@superlinked/sie-llamaindex 0.1.8 → 0.1.10

package/dist/index.cjs

@@ -21,12 +21,180 @@ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: tru
 var index_exports = {};
 __export(index_exports, {
   SIEEmbedding: () => SIEEmbedding,
-  SIESparseEmbeddingFunction: () => SIESparseEmbeddingFunction
+  SIENodePostprocessor: () => SIENodePostprocessor,
+  SIESparseEmbeddingFunction: () => SIESparseEmbeddingFunction,
+  createSIEExtractorTool: () => createSIEExtractorTool
 });
 module.exports = __toCommonJS(index_exports);
+var import_sie_sdk3 = require("@superlinked/sie-sdk");
+var import_llamaindex3 = require("llamaindex");
+
+// src/rerankers.ts
 var import_sie_sdk = require("@superlinked/sie-sdk");
 var import_llamaindex = require("llamaindex");
-var SIEEmbedding = class extends import_llamaindex.BaseEmbedding {
+function extractQueryString(query) {
+  if (typeof query === "string") {
+    return query;
+  }
+  return query.filter(
+    (part) => "type" in part && part.type === "text"
+  ).map((part) => part.text).join(" ");
+}
+var SIENodePostprocessor = class {
+  modelName;
+  topN;
+  _client;
+  _ownsClient;
+  baseUrl;
+  clientOptions;
+  constructor(options = {}) {
+    const {
+      baseUrl = "http://localhost:8080",
+      modelName = "jinaai/jina-reranker-v2-base-multilingual",
+      client,
+      topN,
+      gpu,
+      timeout = 18e4
+    } = options;
+    this.baseUrl = baseUrl;
+    this.modelName = modelName;
+    this.topN = topN;
+    this._client = client;
+    this._ownsClient = !client;
+    this.clientOptions = {
+      timeout,
+      gpu
+    };
+  }
+  /**
+   * Get or create the SIEClient.
+   */
+  get client() {
+    if (!this._client) {
+      this._client = new import_sie_sdk.SIEClient(this.baseUrl, this.clientOptions);
+    }
+    return this._client;
+  }
+  /**
+   * Rerank nodes by relevance to query.
+   *
+   * @param nodes - Nodes with scores to rerank.
+   * @param query - Optional query string or MessageContent.
+   * @returns Reranked nodes with updated scores, sorted by relevance descending.
+   */
+  async postprocessNodes(nodes, query) {
+    if (nodes.length === 0 || query === void 0) {
+      return nodes;
+    }
+    const queryText = extractQueryString(query);
+    if (!queryText) {
+      return nodes;
+    }
+    const queryItem = { text: queryText };
+    const docItems = nodes.map((n) => ({ text: n.node.getContent(import_llamaindex.MetadataMode.NONE) }));
+    const result = await this.client.score(this.modelName, queryItem, docItems);
+    const reranked = [];
+    for (const entry of result.scores) {
+      const idx = Number.parseInt(entry.itemId, 10);
+      const original = nodes[idx];
+      if (original) {
+        reranked.push({
+          node: original.node,
+          score: entry.score
+        });
+      }
+    }
+    if (this.topN !== void 0) {
+      return reranked.slice(0, this.topN);
+    }
+    return reranked;
+  }
+  /**
+   * Close the underlying client connection.
+   */
+  async close() {
+    if (this._client && this._ownsClient) {
+      await this._client.close();
+    }
+  }
+};
+
+// src/extractors.ts
+var import_sie_sdk2 = require("@superlinked/sie-sdk");
+var import_llamaindex2 = require("llamaindex");
+var _SIEExtractor = class {
+  baseUrl;
+  modelName;
+  labels;
+  threshold;
+  clientOptions;
+  _client;
+  constructor(options) {
+    this.baseUrl = options.baseUrl ?? "http://localhost:8080";
+    this.modelName = options.modelName ?? "urchade/gliner_multi-v2.1";
+    this.labels = options.labels ?? ["person", "organization", "location"];
+    this.threshold = options.threshold;
+    this.clientOptions = {
+      timeout: options.timeout ?? 18e4,
+      gpu: options.gpu
+    };
+  }
+  get client() {
+    if (!this._client) {
+      this._client = new import_sie_sdk2.SIEClient(this.baseUrl, this.clientOptions);
+    }
+    return this._client;
+  }
+  async extract(text) {
+    const extractOptions = {
+      labels: this.labels
+    };
+    if (this.threshold !== void 0) {
+      extractOptions.threshold = this.threshold;
+    }
+    const result = await this.client.extract(this.modelName, { text }, extractOptions);
+    return JSON.stringify({
+      entities: result.entities.map((e) => ({
+        text: e.text,
+        label: e.label,
+        score: e.score,
+        ...e.start !== void 0 && { start: e.start },
+        ...e.end !== void 0 && { end: e.end }
+      })),
+      relations: result.relations.map((r) => ({
+        head: r.head,
+        tail: r.tail,
+        relation: r.relation,
+        score: r.score
+      })),
+      classifications: result.classifications.map((c) => ({
+        label: c.label,
+        score: c.score
+      })),
+      objects: result.objects.map((o) => ({
+        label: o.label,
+        score: o.score,
+        bbox: o.bbox
+      }))
+    });
+  }
+};
+function createSIEExtractorTool(options = {}) {
+  const labels = options.labels ?? ["person", "organization", "location"];
+  const name = options.name ?? "sie_extract";
+  const description = options.description ?? `Extract structured information from text. Finds entities of types: ${labels.join(", ")}. Returns entities, relations, classifications, and detected objects.`;
+  const extractor = new _SIEExtractor(options);
+  async function extract(input) {
+    return extractor.extract(input.text);
+  }
+  return import_llamaindex2.FunctionTool.from(extract, {
+    name,
+    description
+  });
+}
+
+// src/index.ts
+var SIEEmbedding = class extends import_llamaindex3.BaseEmbedding {
   modelName;
   instruction;
   outputDtype;
@@ -49,7 +217,7 @@ var SIEEmbedding = class extends import_llamaindex.BaseEmbedding {
       isQuery: false
     };
     const results = await this.client.encode(this.modelName, items, options);
-    return results.map((result) => this.extractDense(result));
+    return results.map((result) => (0, import_sie_sdk3.denseEmbedding)(result));
   };
   constructor(options = {}) {
     super();
@@ -79,7 +247,7 @@ var SIEEmbedding = class extends import_llamaindex.BaseEmbedding {
    */
   get client() {
     if (!this._client) {
-      this._client = new import_sie_sdk.SIEClient(this.baseUrl, this.clientOptions);
+      this._client = new import_sie_sdk3.SIEClient(this.baseUrl, this.clientOptions);
     }
     return this._client;
   }
@@ -97,17 +265,7 @@ var SIEEmbedding = class extends import_llamaindex.BaseEmbedding {
       isQuery: false
     };
     const result = await this.client.encode(this.modelName, { text }, options);
-    return this.extractDense(result);
-  }
-  /**
-   * Extract dense embedding from encode result.
-   */
-  extractDense(result) {
-    const dense = result.dense;
-    if (!dense) {
-      throw new Error("Encode result missing dense embedding");
-    }
-    return (0, import_sie_sdk.toNumberArray)(dense);
+    return (0, import_sie_sdk3.denseEmbedding)(result);
   }
   /**
    * Close the underlying client connection.
@@ -142,7 +300,7 @@ var SIESparseEmbeddingFunction = class {
    */
   get client() {
     if (!this._client) {
-      this._client = new import_sie_sdk.SIEClient(this.baseUrl, this.clientOptions);
+      this._client = new import_sie_sdk3.SIEClient(this.baseUrl, this.clientOptions);
     }
     return this._client;
   }
@@ -165,7 +323,7 @@ var SIESparseEmbeddingFunction = class {
     const indicesList = [];
     const valuesList = [];
     for (const result of results) {
-      const sparse = this.extractSparse(result);
+      const sparse = (0, import_sie_sdk3.sparseEmbedding)(result);
       indicesList.push(sparse.indices);
       valuesList.push(sparse.values);
     }
@@ -190,25 +348,12 @@ var SIESparseEmbeddingFunction = class {
     const indicesList = [];
     const valuesList = [];
     for (const result of results) {
-      const sparse = this.extractSparse(result);
+      const sparse = (0, import_sie_sdk3.sparseEmbedding)(result);
       indicesList.push(sparse.indices);
       valuesList.push(sparse.values);
     }
     return [indicesList, valuesList];
   }
-  /**
-   * Extract sparse embedding from encode result.
-   */
-  extractSparse(result) {
-    const sparse = result.sparse;
-    if (!sparse) {
-      return { indices: [], values: [] };
-    }
-    return {
-      indices: (0, import_sie_sdk.toNumberArray)(sparse.indices),
-      values: (0, import_sie_sdk.toNumberArray)(sparse.values)
-    };
-  }
   /**
    * Close the underlying client connection.
    */
@@ -221,6 +366,8 @@ var SIESparseEmbeddingFunction = class {
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   SIEEmbedding,
-  SIESparseEmbeddingFunction
+  SIENodePostprocessor,
+  SIESparseEmbeddingFunction,
+  createSIEExtractorTool
 });
 //# sourceMappingURL=index.cjs.map

package/dist/index.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/index.ts"],"sourcesContent":["/**\n * SIE embeddings integration for LlamaIndex.TS\n *\n * Provides embedding generation using SIE's encode endpoint:\n * - SIEEmbedding: Dense embeddings implementing BaseEmbedding\n * - SIESparseEmbeddingFunction: Sparse embeddings for hybrid search\n *\n * @example\n * ```typescript\n * import { Settings } from \"llamaindex\";\n * import { SIEEmbedding } from \"@superlinked/sie-llamaindex\";\n *\n * // Set as default embedding model\n * Settings.embedModel = new SIEEmbedding({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n * ```\n */\n\nimport {\n  type DType,\n  type EncodeOptions,\n  type EncodeResult,\n  SIEClient,\n  type SIEClientOptions,\n  toNumberArray,\n} from \"@superlinked/sie-sdk\";\nimport { BaseEmbedding } from \"llamaindex\";\n\n/**\n * Configuration options for SIEEmbedding.\n */\nexport interface SIEEmbeddingOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Model name/ID to use for encoding.\n   * @default \"BAAI/bge-m3\"\n   */\n  modelName?: string;\n\n  /**\n   * Optional pre-configured SIEClient instance.\n   */\n  client?: SIEClient;\n\n  /**\n   * Optional instruction prefix for embedding (model-dependent).\n   */\n  instruction?: string;\n\n  /**\n   * Output dtype: \"float32\" (default), \"float16\", \"int8\", \"binary\".\n   */\n  outputDtype?: DType;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n\n  /**\n   * Batch size for embedding multiple texts.\n   * @default 10\n   */\n  embedBatchSize?: number;\n}\n\n/**\n * LlamaIndex BaseEmbedding implementation using SIE.\n *\n * Wraps SIEClient.encode() to implement the LlamaIndex BaseEmbedding interface.\n *\n * @example\n * ```typescript\n * import { Settings, VectorStoreIndex, Document } from \"llamaindex\";\n * import { SIEEmbedding } from \"@superlinked/sie-llamaindex\";\n *\n * // Set as default embedding model\n * Settings.embedModel = new SIEEmbedding({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n *\n * // Create index with documents\n * const index = await VectorStoreIndex.fromDocuments([\n *   new Document({ text: \"Hello world\" }),\n * ]);\n *\n * // With GPU routing for multi-GPU clusters\n * const embedModel = new SIEEmbedding({\n *   baseUrl: \"https://cluster.example.com\",\n *   modelName: \"BAAI/bge-m3\",\n *   gpu: \"a100-80gb\",\n * });\n * ```\n */\nexport class SIEEmbedding extends BaseEmbedding {\n  readonly modelName: string;\n  private readonly instruction?: string;\n  private readonly outputDtype?: DType;\n  private _client: SIEClient | undefined;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  /**\n   * Get embeddings for multiple text strings (documents).\n   * This is a property (arrow function) to match BaseEmbedding interface.\n   */\n  override getTextEmbeddings = async (texts: string[]): Promise<number[][]> => {\n    if (texts.length === 0) {\n      return [];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"dense\"],\n      instruction: this.instruction,\n      outputDtype: this.outputDtype,\n      isQuery: false,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n    return (results as EncodeResult[]).map((result) => this.extractDense(result));\n  };\n\n  constructor(options: SIEEmbeddingOptions = {}) {\n    super();\n\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"BAAI/bge-m3\",\n      client,\n      instruction,\n      outputDtype,\n      gpu,\n      timeout = 180_000,\n      embedBatchSize = 10,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.instruction = instruction;\n    this.outputDtype = outputDtype;\n    this._client = client;\n    this.embedBatchSize = embedBatchSize;\n\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Get embedding for a single text string (document).\n   *\n   * @param text - Text to embed.\n   * @returns Embedding vector as array of numbers.