@open-mercato/search 0.4.2-canary-c02407ff85

package/dist/strategies/vector.strategy.js

@@ -0,0 +1,137 @@
+import { createHash } from "crypto";
+import { searchDebugWarn } from "../lib/debug.js";
+class VectorSearchStrategy {
+  constructor(embeddingService, vectorDriver, config) {
+    this.embeddingService = embeddingService;
+    this.vectorDriver = vectorDriver;
+    this.id = "vector";
+    this.name = "Vector Search";
+    this.priority = 20;
+    this.ready = false;
+    this.readyPromise = null;
+    this.defaultLimit = config?.defaultLimit ?? 20;
+  }
+  async isAvailable() {
+    return this.embeddingService.available;
+  }
+  async ensureReady() {
+    if (this.ready) return;
+    if (!this.readyPromise) {
+      this.readyPromise = this.vectorDriver.ensureReady().then(() => {
+        this.ready = true;
+      });
+    }
+    return this.readyPromise;
+  }
+  async search(query, options) {
+    await this.ensureReady();
+    const embedding = await this.embeddingService.createEmbedding(query);
+    const filter = {
+      tenantId: options.tenantId,
+      entityIds: options.entityTypes
+    };
+    if (options.organizationId) {
+      filter.organizationId = options.organizationId;
+    }
+    const results = await this.vectorDriver.query({
+      vector: embedding,
+      limit: options.limit ?? this.defaultLimit,
+      filter
+    });
+    return results.map((hit) => ({
+      entityId: hit.entityId,
+      recordId: hit.recordId,
+      score: hit.score,
+      source: this.id,
+      presenter: hit.presenter ?? void 0,
+      url: hit.primaryLinkHref ?? hit.url ?? void 0,
+      links: hit.links?.map((link) => ({
+        href: link.href,
+        label: link.label ?? "",
+        kind: link.kind
+      })),
+      metadata: hit.payload ?? void 0
+    }));
+  }
+  async index(record) {
+    await this.ensureReady();
+    const textContent = record.text ? Array.isArray(record.text) ? record.text.join("\n") : record.text : this.buildTextContent(record);
+    if (!textContent) return;
+    const embedding = await this.embeddingService.createEmbedding(textContent);
+    const doc = {
+      entityId: record.entityId,
+      recordId: record.recordId,
+      tenantId: record.tenantId,
+      organizationId: record.organizationId,
+      checksum: this.computeChecksum(record),
+      embedding,
+      url: record.url,
+      presenter: record.presenter,
+      links: record.links,
+      driverId: this.vectorDriver.id,
+      resultTitle: record.presenter?.title ?? record.recordId,
+      resultSubtitle: record.presenter?.subtitle,
+      resultIcon: record.presenter?.icon,
+      resultBadge: record.presenter?.badge
+    };
+    await this.vectorDriver.upsert(doc);
+  }
+  async delete(entityId, recordId, tenantId) {
+    await this.ensureReady();
+    await this.vectorDriver.delete(entityId, recordId, tenantId);
+  }
+  async purge(entityId, tenantId) {
+    await this.ensureReady();
+    if (this.vectorDriver.purge) {
+      await this.vectorDriver.purge(entityId, tenantId);
+    }
+  }
+  /**
+   * Build text content from record fields for embedding.
+   */
+  buildTextContent(record) {
+    const parts = [];
+    if (record.presenter?.title) {
+      parts.push(record.presenter.title);
+    }
+    if (record.presenter?.subtitle) {
+      parts.push(record.presenter.subtitle);
+    }
+    for (const [, value] of Object.entries(record.fields)) {
+      if (typeof value === "string" && value.trim()) {
+        parts.push(value);
+      }
+    }
+    return parts.join(" ").trim();
+  }
+  /**
+   * Compute a checksum for change detection using SHA-256.
+   * Uses checksumSource from buildSource if available, otherwise uses fields/presenter/url.
+   */
+  computeChecksum(record) {
+    const source = record.checksumSource !== void 0 ? record.checksumSource : {
+      fields: record.fields,
+      presenter: record.presenter,
+      url: record.url
+    };
+    const content = JSON.stringify(source);
+    return createHash("sha256").update(content).digest("hex").slice(0, 16);
+  }
+  /**
+   * List entries in the vector index (for admin/debugging).
+   */
+  async listEntries(options) {
+    await this.ensureReady();
+    const listMethod = this.vectorDriver.list;
+    if (typeof listMethod === "function") {
+      const entries = await listMethod.call(this.vectorDriver, options);
+      return entries;
+    }
+    searchDebugWarn("VectorSearchStrategy", "Vector driver does not support listing entries");
+    return [];
+  }
+}
+export {
+  VectorSearchStrategy
+};
+//# sourceMappingURL=vector.strategy.js.map

