@atom8n/n8n-nodes-langchain 2.2.0

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/insertOperation.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../../nodes/vector_store/shared/createVectorStoreNode/operations/insertOperation.ts"],"sourcesContent":["import type { Document } from '@langchain/core/documents';\nimport type { Embeddings } from '@langchain/core/embeddings';\nimport type { VectorStore } from '@langchain/core/vectorstores';\nimport type { IExecuteFunctions, INodeExecutionData } from 'n8n-workflow';\nimport { NodeConnectionTypes } from 'n8n-workflow';\n\nimport { logAiEvent } from '@utils/helpers';\nimport type { N8nBinaryLoader } from '@utils/N8nBinaryLoader';\nimport type { N8nJsonLoader } from '@utils/N8nJsonLoader';\n\nimport { processDocument } from '../../processDocuments';\nimport type { VectorStoreNodeConstructorArgs } from '../types';\n\n/**\n * Handles the 'insert' operation mode\n * Inserts documents from the input into the vector store\n */\nexport async function handleInsertOperation<T extends VectorStore = VectorStore>(\n\tcontext: IExecuteFunctions,\n\targs: VectorStoreNodeConstructorArgs<T>,\n\tembeddings: Embeddings,\n): Promise<INodeExecutionData[]> {\n\tconst nodeVersion = context.getNode().typeVersion;\n\t// Get the input items and document data\n\tconst items = context.getInputData();\n\tconst documentInput = (await context.getInputConnectionData(NodeConnectionTypes.AiDocument, 0)) as\n\t\t| N8nJsonLoader\n\t\t| N8nBinaryLoader\n\t\t| Array<Document<Record<string, unknown>>>;\n\n\tconst resultData: INodeExecutionData[] = [];\n\tconst documentsForEmbedding: Array<Document<Record<string, unknown>>> = [];\n\n\t// Process each input item\n\tfor (let itemIndex = 0; itemIndex < items.length; itemIndex++) {\n\t\t// Check if execution is being cancelled\n\t\tif (context.getExecutionCancelSignal()?.aborted) {\n\t\t\tbreak;\n\t\t}\n\n\t\tconst itemData = items[itemIndex];\n\n\t\t// Process the document from the input\n\t\tconst processedDocuments = await processDocument(documentInput, itemData, itemIndex);\n\n\t\t// Add the serialized documents to the result\n\t\tresultData.push(...processedDocuments.serializedDocuments);\n\n\t\t// Add the processed documents to the documents to embedd\n\t\tdocumentsForEmbedding.push(...processedDocuments.processedDocuments);\n\n\t\t// For the version 1, we run the populateVectorStore(embedding and insert) function for each item\n\t\tif (nodeVersion === 1) {\n\t\t\tawait args.populateVectorStore(\n\t\t\t\tcontext,\n\t\t\t\tembeddings,\n\t\t\t\tprocessedDocuments.processedDocuments,\n\t\t\t\titemIndex,\n\t\t\t);\n\t\t}\n\t\t// Log the AI event for analytics\n\t\tlogAiEvent(context, 'ai-vector-store-populated');\n\t}\n\n\t// For the version 1.1, we run the populateVectorStore in batches\n\tif (nodeVersion >= 1.1) {\n\t\tconst embeddingBatchSize =\n\t\t\t(context.getNodeParameter('embeddingBatchSize', 0, 200) as number) ?? 200;\n\n\t\t// Populate the vector store with the processed documents in batches\n\t\tfor (let i = 0; i < documentsForEmbedding.length; i += embeddingBatchSize) {\n\t\t\tconst nextBatch = documentsForEmbedding.slice(i, i + embeddingBatchSize);\n\t\t\tawait args.populateVectorStore(context, embeddings, nextBatch, 0);\n\t\t}\n\t}\n\n\treturn