA,eAAe,2BACb,UACA,SACe;AACf,QAAM,SAAS,mBAAmB,QAAQ,KAAK;AAE/C,MAAI,WAAW,QAAW;AACxB;AAAA,EACF;AAEA,QAAM,CAAC,aAAa,OAAO,IAAI,MAAM,QAAQ,IAAI;AAAA,IAC/C,SAAS,SAAS;AAAA,MAChB,+BAA+B,QAAQ,KAAK;AAAA,IAC9C;AAAA,IACA,iBAAiB,UAAU,QAAQ,OAAO;AAAA,MACxC,GAAI,QAAQ,aAAa,OAAO,CAAC,IAAI,EAAE,UAAU,QAAQ,SAAS;AAAA,MAClE,OAAO,eAAe,QAAQ,KAAK;AAAA,MACnC,gBAAgB;AAAA,MAChB,cAAc;AAAA,MACd,OAAO;AAAA,IACT,CAAC;AAAA,EACH,CAAC;AACD,QAAM,iBAAiB;AAAA,IACrB,GAAG,aAAa,QAAQ,EAAE,UAAU,YAAY,CAAC;AAAA,IACjD,GAAI,MAAM,YAAY,UAAU,QAAQ,EAAE,UAAU,YAAY,CAAC;AAAA,EACnE;AACA,QAAM,mBAAmB;AAAA,IACvB;AAAA,IACA;AAAA,EACF;AAEA,QAAM,kCAAkC;AAAA,IACtC,YAAY,iBAAiB;AAAA,IAC7B,gBAAgB,iBAAiB;AAAA,IACjC,WAAW,QAAQ;AAAA,EACrB,CAAC;AACH;AAEA,eAAe,oBACb,UACA,SACA,OACA,OAIC;AACD,QAAM,OAAO,MAAM;AAAA,IACjB,QAAQ;AAAA,IACR;AAAA,IACA;AAAA,EACF;AACA,QAAM,QAAQ,KAAK,MAAM,GAAG,KAAK;AACjC,QAAM,YACJ,MAAM,QAAQ;AAAA,IACZ,MAAM;AAAA,MACJ,OAAO,QAAQ,MAAM,4BAA4B,UAAU,GAAG;AAAA,IAChE;AAAA,EACF,GACA,OAAO,SAAS;AAClB,QAAM,OAAO,MAAM,GAAG,EAAE;AAExB,SAAO;AAAA,IACL,OAAO;AAAA,IACP,YACE,KAAK,SAAS,SAAS,SAAS,SAC5B;AAAA,MACE,QAAQ;AAAA,MACR,KAAK;AAAA,QACH,WAAW,gBAAgB,IAAI;AAAA,QAC/B,SAAS;AAAA,UACP,KAAK,GAAG,MAAM,gBAAgB,MAAM;AAAA,QACtC;AAAA,QACA,OAAO,KAAK,SAAS;AAAA,MACvB;AAAA,IACF,IACA,EAAE,QAAQ,EAAE;AAAA,EACpB;AACF;AAEA,eAAe,mBACb,UACA,SACA,OACA,OAIC;AACD,QAAM,SAAS,MAAM,iBAAiB,UAAU,QAAQ,OAAO;AAAA,IAC7D,GAAI,QAAQ,aAAa,OAAO,CAAC,IAAI,EAAE,UAAU,QAAQ,SAAS;AAAA,IAClE,OAAO,eAAe,QAAQ,KAAK;AAAA,IACnC,gBAAgB;AAAA,IAChB,GAAI,UAAU,SACV,CAAC,IACD;AAAA,MACE,WAAW;AAAA,QACT,WAAW,MAAM;AAAA,QACjB,WAAW,MAAM;AAAA,QACjB,MAAM,MAAM;AAAA,QACZ,MAAM,MAAM;AAAA,QACZ,eAAe,MAAM;AAAA,MACvB;AAAA,IACF;AAAA,IACJ,cAAc,wBAAwB,KAAK;AAAA,IAC3C,OAAO,CAAC,UAAU,SAAS;AAAA,EAC7B,CAAC;AACD,QAAM,WAAW,IAAI;AAAA,KAClB,MAAM,aAAa,QAAQ,GAAG,IAAI,CAAC,YAAY;AAAA,MAC9C,QAAQ;AAAA,MACR;AAAA,IACF,CAAC;AAAA,EACH;AACA,QAAM,OAAO,CAAC,GAAI,QAAQ,YAAY,CAAC,CAAE,EACtC,KAAK,oBAAoB,EACzB,OAAO,CAAC,QAAQ,eAAe,KAAK,KAAK,CAAC;AAC7C,QAAM,OAAO,KAAK,MAAM,GAAG,QAAQ,CAAC;AACpC,QAAM,YACJ,MAAM,QAAQ;AAAA,IACZ,KAAK,MAAM,GAAG,KAAK,EAAE,IAAI,OAAO,QAAQ;AACtC,YAAM,OAAO,MAAM,qBAAqB,UAAU,UAAU,GAAG;AAE/D,aAAO,SAAS,SACZ,SACA,gBAAgB,MAAM,QAAQ,KAAK;AAAA,IACzC,CAAC;AAAA,EACH,GACA,OAAO,SAAS;AAClB,QAAM,OAAO,KAAK,GAAG,QAAQ,CAAC;AAE9B,SAAO;AAAA,IACL,OAAO;AAAA,IACP,YACE,KAAK,SAAS,SAAS,SAAS,SAC5B;AAAA,MACE,QAAQ;AAAA,MACR,KAAK;AAAA,QACH,WAAW,KAAK;AAAA,QAChB,WAAW,KAAK;AAAA,QAChB,MAAM,KAAK;AAAA,QACX,MAAM,KAAK;AAAA,QACX,eAAe,KAAK;AAAA,MACtB;AAAA,IACF,IACA;AAAA,EACR;AACF;AAEA,SAAS,wBAAwB,OAAuB;AACtD,SAAO,KAAK,IAAI,QAAQ,GAAG,QAAQ,IAAI,GAAG,GAAG;AAC/C;AAEA,SAAS,gBAAgB,KAA6B;AACpD,SAAO,IAAI,aAAa;AAC1B;AAEO,SAAS,mCAAmC,QAMrC;AACZ,MAAI;AACF,WAAO,gCAAgC,MAAM;AAAA,EAC/C,QAAQ;AACN,WAAO;AAAA,EACT;AACF;","names":[]}

package/dist/retrieval/query/archive-view/search/cache-input.d.ts

@@ -0,0 +1,45 @@
+import { ReadonlyDocument } from '../../../../document/directory/types.js';
+import { SearchEntityHitInput, SearchEvidenceHitEventInput, SearchChunkHitInput, SearchTripleHitInput } from '../../search-cache/types.js';
+import { SearchIndexQueryResult } from '../../../search-index/search/types.js';
+import { ArchiveFindHit, ArchiveFindOptions } from '../types.js';
+import '../../../../text/source/meta.js';
+import 'zod';
+import '../../../../text/source/toc.js';
+import '../../../../text/source/types.js';
+import '../../../../document/database.js';
+import 'sqlite3';
+import '../../../../document/types.js';
+import '../../../../document/text-streams/serial.js';
+import '../../../../document/text-streams/draft.js';
+import '../../../../document/text-streams/types.js';
+import '../../../../document/stores/chunk.js';
+import '../../../../document/stores/types.js';
+import '../../../../document/stores/fragment-group.js';
+import '../../../../document/stores/graph-build-parameter.js';