package/dist/strategies/vector.strategy.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../src/strategies/vector.strategy.ts"],
+  "sourcesContent": ["import { createHash } from 'crypto'\nimport type {\n  SearchStrategy,\n  SearchStrategyId,\n  SearchOptions,\n  SearchResult,\n  IndexableRecord,\n} from '../types'\nimport type { EntityId } from '@open-mercato/shared/modules/entities'\nimport type { VectorDriver, VectorDriverDocument } from '../vector/types'\nimport { searchDebugWarn } from '../lib/debug'\n\n/**\n * Embedding service interface - minimal subset needed by VectorSearchStrategy.\n */\nexport interface EmbeddingService {\n  createEmbedding(text: string): Promise<number[]>\n  available: boolean\n}\n\n/**\n * Configuration for VectorSearchStrategy.\n */\nexport type VectorStrategyConfig = {\n  /** Default limit for search results */\n  defaultLimit?: number\n}\n\n/**\n * VectorSearchStrategy provides semantic search using embeddings.\n * It wraps the existing vector module infrastructure.\n */\nexport class VectorSearchStrategy implements SearchStrategy {\n  readonly id: SearchStrategyId = 'vector'\n  readonly name = 'Vector Search'\n  readonly priority = 20 // Medium priority\n\n  private readonly defaultLimit: number\n  private ready = false\n  private readyPromise: Promise<void> | null = null\n\n  constructor(\n    private readonly embeddingService: EmbeddingService,\n    private readonly vectorDriver: VectorDriver,\n    config?: VectorStrategyConfig,\n  ) {\n    this.defaultLimit = config?.defaultLimit ?? 20\n  }\n\n  async isAvailable(): Promise<boolean> {\n    return this.embeddingService.available\n  }\n\n  async ensureReady(): Promise<void> {\n    if (this.ready) return\n    if (!this.readyPromise) {\n      this.readyPromise = this.vectorDriver.ensureReady().then(() => {\n        this.ready = true\n      })\n    }\n    return this.readyPromise\n  }\n\n  async search(query: string, options: SearchOptions): Promise<SearchResult[]> {\n    await this.ensureReady()\n    const embedding = await this.embeddingService.createEmbedding(query)\n\n    // Build filter - only include organizationId if it's a real value\n    // The pgvector driver treats null as \"only records with null org_id\",\n    // but we want null/undefined to mean \"no organization filter\"\n    const filter: {\n      tenantId: string\n      organizationId?: string | null\n      entityIds?: EntityId[]\n    } = {\n      tenantId: options.tenantId,\n      entityIds: options.entityTypes as EntityId[],\n    }\n\n    // Only add organizationId filter if it's a real org ID\n    if (options.organizationId) {\n      filter.organizationId = options.organizationId\n    }\n\n    const results = await this.vectorDriver.query({\n      vector: embedding,\n      limit: options.limit ?? this.defaultLimit,\n      filter,\n    })\n\n    return results.map((hit) => ({\n      entityId: hit.entityId,\n      recordId: hit.recordId,\n      score: hit.score,\n      source: this.id,\n      presenter: hit.presenter ?? undefined,\n      url: hit.primaryLinkHref ?? hit.url ?? undefined,\n      links: hit.links?.map((link) => ({\n        href: link.href,\n        label: link.label ?? '',\n        kind: link.kind,\n      })),\n      metadata: hit.payload ?? undefined,\n    }))\n  }\n\n  async index(record: IndexableRecord): Promise<void> {\n    await this.ensureReady()\n    // Use text from buildSource if