resultData;\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAIA,0BAAoC;AAEpC,qBAA2B;AAI3B,8BAAgC;AAOhC,eAAsB,sBACrB,SACA,MACA,YACgC;AAChC,QAAM,cAAc,QAAQ,QAAQ,EAAE;AAEtC,QAAM,QAAQ,QAAQ,aAAa;AACnC,QAAM,gBAAiB,MAAM,QAAQ,uBAAuB,wCAAoB,YAAY,CAAC;AAK7F,QAAM,aAAmC,CAAC;AAC1C,QAAM,wBAAkE,CAAC;AAGzE,WAAS,YAAY,GAAG,YAAY,MAAM,QAAQ,aAAa;AAE9D,QAAI,QAAQ,yBAAyB,GAAG,SAAS;AAChD;AAAA,IACD;AAEA,UAAM,WAAW,MAAM,SAAS;AAGhC,UAAM,qBAAqB,UAAM,yCAAgB,eAAe,UAAU,SAAS;AAGnF,eAAW,KAAK,GAAG,mBAAmB,mBAAmB;AAGzD,0BAAsB,KAAK,GAAG,mBAAmB,kBAAkB;AAGnE,QAAI,gBAAgB,GAAG;AACtB,YAAM,KAAK;AAAA,QACV;AAAA,QACA;AAAA,QACA,mBAAmB;AAAA,QACnB;AAAA,MACD;AAAA,IACD;AAEA,mCAAW,SAAS,2BAA2B;AAAA,EAChD;AAGA,MAAI,eAAe,KAAK;AACvB,UAAM,qBACJ,QAAQ,iBAAiB,sBAAsB,GAAG,GAAG,KAAgB;AAGvE,aAAS,IAAI,GAAG,IAAI,sBAAsB,QAAQ,KAAK,oBAAoB;AAC1E,YAAM,YAAY,sBAAsB,MAAM,GAAG,IAAI,kBAAkB;AACvE,YAAM,KAAK,oBAAoB,SAAS,YAAY,WAAW,CAAC;AAAA,IACjE;AAAA,EACD;AAEA,SAAO;AACR;","names":[]}

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/loadOperation.js

@@ -0,0 +1,80 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var loadOperation_exports = {};
+__export(loadOperation_exports, {
+  handleLoadOperation: () => handleLoadOperation
+});
+module.exports = __toCommonJS(loadOperation_exports);
+var import_n8n_workflow = require("n8n-workflow");
+var import_helpers = require("../../../../../utils/helpers");
+async function handleLoadOperation(context, args, embeddings, itemIndex) {
+  const filter = (0, import_helpers.getMetadataFiltersValues)(context, itemIndex);
+  const vectorStore = await args.getVectorStoreClient(
+    context,
+    // We'll pass filter to similaritySearchVectorWithScore instead of getVectorStoreClient
+    void 0,
+    embeddings,
+    itemIndex
+  );
+  try {
+    const prompt = context.getNodeParameter("prompt", itemIndex);
+    const topK = context.getNodeParameter("topK", itemIndex, 4);
+    const useReranker = context.getNodeParameter("useReranker", itemIndex, false);
+    const includeDocumentMetadata = context.getNodeParameter(
+      "includeDocumentMetadata",
+      itemIndex,
+      true
+    );
+    const embeddedPrompt = await embeddings.embedQuery(prompt);
+    let docs = await vectorStore.similaritySearchVectorWithScore(embeddedPrompt, topK, filter);
+    if (useReranker && docs.length > 0) {
+      const reranker = await context.getInputConnectionData(
+        import_n8n_workflow.NodeConnectionTypes.AiReranker,
+        0
+      );
+      const documents = docs.map(([doc]) => doc);
+      const rerankedDocuments = await reranker.compressDocuments(documents, prompt);
+      docs = rerankedDocuments.map((doc) => {
+        const { relevanceScore, ...metadata } = doc.metadata || {};
+        return [{ ...doc, metadata }, relevanceScore];
+      });
+    }
+    const serializedDocs = docs.map(([doc, score]) => {
+      const document = {
+        pageContent: doc.pageContent,
+        ...includeDocumentMetadata ? { metadata: doc.metadata } : {}
+      };
+      return {
+        json: { document, score },
+        pairedItem: {
+          item: itemIndex
+        }
+      };
+    });
+    (0, import_helpers.logAiEvent)(context, "ai-vector-store-searched", { query: prompt });
+    return serializedDocs;
+  } finally {
+    args.releaseVectorStoreClient?.(vectorStore);
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  handleLoadOperation
+});
+//# sourceMappingURL=loadOperation.js.map

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/loadOperation.