+import '../../../../document/stores/mention-link.js';
+import '../../../../document/stores/mention.js';
+import '../../../../document/stores/object-metadata.js';
+import '../../../../document/stores/reading-edge.js';
+import '../../../../document/stores/serial.js';
+import '../../../../document/stores/snake/chunk.js';
+import '../../../../document/stores/snake/core.js';
+import '../../../../document/stores/snake/edge.js';
+import '../../../../graph/reading.js';
+import '../../../../document/chapter/types.js';
+
+declare function isEntityOnlySearch(options: ArchiveFindOptions): boolean;
+declare function isEntitySearchTypes(types: readonly string[] | null): boolean;
+declare function assertSearchCursorTypesMatch(requestedTypes: readonly string[] | undefined, sessionTypes: readonly string[] | null): void;
+declare function createEntitySearchCacheInput(hits: readonly ArchiveFindHit[], indexResult?: SearchIndexQueryResult): {
+    readonly entityHits: readonly SearchEntityHitInput[];
+    readonly evidenceEvents: readonly SearchEvidenceHitEventInput[];
+};
+declare function createSentenceEvidenceSearchCacheInput(document: ReadonlyDocument, indexResult: SearchIndexQueryResult | undefined, options: ArchiveFindOptions): Promise<{
+    readonly chunkHits: readonly SearchChunkHitInput[];
+    readonly entityHits: readonly SearchEntityHitInput[];
+    readonly evidenceEvents: readonly SearchEvidenceHitEventInput[];
+    readonly tripleHits: readonly SearchTripleHitInput[];
+}>;
+declare function createSentenceHitKey(archiveId: number, chapterId: number, sentenceIndex: number): string;
+
+export { assertSearchCursorTypesMatch, createEntitySearchCacheInput, createSentenceEvidenceSearchCacheInput, createSentenceHitKey, isEntityOnlySearch, isEntitySearchTypes };

package/dist/retrieval/query/archive-view/search/cache-input.js

@@ -0,0 +1,255 @@
+import { SEARCH_EVIDENCE_KIND } from "../../search-cache/index.js";
+import {
+  SEARCH_OBJECT_PROPERTY_OWNER_KIND,
+  TEXT_SENTENCE_KIND
+} from "../../../search-index/search/index.js";
+import { isTextOnlySearch } from "./hydration.js";
+import {
+  formatTripleUri,
+  getMentionForTripleSearch,
+  parseEntityQid
+} from "../knowledge.js";
+function isEntityOnlySearch(options) {
+  return isEntitySearchTypes(options.types ?? null);
+}
+function isEntitySearchTypes(types) {
+  return types !== null && types.length === 1 && types[0] === "entity";
+}
+function assertSearchCursorTypesMatch(requestedTypes, sessionTypes) {
+  if (requestedTypes === void 0) {
+    return;
+  }
+  if (requestedTypes.length !== (sessionTypes?.length ?? 0)) {
+    throw new Error("Search cursor does not match the requested result types.");
+  }
+  const sessionTypeSet = new Set(sessionTypes ?? []);
+  if (requestedTypes.some((type) => !sessionTypeSet.has(type))) {
+    throw new Error("Search cursor does not match the requested result types.");
+  }
+}
+function createEntitySearchCacheInput(hits, indexResult) {
+  const evidenceEvents = [];
+  const evidenceScoresByEntity = /* @__PURE__ */ new Map();
+  const propertyScoresByEntity = /* @__PURE__ */ new Map();
+  const entitiesByKey = /* @__PURE__ */ new Map();
+  for (const hit of indexResult?.objectHits ?? []) {
+    if (hit.ownerKind !== SEARCH_OBJECT_PROPERTY_OWNER_KIND.entity) {
+      continue;
+    }
+    const key = createScopedObjectKey(hit.archiveId, hit.ownerId);
+    const scores = propertyScoresByEntity.get(key) ?? [];
+    scores.push(hit.score);
+    propertyScoresByEntity.set(key, scores);
+    entitiesByKey.set(key, { archiveId: hit.archiveId, qid: hit.ownerId });
+  }
+  for (const hit of hits) {
+    if (hit.type !== "entity" || hit.evidenceMentions === void 0) {