available, otherwise fall back to generic extraction\n    const textContent = record.text\n      ? (Array.isArray(record.text) ? record.text.join('\\n') : record.text)\n      : this.buildTextContent(record)\n    if (!textContent) return\n\n    const embedding = await this.embeddingService.createEmbedding(textContent)\n\n    const doc: VectorDriverDocument = {\n      entityId: record.entityId as EntityId,\n      recordId: record.recordId,\n      tenantId: record.tenantId,\n      organizationId: record.organizationId,\n      checksum: this.computeChecksum(record),\n      embedding,\n      url: record.url,\n      presenter: record.presenter,\n      links: record.links,\n      driverId: this.vectorDriver.id,\n      resultTitle: record.presenter?.title ?? record.recordId,\n      resultSubtitle: record.presenter?.subtitle,\n      resultIcon: record.presenter?.icon,\n      resultBadge: record.presenter?.badge,\n    }\n\n    await this.vectorDriver.upsert(doc)\n  }\n\n  async delete(entityId: EntityId, recordId: string, tenantId: string): Promise<void> {\n    await this.ensureReady()\n    await this.vectorDriver.delete(entityId, recordId, tenantId)\n  }\n\n  async purge(entityId: EntityId, tenantId: string): Promise<void> {\n    await this.ensureReady()\n    if (this.vectorDriver.purge) {\n      await this.vectorDriver.purge(entityId, tenantId)\n    }\n  }\n\n  /**\n   * Build text content from record fields for embedding.\n   */\n  private buildTextContent(record: IndexableRecord): string {\n    const parts: string[] = []\n\n    // Add presenter info\n    if (record.presenter?.title) {\n      parts.push(record.presenter.title)\n    }\n    if (record.presenter?.subtitle) {\n      parts.push(record.presenter.subtitle)\n    }\n\n    // Add string fields from record\n    for (const [, value] of Object.entries(record.fields)) {\n      if (typeof value === 'string' && value.trim()) {\n        parts.push(value)\n      }\n    }\n\n    return parts.join(' ').trim()\n  }\n\n  /**\n   * Compute a checksum for change detection using SHA-256.\n   * Uses checksumSource from buildSource if available, otherwise uses fields/presenter/url.\n   */\n  private computeChecksum(record: IndexableRecord): string {\n    const source = record.checksumSource !== undefined\n      ? record.checksumSource\n      : {\n          fields: record.fields,\n          presenter: record.presenter,\n          url: record.url,\n        }\n    const content = JSON.stringify(source)\n    return createHash('sha256').update(content).digest('hex').slice(0, 16)\n  }\n\n  /**\n   * List entries in the vector index (for admin/debugging).\n   */\n  async listEntries(options: {\n    tenantId: string\n    organizationId?: string | null\n    entityId?: string\n    limit?: number\n    offset?: number\n  }): Promise<Array<{\n    entityId: string\n    recordId: string\n    tenantId: string\n    organizationId: string | null\n    presenter?: unknown\n    url?: string\n  }>> {\n    await this.ensureReady()\n    // Delegate to vector driver's list method if available\n    const listMethod = (this.vectorDriver as unknown as {\n      list?: (options: {\n        tenantId: string\n        organizationId?: string | null\n        entityId?: string\n        limit?: number\n        offset?: number\n      }) => Promise<unknown[]>\n    }).list\n\n    if (typeof listMethod === 'function') {\n      const entries = await listMethod.call(this.vectorDriver, options)\n      return entries as Array<{\n        entityId: string\n        recordId: string\n        tenantId: string\n        organizationId: string | null\n        presenter?: unknown\n        url?: string\n      }>\n    }\n\n    // Fallback: return empty array if driver doesn't support listing\n    searchDebugWarn('VectorSearchStrategy', 'Vector driver does not support listing entries')\n    return []\n  }\n}\n"],