\n   */\n  async getTextEmbedding(text: string): Promise<number[]> {\n    const options: EncodeOptions = {\n      outputTypes: [\"dense\"],\n      instruction: this.instruction,\n      outputDtype: this.outputDtype,\n      isQuery: false,\n    };\n\n    const result = await this.client.encode(this.modelName, { text }, options);\n    return this.extractDense(result as EncodeResult);\n  }\n\n  /**\n   * Extract dense embedding from encode result.\n   */\n  private extractDense(result: EncodeResult): number[] {\n    const dense = result.dense;\n    if (!dense) {\n      throw new Error(\"Encode result missing dense embedding\");\n    }\n    return toNumberArray(dense);\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client) {\n      await this._client.close();\n    }\n  }\n}\n\n/**\n * Configuration options for SIESparseEmbeddingFunction.\n */\nexport interface SIESparseEmbeddingFunctionOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Model name/ID to use for encoding. Must support sparse output.\n   * @default \"BAAI/bge-m3\"\n   */\n  modelName?: string;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n}\n\n/**\n * Sparse embedding function for LlamaIndex hybrid search.\n *\n * Compatible with LlamaIndex vector stores that support hybrid search,\n * such as QdrantVectorStore with enableHybrid: true.\n *\n * @example\n * ```typescript\n * import { QdrantVectorStore } from \"llamaindex\";\n * import { SIESparseEmbeddingFunction } from \"@superlinked/sie-llamaindex\";\n *\n * const sparseEmbedFn = new SIESparseEmbeddingFunction({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n *\n * const vectorStore = new QdrantVectorStore({\n *   client: qdrantClient,\n *   collectionName: \"hybrid_docs\",\n *   enableHybrid: true,\n *   sparseEmbeddingFunction: sparseEmbedFn,\n * });\n * ```\n */\nexport class SIESparseEmbeddingFunction {\n  private readonly modelName: string;\n  private _client: SIEClient | undefined;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  constructor(options: SIESparseEmbeddingFunctionOptions = {}) {\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"BAAI/bge-m3\",\n      gpu,\n      timeout = 180_000,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Encode query texts to sparse vectors.\n   *\n   * @param texts - List of query texts to encode.\n   * @returns Tuple of [indices_list, values_list].\n   */\n  async encodeQueries(texts: string[]): Promise<[number[][], number[][]]> {\n    if (texts.length === 0) {\n      return [[], []];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"sparse\"],\n      isQuery: true,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n\n    const indicesList: number[][] = [];\n    const valuesList: number[][] = [];\n\n    for (const result of results as EncodeResult[]) {\n      const sparse = this.extractSparse(result);\n      indicesList.push(sparse.indices);\n      valuesList.push(sparse.values);\n    }\n\n    return [indicesList, valuesList];\n  }\n\n  /**\n   * Encode document texts to sparse vectors.\n   *\n   * @param texts - List of document texts to encode.\n   * @returns Tuple of [indices_list, values_list].\n   */\n  async encodeDocuments(texts: string[]): Promise<[number[][], number[][]]> {\n    if (texts.length === 0) {\n      return [[], []];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"sparse\"],\n      isQuery: false,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n\n    const indicesList: number[][] = [];\n    const valuesList: number[][] = [];\n\n    for (const result of results as EncodeResult[]) {\n      const sparse = this.extractSparse(result);\n      indicesList.push(sparse.indices);\n      valuesList.push(sparse.values);\n    }\n\n    return [indicesList, valuesList];\n  }\n\n  /**\n   * Extract sparse embedding from encode result.\n   */\n  private extractSparse(result: EncodeResult): { indices: number[]; values: number[] } {\n    const sparse = result.sparse;\n    if (!sparse) {\n      return { indices: [], values: [] };\n    }\n\n    return {\n      indices: toNumberArray(sparse.indices),\n      values: toNumberArray(sparse.values),\n    };\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client) {\n      await this._client.close();\n    }\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAoBA,qBAOO;AACP,wBAA8B;AAgFvB,IAAM,eAAN,cAA2B,gCAAc;AAAA,EACrC;AAAA,EACQ;AAAA,EACA;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA;AAAA;AAAA;AAAA;AAAA,EAMR,oBAAoB,OAAO,UAAyC;AAC3E,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC;AAAA,IACV;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,OAAO;AAAA,MACrB,aAAa,KAAK;AAAA,MAClB,aAAa,KAAK;AAAA,MAClB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AACvE,WAAQ,QAA2B,IAAI,CAAC,WAAW,KAAK,aAAa,MAAM,CAAC;AAAA,EAC9E;AAAA,EAEA,YAAY,UAA+B,CAAC,GAAG;AAC7C,UAAM;AAEN,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA,UAAU;AAAA,MACV,iBAAiB;AAAA,IACnB,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,cAAc;AACnB,SAAK,cAAc;AACnB,SAAK,UAAU;AACf,SAAK,iBAAiB;AAEtB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAI,yBAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,iBAAiB,MAAiC;AACtD,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,OAAO;AAAA,MACrB,aAAa,KAAK;AAAA,MAClB,aAAa,KAAK;AAAA,MAClB,SAAS;AAAA,IACX;AAEA,UAAM,SAAS,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,EAAE,KAAK,GAAG,OAAO;AACzE,WAAO,KAAK,aAAa,MAAsB;AAAA,EACjD;AAAA;AAAA;AAAA;AAAA,EAKQ,aAAa,QAAgC;AACnD,UAAM,QAAQ,OAAO;AACrB,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,uCAAuC;AAAA,IACzD;AACA,eAAO,8BAAc,KAAK;AAAA,EAC5B;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,SAAS;AAChB,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;AAsDO,IAAM,6BAAN,MAAiC;AAAA,EACrB;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA,EAEjB,YAAY,UAA6C,CAAC,GAAG;AAC3D,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA,UAAU;AAAA,IACZ,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAI,yBAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,cAAc,OAAoD;AACtE,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,IAChB;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,QAAQ;AAAA,MACtB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AAEvE,UAAM,cAA0B,CAAC;AACjC,UAAM,aAAyB,CAAC;AAEhC,eAAW,UAAU,SAA2B;AAC9C,YAAM,SAAS,KAAK,cAAc,MAAM;AACxC,kBAAY,KAAK,OAAO,OAAO;AAC/B,iBAAW,KAAK,OAAO,MAAM;AAAA,IAC/B;AAEA,WAAO,CAAC,aAAa,UAAU;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,gBAAgB,OAAoD;AACxE,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,IAChB;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,QAAQ;AAAA,MACtB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AAEvE,UAAM,cAA0B,CAAC;AACjC,UAAM,aAAyB,CAAC;AAEhC,eAAW,UAAU,SAA2B;AAC9C,YAAM,SAAS,KAAK,cAAc,MAAM;AACxC,kBAAY,KAAK,OAAO,OAAO;AAC/B,iBAAW,KAAK,OAAO,MAAM;AAAA,IAC/B;AAEA,WAAO,CAAC,aAAa,UAAU;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA,EAKQ,cAAc,QAA+D;AACnF,UAAM,SAAS,OAAO;AACtB,QAAI,CAAC,QAAQ;AACX,aAAO,EAAE,SAAS,CAAC,GAAG,QAAQ,CAAC,EAAE;AAAA,IACnC;AAEA,WAAO;AAAA,MACL,aAAS,8BAAc,OAAO,OAAO;AAAA,MACrC,YAAQ,8BAAc,OAAO,MAAM;AAAA,IACrC;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,SAAS;AAChB,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;","names":[]}
+{"version":3,"sources":["../src/index.ts","../src/rerankers.ts","../src/extractors.ts"],"sourcesContent":["/**\n * SIE embeddings integration for LlamaIndex.TS\n *\n * Provides embedding generation using SIE's encode endpoint:\n * - SIEEmbedding: Dense embeddings implementing BaseEmbedding\n * - SIESparseEmbeddingFunction: Sparse embeddings for hybrid search\n * - SIENodePostprocessor: Cross-encoder reranking for query pipelines\n * - createSIEExtractorTool: Entity extraction tool for agents\n *\n * @example\n * ```typescript\n * import { Settings } from \"llamaindex\";\n * import { SIEEmbedding } from \"@superlinked/sie-llamaindex\";\n *\n * // Set as default embedding model\n * Settings.embedModel = new SIEEmbedding({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n * ```\n */\n\nimport {\n  type DType,\n  type EncodeOptions,\n  type EncodeResult,\n  SIEClient,\n  type SIEClientOptions,\n  denseEmbedding,\n  sparseEmbedding,\n} from \"@superlinked/sie-sdk\";\nimport { BaseEmbedding } from \"llamaindex\";\n\n/**\n * Configuration options for SIEEmbedding.\n */\nexport interface SIEEmbeddingOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Model name/ID to use for encoding.\n   * @default \"BAAI/bge-m3\"\n   */\n  modelName?: string;\n\n  /**\n   * Optional pre-configured SIEClient instance.\n   */\n  client?: SIEClient;\n\n  /**\n   * Optional instruction prefix for embedding (model-dependent).\n   */\n  instruction?: string;\n\n  /**\n   * Output dtype: \"float32\" (default), \"float16\", \"int8\", \"binary\".\n   */\n  outputDtype?: DType;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n\n  /**\n   * Batch size for embedding multiple texts.\n   * @default 10\n   */\n  embedBatchSize?: number;\n}\n\n/**\n * LlamaIndex BaseEmbedding implementation using SIE.\n *\n * Wraps SIEClient.encode() to implement the LlamaIndex BaseEmbedding interface.\n *\n * @example\n * ```typescript\n * import { Settings, VectorStoreIndex, Document } from \"llamaindex\";\n * import { SIEEmbedding } from \"@superlinked/sie-llamaindex\";\n *\n * // Set as default embedding model\n * Settings.embedModel = new SIEEmbedding({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n *\n * // Create index with documents\n * const index = await VectorStoreIndex.fromDocuments([\n *   new Document({ text: \"Hello world\" }),\n * ]);\n *\n * // With GPU routing for multi-GPU clusters\n * const embedModel = new SIEEmbedding({\n *   baseUrl: \"https://cluster.example.com\",\n *   modelName: \"BAAI/bge-m3\",\n *   gpu: \"a100-80gb\",\n * });\n * ```\n */\nexport class SIEEmbedding extends BaseEmbedding {\n  readonly modelName: string;\n  private readonly instruction?: string;\n  private readonly outputDtype?: DType;\n  private _client: SIEClient | undefined;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  /**\n   * Get embeddings for multiple text strings (documents).\n   * This is a property (arrow function) to match BaseEmbedding interface.\n   */\n  override getTextEmbeddings = async (texts: string[]): Promise<number[][]> => {\n    if (texts.length === 0) {\n      return [];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"dense\"],\n      instruction: this.instruction,\n      outputDtype: this.outputDtype,\n      isQuery: false,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n    return (results as EncodeResult[]).map((result) => denseEmbedding(result));\n  };\n\n  constructor(options: SIEEmbeddingOptions = {}) {\n    super();\n\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"BAAI/bge-m3\",\n      client,\n      instruction,\n      outputDtype,\n      gpu,\n      timeout = 180_000,\n      embedBatchSize = 10,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.instruction = instruction;\n    this.outputDtype = outputDtype;\n    this._client = client;\n    this.embedBatchSize = embedBatchSize;\n\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Get embedding for a single text string (document).\n   *\n   * @param text - Text to embed.\n   * @returns Embedding vector as array of numbers.\n   */\n  async getTextEmbedding(text: string): Promise<number[]> {\n    const options: EncodeOptions = {\n      outputTypes: [\"dense\"],\n      instruction: this.instruction,\n      outputDtype: this.outputDtype,\n      isQuery: false,\n    };\n\n    const result = await this.client.encode(this.modelName, { text }, options);\n    return denseEmbedding(result as EncodeResult);\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client) {\n      await this._client.close();\n    }\n  }\n}\n\n/**\n * Configuration options for SIESparseEmbeddingFunction.\n */\nexport interface SIESparseEmbeddingFunctionOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Model name/ID to use for encoding. Must support sparse output.\n   * @default \"BAAI/bge-m3\"\n   */\n  modelName?: string;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n}\n\n/**\n * Sparse embedding function for LlamaIndex hybrid search.\n *\n * Compatible with LlamaIndex vector stores that support hybrid search,\n * such as QdrantVectorStore with enableHybrid: true.\n *\n * @example\n * ```typescript\n * import { QdrantVectorStore } from \"llamaindex\";\n * import { SIESparseEmbeddingFunction } from \"@superlinked/sie-llamaindex\";\n *\n * const sparseEmbedFn = new SIESparseEmbeddingFunction({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n *\n * const vectorStore = new QdrantVectorStore({\n *   client: qdrantClient,\n *   collectionName: \"hybrid_docs\",\n *   enableHybrid: true,\n *   sparseEmbeddingFunction: sparseEmbedFn,\n * });\n * ```\n */\nexport class SIESparseEmbeddingFunction {\n  private readonly modelName: string;\n  private _client: SIEClient | undefined;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  constructor(options: SIESparseEmbeddingFunctionOptions = {}) {\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"BAAI/bge-m3\",\n      gpu,\n      timeout = 180_000,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Encode query texts to sparse vectors.