+      continue;
+    }
+    const qid = parseEntityQid(hit.id);
+    if (qid === void 0) {
+      continue;
+    }
+    const archiveId = hit.archiveId ?? 0;
+    const key = createScopedObjectKey(archiveId, qid);
+    const scores = evidenceScoresByEntity.get(key) ?? [];
+    for (const evidenceMention of hit.evidenceMentions) {
+      const sentenceIndex = evidenceMention.mention.sentenceIndex;
+      if (sentenceIndex === void 0) {
+        continue;
+      }
+      const score = evidenceMention.match.score ?? hit.score ?? 0;
+      evidenceEvents.push({
+        archiveId,
+        chapterId: evidenceMention.mention.chapterId,
+        evidenceId: evidenceMention.mention.id,
+        evidenceKind: SEARCH_EVIDENCE_KIND.mention,
+        score,
+        sentenceIndex
+      });
+      scores.push(score);
+    }
+    evidenceScoresByEntity.set(key, scores);
+    entitiesByKey.set(key, { archiveId, qid });
+  }
+  const qids = /* @__PURE__ */ new Set([
+    ...evidenceScoresByEntity.keys(),
+    ...propertyScoresByEntity.keys()
+  ]);
+  return {
+    entityHits: [...qids].map((key) => {
+      const entity = entitiesByKey.get(key);
+      if (entity === void 0) {
+        throw new Error("Internal error: missing scoped entity key.");
+      }
+      return {
+        archiveId: entity.archiveId,
+        evidenceTopScores: evidenceScoresByEntity.get(key) ?? [],
+        propertyTopScores: propertyScoresByEntity.get(key) ?? [],
+        qid: entity.qid
+      };
+    }),
+    evidenceEvents
+  };
+}
+async function createSentenceEvidenceSearchCacheInput(document, indexResult, options) {
+  if (indexResult === void 0 || isTextOnlySearch(options)) {
+    return {
+      chunkHits: [],
+      entityHits: [],
+      evidenceEvents: [],
+      tripleHits: []
+    };
+  }
+  const sourceHits = indexResult.textHits.filter(
+    (hit) => hit.kind === TEXT_SENTENCE_KIND.source
+  );
+  if (sourceHits.length === 0) {
+    return {
+      chunkHits: [],
+      entityHits: [],
+      evidenceEvents: [],
+      tripleHits: []
+    };
+  }
+  const sourceHitScores = /* @__PURE__ */ new Map();
+  const chapterIds = /* @__PURE__ */ new Set();
+  const archiveIds = /* @__PURE__ */ new Set();
+  for (const hit of sourceHits) {
+    chapterIds.add(hit.chapterId);
+    archiveIds.add(hit.archiveId);
+  }
+  const archiveId = archiveIds.size === 1 ? [...archiveIds][0] ?? 0 : 0;
+  for (const hit of sourceHits) {
+    sourceHitScores.set(
+      createSentenceHitKey(archiveId, hit.chapterId, hit.sentenceIndex),
+      hit.score
+    );
+  }
+  const evidenceEvents = [];
+  const entityEvidenceScoresByQid = /* @__PURE__ */ new Map();
+  const chunkEvidenceScoresById = /* @__PURE__ */ new Map();
+  const tripleEvidenceScoresByKey = /* @__PURE__ */ new Map();
+  for (const chapterId of chapterIds) {
+    for (const mention of await document.mentions.listByChapter(chapterId)) {
+      if (mention.sentenceIndex === void 0) {
+        continue;
+      }
+      const score = sourceHitScores.get(
+        createSentenceHitKey(archiveId, chapterId, mention.sentenceIndex)
+      );
+      if (score === void 0) {
+        continue;
+      }
+      evidenceEvents.push({
+        archiveId,
+        chapterId,
+        evidenceId: mention.id,
+        evidenceKind: SEARCH_EVIDENCE_KIND.mention,
+        score,
+        sentenceIndex: mention.sentenceIndex
+      });
+      const scores = entityEvidenceScoresByQid.get(mention.qid) ?? [];
+      scores.push(score);
+      entityEvidenceScoresByQid.set(mention.qid, scores);
+    }
+    for (const chunk of await document.chunks.listBySerial(chapterId)) {
+      for (const [, sentenceIndex] of chunk.sentenceIds) {