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../../nodes/vector_store/shared/createVectorStoreNode/operations/loadOperation.ts"],"sourcesContent":["import type { Embeddings } from '@langchain/core/embeddings';\nimport type { BaseDocumentCompressor } from '@langchain/core/retrievers/document_compressors';\nimport type { VectorStore } from '@langchain/core/vectorstores';\nimport { NodeConnectionTypes, type IExecuteFunctions, type INodeExecutionData } from 'n8n-workflow';\n\nimport { getMetadataFiltersValues, logAiEvent } from '@utils/helpers';\n\nimport type { VectorStoreNodeConstructorArgs } from '../types';\n\n/**\n * Handles the 'load' operation mode\n * Searches the vector store for documents similar to a query\n */\nexport async function handleLoadOperation<T extends VectorStore = VectorStore>(\n\tcontext: IExecuteFunctions,\n\targs: VectorStoreNodeConstructorArgs<T>,\n\tembeddings: Embeddings,\n\titemIndex: number,\n): Promise<INodeExecutionData[]> {\n\tconst filter = getMetadataFiltersValues(context, itemIndex);\n\tconst vectorStore = await args.getVectorStoreClient(\n\t\tcontext,\n\t\t// We'll pass filter to similaritySearchVectorWithScore instead of getVectorStoreClient\n\t\tundefined,\n\t\tembeddings,\n\t\titemIndex,\n\t);\n\n\ttry {\n\t\t// Get the search parameters from the node\n\t\tconst prompt = context.getNodeParameter('prompt', itemIndex) as string;\n\t\tconst topK = context.getNodeParameter('topK', itemIndex, 4) as number;\n\t\tconst useReranker = context.getNodeParameter('useReranker', itemIndex, false) as boolean;\n\n\t\tconst includeDocumentMetadata = context.getNodeParameter(\n\t\t\t'includeDocumentMetadata',\n\t\t\titemIndex,\n\t\t\ttrue,\n\t\t) as boolean;\n\n\t\t// Embed the prompt to prepare for vector similarity search\n\t\tconst embeddedPrompt = await embeddings.embedQuery(prompt);\n\n\t\t// Get the most similar documents to the embedded prompt\n\t\tlet docs = await vectorStore.similaritySearchVectorWithScore(embeddedPrompt, topK, filter);\n\n\t\t// If reranker is used, rerank the documents\n\t\tif (useReranker && docs.length > 0) {\n\t\t\tconst reranker = (await context.getInputConnectionData(\n\t\t\t\tNodeConnectionTypes.AiReranker,\n\t\t\t\t0,\n\t\t\t)) as BaseDocumentCompressor;\n\t\t\tconst documents = docs.map(([doc]) => doc);\n\n\t\t\tconst rerankedDocuments = await reranker.compressDocuments(documents, prompt);\n\t\t\tdocs = rerankedDocuments.map((doc) => {\n\t\t\t\tconst { relevanceScore, ...metadata } = doc.metadata || {};\n\t\t\t\treturn [{ ...doc, metadata }, relevanceScore];\n\t\t\t});\n\t\t}\n\n\t\t// Format the documents for the output\n\t\tconst serializedDocs = docs.map(([doc, score]) => {\n\t\t\tconst document = {\n\t\t\t\tpageContent: doc.pageContent,\n\t\t\t\t...(includeDocumentMetadata ? { metadata: doc.metadata } : {}),\n\t\t\t};\n\n\t\t\treturn {\n\t\t\t\tjson: { document, score },\n\t\t\t\tpairedItem: {\n\t\t\t\t\titem: itemIndex,\n\t\t\t\t},\n\t\t\t};\n\t\t});\n\n\t\t// Log the AI event for analytics\n\t\tlogAiEvent(context, 'ai-vector-store-searched', { query: prompt });\n\n\t\treturn serializedDocs;\n\t} finally {\n\t\t// Release the vector store client if a release method was provided\n\t\targs.releaseVectorStoreClient?.(vectorStore);\n\t}\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,0BAAqF;AAErF,qBAAqD;AAQrD,eAAsB,oBACrB,SACA,MACA,YACA,WACgC;AAChC,QAAM,aAAS,yCAAyB,SAAS,SAAS;AAC1D,QAAM,cAAc,MAAM,KAAK;AAAA,IAC9B;AAAA;AAAA,IAEA;AAAA,IACA;AAAA,IACA;AAAA,EACD;AAEA,MAAI;AAEH,UAAM,SAAS,QAAQ,iBAAiB,UAAU,SAAS;AAC3D,UAAM,OAAO,QAAQ,iBAAiB,QAAQ,WAAW,CAAC;AAC1D,UAAM,cAAc,QAAQ,iBAAiB,eAAe,WAAW,KAAK;AAE5E,UAAM,0BAA0B,QAAQ;AAAA,MACvC;AAAA,MACA;AAAA,MACA;AAAA,IACD;AAGA,UAAM,iBAAiB,MAAM,WAAW,WAAW,MAAM;AAGzD,QAAI,OAAO,MAAM,YAAY,gCAAgC,gBAAgB,MAAM,MAAM;AAGzF,QAAI,eAAe,KAAK,SAAS,GAAG;AACnC,YAAM,WAAY,MAAM,QAAQ;AAAA,QAC/B,wCAAoB;AAAA,QACpB;AAAA,MACD;AACA,YAAM,YAAY,KAAK,IAAI,CAAC,CAAC,GAAG,MAAM,GAAG;AAEzC,YAAM,oBAAoB,MAAM,SAAS,kBAAkB,WAAW,MAAM;AAC5E,aAAO,kBAAkB,IAAI,CAAC,QAAQ;AACrC,cAAM,EAAE,gBAAgB,GAAG,SAAS,IAAI,IAAI,YAAY,CAAC;AACzD,eAAO,CAAC,EAAE,GAAG,KAAK,SAAS,GAAG,cAAc;AAAA,MAC7C,CAAC;AAAA,IACF;AAGA,UAAM,iBAAiB,KAAK,IAAI,CAAC,CAAC,KAAK,KAAK,MAAM;AACjD,YAAM,WAAW;AAAA,QAChB,aAAa,IAAI;AAAA,QACjB,GAAI,0BAA0B,EAAE,UAAU,IAAI,SAAS,IAAI,CAAC;AAAA,MAC7D;AAEA,aAAO;AAAA,QACN,MAAM,EAAE,UAAU,MAAM;AAAA,QACxB,YAAY;AAAA,UACX,MAAM;AAAA,QACP;AAAA,MACD;AAAA,IACD,CAAC;AAGD,mCAAW,SAAS,4BAA4B,EAAE,OAAO,OAAO,CAAC;AAEjE,WAAO;AAAA,EACR,UAAE;AAED,SAAK,2BAA2B,WAAW;AAAA,EAC5C;AACD;","names":[]}