\n   *\n   * @param texts - List of query texts to encode.\n   * @returns Tuple of [indices_list, values_list].\n   */\n  async encodeQueries(texts: string[]): Promise<[number[][], number[][]]> {\n    if (texts.length === 0) {\n      return [[], []];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"sparse\"],\n      isQuery: true,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n\n    const indicesList: number[][] = [];\n    const valuesList: number[][] = [];\n\n    for (const result of results as EncodeResult[]) {\n      const sparse = sparseEmbedding(result);\n      indicesList.push(sparse.indices);\n      valuesList.push(sparse.values);\n    }\n\n    return [indicesList, valuesList];\n  }\n\n  /**\n   * Encode document texts to sparse vectors.\n   *\n   * @param texts - List of document texts to encode.\n   * @returns Tuple of [indices_list, values_list].\n   */\n  async encodeDocuments(texts: string[]): Promise<[number[][], number[][]]> {\n    if (texts.length === 0) {\n      return [[], []];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"sparse\"],\n      isQuery: false,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n\n    const indicesList: number[][] = [];\n    const valuesList: number[][] = [];\n\n    for (const result of results as EncodeResult[]) {\n      const sparse = sparseEmbedding(result);\n      indicesList.push(sparse.indices);\n      valuesList.push(sparse.values);\n    }\n\n    return [indicesList, valuesList];\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client) {\n      await this._client.close();\n    }\n  }\n}\n\nexport { SIENodePostprocessor, type SIENodePostprocessorOptions } from \"./rerankers.js\";\nexport { createSIEExtractorTool, type SIEExtractorToolOptions } from \"./extractors.js\";\n","/**\n * SIE reranker integration for LlamaIndex.TS\n *\n * Provides node reranking using SIE's score endpoint:\n * - SIENodePostprocessor: Cross-encoder reranking implementing BaseNodePostprocessor\n *\n * @example\n * ```typescript\n * import { SIENodePostprocessor } from \"@superlinked/sie-llamaindex\";\n *\n * const reranker = new SIENodePostprocessor({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"jinaai/jina-reranker-v2-base-multilingual\",\n *   topN: 3,\n * });\n *\n * const reranked = await reranker.postprocessNodes(nodes, \"search query\");\n * ```\n */\n\nimport { SIEClient, type SIEClientOptions } from \"@superlinked/sie-sdk\";\nimport { MetadataMode } from \"llamaindex\";\nimport type { BaseNodePostprocessor, MessageContent, NodeWithScore } from \"llamaindex\";\n\n/**\n * Configuration options for SIENodePostprocessor.\n */\nexport interface SIENodePostprocessorOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Reranker model name/ID.\n   * @default \"jinaai/jina-reranker-v2-base-multilingual\"\n   */\n  modelName?: string;\n\n  /**\n   * Optional pre-configured SIEClient instance.\n   * If provided, baseUrl and other connection options are ignored.\n   */\n  client?: SIEClient;\n\n  /**\n   * Number of top nodes to return. If undefined, returns all nodes.\n   */\n  topN?: number;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n}\n\n/**\n * Extract a plain query string from LlamaIndex's MessageContent type.\n *\n * MessageContent is `string | MessageContentDetail[]`. For reranking\n * we need a plain string.\n */\nfunction extractQueryString(query: MessageContent): string {\n  if (typeof query === \"string\") {\n    return query;\n  }\n  // MessageContentDetail[] — concatenate text parts\n  return query\n    .filter(\n      (part): part is { type: \"text\"; text: string } => \"type\" in part && part.type === \"text\",\n    )\n    .map((part) => part.text)\n    .join(\" \");\n}\n\n/**\n * LlamaIndex node postprocessor using SIE's reranking.\n *\n * Wraps SIEClient.score() to implement the BaseNodePostprocessor interface.\n *\n * @example\n * ```typescript\n * import { VectorStoreIndex } from \"llamaindex\";\n * import { SIENodePostprocessor } from \"@superlinked/sie-llamaindex\";\n *\n * const reranker = new SIENodePostprocessor({\n *   modelName: \"jinaai/jina-reranker-v2-base-multilingual\",\n *   topN: 3,\n * });\n *\n * const queryEngine = index.asQueryEngine({\n *   nodePostprocessors: [reranker],\n * });\n *\n * const response = await queryEngine.query({ query: \"What is the topic?\" });\n * ```\n */\nexport class SIENodePostprocessor implements BaseNodePostprocessor {\n  readonly modelName: string;\n  private readonly topN?: number;\n  private _client: SIEClient | undefined;\n  private readonly _ownsClient: boolean;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  constructor(options: SIENodePostprocessorOptions = {}) {\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"jinaai/jina-reranker-v2-base-multilingual\",\n      client,\n      topN,\n      gpu,\n      timeout = 180_000,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.topN = topN;\n    this._client = client;\n    this._ownsClient = !client;\n\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Rerank nodes by relevance to query.\n   *\n   * @param nodes - Nodes with scores to rerank.\n   * @param query - Optional query string or MessageContent.\n   * @returns Reranked nodes with updated scores, sorted by relevance descending.\n   */\n  async postprocessNodes(nodes: NodeWithScore[], query?: MessageContent): Promise<NodeWithScore[]> {\n    if (nodes.length === 0 || query === undefined) {\n      return nodes;\n    }\n\n    const queryText = extractQueryString(query);\n    if (!queryText) {\n      return nodes;\n    }\n\n    const queryItem = { text: queryText };\n    const docItems = nodes.map((n) => ({ text: n.node.getContent(MetadataMode.NONE) }));\n\n    const result = await this.client.score(this.modelName, queryItem, docItems);\n\n    // Map score entries back to NodeWithScore with updated scores.\n    // ScoreResult.scores are already sorted by score descending.\n    const reranked: NodeWithScore[] = [];\n    for (const entry of result.scores) {\n      const idx = Number.parseInt(entry.itemId, 10);\n      const original = nodes[idx];\n      if (original) {\n        reranked.push({\n          node: original.node,\n          score: entry.score,\n        });\n      }\n    }\n\n    if (this.topN !== undefined) {\n      return reranked.slice(0, this.topN);\n    }\n    return reranked;\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client && this._ownsClient) {\n      await this._client.close();\n    }\n  }\n}\n","/**\n * SIE extraction tool for LlamaIndex.TS\n *\n * Provides extraction using SIE's extract endpoint:\n * - createSIEExtractorTool: Factory that creates a FunctionTool for extraction\n *\n * Returns entities, relations, classifications, and detected objects.\n *\n * @example\n * ```typescript\n * import { createSIEExtractorTool } from \"@superlinked/sie-llamaindex\";\n *\n * const extractor = createSIEExtractorTool({\n *   modelName: \"urchade/gliner_multi-v2.1\",\n *   labels: [\"person\", \"organization\", \"location\"],\n * });\n *\n * const result = await extractor.call({ text: \"John Smith works at Acme Corp\" });\n * const parsed = JSON.parse(result);\n * console.log(parsed.entities);\n * ```\n */\n\nimport {\n  type ExtractOptions,\n  type ExtractResult,\n  SIEClient,\n  type SIEClientOptions,\n} from \"@superlinked/sie-sdk\";\nimport { FunctionTool } from \"llamaindex\";\n\n/**\n * Configuration options for createSIEExtractorTool.\n */\nexport interface SIEExtractorToolOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Extraction model name/ID.\n   * @default \"urchade/gliner_multi-v2.1\"\n   */\n  modelName?: string;\n\n  /**\n   * Labels to extract.\n   * @default [\"person\", \"organization\", \"location\"]\n   */\n  labels?: string[];\n\n  /**\n   * Minimum confidence threshold (0-1).\n   */\n  threshold?: number;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n\n  /**\n   * Tool name for the agent.\n   * @default \"sie_extract\"\n   */\n  name?: string;\n\n  /**\n   * Tool description for the agent.\n   */\n  description?: string;\n}\n\n/**\n * Internal class to hold SIE extractor state.\n */\nclass _SIEExtractor {\n  private readonly baseUrl: string;\n  private readonly modelName: string;\n  private readonly labels: string[];\n  private readonly threshold?: number;\n  private readonly clientOptions: SIEClientOptions;\n  private _client: SIEClient | undefined;\n\n  constructor(options: SIEExtractorToolOptions) {\n    this.baseUrl = options.baseUrl ?? \"http://localhost:8080\";\n    this.modelName = options.modelName ?? \"urchade/gliner_multi-v2.1\";\n    this.labels = options.labels ?? [\"person\", \"organization\", \"location\"];\n    this.threshold = options.threshold;\n    this.clientOptions = {\n      timeout: options.timeout ?? 180_000,\n      gpu: options.gpu,\n    };\n  }\n\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  async extract(text: string): Promise<string> {\n    const extractOptions: ExtractOptions = {\n      labels: this.labels,\n    };\n    if (this.threshold !== undefined) {\n      extractOptions.threshold = this.threshold;\n    }\n\n    const result: ExtractResult = await this.client.extract(this.modelName, { text }, extractOptions);\n\n    return JSON.stringify({\n      entities: result.entities.map((e) => ({\n        text: e.text,\n        label: e.label,\n        score: e.score,\n        ...(e.start !== undefined && { start: e.start }),\n        ...(e.end !== undefined && { end: e.end }),\n      })),\n      relations: result.relations.map((r) => ({\n        head: r.head,\n        tail: r.tail,\n        relation: r.relation,\n        score: r.score,\n      })),\n      classifications: result.classifications.map((c) => ({\n        label: c.label,\n        score: c.score,\n      })),\n      objects: result.objects.map((o) => ({\n        label: o.label,\n        score: o.score,\n        bbox: o.bbox,\n      })),\n    });\n  }\n}\n\n/**\n * Create a LlamaIndex FunctionTool for extraction.\n *\n * Creates a tool that wraps SIE's extract endpoint for use with\n * LlamaIndex agents and workflows. Returns JSON with entities,\n * relations, classifications, and detected objects.\n *\n * @example\n * ```typescript\n * import { OpenAI, ReActAgent } from \"llamaindex\";\n * import { createSIEExtractorTool } from \"@superlinked/sie-llamaindex\";\n *\n * const extractor = createSIEExtractorTool({\n *   modelName: \"urchade/gliner_multi-v2.1\",\n *   labels: [\"person\", \"organization\", \"location\"],\n * });\n *\n * const agent = new ReActAgent({\n *   tools: [extractor],\n *   llm: new OpenAI(),\n * });\n *\n * const response = await agent.chat({\n *   message: \"Extract entities from: John works at Google in NYC\",\n * });\n * ```\n *\n * @param options - Configuration for the extractor tool.\n * @returns FunctionTool wrapping SIE extraction.\n */\nexport function createSIEExtractorTool(options: SIEExtractorToolOptions = {}) {\n  const labels = options.labels ?? [\"person\", \"organization\", \"location\"];\n  const name = options.name ?? \"sie_extract\";\n  const description =\n    options.description ??\n    `Extract structured information from text. Finds entities of types: ${labels.join(\", \")}. Returns entities, relations, classifications, and detected objects.