+        const score = sourceHitScores.get(
+          createSentenceHitKey(archiveId, chapterId, sentenceIndex)
+        );
+        if (score === void 0) {
+          continue;
+        }
+        evidenceEvents.push({
+          archiveId,
+          chapterId,
+          evidenceId: String(chunk.id),
+          evidenceKind: SEARCH_EVIDENCE_KIND.chunk,
+          score,
+          sentenceIndex
+        });
+        const scores = chunkEvidenceScoresById.get(chunk.id) ?? [];
+        scores.push(score);
+        chunkEvidenceScoresById.set(chunk.id, scores);
+      }
+    }
+    const mentionCache = /* @__PURE__ */ new Map();
+    for (const link of await document.mentionLinks.listByChapter(chapterId)) {
+      const [source, target] = await Promise.all([
+        getMentionForTripleSearch(document, mentionCache, link.sourceMentionId),
+        getMentionForTripleSearch(document, mentionCache, link.targetMentionId)
+      ]);
+      if (source === void 0 || target === void 0) {
+        continue;
+      }
+      for (const [
+        evidenceChapterId,
+        sentenceIndex
+      ] of link.evidenceSentenceIds) {
+        const score = sourceHitScores.get(
+          createSentenceHitKey(archiveId, evidenceChapterId, sentenceIndex)
+        );
+        if (score === void 0) {
+          continue;
+        }
+        evidenceEvents.push({
+          archiveId,
+          chapterId: evidenceChapterId,
+          evidenceId: link.id,
+          evidenceKind: SEARCH_EVIDENCE_KIND.mentionLink,
+          score,
+          sentenceIndex
+        });
+        const key = formatTripleUri(source.qid, link.predicate, target.qid);
+        const current = tripleEvidenceScoresByKey.get(key) ?? {
+          objectQid: target.qid,
+          predicate: link.predicate,
+          scores: [],
+          subjectQid: source.qid
+        };
+        current.scores.push(score);
+        tripleEvidenceScoresByKey.set(key, current);
+      }
+    }
+  }
+  return {
+    chunkHits: [...chunkEvidenceScoresById.entries()].map(
+      ([chunkId, scores]) => ({
+        archiveId,
+        chunkId,
+        evidenceTopScores: scores
+      })
+    ),
+    entityHits: [...entityEvidenceScoresByQid.entries()].map(
+      ([qid, scores]) => ({
+        archiveId,
+        evidenceTopScores: scores,
+        qid
+      })
+    ),
+    evidenceEvents,
+    tripleHits: [...tripleEvidenceScoresByKey.values()].map((hit) => ({
+      archiveId,
+      evidenceTopScores: hit.scores,
+      objectQid: hit.objectQid,
+      predicate: hit.predicate,
+      subjectQid: hit.subjectQid
+    }))
+  };
+}
+function createSentenceHitKey(archiveId, chapterId, sentenceIndex) {
+  return `${archiveId}:${chapterId}:${sentenceIndex}`;
+}
+function createScopedObjectKey(archiveId, objectId) {
+  return `${archiveId}:${objectId}`;
+}
+export {
+  assertSearchCursorTypesMatch,
+  createEntitySearchCacheInput,
+  createSentenceEvidenceSearchCacheInput,
+  createSentenceHitKey,
+  isEntityOnlySearch,
+  isEntitySearchTypes
+};
+//# sourceMappingURL=cache-input.js.map

package/dist/retrieval/query/archive-view/search/cache-input.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../src/retrieval/query/archive-view/search/cache-input.ts"],"sourcesContent":["import type {\n  MentionRecord,\n  ReadonlyDocument,\n} from \"../../../../document/index.js\";\n\nimport { SEARCH_EVIDENCE_KIND } from \"../../search-cache/index.js\";\nimport type {\n  SearchChunkHitInput,\n  SearchEntityHitInput,\n  SearchEvidenceHitEventInput,\n  SearchTripleHitInput,\n} from \"../../search-cache/types.js\";\nimport {\n  SEARCH_OBJECT_PROPERTY_OWNER_KIND,\n  TEXT_SENTENCE_KIND,\n  type SearchIndexQueryResult,\n} from \"../../../search-index/search/index.js\";\nimport { isTextOnlySearch } from \"./hydration.js\";\nimport type { ArchiveFindHit, ArchiveFindOptions } from \"../types.js\";\nimport {\n  formatTripleUri,\n  getMentionForTripleSearch,\n  parseEntityQid,\n} from \"../knowledge.js\";\n\nexport function isEntityOnlySearch(options: ArchiveFindOptions): boolean {\n  return isEntitySearchTypes(options.types ?? null);\n}\n\nexport function isEntitySearchTypes(types: readonly string[] | null): boolean {\n  return types !