+  "mappings": "AAAA,SAAS,kBAAkB;AAU3B,SAAS,uBAAuB;AAsBzB,MAAM,qBAA+C;AAAA,EAS1D,YACmB,kBACA,cACjB,QACA;AAHiB;AACA;AAVnB,SAAS,KAAuB;AAChC,SAAS,OAAO;AAChB,SAAS,WAAW;AAGpB,SAAQ,QAAQ;AAChB,SAAQ,eAAqC;AAO3C,SAAK,eAAe,QAAQ,gBAAgB;AAAA,EAC9C;AAAA,EAEA,MAAM,cAAgC;AACpC,WAAO,KAAK,iBAAiB;AAAA,EAC/B;AAAA,EAEA,MAAM,cAA6B;AACjC,QAAI,KAAK,MAAO;AAChB,QAAI,CAAC,KAAK,cAAc;AACtB,WAAK,eAAe,KAAK,aAAa,YAAY,EAAE,KAAK,MAAM;AAC7D,aAAK,QAAQ;AAAA,MACf,CAAC;AAAA,IACH;AACA,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,MAAM,OAAO,OAAe,SAAiD;AAC3E,UAAM,KAAK,YAAY;AACvB,UAAM,YAAY,MAAM,KAAK,iBAAiB,gBAAgB,KAAK;AAKnE,UAAM,SAIF;AAAA,MACF,UAAU,QAAQ;AAAA,MAClB,WAAW,QAAQ;AAAA,IACrB;AAGA,QAAI,QAAQ,gBAAgB;AAC1B,aAAO,iBAAiB,QAAQ;AAAA,IAClC;AAEA,UAAM,UAAU,MAAM,KAAK,aAAa,MAAM;AAAA,MAC5C,QAAQ;AAAA,MACR,OAAO,QAAQ,SAAS,KAAK;AAAA,MAC7B;AAAA,IACF,CAAC;AAED,WAAO,QAAQ,IAAI,CAAC,SAAS;AAAA,MAC3B,UAAU,IAAI;AAAA,MACd,UAAU,IAAI;AAAA,MACd,OAAO,IAAI;AAAA,MACX,QAAQ,KAAK;AAAA,MACb,WAAW,IAAI,aAAa;AAAA,MAC5B,KAAK,IAAI,mBAAmB,IAAI,OAAO;AAAA,MACvC,OAAO,IAAI,OAAO,IAAI,CAAC,UAAU;AAAA,QAC/B,MAAM,KAAK;AAAA,QACX,OAAO,KAAK,SAAS;AAAA,QACrB,MAAM,KAAK;AAAA,MACb,EAAE;AAAA,MACF,UAAU,IAAI,WAAW;AAAA,IAC3B,EAAE;AAAA,EACJ;AAAA,EAEA,MAAM,MAAM,QAAwC;AAClD,UAAM,KAAK,YAAY;AAEvB,UAAM,cAAc,OAAO,OACtB,MAAM,QAAQ,OAAO,IAAI,IAAI,OAAO,KAAK,KAAK,IAAI,IAAI,OAAO,OAC9D,KAAK,iBAAiB,MAAM;AAChC,QAAI,CAAC,YAAa;AAElB,UAAM,YAAY,MAAM,KAAK,iBAAiB,gBAAgB,WAAW;AAEzE,UAAM,MAA4B;AAAA,MAChC,UAAU,OAAO;AAAA,MACjB,UAAU,OAAO;AAAA,MACjB,UAAU,OAAO;AAAA,MACjB,gBAAgB,OAAO;AAAA,MACvB,UAAU,KAAK,gBAAgB,MAAM;AAAA,MACrC;AAAA,MACA,KAAK,OAAO;AAAA,MACZ,WAAW,OAAO;AAAA,MAClB,OAAO,OAAO;AAAA,MACd,UAAU,KAAK,aAAa;AAAA,MAC5B,aAAa,OAAO,WAAW,SAAS,OAAO;AAAA,MAC/C,gBAAgB,OAAO,WAAW;AAAA,MAClC,YAAY,OAAO,WAAW;AAAA,MAC9B,aAAa,OAAO,WAAW;AAAA,IACjC;AAEA,UAAM,KAAK,aAAa,OAAO,GAAG;AAAA,EACpC;AAAA,EAEA,MAAM,OAAO,UAAoB,UAAkB,UAAiC;AAClF,UAAM,KAAK,YAAY;AACvB,UAAM,KAAK,aAAa,OAAO,UAAU,UAAU,QAAQ;AAAA,EAC7D;AAAA,EAEA,MAAM,MAAM,UAAoB,UAAiC;AAC/D,UAAM,KAAK,YAAY;AACvB,QAAI,KAAK,aAAa,OAAO;AAC3B,YAAM,KAAK,aAAa,MAAM,UAAU,QAAQ;AAAA,IAClD;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKQ,iBAAiB,QAAiC;AACxD,UAAM,QAAkB,CAAC;AAGzB,QAAI,OAAO,WAAW,OAAO;AAC3B,YAAM,KAAK,OAAO,UAAU,KAAK;AAAA,IACnC;AACA,QAAI,OAAO,WAAW,UAAU;AAC9B,YAAM,KAAK,OAAO,UAAU,QAAQ;AAAA,IACtC;AAGA,eAAW,CAAC,EAAE,KAAK,KAAK,OAAO,QAAQ,OAAO,MAAM,GAAG;AACrD,UAAI,OAAO,UAAU,YAAY,MAAM,KAAK,GAAG;AAC7C,cAAM,KAAK,KAAK;AAAA,MAClB;AAAA,IACF;AAEA,WAAO,MAAM,KAAK,GAAG,EAAE,KAAK;AAAA,EAC9B;AAAA;AAAA;AAAA;AAAA;AAAA,EAMQ,gBAAgB,QAAiC;AACvD,UAAM,SAAS,OAAO,mBAAmB,SACrC,OAAO,iBACP;AAAA,MACE,QAAQ,OAAO;AAAA,MACf,WAAW,OAAO;AAAA,MAClB,KAAK,OAAO;AAAA,IACd;AACJ,UAAM,UAAU,KAAK,UAAU,MAAM;AACrC,WAAO,WAAW,QAAQ,EAAE,OAAO,OAAO,EAAE,OAAO,KAAK,EAAE,MAAM,GAAG,EAAE;AAAA,EACvE;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,YAAY,SAad;AACF,UAAM,KAAK,YAAY;AAEvB,UAAM,aAAc,KAAK,aAQtB;AAEH,QAAI,OAAO,eAAe,YAAY;AACpC,YAAM,UAAU,MAAM,WAAW,KAAK,KAAK,cAAc,OAAO;AAChE,aAAO;AAAA,IAQT;AAGA,oBAAgB,wBAAwB,gDAAgD;AACxF,WAAO,CAAC;AAAA,EACV;AACF;",
+  "names": []
+}