`;\n\n  const extractor = new _SIEExtractor(options);\n\n  async function extract(input: { text: string }): Promise<string> {\n    return extractor.extract(input.text);\n  }\n\n  return FunctionTool.from(extract, {\n    name,\n    description,\n  });\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAsBA,IAAAA,kBAQO;AACP,IAAAC,qBAA8B;;;ACX9B,qBAAiD;AACjD,wBAA6B;AAgD7B,SAAS,mBAAmB,OAA+B;AACzD,MAAI,OAAO,UAAU,UAAU;AAC7B,WAAO;AAAA,EACT;AAEA,SAAO,MACJ;AAAA,IACC,CAAC,SAAiD,UAAU,QAAQ,KAAK,SAAS;AAAA,EACpF,EACC,IAAI,CAAC,SAAS,KAAK,IAAI,EACvB,KAAK,GAAG;AACb;AAwBO,IAAM,uBAAN,MAA4D;AAAA,EACxD;AAAA,EACQ;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA,EACA;AAAA,EAEjB,YAAY,UAAuC,CAAC,GAAG;AACrD,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA;AAAA,MACA;AAAA,MACA,UAAU;AAAA,IACZ,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,OAAO;AACZ,SAAK,UAAU;AACf,SAAK,cAAc,CAAC;AAEpB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAI,yBAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,MAAM,iBAAiB,OAAwB,OAAkD;AAC/F,QAAI,MAAM,WAAW,KAAK,UAAU,QAAW;AAC7C,aAAO;AAAA,IACT;AAEA,UAAM,YAAY,mBAAmB,KAAK;AAC1C,QAAI,CAAC,WAAW;AACd,aAAO;AAAA,IACT;AAEA,UAAM,YAAY,EAAE,MAAM,UAAU;AACpC,UAAM,WAAW,MAAM,IAAI,CAAC,OAAO,EAAE,MAAM,EAAE,KAAK,WAAW,+BAAa,IAAI,EAAE,EAAE;AAElF,UAAM,SAAS,MAAM,KAAK,OAAO,MAAM,KAAK,WAAW,WAAW,QAAQ;AAI1E,UAAM,WAA4B,CAAC;AACnC,eAAW,SAAS,OAAO,QAAQ;AACjC,YAAM,MAAM,OAAO,SAAS,MAAM,QAAQ,EAAE;AAC5C,YAAM,WAAW,MAAM,GAAG;AAC1B,UAAI,UAAU;AACZ,iBAAS,KAAK;AAAA,UACZ,MAAM,SAAS;AAAA,UACf,OAAO,MAAM;AAAA,QACf,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,KAAK,SAAS,QAAW;AAC3B,aAAO,SAAS,MAAM,GAAG,KAAK,IAAI;AAAA,IACpC;AACA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,WAAW,KAAK,aAAa;AACpC,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;;;AC3KA,IAAAC,kBAKO;AACP,IAAAC,qBAA6B;AAuD7B,IAAM,gBAAN,MAAoB;AAAA,EACD;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACT;AAAA,EAER,YAAY,SAAkC;AAC5C,SAAK,UAAU,QAAQ,WAAW;AAClC,SAAK,YAAY,QAAQ,aAAa;AACtC,SAAK,SAAS,QAAQ,UAAU,CAAC,UAAU,gBAAgB,UAAU;AACrE,SAAK,YAAY,QAAQ;AACzB,SAAK,gBAAgB;AAAA,MACnB,SAAS,QAAQ,WAAW;AAAA,MAC5B,KAAK,QAAQ;AAAA,IACf;AAAA,EACF;AAAA,EAEA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAI,0BAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,MAAM,QAAQ,MAA+B;AAC3C,UAAM,iBAAiC;AAAA,MACrC,QAAQ,KAAK;AAAA,IACf;AACA,QAAI,KAAK,cAAc,QAAW;AAChC,qBAAe,YAAY,KAAK;AAAA,IAClC;AAEA,UAAM,SAAwB,MAAM,KAAK,OAAO,QAAQ,KAAK,WAAW,EAAE,KAAK,GAAG,cAAc;AAEhG,WAAO,KAAK,UAAU;AAAA,MACpB,UAAU,OAAO,SAAS,IAAI,CAAC,OAAO;AAAA,QACpC,MAAM,EAAE;AAAA,QACR,OAAO,EAAE;AAAA,QACT,OAAO,EAAE;AAAA,QACT,GAAI,EAAE,UAAU,UAAa,EAAE,OAAO,EAAE,MAAM;AAAA,QAC9C,GAAI,EAAE,QAAQ,UAAa,EAAE,KAAK,EAAE,IAAI;AAAA,MAC1C,EAAE;AAAA,MACF,WAAW,OAAO,UAAU,IAAI,CAAC,OAAO;AAAA,QACtC,MAAM,EAAE;AAAA,QACR,MAAM,EAAE;AAAA,QACR,UAAU,EAAE;AAAA,QACZ,OAAO,EAAE;AAAA,MACX,EAAE;AAAA,MACF,iBAAiB,OAAO,gBAAgB,IAAI,CAAC,OAAO;AAAA,QAClD,OAAO,EAAE;AAAA,QACT,OAAO,EAAE;AAAA,MACX,EAAE;AAAA,MACF,SAAS,OAAO,QAAQ,IAAI,CAAC,OAAO;AAAA,QAClC,OAAO,EAAE;AAAA,QACT,OAAO,EAAE;AAAA,QACT,MAAM,EAAE;AAAA,MACV,EAAE;AAAA,IACJ,CAAC;AAAA,EACH;AACF;AAgCO,SAAS,uBAAuB,UAAmC,CAAC,GAAG;AAC5E,QAAM,SAAS,QAAQ,UAAU,CAAC,UAAU,gBAAgB,UAAU;AACtE,QAAM,OAAO,QAAQ,QAAQ;AAC7B,QAAM,cACJ,QAAQ,eACR,sEAAsE,OAAO,KAAK,IAAI,CAAC;AAEzF,QAAM,YAAY,IAAI,cAAc,OAAO;AAE3C,iBAAe,QAAQ,OAA0C;AAC/D,WAAO,UAAU,QAAQ,MAAM,IAAI;AAAA,EACrC;AAEA,SAAO,gCAAa,KAAK,SAAS;AAAA,IAChC;AAAA,IACA;AAAA,EACF,CAAC;AACH;;;AFnFO,IAAM,eAAN,cAA2B,iCAAc;AAAA,EACrC;AAAA,EACQ;AAAA,EACA;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA;AAAA;AAAA;AAAA;AAAA,EAMR,oBAAoB,OAAO,UAAyC;AAC3E,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC;AAAA,IACV;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,OAAO;AAAA,MACrB,aAAa,KAAK;AAAA,MAClB,aAAa,KAAK;AAAA,MAClB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AACvE,WAAQ,QAA2B,IAAI,CAAC,eAAW,gCAAe,MAAM,CAAC;AAAA,EAC3E;AAAA,EAEA,YAAY,UAA+B,CAAC,GAAG;AAC7C,UAAM;AAEN,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA,UAAU;AAAA,MACV,iBAAiB;AAAA,IACnB,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,cAAc;AACnB,SAAK,cAAc;AACnB,SAAK,UAAU;AACf,SAAK,iBAAiB;AAEtB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAI,0BAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,iBAAiB,MAAiC;AACtD,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,OAAO;AAAA,MACrB,aAAa,KAAK;AAAA,MAClB,aAAa,KAAK;AAAA,MAClB,SAAS;AAAA,IACX;AAEA,UAAM,SAAS,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,EAAE,KAAK,GAAG,OAAO;AACzE,eAAO,gCAAe,MAAsB;AAAA,EAC9C;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,SAAS;AAChB,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;AAsDO,IAAM,6BAAN,MAAiC;AAAA,EACrB;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA,EAEjB,YAAY,UAA6C,CAAC,GAAG;AAC3D,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA,UAAU;AAAA,IACZ,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAI,0BAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,cAAc,OAAoD;AACtE,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,IAChB;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,QAAQ;AAAA,MACtB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AAEvE,UAAM,cAA0B,CAAC;AACjC,UAAM,aAAyB,CAAC;AAEhC,eAAW,UAAU,SAA2B;AAC9C,YAAM,aAAS,iCAAgB,MAAM;AACrC,kBAAY,KAAK,OAAO,OAAO;AAC/B,iBAAW,KAAK,OAAO,MAAM;AAAA,IAC/B;AAEA,WAAO,CAAC,aAAa,UAAU;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,gBAAgB,OAAoD;AACxE,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,IAChB;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,QAAQ;AAAA,MACtB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AAEvE,UAAM,cAA0B,CAAC;AACjC,UAAM,aAAyB,CAAC;AAEhC,eAAW,UAAU,SAA2B;AAC9C,YAAM,aAAS,iCAAgB,MAAM;AACrC,kBAAY,KAAK,OAAO,OAAO;AAC/B,iBAAW,KAAK,OAAO,MAAM;AAAA,IAC/B;AAEA,WAAO,CAAC,aAAa,UAAU;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,SAAS;AAChB,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;","names":["import_sie_sdk","import_llamaindex","import_sie_sdk","import_llamaindex"]}

package/dist/index.d.cts

@@ -1,5 +1,183 @@
 import { SIEClient, DType } from '@superlinked/sie-sdk';
-import { BaseEmbedding } from 'llamaindex';
+import * as llamaindex from 'llamaindex';
+import { BaseNodePostprocessor, NodeWithScore, MessageContent, FunctionTool, BaseEmbedding } from 'llamaindex';
+
+/**
+ * SIE reranker integration for LlamaIndex.TS
+ *
+ * Provides node reranking using SIE's score endpoint:
+ * - SIENodePostprocessor: Cross-encoder reranking implementing BaseNodePostprocessor
+ *
+ * @example
+ * ```typescript
+ * import { SIENodePostprocessor } from "@superlinked/sie-llamaindex";
+ *
+ * const reranker = new SIENodePostprocessor({
+ *   baseUrl: "http://localhost:8080",
+ *   modelName: "jinaai/jina-reranker-v2-base-multilingual",
+ *   topN: 3,
+ * });
+ *
+ * const reranked = await reranker.postprocessNodes(nodes, "search query");
+ * ```
+ */
+
+/**
+ * Configuration options for SIENodePostprocessor.
+ */
+interface SIENodePostprocessorOptions {
+    /**
+     * URL of the SIE server.
+     * @default "http://localhost:8080"
+     */
+    baseUrl?: string;
+    /**
+     * Reranker model name/ID.
+     * @default "jinaai/jina-reranker-v2-base-multilingual"
+     */
+    modelName?: string;
+    /**
+     * Optional pre-configured SIEClient instance.
+     * If provided, baseUrl and other connection options are ignored.
+     */
+    client?: SIEClient;
+    /**
+     * Number of top nodes to return. If undefined, returns all nodes.
+     */
+    topN?: number;
+    /**
+     * Target GPU type for routing (e.g., "l4", "a100-80gb").
+     */
+    gpu?: string;
+    /**
+     * Request timeout in milliseconds.
+     * @default 180000 (3 minutes)
+     */
+    timeout?: number;
+}
+/**
+ * LlamaIndex node postprocessor using SIE's reranking.
+ *
+ * Wraps SIEClient.score() to implement the BaseNodePostprocessor interface.
+ *
+ * @example
+ * ```typescript
+ * import { VectorStoreIndex } from "llamaindex";
+ * import { SIENodePostprocessor } from "@superlinked/sie-llamaindex";
+ *
+ * const reranker = new SIENodePostprocessor({
+ *   modelName: "jinaai/jina-reranker-v2-base-multilingual",
+ *   topN: 3,
+ * });
+ *
+ * const queryEngine = index.asQueryEngine({
+ *   nodePostprocessors: [reranker],
+ * });
+ *
+ * const response = await queryEngine.query({ query: "What is the topic?" });
+ * ```
+ */
+declare class SIENodePostprocessor implements BaseNodePostprocessor {
+    readonly modelName: string;
+    private readonly topN?;
+    private _client;
+    private readonly _ownsClient;
+    private readonly baseUrl;
+    private readonly clientOptions;
+    constructor(options?: SIENodePostprocessorOptions);
+    /**
+     * Get or create the SIEClient.
+     */
+    private get client();
+    /**
+     * Rerank nodes by relevance to query.
+     *
+     * @param nodes - Nodes with scores to rerank.
+     * @param query - Optional query string or MessageContent.
+     * @returns Reranked nodes with updated scores, sorted by relevance descending.
+     */
+    postprocessNodes(nodes: NodeWithScore[], query?: MessageContent): Promise<NodeWithScore[]>;
+    /**
+     * Close the underlying client connection.
+     */
+    close(): Promise<void>;
+}
+
+/**
+ * Configuration options for createSIEExtractorTool.
+ */
+interface SIEExtractorToolOptions {
+    /**
+     * URL of the SIE server.
+     * @default "http://localhost:8080"
+     */
+    baseUrl?: string;
+    /**
+     * Extraction model name/ID.
+     * @default "urchade/gliner_multi-v2.1"
+     */
+    modelName?: string;
+    /**
+     * Labels to extract.
+     * @default ["person", "organization", "location"]
+     */
+    labels?: string[];
+    /**
+     * Minimum confidence threshold (0-1).
+     */
+    threshold?: number;
+    /**
+     * Target GPU type for routing (e.g., "l4", "a100-80gb").
+     */
+    gpu?: string;
+    /**
+     * Request timeout in milliseconds.
+     * @default 180000 (3 minutes)
+     */
+    timeout?: number;
+    /**
+     * Tool name for the agent.
+     * @default "sie_extract"
+     */
+    name?: string;
+    /**
+     * Tool description for the agent.
+     */
+    description?: string;
+}
+/**
+ * Create a LlamaIndex FunctionTool for extraction.
+ *
+ * Creates a tool that wraps SIE's extract endpoint for use with
+ * LlamaIndex agents and workflows. Returns JSON with entities,
+ * relations, classifications, and detected objects.
+ *
+ * @example
+ * ```typescript
+ * import { OpenAI, ReActAgent } from "llamaindex";
+ * import { createSIEExtractorTool } from "@superlinked/sie-llamaindex";
+ *
+ * const extractor = createSIEExtractorTool({
+ *   modelName: "urchade/gliner_multi-v2.1",
+ *   labels: ["person", "organization", "location"],
+ * });
+ *
+ * const agent = new ReActAgent({
+ *   tools: [extractor],
+ *   llm: new OpenAI(),
+ * });
+ *
+ * const response = await agent.chat({
+ *   message: "Extract entities from: John works at Google in NYC",
+ * });
+ * ```
+ *
+ * @param options - Configuration for the extractor tool.
+ * @returns FunctionTool wrapping SIE extraction.
+ */
+declare function createSIEExtractorTool(options?: SIEExtractorToolOptions): FunctionTool<{
+    text: string;
+}, llamaindex.JSONValue | Promise<llamaindex.JSONValue>>;
 
 /**
  * SIE embeddings integration for LlamaIndex.TS
@@ -7,6 +185,8 @@ import { BaseEmbedding } from 'llamaindex';
  * Provides embedding generation using SIE's encode endpoint:
  * - SIEEmbedding: Dense embeddings implementing BaseEmbedding
  * - SIESparseEmbeddingFunction: Sparse embeddings for hybrid search
+ * - SIENodePostprocessor: Cross-encoder reranking for query pipelines
+ * - createSIEExtractorTool: Entity extraction tool for agents
  *
  * @example
  * ```typescript
@@ -115,10 +295,6 @@ declare class SIEEmbedding extends BaseEmbedding {
      * @returns Embedding vector as array of numbers.
      */
     getTextEmbedding(text: string): Promise<number[]>;
-    /**
-     * Extract dense embedding from encode result.
-     */
-    private extractDense;
     /**
      * Close the underlying client connection.
      */
@@ -196,14 +372,10 @@ declare class SIESparseEmbeddingFunction {
      * @returns Tuple of [indices_list, values_list].
      */
     encodeDocuments(texts: string[]): Promise<[number[][], number[][]]>;
-    /**
-     * Extract sparse embedding from encode result.
-     */
-    private extractSparse;
     /**
      * Close the underlying client connection.
      */
     close(): Promise<void>;
 }
 
-export { SIEEmbedding, type SIEEmbeddingOptions, SIESparseEmbeddingFunction, type SIESparseEmbeddingFunctionOptions };
+export { SIEEmbedding, type SIEEmbeddingOptions, type SIEExtractorToolOptions, SIENodePostprocessor, type SIENodePostprocessorOptions, SIESparseEmbeddingFunction, type SIESparseEmbeddingFunctionOptions, createSIEExtractorTool };

package/dist/index.d.ts

@@ -1,5 +1,183 @@
 import { SIEClient, DType } from '@superlinked/sie-sdk';
-import { BaseEmbedding } from 'llamaindex';
+import * as llamaindex from 'llamaindex';
+import { BaseNodePostprocessor, NodeWithScore, MessageContent, FunctionTool, BaseEmbedding } from 'llamaindex';
+
+/**
+ * SIE reranker integration for LlamaIndex.TS
+ *
+ * Provides node reranking using SIE's score endpoint:
+ * - SIENodePostprocessor: Cross-encoder reranking implementing BaseNodePostprocessor
+ *
+ * @example
+ * ```typescript
+ * import { SIENodePostprocessor } from "@superlinked/sie-llamaindex";
+ *
+ * const reranker = new SIENodePostprocessor({
+ *   baseUrl: "http://localhost:8080",
+ *   modelName: "jinaai/jina-reranker-v2-base-multilingual",
+ *   topN: 3,
+ * });
+ *
+ * const reranked = await reranker.postprocessNodes(nodes, "search query");
+ * ```
+ */
+
+/**
+ * Configuration options for SIENodePostprocessor.