== null && types.length === 1 && types[0] === \"entity\";\n}\n\nexport function assertSearchCursorTypesMatch(\n  requestedTypes: readonly string[] | undefined,\n  sessionTypes: readonly string[] | null,\n): void {\n  if (requestedTypes === undefined) {\n    return;\n  }\n  if (requestedTypes.length !== (sessionTypes?.length ?? 0)) {\n    throw new Error(\"Search cursor does not match the requested result types.\");\n  }\n  const sessionTypeSet = new Set(sessionTypes ?? []);\n\n  if (requestedTypes.some((type) => !sessionTypeSet.has(type))) {\n    throw new Error(\"Search cursor does not match the requested result types.\");\n  }\n}\n\nexport function createEntitySearchCacheInput(\n  hits: readonly ArchiveFindHit[],\n  indexResult?: SearchIndexQueryResult,\n): {\n  readonly entityHits: readonly SearchEntityHitInput[];\n  readonly evidenceEvents: readonly SearchEvidenceHitEventInput[];\n} {\n  const evidenceEvents: SearchEvidenceHitEventInput[] = [];\n  const evidenceScoresByEntity = new Map<string, number[]>();\n  const propertyScoresByEntity = new Map<string, number[]>();\n  const entitiesByKey = new Map<\n    string,\n    { readonly archiveId: number; readonly qid: string }\n  >();\n\n  for (const hit of indexResult?.objectHits ?? []) {\n    if (hit.ownerKind !== SEARCH_OBJECT_PROPERTY_OWNER_KIND.entity) {\n      continue;\n    }\n\n    const key = createScopedObjectKey(hit.archiveId, hit.ownerId);\n    const scores = propertyScoresByEntity.get(key) ?? [];\n\n    scores.push(hit.score);\n    propertyScoresByEntity.set(key, scores);\n    entitiesByKey.set(key, { archiveId: hit.archiveId, qid: hit.ownerId });\n  }\n\n  for (const hit of hits) {\n    if (hit.type !== \"entity\" || hit.evidenceMentions === undefined) {\n      continue;\n    }\n\n    const qid = parseEntityQid(hit.id);\n\n    if (qid === undefined) {\n      continue;\n    }\n\n    const archiveId = hit.archiveId ?? 0;\n    const key = createScopedObjectKey(archiveId, qid);\n    const scores = evidenceScoresByEntity.get(key) ?? [];\n\n    for (const evidenceMention of hit.evidenceMentions) {\n      const sentenceIndex = evidenceMention.mention.sentenceIndex;\n\n      if (sentenceIndex === undefined) {\n        continue;\n      }\n\n      const score = evidenceMention.match.score ?? hit.score ?? 0;\n\n      evidenceEvents.push({\n        archiveId,\n        chapterId: evidenceMention.mention.chapterId,\n        evidenceId: evidenceMention.mention.id,\n        evidenceKind: SEARCH_EVIDENCE_KIND.mention,\n        score,\n        sentenceIndex,\n      });\n      scores.push(score);\n    }\n\n    evidenceScoresByEntity.set(key, scores);\n    entitiesByKey.set(key, { archiveId, qid });\n  }\n\n  const qids = new Set([\n    ...evidenceScoresByEntity.keys(),\n    ...propertyScoresByEntity.keys(),\n  ]);\n\n  return {\n    entityHits: [...qids].map((key) => {\n      const entity = entitiesByKey.get(key);\n\n      if (entity === undefined) {\n        throw new Error(\"Internal error: missing scoped entity key.