package/dist/types.js

@@ -0,0 +1 @@
+//# sourceMappingURL=types.js.map

package/dist/types.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": [],
+  "sourcesContent": [],
+  "mappings": "",
+  "names": []
+}

package/dist/vector/drivers/chromadb/index.js

@@ -0,0 +1,44 @@
+function notImplemented(method) {
+  throw new Error(`[vector.chromadb] ${method} not implemented yet`);
+}
+function createChromaDbDriver() {
+  return {
+    id: "chromadb",
+    async ensureReady() {
+      notImplemented("ensureReady");
+    },
+    async upsert(doc) {
+      void doc;
+      notImplemented("upsert");
+    },
+    async delete(entityId, recordId, tenantId) {
+      void entityId;
+      void recordId;
+      void tenantId;
+      notImplemented("delete");
+    },
+    async getChecksum(entityId, recordId, tenantId) {
+      void entityId;
+      void recordId;
+      void tenantId;
+      notImplemented("getChecksum");
+    },
+    async query(input) {
+      void input;
+      notImplemented("query");
+    },
+    async purge(entityId, tenantId) {
+      void entityId;
+      void tenantId;
+      notImplemented("purge");
+    },
+    async count(params) {
+      void params;
+      notImplemented("count");
+    }
+  };
+}
+export {
+  createChromaDbDriver
+};
+//# sourceMappingURL=index.js.map