+ */
+interface SIENodePostprocessorOptions {
+    /**
+     * URL of the SIE server.
+     * @default "http://localhost:8080"
+     */
+    baseUrl?: string;
+    /**
+     * Reranker model name/ID.
+     * @default "jinaai/jina-reranker-v2-base-multilingual"
+     */
+    modelName?: string;
+    /**
+     * Optional pre-configured SIEClient instance.
+     * If provided, baseUrl and other connection options are ignored.
+     */
+    client?: SIEClient;
+    /**
+     * Number of top nodes to return. If undefined, returns all nodes.
+     */
+    topN?: number;
+    /**
+     * Target GPU type for routing (e.g., "l4", "a100-80gb").
+     */
+    gpu?: string;
+    /**
+     * Request timeout in milliseconds.
+     * @default 180000 (3 minutes)
+     */
+    timeout?: number;
+}
+/**
+ * LlamaIndex node postprocessor using SIE's reranking.
+ *
+ * Wraps SIEClient.score() to implement the BaseNodePostprocessor interface.
+ *
+ * @example
+ * ```typescript
+ * import { VectorStoreIndex } from "llamaindex";
+ * import { SIENodePostprocessor } from "@superlinked/sie-llamaindex";
+ *
+ * const reranker = new SIENodePostprocessor({
+ *   modelName: "jinaai/jina-reranker-v2-base-multilingual",
+ *   topN: 3,
+ * });
+ *
+ * const queryEngine = index.asQueryEngine({
+ *   nodePostprocessors: [reranker],
+ * });
+ *
+ * const response = await queryEngine.query({ query: "What is the topic?" });
+ * ```
+ */
+declare class SIENodePostprocessor implements BaseNodePostprocessor {
+    readonly modelName: string;
+    private readonly topN?;
+    private _client;
+    private readonly _ownsClient;
+    private readonly baseUrl;
+    private readonly clientOptions;
+    constructor(options?: SIENodePostprocessorOptions);
+    /**
+     * Get or create the SIEClient.
+     */
+    private get client();
+    /**
+     * Rerank nodes by relevance to query.
+     *
+     * @param nodes - Nodes with scores to rerank.
+     * @param query - Optional query string or MessageContent.
+     * @returns Reranked nodes with updated scores, sorted by relevance descending.
+     */
+    postprocessNodes(nodes: NodeWithScore[], query?: MessageContent): Promise<NodeWithScore[]>;
+    /**
+     * Close the underlying client connection.
+     */
+    close(): Promise<void>;
+}
+
+/**
+ * Configuration options for createSIEExtractorTool.
+ */
+interface SIEExtractorToolOptions {
+    /**
+     * URL of the SIE server.
+     * @default "http://localhost:8080"
+     */
+    baseUrl?: string;
+    /**
+     * Extraction model name/ID.
+     * @default "urchade/gliner_multi-v2.1"
+     */
+    modelName?: string;
+    /**
+     * Labels to extract.
+     * @default ["person", "organization", "location"]
+     */
+    labels?: string[];
+    /**
+     * Minimum confidence threshold (0-1).
+     */
+    threshold?: number;
+    /**
+     * Target GPU type for routing (e.g., "l4", "a100-80gb").
+     */
+    gpu?: string;
+    /**
+     * Request timeout in milliseconds.
+     * @default 180000 (3 minutes)
+     */
+    timeout?: number;
+    /**
+     * Tool name for the agent.
+     * @default "sie_extract"
+     */
+    name?: string;
+    /**
+     * Tool description for the agent.
+     */
+    description?: string;
+}
+/**
+ * Create a LlamaIndex FunctionTool for extraction.
+ *
+ * Creates a tool that wraps SIE's extract endpoint for use with
+ * LlamaIndex agents and workflows. Returns JSON with entities,
+ * relations, classifications, and detected objects.
+ *
+ * @example
+ * ```typescript
+ * import { OpenAI, ReActAgent } from "llamaindex";
+ * import { createSIEExtractorTool } from "@superlinked/sie-llamaindex";
+ *
+ * const extractor = createSIEExtractorTool({
+ *   modelName: "urchade/gliner_multi-v2.1",
+ *   labels: ["person", "organization", "location"],
+ * });
+ *
+ * const agent = new ReActAgent({
+ *   tools: [extractor],
+ *   llm: new OpenAI(),
+ * });
+ *
+ * const response = await agent.chat({
+ *   message: "Extract entities from: John works at Google in NYC",
+ * });
+ * ```
+ *
+ * @param options - Configuration for the extractor tool.
+ * @returns FunctionTool wrapping SIE extraction.
+ */
+declare function createSIEExtractorTool(options?: SIEExtractorToolOptions): FunctionTool<{
+    text: string;
+}, llamaindex.JSONValue | Promise<llamaindex.JSONValue>>;
 
 /**
  * SIE embeddings integration for LlamaIndex.TS
@@ -7,6 +185,8 @@ import { BaseEmbedding } from 'llamaindex';
  * Provides embedding generation using SIE's encode endpoint:
  * - SIEEmbedding: Dense embeddings implementing BaseEmbedding
  * - SIESparseEmbeddingFunction: Sparse embeddings for hybrid search
+ * - SIENodePostprocessor: Cross-encoder reranking for query pipelines
+ * - createSIEExtractorTool: Entity extraction tool for agents
  *
  * @example
  * ```typescript
@@ -115,10 +295,6 @@ declare class SIEEmbedding extends BaseEmbedding {
      * @returns Embedding vector as array of numbers.
      */
     getTextEmbedding(text: string): Promise<number[]>;
-    /**
-     * Extract dense embedding from encode result.
-     */
-    private extractDense;
     /**
      * Close the underlying client connection.
      */
@@ -196,14 +372,10 @@ declare class SIESparseEmbeddingFunction {
      * @returns Tuple of [indices_list, values_list].
      */
     encodeDocuments(texts: string[]): Promise<[number[][], number[][]]>;
-    /**
-     * Extract sparse embedding from encode result.
-     */
-    private extractSparse;
     /**
      * Close the underlying client connection.
      */
     close(): Promise<void>;
 }
 
-export { SIEEmbedding, type SIEEmbeddingOptions, SIESparseEmbeddingFunction, type SIESparseEmbeddingFunctionOptions };
+export { SIEEmbedding, type SIEEmbeddingOptions, type SIEExtractorToolOptions, SIENodePostprocessor, type SIENodePostprocessorOptions, SIESparseEmbeddingFunction, type SIESparseEmbeddingFunctionOptions, createSIEExtractorTool };

package/dist/index.js

@@ -1,9 +1,178 @@
 // src/index.ts
 import {
-  SIEClient,
-  toNumberArray
+  SIEClient as SIEClient3,
+  denseEmbedding,
+  sparseEmbedding
 } from "@superlinked/sie-sdk";
 import { BaseEmbedding } from "llamaindex";
+
+// src/rerankers.ts
+import { SIEClient } from "@superlinked/sie-sdk";
+import { MetadataMode } from "llamaindex";
+function extractQueryString(query) {
+  if (typeof query === "string") {
+    return query;
+  }
+  return query.filter(
+    (part) => "type" in part && part.type === "text"
+  ).map((part) => part.text).join(" ");
+}
+var SIENodePostprocessor = class {
+  modelName;
+  topN;
+  _client;
+  _ownsClient;
+  baseUrl;
+  clientOptions;
+  constructor(options = {}) {
+    const {
+      baseUrl = "http://localhost:8080",
+      modelName = "jinaai/jina-reranker-v2-base-multilingual",
+      client,
+      topN,
+      gpu,
+      timeout = 18e4
+    } = options;
+    this.baseUrl = baseUrl;
+    this.modelName = modelName;
+    this.topN = topN;
+    this._client = client;
+    this._ownsClient = !client;
+    this.clientOptions = {
+      timeout,
+      gpu
+    };
+  }
+  /**
+   * Get or create the SIEClient.
+   */
+  get client() {
+    if (!this._client) {
+      this._client = new SIEClient(this.baseUrl, this.clientOptions);
+    }
+    return this._client;
+  }
+  /**
+   * Rerank nodes by relevance to query.
+   *
+   * @param nodes - Nodes with scores to rerank.
+   * @param query - Optional query string or MessageContent.
+   * @returns Reranked nodes with updated scores, sorted by relevance descending.
+   */
+  async postprocessNodes(nodes, query) {
+    if (nodes.length === 0 || query === void 0) {
+      return nodes;
+    }
+    const queryText = extractQueryString(query);
+    if (!queryText) {
+      return nodes;
+    }
+    const queryItem = { text: queryText };
+    const docItems = nodes.map((n) => ({ text: n.node.getContent(MetadataMode.NONE) }));
+    const result = await this.client.score(this.modelName, queryItem, docItems);
+    const reranked = [];
+    for (const entry of result.scores) {
+      const idx = Number.parseInt(entry.itemId, 10);
+      const original = nodes[idx];
+      if (original) {
+        reranked.push({
+          node: original.node,
+          score: entry.score
+        });
+      }
+    }
+    if (this.topN !== void 0) {
+      return reranked.slice(0, this.topN);
+    }
+    return reranked;
+  }
+  /**
+   * Close the underlying client connection.
+   */
+  async close() {
+    if (this._client && this._ownsClient) {
+      await this._client.close();
+    }
+  }
+};
+
+// src/extractors.ts
+import {
+  SIEClient as SIEClient2
+} from "@superlinked/sie-sdk";
+import { FunctionTool } from "llamaindex";
+var _SIEExtractor = class {
+  baseUrl;
+  modelName;
+  labels;
+  threshold;
+  clientOptions;
+  _client;
+  constructor(options) {
+    this.baseUrl = options.baseUrl ?? "http://localhost:8080";
+    this.modelName = options.modelName ?? "urchade/gliner_multi-v2.1";
+    this.labels = options.labels ?? ["person", "organization", "location"];
+    this.threshold = options.threshold;
+    this.clientOptions = {
+      timeout: options.timeout ?? 18e4,
+      gpu: options.gpu
+    };
+  }
+  get client() {
+    if (!this._client) {
+      this._client = new SIEClient2(this.baseUrl, this.clientOptions);
+    }
+    return this._client;
+  }
+  async extract(text) {
+    const extractOptions = {
+      labels: this.labels
+    };
+    if (this.threshold !== void 0) {
+      extractOptions.threshold = this.threshold;
+    }
+    const result = await this.client.extract(this.modelName, { text }, extractOptions);
+    return JSON.stringify({
+      entities: result.entities.map((e) => ({
+        text: e.text,
+        label: e.label,
+        score: e.score,
+        ...e.start !== void 0 && { start: e.start },
+        ...e.end !== void 0 && { end: e.end }
+      })),
+      relations: result.relations.map((r) => ({
+        head: r.head,
+        tail: r.tail,
+        relation: r.relation,
+        score: r.score
+      })),
+      classifications: result.classifications.map((c) => ({
+        label: c.label,
+        score: c.score
+      })),
+      objects: result.objects.map((o) => ({
+        label: o.label,
+        score: o.score,
+        bbox: o.bbox
+      }))
+    });
+  }
+};
+function createSIEExtractorTool(options = {}) {
+  const labels = options.labels ?? ["person", "organization", "location"];
+  const name = options.name ?? "sie_extract";
+  const description = options.description ?? `Extract structured information from text. Finds entities of types: ${labels.join(", ")}. Returns entities, relations, classifications, and detected objects.`;
+  const extractor = new _SIEExtractor(options);
+  async function extract(input) {
+    return extractor.extract(input.text);
+  }
+  return FunctionTool.from(extract, {
+    name,
+    description
+  });
+}
+
+// src/index.ts
 var SIEEmbedding = class extends BaseEmbedding {
   modelName;
   instruction;
@@ -27,7 +196,7 @@ var SIEEmbedding = class extends BaseEmbedding {
       isQuery: false
     };
     const results = await this.client.encode(this.modelName, items, options);
-    return results.map((result) => this.extractDense(result));
+    return results.map((result) => denseEmbedding(result));
   };
   constructor(options = {}) {
     super();
@@ -57,7 +226,7 @@ var SIEEmbedding = class extends BaseEmbedding {
    */
   get client() {
     if (!this._client) {
-      this._client = new SIEClient(this.baseUrl, this.clientOptions);
+      this._client = new SIEClient3(this.baseUrl, this.clientOptions);
     }
     return this._client;
   }
@@ -75,17 +244,7 @@ var SIEEmbedding = class extends BaseEmbedding {
       isQuery: false
     };
     const result = await this.client.encode(this.modelName, { text }, options);
-    return this.extractDense(result);
-  }
-  /**
-   * Extract dense embedding from encode result.
-   */
-  extractDense(result) {
-    const dense = result.dense;
-    if (!dense) {
-      throw new Error("Encode result missing dense embedding");
-    }
-    return toNumberArray(dense);
+    return denseEmbedding(result);
   }
   /**
    * Close the underlying client connection.
@@ -120,7 +279,7 @@ var SIESparseEmbeddingFunction = class {
    */
   get client() {
     if (!this._client) {
-      this._client = new SIEClient(this.baseUrl, this.clientOptions);
+      this._client = new SIEClient3(this.baseUrl, this.clientOptions);
     }
     return this._client;
   }
@@ -143,7 +302,7 @@ var SIESparseEmbeddingFunction = class {
     const indicesList = [];
     const valuesList = [];
     for (const result of results) {
-      const sparse = this.extractSparse(result);
+      const sparse = sparseEmbedding(result);
       indicesList.push(sparse.indices);
       valuesList.push(sparse.values);
     }
@@ -168,25 +327,12 @@ var SIESparseEmbeddingFunction = class {
     const indicesList = [];
     const valuesList = [];
     for (const result of results) {
-      const sparse = this.extractSparse(result);
+      const sparse = sparseEmbedding(result);
       indicesList.push(sparse.indices);
       valuesList.push(sparse.values);
     }
     return [indicesList, valuesList];
   }
-  /**
-   * Extract sparse embedding from encode result.