\");\n      }\n\n      return {\n        archiveId: entity.archiveId,\n        evidenceTopScores: evidenceScoresByEntity.get(key) ?? [],\n        propertyTopScores: propertyScoresByEntity.get(key) ?? [],\n        qid: entity.qid,\n      };\n    }),\n    evidenceEvents,\n  };\n}\n\nexport async function createSentenceEvidenceSearchCacheInput(\n  document: ReadonlyDocument,\n  indexResult: SearchIndexQueryResult | undefined,\n  options: ArchiveFindOptions,\n): Promise<{\n  readonly chunkHits: readonly SearchChunkHitInput[];\n  readonly entityHits: readonly SearchEntityHitInput[];\n  readonly evidenceEvents: readonly SearchEvidenceHitEventInput[];\n  readonly tripleHits: readonly SearchTripleHitInput[];\n}> {\n  if (indexResult === undefined || isTextOnlySearch(options)) {\n    return {\n      chunkHits: [],\n      entityHits: [],\n      evidenceEvents: [],\n      tripleHits: [],\n    };\n  }\n\n  const sourceHits = indexResult.textHits.filter(\n    (hit) => hit.kind === TEXT_SENTENCE_KIND.source,\n  );\n\n  if (sourceHits.length === 0) {\n    return {\n      chunkHits: [],\n      entityHits: [],\n      evidenceEvents: [],\n      tripleHits: [],\n    };\n  }\n\n  const sourceHitScores = new Map<string, number>();\n  const chapterIds = new Set<number>();\n  const archiveIds = new Set<number>();\n\n  for (const hit of sourceHits) {\n    chapterIds.add(hit.chapterId);\n    archiveIds.add(hit.archiveId);\n  }\n  const archiveId = archiveIds.size === 1 ? ([...archiveIds][0] ?? 0) : 0;\n\n  for (const hit of sourceHits) {\n    sourceHitScores.set(\n      createSentenceHitKey(archiveId, hit.chapterId, hit.sentenceIndex),\n      hit.score,\n    );\n  }\n\n  const evidenceEvents: SearchEvidenceHitEventInput[] = [];\n  const entityEvidenceScoresByQid = new Map<string, number[]>();\n  const chunkEvidenceScoresById = new Map<number, number[]>();\n  const tripleEvidenceScoresByKey = new Map<\n    string,\n    {\n      readonly objectQid: string;\n      readonly predicate: string;\n      readonly scores: number[];\n      readonly subjectQid: string;\n    }\n  >();\n\n  for (const chapterId of chapterIds) {\n    for (const mention of await document.mentions.listByChapter(chapterId)) {\n      if (mention.sentenceIndex === undefined) {\n        continue;\n      }\n\n      const score = sourceHitScores.get(\n        createSentenceHitKey(archiveId, chapterId, mention.sentenceIndex),\n      );\n\n      if (score === undefined) {\n        continue;\n      }\n\n      evidenceEvents.push({\n        archiveId,\n        chapterId,\n        evidenceId: mention.id,\n        evidenceKind: SEARCH_EVIDENCE_KIND.mention,\n        score,\n        sentenceIndex: mention.sentenceIndex,\n      });\n      const scores = entityEvidenceScoresByQid.get(mention.qid) ?? [];\n\n      scores.push(score);\n      entityEvidenceScoresByQid.set(mention.qid, scores);\n    }\n\n    for (const chunk of await document.chunks.listBySerial(chapterId)) {\n      for (const [, sentenceIndex] of chunk.sentenceIds) {\n        const score = sourceHitScores.get(\n          createSentenceHitKey(archiveId, chapterId, sentenceIndex),\n        );\n\n        if (score === undefined) {\n          continue;\n        }\n\n        evidenceEvents.push({\n          archiveId,\n          chapterId,\n          evidenceId: String(chunk.id),\n          evidenceKind: SEARCH_EVIDENCE_KIND.chunk,\n          score,\n          sentenceIndex,\n        });\n        const scores = chunkEvidenceScoresById.get(chunk.id) ?? [];\n\n        scores.push(score);\n        chunkEvidenceScoresById.set(chunk.id, scores);\n      }\n    }\n\n    const mentionCache = new Map<string, MentionRecord>();\n\n    for (const link of await document.mentionLinks.listByChapter(chapterId)) {\n      const [source, target] = await Promise.all([\n        getMentionForTripleSearch(document, mentionCache, link.sourceMentionId),\n        