package/dist/vector/drivers/chromadb/index.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../../../src/vector/drivers/chromadb/index.ts"],
+  "sourcesContent": ["import type {\n  VectorDriver,\n  VectorDriverDocument,\n  VectorDriverQuery,\n  VectorDriverQueryResult,\n  VectorDriverCountParams,\n} from '../../types'\n\nfunction notImplemented(method: string): never {\n  throw new Error(`[vector.chromadb] ${method} not implemented yet`)\n}\n\nexport function createChromaDbDriver(): VectorDriver {\n  return {\n    id: 'chromadb',\n    async ensureReady() {\n      notImplemented('ensureReady')\n    },\n    async upsert(doc: VectorDriverDocument) {\n      void doc\n      notImplemented('upsert')\n    },\n    async delete(entityId: string, recordId: string, tenantId: string) {\n      void entityId\n      void recordId\n      void tenantId\n      notImplemented('delete')\n    },\n    async getChecksum(entityId: string, recordId: string, tenantId: string) {\n      void entityId\n      void recordId\n      void tenantId\n      notImplemented('getChecksum')\n    },\n    async query(input: VectorDriverQuery): Promise<VectorDriverQueryResult[]> {\n      void input\n      notImplemented('query')\n    },\n    async purge(entityId: string, tenantId: string) {\n      void entityId\n      void tenantId\n      notImplemented('purge')\n    },\n    async count(params: VectorDriverCountParams) {\n      void params\n      notImplemented('count')\n    },\n  }\n}\n"],
+  "mappings": "AAQA,SAAS,eAAe,QAAuB;AAC7C,QAAM,IAAI,MAAM,qBAAqB,MAAM,sBAAsB;AACnE;AAEO,SAAS,uBAAqC;AACnD,SAAO;AAAA,IACL,IAAI;AAAA,IACJ,MAAM,cAAc;AAClB,qBAAe,aAAa;AAAA,IAC9B;AAAA,IACA,MAAM,OAAO,KAA2B;AACtC,WAAK;AACL,qBAAe,QAAQ;AAAA,IACzB;AAAA,IACA,MAAM,OAAO,UAAkB,UAAkB,UAAkB;AACjE,WAAK;AACL,WAAK;AACL,WAAK;AACL,qBAAe,QAAQ;AAAA,IACzB;AAAA,IACA,MAAM,YAAY,UAAkB,UAAkB,UAAkB;AACtE,WAAK;AACL,WAAK;AACL,WAAK;AACL,qBAAe,aAAa;AAAA,IAC9B;AAAA,IACA,MAAM,MAAM,OAA8D;AACxE,WAAK;AACL,qBAAe,OAAO;AAAA,IACxB;AAAA,IACA,MAAM,MAAM,UAAkB,UAAkB;AAC9C,WAAK;AACL,WAAK;AACL,qBAAe,OAAO;AAAA,IACxB;AAAA,IACA,MAAM,MAAM,QAAiC;AAC3C,WAAK;AACL,qBAAe,OAAO;AAAA,IACxB;AAAA,EACF;AACF;",
+  "names": []
+}

package/dist/vector/drivers/index.js

@@ -0,0 +1,9 @@
+import { createPgVectorDriver } from "./pgvector/index.js";
+import { createQdrantDriver } from "./qdrant/index.js";
+import { createChromaDbDriver } from "./chromadb/index.js";
+export {
+  createChromaDbDriver,
+  createPgVectorDriver,
+  createQdrantDriver
+};
+//# sourceMappingURL=index.js.map

package/dist/vector/drivers/index.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../../src/vector/drivers/index.ts"],
+  "sourcesContent": ["export { createPgVectorDriver } from './pgvector'\nexport { createQdrantDriver } from './qdrant'\nexport { createChromaDbDriver } from './chromadb'\nexport type { VectorDriver } from '../types'\n"],
+  "mappings": "AAAA,SAAS,4BAA4B;AACrC,SAAS,0BAA0B;AACnC,SAAS,4BAA4B;",
+  "names": []
+}