-   */
-  extractSparse(result) {
-    const sparse = result.sparse;
-    if (!sparse) {
-      return { indices: [], values: [] };
-    }
-    return {
-      indices: toNumberArray(sparse.indices),
-      values: toNumberArray(sparse.values)
-    };
-  }
   /**
    * Close the underlying client connection.
    */
@@ -198,6 +344,8 @@ var SIESparseEmbeddingFunction = class {
 };
 export {
   SIEEmbedding,
-  SIESparseEmbeddingFunction
+  SIENodePostprocessor,
+  SIESparseEmbeddingFunction,
+  createSIEExtractorTool
 };
 //# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/index.ts"],"sourcesContent":["/**\n * SIE embeddings integration for LlamaIndex.TS\n *\n * Provides embedding generation using SIE's encode endpoint:\n * - SIEEmbedding: Dense embeddings implementing BaseEmbedding\n * - SIESparseEmbeddingFunction: Sparse embeddings for hybrid search\n *\n * @example\n * ```typescript\n * import { Settings } from \"llamaindex\";\n * import { SIEEmbedding } from \"@superlinked/sie-llamaindex\";\n *\n * // Set as default embedding model\n * Settings.embedModel = new SIEEmbedding({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n * ```\n */\n\nimport {\n  type DType,\n  type EncodeOptions,\n  type EncodeResult,\n  SIEClient,\n  type SIEClientOptions,\n  toNumberArray,\n} from \"@superlinked/sie-sdk\";\nimport { BaseEmbedding } from \"llamaindex\";\n\n/**\n * Configuration options for SIEEmbedding.\n */\nexport interface SIEEmbeddingOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Model name/ID to use for encoding.\n   * @default \"BAAI/bge-m3\"\n   */\n  modelName?: string;\n\n  /**\n   * Optional pre-configured SIEClient instance.\n   */\n  client?: SIEClient;\n\n  /**\n   * Optional instruction prefix for embedding (model-dependent).\n   */\n  instruction?: string;\n\n  /**\n   * Output dtype: \"float32\" (default), \"float16\", \"int8\", \"binary\".\n   */\n  outputDtype?: DType;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n\n  /**\n   * Batch size for embedding multiple texts.\n   * @default 10\n   */\n  embedBatchSize?: number;\n}\n\n/**\n * LlamaIndex BaseEmbedding implementation using SIE.\n *\n * Wraps SIEClient.encode() to implement the LlamaIndex BaseEmbedding interface.\n *\n * @example\n * ```typescript\n * import { Settings, VectorStoreIndex, Document } from \"llamaindex\";\n * import { SIEEmbedding } from \"@superlinked/sie-llamaindex\";\n *\n * // Set as default embedding model\n * Settings.embedModel = new SIEEmbedding({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n *\n * // Create index with documents\n * const index = await VectorStoreIndex.fromDocuments([\n *   new Document({ text: \"Hello world\" }),\n * ]);\n *\n * // With GPU routing for multi-GPU clusters\n * const embedModel = new SIEEmbedding({\n *   baseUrl: \"https://cluster.example.com\",\n *   modelName: \"BAAI/bge-m3\",\n *   gpu: \"a100-80gb\",\n * });\n * ```\n */\nexport class SIEEmbedding extends BaseEmbedding {\n  readonly modelName: string;\n  private readonly instruction?: string;\n  private readonly outputDtype?: DType;\n  private _client: SIEClient | undefined;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  /**\n   * Get embeddings for multiple text strings (documents).\n   * This is a property (arrow function) to match BaseEmbedding interface.\n   */\n  override getTextEmbeddings = async (texts: string[]): Promise<number[][]> => {\n    if (texts.length === 0) {\n      return [];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"dense\"],\n      instruction: this.instruction,\n      outputDtype: this.outputDtype,\n      isQuery: false,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n    return (results as EncodeResult[]).map((result) => this.extractDense(result));\n  };\n\n  constructor(options: SIEEmbeddingOptions = {}) {\n    super();\n\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"BAAI/bge-m3\",\n      client,\n      instruction,\n      outputDtype,\n      gpu,\n      timeout = 180_000,\n      embedBatchSize = 10,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.instruction = instruction;\n    this.outputDtype = outputDtype;\n    this._client = client;\n    this.embedBatchSize = embedBatchSize;\n\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Get embedding for a single text string (document).\n   *\n   * @param text - Text to embed.\n   * @returns Embedding vector as array of numbers.\n   */\n  async getTextEmbedding(text: string): Promise<number[]> {\n    const options: EncodeOptions = {\n      outputTypes: [\"dense\"],\n      instruction: this.instruction,\n      outputDtype: this.outputDtype,\n      isQuery: false,\n    };\n\n    const result = await this.client.encode(this.modelName, { text }, options);\n    return this.extractDense(result as EncodeResult);\n  }\n\n  /**\n   * Extract dense embedding from encode result.\n   */\n  private extractDense(result: EncodeResult): number[] {\n    const dense = result.dense;\n    if (!dense) {\n      throw new Error(\"Encode result missing dense embedding\");\n    }\n    return toNumberArray(dense);\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client) {\n      await this._client.close();\n    }\n  }\n}\n\n/**\n * Configuration options for SIESparseEmbeddingFunction.\n */\nexport interface SIESparseEmbeddingFunctionOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Model name/ID to use for encoding. Must support sparse output.\n   * @default \"BAAI/bge-m3\"\n   */\n  modelName?: string;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n}\n\n/**\n * Sparse embedding function for LlamaIndex hybrid search.\n *\n * Compatible with LlamaIndex vector stores that support hybrid search,\n * such as QdrantVectorStore with enableHybrid: true.\n *\n * @example\n * ```typescript\n * import { QdrantVectorStore } from \"llamaindex\";\n * import { SIESparseEmbeddingFunction } from \"@superlinked/sie-llamaindex\";\n *\n * const sparseEmbedFn = new SIESparseEmbeddingFunction({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n *\n * const vectorStore = new QdrantVectorStore({\n *   client: qdrantClient,\n *   collectionName: \"hybrid_docs\",\n *   enableHybrid: true,\n *   sparseEmbeddingFunction: sparseEmbedFn,\n * });\n * ```\n */\nexport class SIESparseEmbeddingFunction {\n  private readonly modelName: string;\n  private _client: SIEClient | undefined;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  constructor(options: SIESparseEmbeddingFunctionOptions = {}) {\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"BAAI/bge-m3\",\n      gpu,\n      timeout = 180_000,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Encode query texts to sparse vectors.\n   *\n   * @param texts - List of query texts to encode.\n   * @returns Tuple of [indices_list, values_list].\n   */\n  async encodeQueries(texts: string[]): Promise<[number[][], number[][]]> {\n    if (texts.length === 0) {\n      return [[], []];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"sparse\"],\n      isQuery: true,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n\n    const indicesList: number[][] = [];\n    const valuesList: number[][] = [];\n\n    for (const result of results as EncodeResult[]) {\n      const sparse = this.extractSparse(result);\n      indicesList.push(sparse.indices);\n      valuesList.push(sparse.values);\n    }\n\n    return [indicesList, valuesList];\n  }\n\n  /**\n   * Encode document texts to sparse vectors.\n   *\n   * @param texts - List of document texts to encode.\n   * @returns Tuple of [indices_list, values_list].\n   */\n  async encodeDocuments(texts: string[]): Promise<[number[][], number[][]]> {\n    if (texts.length === 0) {\n      return [[], []];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"sparse\"],\n      isQuery: false,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n\n    const indicesList: number[][] = [];\n    const valuesList: number[][] = [];\n\n    for (const result of results as EncodeResult[]) {\n      const sparse = this.extractSparse(result);\n      indicesList.push(sparse.indices);\n      valuesList.push(sparse.values);\n    }\n\n    return [indicesList, valuesList];\n  }\n\n  /**\n   * Extract sparse embedding from encode result.\n   */\n  private extractSparse(result: EncodeResult): { indices: number[]; values: number[] } {\n    const sparse = result.sparse;\n    if (!sparse) {\n      return { indices: [], values: [] };\n    }\n\n    return {\n      indices: toNumberArray(sparse.indices),\n      values: toNumberArray(sparse.values),\n    };\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client) {\n      await this._client.close();\n    }\n  }\n}\n"],"mappings":";AAoBA;AAAA,EAIE;AAAA,EAEA;AAAA,OACK;AACP,SAAS,qBAAqB;AAgFvB,IAAM,eAAN,cAA2B,cAAc;AAAA,EACrC;AAAA,EACQ;AAAA,EACA;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA;AAAA;AAAA;AAAA;AAAA,EAMR,oBAAoB,OAAO,UAAyC;AAC3E,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC;AAAA,IACV;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,OAAO;AAAA,MACrB,aAAa,KAAK;AAAA,MAClB,aAAa,KAAK;AAAA,MAClB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AACvE,WAAQ,QAA2B,IAAI,CAAC,WAAW,KAAK,aAAa,MAAM,CAAC;AAAA,EAC9E;AAAA,EAEA,YAAY,UAA+B,CAAC,GAAG;AAC7C,UAAM;AAEN,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA,UAAU;AAAA,MACV,iBAAiB;AAAA,IACnB,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,cAAc;AACnB,SAAK,cAAc;AACnB,SAAK,UAAU;AACf,SAAK,iBAAiB;AAEtB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAI,UAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,iBAAiB,MAAiC;AACtD,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,OAAO;AAAA,MACrB,aAAa,KAAK;AAAA,MAClB,aAAa,KAAK;AAAA,MAClB,SAAS;AAAA,IACX;AAEA,UAAM,SAAS,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,EAAE,KAAK,GAAG,OAAO;AACzE,WAAO,KAAK,aAAa,MAAsB;AAAA,EACjD;AAAA;AAAA;AAAA;AAAA,EAKQ,aAAa,QAAgC;AACnD,UAAM,QAAQ,OAAO;AACrB,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,uCAAuC;AAAA,IACzD;AACA,WAAO,cAAc,KAAK;AAAA,EAC5B;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,SAAS;AAChB,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;AAsDO,IAAM,6BAAN,MAAiC;AAAA,EACrB;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA,EAEjB,YAAY,UAA6C,CAAC,GAAG;AAC3D,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA,UAAU;AAAA,IACZ,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAI,UAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,cAAc,OAAoD;AACtE,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,IAChB;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,QAAQ;AAAA,MACtB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AAEvE,UAAM,cAA0B,CAAC;AACjC,UAAM,aAAyB,CAAC;AAEhC,eAAW,UAAU,SAA2B;AAC9C,YAAM,SAAS,KAAK,cAAc,MAAM;AACxC,kBAAY,KAAK,OAAO,OAAO;AAC/B,iBAAW,KAAK,OAAO,MAAM;AAAA,IAC/B;AAEA,WAAO,CAAC,aAAa,UAAU;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,gBAAgB,OAAoD;AACxE,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,IAChB;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,QAAQ;AAAA,MACtB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AAEvE,UAAM,cAA0B,CAAC;AACjC,UAAM,aAAyB,CAAC;AAEhC,eAAW,UAAU,SAA2B;AAC9C,YAAM,SAAS,KAAK,cAAc,MAAM;AACxC,kBAAY,KAAK,OAAO,OAAO;AAC/B,iBAAW,KAAK,OAAO,MAAM;AAAA,IAC/B;AAEA,WAAO,CAAC,aAAa,UAAU;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA,EAKQ,cAAc,QAA+D;AACnF,UAAM,SAAS,OAAO;AACtB,QAAI,CAAC,QAAQ;AACX,aAAO,EAAE,SAAS,CAAC,GAAG,QAAQ,CAAC,EAAE;AAAA,IACnC;AAEA,WAAO;AAAA,MACL,SAAS,cAAc,OAAO,OAAO;AAAA,MACrC,QAAQ,cAAc,OAAO,MAAM;AAAA,IACrC;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,SAAS;AAChB,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;","names":[]}