getMentionForTripleSearch(document, mentionCache, link.targetMentionId),\n      ]);\n\n      if (source === undefined || target === undefined) {\n        continue;\n      }\n\n      for (const [\n        evidenceChapterId,\n        sentenceIndex,\n      ] of link.evidenceSentenceIds) {\n        const score = sourceHitScores.get(\n          createSentenceHitKey(archiveId, evidenceChapterId, sentenceIndex),\n        );\n\n        if (score === undefined) {\n          continue;\n        }\n\n        evidenceEvents.push({\n          archiveId,\n          chapterId: evidenceChapterId,\n          evidenceId: link.id,\n          evidenceKind: SEARCH_EVIDENCE_KIND.mentionLink,\n          score,\n          sentenceIndex,\n        });\n\n        const key = formatTripleUri(source.qid, link.predicate, target.qid);\n        const current = tripleEvidenceScoresByKey.get(key) ?? {\n          objectQid: target.qid,\n          predicate: link.predicate,\n          scores: [],\n          subjectQid: source.qid,\n        };\n\n        current.scores.push(score);\n        tripleEvidenceScoresByKey.set(key, current);\n      }\n    }\n  }\n\n  return {\n    chunkHits: [...chunkEvidenceScoresById.entries()].map(\n      ([chunkId, scores]) => ({\n        archiveId,\n        chunkId,\n        evidenceTopScores: scores,\n      }),\n    ),\n    entityHits: [...entityEvidenceScoresByQid.entries()].map(\n      ([qid, scores]) => ({\n        archiveId,\n        evidenceTopScores: scores,\n        qid,\n      }),\n    ),\n    evidenceEvents,\n    tripleHits: [...tripleEvidenceScoresByKey.values()].map((hit) => ({\n      archiveId,\n      evidenceTopScores: hit.scores,\n      objectQid: hit.objectQid,\n      predicate: hit.predicate,\n      subjectQid: hit.subjectQid,\n    })),\n  };\n}\n\nexport function createSentenceHitKey(\n  archiveId: number,\n  chapterId: number,\n  sentenceIndex: number,\n): string {\n  return `${archiveId}:${chapterId}:${sentenceIndex}`;\n}\n\nfunction createScopedObjectKey(archiveId: number, objectId: string): string {\n  return `${archiveId}:${objectId}`;\n}\n"],"mappings":"AAKA,SAAS,4BAA4B;AAOrC;AAAA,EACE;AAAA,EACA;AAAA,OAEK;AACP,SAAS,wBAAwB;AAEjC;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,OACK;AAEA,SAAS,mBAAmB,SAAsC;AACvE,SAAO,oBAAoB,QAAQ,SAAS,IAAI;AAClD;AAEO,SAAS,oBAAoB,OAA0C;AAC5E,SAAO,UAAU,QAAQ,MAAM,WAAW,KAAK,MAAM,CAAC,MAAM;AAC9D;AAEO,SAAS,6BACd,gBACA,cACM;AACN,MAAI,mBAAmB,QAAW;AAChC;AAAA,EACF;AACA,MAAI,eAAe,YAAY,cAAc,UAAU,IAAI;AACzD,UAAM,IAAI,MAAM,0DAA0D;AAAA,EAC5E;AACA,QAAM,iBAAiB,IAAI,IAAI,gBAAgB,CAAC,CAAC;AAEjD,MAAI,eAAe,KAAK,CAAC,SAAS,CAAC,eAAe,IAAI,IAAI,CAAC,GAAG;AAC5D,UAAM,IAAI,MAAM,0DAA0D;AAAA,EAC5E;AACF;AAEO,SAAS,6BACd,MACA,aAIA;AACA,QAAM,iBAAgD,CAAC;AACvD,QAAM,yBAAyB,oBAAI,IAAsB;AACzD,QAAM,yBAAyB,oBAAI,IAAsB;AACzD,QAAM,gBAAgB,oBAAI,IAGxB;AAEF,aAAW,OAAO,aAAa,cAAc,CAAC,GAAG;AAC/C,QAAI,IAAI,cAAc,kCAAkC,QAAQ;AAC9D;AAAA,IACF;AAEA,UAAM,MAAM,sBAAsB,IAAI,WAAW,IAAI,OAAO;AAC5D,UAAM,SAAS,uBAAuB,IAAI,GAAG,KAAK,CAAC;AAEnD,WAAO,KAAK,IAAI,KAAK;AACrB,2BAAuB,IAAI,KAAK,MAAM;AACtC,kBAAc,IAAI,KAAK,EAAE,WAAW,IAAI,WAAW,KAAK,IAAI,QAAQ,CAAC;AAAA,EACvE;AAEA,aAAW,OAAO,MAAM;AACtB,QAAI,IAAI,SAAS,YAAY,IAAI,qBAAqB,QAAW;AAC/D;AAAA,IACF;AAEA,UAAM,MAAM,eAAe,IAAI,EAAE;AAEjC,QAAI,QAAQ,QAAW;AACrB;AAAA,IACF;AAEA,UAAM,YAAY,IAAI,aAAa;AACnC,UAAM,MAAM,sBAAsB,WAAW,GAAG;AAChD,UAAM,SAAS,uBAAuB,IAAI,GAAG,KAAK,CAAC;AAEnD,eAAW,mBAAmB,IAAI,kBAAkB;AAClD,YAAM,gBAAgB,gBAAgB,QAAQ;AAE9C,UAAI,kBAAkB,QAAW;AAC/B;AAAA,MACF;AAEA,YAAM,QAAQ,gBAAgB,MAAM,SAAS,IAAI,SAAS;AAE1D,qBAAe,KAAK;AAAA,QAClB;AAAA,QACA,WAAW,gBAAgB,QAAQ;AAAA,QACnC,YAAY,gBAAgB,QAAQ;AAAA,QACpC,cAAc,qBAAqB;AAAA,QACnC;AAAA,QACA;AAAA,MACF,CAAC;AACD,aAAO,KAAK,KAAK;AAAA,IACnB;AAEA,2BAAuB,IAAI,KAAK,MAAM;AACtC,kBAAc,IAAI,KAAK,EAAE,WAAW,IAAI,CAAC;AAAA,EAC3C;AAEA,QAAM,OAAO,oBAAI,IAAI;AAAA,IACnB,GAAG,uBAAuB,KAAK;AAAA,IAC/B,GAAG,uBAAuB,KAAK;AAAA,EACjC,CAAC;AAED,SAAO;AAAA,IACL,YAAY,CAAC,GAAG,IAAI,EAAE,IAAI,CAAC,QAAQ;AACjC,YAAM,SAAS,cAAc,IAAI,GAAG;AAEpC,UAAI,WAAW,QAAW;AACxB,cAAM,IAAI,MAAM,4CAA4C;AAAA,MAC9D;AAEA,aAAO;AAAA,QACL,WAAW,OAAO;AAAA,QAClB,mBAAmB,uBAAuB,IAAI,GAAG,KAAK,CAAC;AAAA,QACvD,mBAAmB,uBAAuB,IAAI,GAAG,KAAK,CAAC;AAAA,QACvD,KAAK,OAAO;AAAA,MACd;AAAA,IACF,CAAC;AAAA,IACD;AAAA,EACF;AACF;AAEA,eAAsB,uCACpB,UACA,aACA,SAMC;AACD,MAAI,gBAAgB,UAAa,iBAAiB,OAAO,GAAG;AAC1D,WAAO;AAAA,MACL,WAAW,CAAC;AAAA,MACZ,YAAY,CAAC;AAAA,MACb,gBAAgB,CAAC;AAAA,MACjB,YAAY,CAAC;AAAA,IACf;AAAA,EACF;AAEA,QAAM,aAAa,YAAY,SAAS;AAAA,IACtC,CAAC,QAAQ,IAAI,SAAS,mBAAmB;AAAA,EAC3C;AAEA,MAAI,WAAW,WAAW,GAAG;AAC3B,WAAO;AAAA,MACL,WAAW,CAAC;AAAA,MACZ,YAAY,CAAC;AAAA,MACb,gBAAgB,CAAC;AAAA,MACjB,YAAY,CAAC;AAAA,IACf;AAAA,EACF;AAEA,QAAM,kBAAkB,oBAAI,IAAoB;AAChD,QAAM,aAAa,oBAAI,IAAY;AACnC,QAAM,aAAa,oBAAI,IAAY;AAEnC,aAAW,OAAO,YAAY;AAC5B,eAAW,IAAI,IAAI,SAAS;AAC5B,eAAW,IAAI,IAAI,SAAS;AAAA,EAC9B;AACA,QAAM,YAAY,WAAW,SAAS,IAAK,CAAC,GAAG,UAAU,EAAE,CAAC,KAAK,IAAK;AAEtE,aAAW,OAAO,YAAY;AAC5B,oBAAgB;AAAA,MACd,qBAAqB,WAAW,IAAI,WAAW,IAAI,aAAa;AAAA,MAChE,IAAI;AAAA,IACN;AAAA,EACF;AAEA,QAAM,iBAAgD,CAAC;AACvD,QAAM,4BAA4B,oBAAI,IAAsB;AAC5D,QAAM,0BAA0B,oBAAI,IAAsB;AAC1D,QAAM,4BAA4B,oBAAI,IAQpC;AAEF,aAAW,aAAa,YAAY;AAClC,eAAW,WAAW,MAAM,SAAS,SAAS,cAAc,SAAS,GAAG;AACtE,UAAI,QAAQ,kBAAkB,QAAW;AACvC;AAAA,MACF;AAEA,YAAM,QAAQ,gBAAgB;AAAA,QAC5B,qBAAqB,WAAW,WAAW,QAAQ,aAAa;AAAA,MAClE;AAEA,UAAI,UAAU,QAAW;AACvB;AAAA,MACF;AAEA,qBAAe,KAAK;AAAA,QAClB;AAAA,QACA;AAAA,QACA,YAAY,QAAQ;AAAA,QACpB,cAAc,qBAAqB;AAAA,QACnC;AAAA,QACA,eAAe,QAAQ;AAAA,MACzB,CAAC;AACD,YAAM,SAAS,0BAA0B,IAAI,QAAQ,GAAG,KAAK,CAAC;AAE9D,aAAO,KAAK,KAAK;AACjB,gCAA0B,IAAI,QAAQ,KAAK,MAAM;AAAA,IACnD;AAEA,eAAW,SAAS,MAAM,SAAS,OAAO,aAAa,SAAS,GAAG;AACjE,iBAAW,CAAC,EAAE,aAAa,KAAK,MAAM,aAAa;AACjD,cAAM,QAAQ,gBAAgB;AAAA,UAC5B,qBAAqB,WAAW,WAAW,aAAa;AAAA,QAC1D;AAEA,YAAI,UAAU,QAAW;AACvB;AAAA,QACF;AAEA,uBAAe,KAAK;AAAA,UAClB;AAAA,UACA;AAAA,UACA,YAAY,OAAO,MAAM,EAAE;AAAA,UAC3B,cAAc,qBAAqB;AAAA,UACnC;AAAA,UACA;AAAA,QACF,CAAC;AACD,cAAM,SAAS,wBAAwB,IAAI,MAAM,EAAE,KAAK,CAAC;AAEzD,eAAO,KAAK,KAAK;AACjB,gCAAwB,IAAI,MAAM,IAAI,MAAM;AAAA,MAC9C;AAAA,IACF;AAEA,UAAM,eAAe,oBAAI,IAA2B;AAEpD,eAAW,QAAQ,MAAM,SAAS,aAAa,cAAc,SAAS,GAAG;AACvE,YAAM,CAAC,QAAQ,MAAM,IAAI,MAAM,QAAQ,IAAI;AAAA,QACzC,0BAA0B,UAAU,cAAc,KAAK,eAAe;AAAA,QACtE,0BAA0B,UAAU,cAAc,KAAK,eAAe;AAAA,MACxE,CAAC;AAED,UAAI,WAAW,UAAa,WAAW,QAAW;AAChD;AAAA,MACF;AAEA,iBAAW;AAAA,QACT;AAAA,QACA;AAAA,MACF,KAAK,KAAK,qBAAqB;AAC7B,cAAM,QAAQ,gBAAgB;AAAA,UAC5B,qBAAqB,WAAW,mBAAmB,aAAa;AAAA,QAClE;AAEA,YAAI,UAAU,QAAW;AACvB;AAAA,QACF;AAEA,uBAAe,KAAK;AAAA,UAClB;AAAA,UACA,WAAW;AAAA,UACX,YAAY,KAAK;AAAA,UACjB,cAAc,qBAAqB;AAAA,UACnC;AAAA,UACA;AAAA,QACF,CAAC;AAED,cAAM,MAAM,gBAAgB,OAAO,KAAK,KAAK,WAAW,OAAO,GAAG;AAClE,cAAM,UAAU,0BAA0B,IAAI,GAAG,KAAK;AAAA,UACpD,WAAW,OAAO;AAAA,UAClB,WAAW,KAAK;AAAA,UAChB,QAAQ,CAAC;AAAA,UACT,YAAY,OAAO;AAAA,QACrB;AAEA,gBAAQ,OAAO,KAAK,KAAK;AACzB,kCAA0B,IAAI,KAAK,OAAO;AAAA,MAC5C;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AAAA,IACL,WAAW,CAAC,GAAG,wBAAwB,QAAQ,CAAC,EAAE;AAAA,MAChD,CAAC,CAAC,SAAS,MAAM,OAAO;AAAA,QACtB;AAAA,QACA;AAAA,QACA,mBAAmB;AAAA,MACrB;AAAA,IACF;AAAA,IACA,YAAY,CAAC,GAAG,0BAA0B,QAAQ,CAAC,EAAE;AAAA,MACnD,CAAC,CAAC,KAAK,MAAM,OAAO;AAAA,QAClB;AAAA,QACA,mBAAmB;AAAA,QACnB;AAAA,MACF;AAAA,IACF;AAAA,IACA;AAAA,IACA,YAAY,CAAC,GAAG,0BAA0B,OAAO,CAAC,EAAE,IAAI,CAAC,SAAS;AAAA,MAChE;AAAA,MACA,mBAAmB,IAAI;AAAA,MACvB,WAAW,IAAI;AAAA,MACf,WAAW,IAAI;AAAA,MACf,YAAY,IAAI;AAAA,IAClB,EAAE;AAAA,EACJ;AACF;AAEO,SAAS,qBACd,WACA,WACA,eACQ;AACR,SAAO,GAAG,SAAS,IAAI,SAAS,IAAI,aAAa;AACnD;AAEA,SAAS,sBAAsB,WAAmB,UAA0B;AAC1E,SAAO,GAAG,SAAS,IAAI,QAAQ;AACjC;","names":[]}