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/retrieveAsToolExecuteOperation.js

@@ -0,0 +1,98 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var retrieveAsToolExecuteOperation_exports = {};
+__export(retrieveAsToolExecuteOperation_exports, {
+  handleRetrieveAsToolExecuteOperation: () => handleRetrieveAsToolExecuteOperation
+});
+module.exports = __toCommonJS(retrieveAsToolExecuteOperation_exports);
+var import_n8n_workflow = require("n8n-workflow");
+var import_helpers = require("../../../../../utils/helpers");
+async function handleRetrieveAsToolExecuteOperation(context, args, embeddings, itemIndex) {
+  const filter = (0, import_helpers.getMetadataFiltersValues)(context, itemIndex);
+  const vectorStore = await args.getVectorStoreClient(
+    context,
+    // We'll pass filter to similaritySearchVectorWithScore instead of getVectorStoreClient
+    void 0,
+    embeddings,
+    itemIndex
+  );
+  try {
+    const inputData = context.getInputData();
+    const item = inputData[itemIndex];
+    const query = typeof item.json.input === "string" ? item.json.input : void 0;
+    if (!query || typeof query !== "string") {
+      throw new Error('Input data must contain a "input" field with the search query');
+    }
+    const topK = context.getNodeParameter("topK", itemIndex, 4);
+    (0, import_n8n_workflow.assertParamIsNumber)("topK", topK, context.getNode());
+    const useReranker = context.getNodeParameter("useReranker", itemIndex, false);
+    (0, import_n8n_workflow.assertParamIsBoolean)("useReranker", useReranker, context.getNode());
+    const includeDocumentMetadata = context.getNodeParameter(
+      "includeDocumentMetadata",
+      itemIndex,
+      true
+    );
+    (0, import_n8n_workflow.assertParamIsBoolean)("includeDocumentMetadata", includeDocumentMetadata, context.getNode());
+    const embeddedQuery = await embeddings.embedQuery(query);
+    let docs = await vectorStore.similaritySearchVectorWithScore(embeddedQuery, topK, filter);
+    if (useReranker && docs.length > 0) {
+      const reranker = await context.getInputConnectionData(
+        import_n8n_workflow.NodeConnectionTypes.AiReranker,
+        0
+      );
+      const documents = docs.map(([doc]) => doc);
+      const rerankedDocuments = await reranker.compressDocuments(documents, query);
+      docs = rerankedDocuments.map((doc) => {
+        const { relevanceScore, ...metadata } = doc.metadata || {};
+        return [{ ...doc, metadata }, relevanceScore ?? 0];
+      });
+    }
+    const serializedDocs = docs.map(([doc]) => {
+      if (includeDocumentMetadata) {
+        return {
+          type: "text",
+          text: JSON.stringify({ ...doc })
+        };
+      } else {
+        return {
+          type: "text",
+          pageContent: JSON.stringify({ pageContent: doc.pageContent })
+        };
+      }
+    });
+    (0, import_helpers.logAiEvent)(context, "ai-vector-store-searched", { input: query });
+    return [
+      {
+        json: {
+          response: serializedDocs
+        },
+        pairedItem: {
+          item: itemIndex
+        }
+      }
+    ];
+  } finally {
+    args.releaseVectorStoreClient?.(vectorStore);
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  handleRetrieveAsToolExecuteOperation
+});