+{"version":3,"sources":["../src/index.ts","../src/rerankers.ts","../src/extractors.ts"],"sourcesContent":["/**\n * SIE embeddings integration for LlamaIndex.TS\n *\n * Provides embedding generation using SIE's encode endpoint:\n * - SIEEmbedding: Dense embeddings implementing BaseEmbedding\n * - SIESparseEmbeddingFunction: Sparse embeddings for hybrid search\n * - SIENodePostprocessor: Cross-encoder reranking for query pipelines\n * - createSIEExtractorTool: Entity extraction tool for agents\n *\n * @example\n * ```typescript\n * import { Settings } from \"llamaindex\";\n * import { SIEEmbedding } from \"@superlinked/sie-llamaindex\";\n *\n * // Set as default embedding model\n * Settings.embedModel = new SIEEmbedding({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n * ```\n */\n\nimport {\n  type DType,\n  type EncodeOptions,\n  type EncodeResult,\n  SIEClient,\n  type SIEClientOptions,\n  denseEmbedding,\n  sparseEmbedding,\n} from \"@superlinked/sie-sdk\";\nimport { BaseEmbedding } from \"llamaindex\";\n\n/**\n * Configuration options for SIEEmbedding.\n */\nexport interface SIEEmbeddingOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Model name/ID to use for encoding.\n   * @default \"BAAI/bge-m3\"\n   */\n  modelName?: string;\n\n  /**\n   * Optional pre-configured SIEClient instance.\n   */\n  client?: SIEClient;\n\n  /**\n   * Optional instruction prefix for embedding (model-dependent).\n   */\n  instruction?: string;\n\n  /**\n   * Output dtype: \"float32\" (default), \"float16\", \"int8\", \"binary\".\n   */\n  outputDtype?: DType;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n\n  /**\n   * Batch size for embedding multiple texts.\n   * @default 10\n   */\n  embedBatchSize?: number;\n}\n\n/**\n * LlamaIndex BaseEmbedding implementation using SIE.\n *\n * Wraps SIEClient.encode() to implement the LlamaIndex BaseEmbedding interface.\n *\n * @example\n * ```typescript\n * import { Settings, VectorStoreIndex, Document } from \"llamaindex\";\n * import { SIEEmbedding } from \"@superlinked/sie-llamaindex\";\n *\n * // Set as default embedding model\n * Settings.embedModel = new SIEEmbedding({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n *\n * // Create index with documents\n * const index = await VectorStoreIndex.fromDocuments([\n *   new Document({ text: \"Hello world\" }),\n * ]);\n *\n * // With GPU routing for multi-GPU clusters\n * const embedModel = new SIEEmbedding({\n *   baseUrl: \"https://cluster.example.com\",\n *   modelName: \"BAAI/bge-m3\",\n *   gpu: \"a100-80gb\",\n * });\n * ```\n */\nexport class SIEEmbedding extends BaseEmbedding {\n  readonly modelName: string;\n  private readonly instruction?: string;\n  private readonly outputDtype?: DType;\n  private _client: SIEClient | undefined;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  /**\n   * Get embeddings for multiple text strings (documents).\n   * This is a property (arrow function) to match BaseEmbedding interface.\n   */\n  override getTextEmbeddings = async (texts: string[]): Promise<number[][]> => {\n    if (texts.length === 0) {\n      return [];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"dense\"],\n      instruction: this.instruction,\n      outputDtype: this.outputDtype,\n      isQuery: false,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n    return (results as EncodeResult[]).map((result) => denseEmbedding(result));\n  };\n\n  constructor(options: SIEEmbeddingOptions = {}) {\n    super();\n\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"BAAI/bge-m3\",\n      client,\n      instruction,\n      outputDtype,\n      gpu,\n      timeout = 180_000,\n      embedBatchSize = 10,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.instruction = instruction;\n    this.outputDtype = outputDtype;\n    this._client = client;\n    this.embedBatchSize = embedBatchSize;\n\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Get embedding for a single text string (document).\n   *\n   * @param text - Text to embed.\n   * @returns Embedding vector as array of numbers.\n   */\n  async getTextEmbedding(text: string): Promise<number[]> {\n    const options: EncodeOptions = {\n      outputTypes: [\"dense\"],\n      instruction: this.instruction,\n      outputDtype: this.outputDtype,\n      isQuery: false,\n    };\n\n    const result = await this.client.encode(this.modelName, { text }, options);\n    return denseEmbedding(result as EncodeResult);\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client) {\n      await this._client.close();\n    }\n  }\n}\n\n/**\n * Configuration options for SIESparseEmbeddingFunction.\n */\nexport interface SIESparseEmbeddingFunctionOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Model name/ID to use for encoding. Must support sparse output.\n   * @default \"BAAI/bge-m3\"\n   */\n  modelName?: string;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n}\n\n/**\n * Sparse embedding function for LlamaIndex hybrid search.\n *\n * Compatible with LlamaIndex vector stores that support hybrid search,\n * such as QdrantVectorStore with enableHybrid: true.\n *\n * @example\n * ```typescript\n * import { QdrantVectorStore } from \"llamaindex\";\n * import { SIESparseEmbeddingFunction } from \"@superlinked/sie-llamaindex\";\n *\n * const sparseEmbedFn = new SIESparseEmbeddingFunction({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"BAAI/bge-m3\",\n * });\n *\n * const vectorStore = new QdrantVectorStore({\n *   client: qdrantClient,\n *   collectionName: \"hybrid_docs\",\n *   enableHybrid: true,\n *   sparseEmbeddingFunction: sparseEmbedFn,\n * });\n * ```\n */\nexport class SIESparseEmbeddingFunction {\n  private readonly modelName: string;\n  private _client: SIEClient | undefined;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  constructor(options: SIESparseEmbeddingFunctionOptions = {}) {\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"BAAI/bge-m3\",\n      gpu,\n      timeout = 180_000,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Encode query texts to sparse vectors.\n   *\n   * @param texts - List of query texts to encode.\n   * @returns Tuple of [indices_list, values_list].\n   */\n  async encodeQueries(texts: string[]): Promise<[number[][], number[][]]> {\n    if (texts.length === 0) {\n      return [[], []];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"sparse\"],\n      isQuery: true,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n\n    const indicesList: number[][] = [];\n    const valuesList: number[][] = [];\n\n    for (const result of results as EncodeResult[]) {\n      const sparse = sparseEmbedding(result);\n      indicesList.push(sparse.indices);\n      valuesList.push(sparse.values);\n    }\n\n    return [indicesList, valuesList];\n  }\n\n  /**\n   * Encode document texts to sparse vectors.\n   *\n   * @param texts - List of document texts to encode.\n   * @returns Tuple of [indices_list, values_list].\n   */\n  async encodeDocuments(texts: string[]): Promise<[number[][], number[][]]> {\n    if (texts.length === 0) {\n      return [[], []];\n    }\n\n    const items = texts.map((text) => ({ text }));\n    const options: EncodeOptions = {\n      outputTypes: [\"sparse\"],\n      isQuery: false,\n    };\n\n    const results = await this.client.encode(this.modelName, items, options);\n\n    const indicesList: number[][] = [];\n    const valuesList: number[][] = [];\n\n    for (const result of results as EncodeResult[]) {\n      const sparse = sparseEmbedding(result);\n      indicesList.push(sparse.indices);\n      valuesList.push(sparse.values);\n    }\n\n    return [indicesList, valuesList];\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client) {\n      await this._client.close();\n    }\n  }\n}\n\nexport { SIENodePostprocessor, type SIENodePostprocessorOptions } from \"./rerankers.js\";\nexport { createSIEExtractorTool, type SIEExtractorToolOptions } from \"./extractors.js\";\n","/**\n * SIE reranker integration for LlamaIndex.TS\n *\n * Provides node reranking using SIE's score endpoint:\n * - SIENodePostprocessor: Cross-encoder reranking implementing BaseNodePostprocessor\n *\n * @example\n * ```typescript\n * import { SIENodePostprocessor } from \"@superlinked/sie-llamaindex\";\n *\n * const reranker = new SIENodePostprocessor({\n *   baseUrl: \"http://localhost:8080\",\n *   modelName: \"jinaai/jina-reranker-v2-base-multilingual\",\n *   topN: 3,\n * });\n *\n * const reranked = await reranker.postprocessNodes(nodes, \"search query\");\n * ```\n */\n\nimport { SIEClient, type SIEClientOptions } from \"@superlinked/sie-sdk\";\nimport { MetadataMode } from \"llamaindex\";\nimport type { BaseNodePostprocessor, MessageContent, NodeWithScore } from \"llamaindex\";\n\n/**\n * Configuration options for SIENodePostprocessor.\n */\nexport interface SIENodePostprocessorOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Reranker model name/ID.\n   * @default \"jinaai/jina-reranker-v2-base-multilingual\"\n   */\n  modelName?: string;\n\n  /**\n   * Optional pre-configured SIEClient instance.\n   * If provided, baseUrl and other connection options are ignored.\n   */\n  client?: SIEClient;\n\n  /**\n   * Number of top nodes to return. If undefined, returns all nodes.\n   */\n  topN?: number;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n}\n\n/**\n * Extract a plain query string from LlamaIndex's MessageContent type.\n *\n * MessageContent is `string | MessageContentDetail[]`. For reranking\n * we need a plain string.\n */\nfunction extractQueryString(query: MessageContent): string {\n  if (typeof query === \"string\") {\n    return query;\n  }\n  // MessageContentDetail[] — concatenate text parts\n  return query\n    .filter(\n      (part): part is { type: \"text\"; text: string } => \"type\" in part && part.type === \"text\",\n    )\n    .map((part) => part.text)\n    .join(\" \");\n}\n\n/**\n * LlamaIndex node postprocessor using SIE's reranking.\n *\n * Wraps SIEClient.score() to implement the BaseNodePostprocessor interface.\n *\n * @example\n * ```typescript\n * import { VectorStoreIndex } from \"llamaindex\";\n * import { SIENodePostprocessor } from \"@superlinked/sie-llamaindex\";\n *\n * const reranker = new SIENodePostprocessor({\n *   modelName: \"jinaai/jina-reranker-v2-base-multilingual\",\n *   topN: 3,\n * });\n *\n * const queryEngine = index.asQueryEngine({\n *   nodePostprocessors: [reranker],\n * });\n *\n * const response = await queryEngine.query({ query: \"What is the topic?\" });\n * ```\n */\nexport class SIENodePostprocessor implements BaseNodePostprocessor {\n  readonly modelName: string;\n  private readonly topN?: number;\n  private _client: SIEClient | undefined;\n  private readonly _ownsClient: boolean;\n  private readonly baseUrl: string;\n  private readonly clientOptions: SIEClientOptions;\n\n  constructor(options: SIENodePostprocessorOptions = {}) {\n    const {\n      baseUrl = \"http://localhost:8080\",\n      modelName = \"jinaai/jina-reranker-v2-base-multilingual\",\n      client,\n      topN,\n      gpu,\n      timeout = 180_000,\n    } = options;\n\n    this.baseUrl = baseUrl;\n    this.modelName = modelName;\n    this.topN = topN;\n    this._client = client;\n    this._ownsClient = !client;\n\n    this.clientOptions = {\n      timeout,\n      gpu,\n    };\n  }\n\n  /**\n   * Get or create the SIEClient.\n   */\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  /**\n   * Rerank nodes by relevance to query.\n   *\n   * @param nodes - Nodes with scores to rerank.\n   * @param query - Optional query string or MessageContent.\n   * @returns Reranked nodes with updated scores, sorted by relevance descending.\n   */\n  async postprocessNodes(nodes: NodeWithScore[], query?: MessageContent): Promise<NodeWithScore[]> {\n    if (nodes.length === 0 || query === undefined) {\n      return nodes;\n    }\n\n    const queryText = extractQueryString(query);\n    if (!queryText) {\n      return nodes;\n    }\n\n    const queryItem = { text: queryText };\n    const docItems = nodes.map((n) => ({ text: n.node.getContent(MetadataMode.NONE) }));\n\n    const result = await this.client.score(this.modelName, queryItem, docItems);\n\n    // Map score entries back to NodeWithScore with updated scores.\n    // ScoreResult.scores are already sorted by score descending.\n    const reranked: NodeWithScore[] = [];\n    for (const entry of result.scores) {\n      const idx = Number.parseInt(entry.itemId, 10);\n      const original = nodes[idx];\n      if (original) {\n        reranked.push({\n          node: original.node,\n          score: entry.score,\n        });\n      }\n    }\n\n    if (this.topN !== undefined) {\n      return reranked.slice(0, this.topN);\n    }\n    return reranked;\n  }\n\n  /**\n   * Close the underlying client connection.\n   */\n  async close(): Promise<void> {\n    if (this._client && this._ownsClient) {\n      await this._client.close();\n    }\n  }\n}\n","/**\n * SIE extraction tool for LlamaIndex.TS\n *\n * Provides extraction using SIE's extract endpoint:\n * - createSIEExtractorTool: Factory that creates a FunctionTool for extraction\n *\n * Returns entities, relations, classifications, and detected objects.\n *\n * @example\n * ```typescript\n * import { createSIEExtractorTool } from \"@superlinked/sie-llamaindex\";\n *\n * const extractor = createSIEExtractorTool({\n *   modelName: \"urchade/gliner_multi-v2.1\",\n *   labels: [\"person\", \"organization\", \"location\"],\n * });\n *\n * const result = await extractor.call({ text: \"John Smith works at Acme Corp\" });\n * const parsed = JSON.parse(result);\n * console.log(parsed.entities);\n * ```\n */\n\nimport {\n  type ExtractOptions,\n  type ExtractResult,\n  SIEClient,\n  type SIEClientOptions,\n} from \"@superlinked/sie-sdk\";\nimport { FunctionTool } from \"llamaindex\";\n\n/**\n * Configuration options for createSIEExtractorTool.\n */\nexport interface SIEExtractorToolOptions {\n  /**\n   * URL of the SIE server.\n   * @default \"http://localhost:8080\"\n   */\n  baseUrl?: string;\n\n  /**\n   * Extraction model name/ID.\n   * @default \"urchade/gliner_multi-v2.1\"\n   */\n  modelName?: string;\n\n  /**\n   * Labels to extract.\n   * @default [\"person\", \"organization\", \"location\"]\n   */\n  labels?: string[];\n\n  /**\n   * Minimum confidence threshold (0-1).\n   */\n  threshold?: number;\n\n  /**\n   * Target GPU type for routing (e.g., \"l4\", \"a100-80gb\").\n   */\n  gpu?: string;\n\n  /**\n   * Request timeout in milliseconds.\n   * @default 180000 (3 minutes)\n   */\n  timeout?: number;\n\n  /**\n   * Tool name for the agent.\n   * @default \"sie_extract\"\n   */\n  name?: string;\n\n  /**\n   * Tool description for the agent.\n   */\n  description?: string;\n}\n\n/**\n * Internal class to hold SIE extractor state.\n */\nclass _SIEExtractor {\n  private readonly baseUrl: string;\n  private readonly modelName: string;\n  private readonly labels: string[];\n  private readonly threshold?: number;\n  private readonly clientOptions: SIEClientOptions;\n  private _client: SIEClient | undefined;\n\n  constructor(options: SIEExtractorToolOptions) {\n    this.baseUrl = options.baseUrl ?? \"http://localhost:8080\";\n    this.modelName = options.modelName ?? \"urchade/gliner_multi-v2.1\";\n    this.labels = options.labels ?? [\"person\", \"organization\", \"location\"];\n    this.threshold = options.threshold;\n    this.clientOptions = {\n      timeout: options.timeout ?? 180_000,\n      gpu: options.gpu,\n    };\n  }\n\n  private get client(): SIEClient {\n    if (!this._client) {\n      this._client = new SIEClient(this.baseUrl, this.clientOptions);\n    }\n    return this._client;\n  }\n\n  async extract(text: string): Promise<string> {\n    const extractOptions: ExtractOptions = {\n      labels: this.labels,\n    };\n    if (this.threshold !== undefined) {\n      extractOptions.threshold = this.threshold;\n    }\n\n    const result: ExtractResult = await this.client.extract(this.modelName, { text }, extractOptions);\n\n    return JSON.stringify({\n      entities: result.entities.map((e) => ({\n        text: e.text,\n        label: e.label,\n        score: e.score,\n        ...(e.start !== undefined && { start: e.start }),\n        ...(e.end !== undefined && { end: e.end }),\n      })),\n      relations: result.relations.map((r) => ({\n        head: r.head,\n        tail: r.tail,\n        relation: r.relation,\n        score: r.score,\n      })),\n      classifications: result.classifications.map((c) => ({\n        label: c.label,\n        score: c.score,\n      })),\n      objects: result.objects.map((o) => ({\n        label: o.label,\n        score: o.score,\n        bbox: o.bbox,\n      })),\n    });\n  }\n}\n\n/**\n * Create a LlamaIndex FunctionTool for extraction.\n *\n * Creates a tool that wraps SIE's extract endpoint for use with\n * LlamaIndex agents and workflows. Returns JSON with entities,\n * relations, classifications, and detected objects.\n *\n * @example\n * ```typescript\n * import { OpenAI, ReActAgent } from \"llamaindex\";\n * import { createSIEExtractorTool } from \"@superlinked/sie-llamaindex\";\n *\n * const extractor = createSIEExtractorTool({\n *   modelName: \"urchade/gliner_multi-v2.1\",\n *   labels: [\"person\", \"organization\", \"location\"],\n * });\n *\n * const agent = new ReActAgent({\n *   tools: [extractor],\n *   llm: new OpenAI(),\n * });\n *\n * const response = await agent.chat({\n *   message: \"Extract entities from: John works at Google in NYC\",\n * });\n * ```\n *\n * @param options - Configuration for the extractor tool.\n * @returns FunctionTool wrapping SIE extraction.\n */\nexport function createSIEExtractorTool(options: SIEExtractorToolOptions = {}) {\n  const labels = options.labels ?? [\"person\", \"organization\", \"location\"];\n  const name = options.name ?? \"sie_extract\";\n  const description =\n    options.description ??\n    `Extract structured information from text. Finds entities of types: ${labels.join(\", \")}. Returns entities, relations, classifications, and detected objects.`;\n\n  const extractor = new _SIEExtractor(options);\n\n  async function extract(input: { text: string }): Promise<string> {\n    return extractor.extract(input.text);\n  }\n\n  return FunctionTool.from(extract, {\n    name,\n    description,\n  });\n}\n"],"mappings":";AAsBA;AAAA,EAIE,aAAAA;AAAA,EAEA;AAAA,EACA;AAAA,OACK;AACP,SAAS,qBAAqB;;;ACX9B,SAAS,iBAAwC;AACjD,SAAS,oBAAoB;AAgD7B,SAAS,mBAAmB,OAA+B;AACzD,MAAI,OAAO,UAAU,UAAU;AAC7B,WAAO;AAAA,EACT;AAEA,SAAO,MACJ;AAAA,IACC,CAAC,SAAiD,UAAU,QAAQ,KAAK,SAAS;AAAA,EACpF,EACC,IAAI,CAAC,SAAS,KAAK,IAAI,EACvB,KAAK,GAAG;AACb;AAwBO,IAAM,uBAAN,MAA4D;AAAA,EACxD;AAAA,EACQ;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA,EACA;AAAA,EAEjB,YAAY,UAAuC,CAAC,GAAG;AACrD,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA;AAAA,MACA;AAAA,MACA,UAAU;AAAA,IACZ,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,OAAO;AACZ,SAAK,UAAU;AACf,SAAK,cAAc,CAAC;AAEpB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAI,UAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,MAAM,iBAAiB,OAAwB,OAAkD;AAC/F,QAAI,MAAM,WAAW,KAAK,UAAU,QAAW;AAC7C,aAAO;AAAA,IACT;AAEA,UAAM,YAAY,mBAAmB,KAAK;AAC1C,QAAI,CAAC,WAAW;AACd,aAAO;AAAA,IACT;AAEA,UAAM,YAAY,EAAE,MAAM,UAAU;AACpC,UAAM,WAAW,MAAM,IAAI,CAAC,OAAO,EAAE,MAAM,EAAE,KAAK,WAAW,aAAa,IAAI,EAAE,EAAE;AAElF,UAAM,SAAS,MAAM,KAAK,OAAO,MAAM,KAAK,WAAW,WAAW,QAAQ;AAI1E,UAAM,WAA4B,CAAC;AACnC,eAAW,SAAS,OAAO,QAAQ;AACjC,YAAM,MAAM,OAAO,SAAS,MAAM,QAAQ,EAAE;AAC5C,YAAM,WAAW,MAAM,GAAG;AAC1B,UAAI,UAAU;AACZ,iBAAS,KAAK;AAAA,UACZ,MAAM,SAAS;AAAA,UACf,OAAO,MAAM;AAAA,QACf,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,KAAK,SAAS,QAAW;AAC3B,aAAO,SAAS,MAAM,GAAG,KAAK,IAAI;AAAA,IACpC;AACA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,WAAW,KAAK,aAAa;AACpC,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;;;AC3KA;AAAA,EAGE,aAAAC;AAAA,OAEK;AACP,SAAS,oBAAoB;AAuD7B,IAAM,gBAAN,MAAoB;AAAA,EACD;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACT;AAAA,EAER,YAAY,SAAkC;AAC5C,SAAK,UAAU,QAAQ,WAAW;AAClC,SAAK,YAAY,QAAQ,aAAa;AACtC,SAAK,SAAS,QAAQ,UAAU,CAAC,UAAU,gBAAgB,UAAU;AACrE,SAAK,YAAY,QAAQ;AACzB,SAAK,gBAAgB;AAAA,MACnB,SAAS,QAAQ,WAAW;AAAA,MAC5B,KAAK,QAAQ;AAAA,IACf;AAAA,EACF;AAAA,EAEA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAIA,WAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,MAAM,QAAQ,MAA+B;AAC3C,UAAM,iBAAiC;AAAA,MACrC,QAAQ,KAAK;AAAA,IACf;AACA,QAAI,KAAK,cAAc,QAAW;AAChC,qBAAe,YAAY,KAAK;AAAA,IAClC;AAEA,UAAM,SAAwB,MAAM,KAAK,OAAO,QAAQ,KAAK,WAAW,EAAE,KAAK,GAAG,cAAc;AAEhG,WAAO,KAAK,UAAU;AAAA,MACpB,UAAU,OAAO,SAAS,IAAI,CAAC,OAAO;AAAA,QACpC,MAAM,EAAE;AAAA,QACR,OAAO,EAAE;AAAA,QACT,OAAO,EAAE;AAAA,QACT,GAAI,EAAE,UAAU,UAAa,EAAE,OAAO,EAAE,MAAM;AAAA,QAC9C,GAAI,EAAE,QAAQ,UAAa,EAAE,KAAK,EAAE,IAAI;AAAA,MAC1C,EAAE;AAAA,MACF,WAAW,OAAO,UAAU,IAAI,CAAC,OAAO;AAAA,QACtC,MAAM,EAAE;AAAA,QACR,MAAM,EAAE;AAAA,QACR,UAAU,EAAE;AAAA,QACZ,OAAO,EAAE;AAAA,MACX,EAAE;AAAA,MACF,iBAAiB,OAAO,gBAAgB,IAAI,CAAC,OAAO;AAAA,QAClD,OAAO,EAAE;AAAA,QACT,OAAO,EAAE;AAAA,MACX,EAAE;AAAA,MACF,SAAS,OAAO,QAAQ,IAAI,CAAC,OAAO;AAAA,QAClC,OAAO,EAAE;AAAA,QACT,OAAO,EAAE;AAAA,QACT,MAAM,EAAE;AAAA,MACV,EAAE;AAAA,IACJ,CAAC;AAAA,EACH;AACF;AAgCO,SAAS,uBAAuB,UAAmC,CAAC,GAAG;AAC5E,QAAM,SAAS,QAAQ,UAAU,CAAC,UAAU,gBAAgB,UAAU;AACtE,QAAM,OAAO,QAAQ,QAAQ;AAC7B,QAAM,cACJ,QAAQ,eACR,sEAAsE,OAAO,KAAK,IAAI,CAAC;AAEzF,QAAM,YAAY,IAAI,cAAc,OAAO;AAE3C,iBAAe,QAAQ,OAA0C;AAC/D,WAAO,UAAU,QAAQ,MAAM,IAAI;AAAA,EACrC;AAEA,SAAO,aAAa,KAAK,SAAS;AAAA,IAChC;AAAA,IACA;AAAA,EACF,CAAC;AACH;;;AFnFO,IAAM,eAAN,cAA2B,cAAc;AAAA,EACrC;AAAA,EACQ;AAAA,EACA;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA;AAAA;AAAA;AAAA;AAAA,EAMR,oBAAoB,OAAO,UAAyC;AAC3E,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC;AAAA,IACV;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,OAAO;AAAA,MACrB,aAAa,KAAK;AAAA,MAClB,aAAa,KAAK;AAAA,MAClB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AACvE,WAAQ,QAA2B,IAAI,CAAC,WAAW,eAAe,MAAM,CAAC;AAAA,EAC3E;AAAA,EAEA,YAAY,UAA+B,CAAC,GAAG;AAC7C,UAAM;AAEN,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA,UAAU;AAAA,MACV,iBAAiB;AAAA,IACnB,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,cAAc;AACnB,SAAK,cAAc;AACnB,SAAK,UAAU;AACf,SAAK,iBAAiB;AAEtB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAIC,WAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,iBAAiB,MAAiC;AACtD,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,OAAO;AAAA,MACrB,aAAa,KAAK;AAAA,MAClB,aAAa,KAAK;AAAA,MAClB,SAAS;AAAA,IACX;AAEA,UAAM,SAAS,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,EAAE,KAAK,GAAG,OAAO;AACzE,WAAO,eAAe,MAAsB;AAAA,EAC9C;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,SAAS;AAChB,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;AAsDO,IAAM,6BAAN,MAAiC;AAAA,EACrB;AAAA,EACT;AAAA,EACS;AAAA,EACA;AAAA,EAEjB,YAAY,UAA6C,CAAC,GAAG;AAC3D,UAAM;AAAA,MACJ,UAAU;AAAA,MACV,YAAY;AAAA,MACZ;AAAA,MACA,UAAU;AAAA,IACZ,IAAI;AAEJ,SAAK,UAAU;AACf,SAAK,YAAY;AACjB,SAAK,gBAAgB;AAAA,MACnB;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,IAAY,SAAoB;AAC9B,QAAI,CAAC,KAAK,SAAS;AACjB,WAAK,UAAU,IAAIA,WAAU,KAAK,SAAS,KAAK,aAAa;AAAA,IAC/D;AACA,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,cAAc,OAAoD;AACtE,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,IAChB;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,QAAQ;AAAA,MACtB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AAEvE,UAAM,cAA0B,CAAC;AACjC,UAAM,aAAyB,CAAC;AAEhC,eAAW,UAAU,SAA2B;AAC9C,YAAM,SAAS,gBAAgB,MAAM;AACrC,kBAAY,KAAK,OAAO,OAAO;AAC/B,iBAAW,KAAK,OAAO,MAAM;AAAA,IAC/B;AAEA,WAAO,CAAC,aAAa,UAAU;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,MAAM,gBAAgB,OAAoD;AACxE,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,IAChB;AAEA,UAAM,QAAQ,MAAM,IAAI,CAAC,UAAU,EAAE,KAAK,EAAE;AAC5C,UAAM,UAAyB;AAAA,MAC7B,aAAa,CAAC,QAAQ;AAAA,MACtB,SAAS;AAAA,IACX;AAEA,UAAM,UAAU,MAAM,KAAK,OAAO,OAAO,KAAK,WAAW,OAAO,OAAO;AAEvE,UAAM,cAA0B,CAAC;AACjC,UAAM,aAAyB,CAAC;AAEhC,eAAW,UAAU,SAA2B;AAC9C,YAAM,SAAS,gBAAgB,MAAM;AACrC,kBAAY,KAAK,OAAO,OAAO;AAC/B,iBAAW,KAAK,OAAO,MAAM;AAAA,IAC/B;AAEA,WAAO,CAAC,aAAa,UAAU;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,QAAuB;AAC3B,QAAI,KAAK,SAAS;AAChB,YAAM,KAAK,QAAQ,MAAM;AAAA,IAC3B;AAAA,EACF;AACF;","names":["SIEClient","SIEClient","SIEClient"]}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@superlinked/sie-llamaindex",
-  "version": "0.1.8",
+  "version": "0.1.10",
   "description": "SIE embeddings integration for LlamaIndex.TS",
   "type": "module",
   "main": "./dist/index.cjs",
@@ -17,7 +17,7 @@
     "dist"
   ],
   "dependencies": {
-    "@superlinked/sie-sdk": "0.1.8"
+    "@superlinked/sie-sdk": "0.1.10"
   },
   "peerDependencies": {
     "llamaindex": ">=0.6.0"