+//# sourceMappingURL=retrieveAsToolExecuteOperation.js.map

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/retrieveAsToolExecuteOperation.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../../nodes/vector_store/shared/createVectorStoreNode/operations/retrieveAsToolExecuteOperation.ts"],"sourcesContent":["import type { Embeddings } from '@langchain/core/embeddings';\nimport type { BaseDocumentCompressor } from '@langchain/core/retrievers/document_compressors';\nimport type { VectorStore } from '@langchain/core/vectorstores';\nimport {\n\tassertParamIsBoolean,\n\tassertParamIsNumber,\n\tNodeConnectionTypes,\n\ttype IExecuteFunctions,\n\ttype INodeExecutionData,\n} from 'n8n-workflow';\n\nimport { getMetadataFiltersValues, logAiEvent } from '@utils/helpers';\n\nimport type { VectorStoreNodeConstructorArgs } from '../types';\n\n/**\n * Handles the 'retrieve-as-tool' operation mode in execute context\n * Searches the vector store for documents similar to a query and returns execution data\n * This is similar to the load operation but designed to work with the new tool execution pattern\n */\nexport async function handleRetrieveAsToolExecuteOperation<T extends VectorStore = VectorStore>(\n\tcontext: IExecuteFunctions,\n\targs: VectorStoreNodeConstructorArgs<T>,\n\tembeddings: Embeddings,\n\titemIndex: number,\n): Promise<INodeExecutionData[]> {\n\tconst filter = getMetadataFiltersValues(context, itemIndex);\n\tconst vectorStore = await args.getVectorStoreClient(\n\t\tcontext,\n\t\t// We'll pass filter to similaritySearchVectorWithScore instead of getVectorStoreClient\n\t\tundefined,\n\t\tembeddings,\n\t\titemIndex,\n\t);\n\n\ttry {\n\t\t// Get the search parameters - query from input data, others from node parameters\n\t\tconst inputData = context.getInputData();\n\t\tconst item = inputData[itemIndex];\n\t\tconst query = typeof item.json.input === 'string' ? item.json.input : undefined;\n\n\t\tif (!query || typeof query !== 'string') {\n\t\t\tthrow new Error('Input data must contain a \"input\" field with the search query');\n\t\t}\n\n\t\tconst topK = context.getNodeParameter('topK', itemIndex, 4);\n\t\tassertParamIsNumber('topK', topK, context.getNode());\n\t\tconst useReranker = context.getNodeParameter('useReranker', itemIndex, false);\n\t\tassertParamIsBoolean('useReranker', useReranker, context.getNode());\n\n\t\tconst includeDocumentMetadata = context.getNodeParameter(\n\t\t\t'includeDocumentMetadata',\n\t\t\titemIndex,\n\t\t\ttrue,\n\t\t);\n\t\tassertParamIsBoolean('includeDocumentMetadata', includeDocumentMetadata, context.getNode());\n\n\t\t// Embed the query to prepare for vector similarity search\n\t\tconst embeddedQuery = await embeddings.embedQuery(query);\n\n\t\t// Get the most similar documents to the embedded query\n\t\tlet docs = await vectorStore.similaritySearchVectorWithScore(embeddedQuery, topK, filter);\n\n\t\t// If reranker is used, rerank the documents\n\t\tif (useReranker && docs.length > 0) {\n\t\t\tconst reranker = (await context.getInputConnectionData(\n\t\t\t\tNodeConnectionTypes.AiReranker,\n\t\t\t\t0,\n\t\t\t)) as BaseDocumentCompressor;\n\t\t\tconst documents = docs.map(([doc]) => doc);\n\n\t\t\tconst rerankedDocuments = await reranker.compressDocuments(documents, query);\n\t\t\tdocs = rerankedDocuments.map((doc) => {\n\t\t\t\tconst { relevanceScore, ...metadata } = doc.metadata || {};\n\t\t\t\treturn [{ ...doc, metadata }, relevanceScore ?? 0];\n\t\t\t});\n\t\t}\n\n\t\t// Format the documents for the output similar to the original tool format\n\t\tconst serializedDocs = docs.map(([doc]) => {\n\t\t\tif (includeDocumentMetadata) {\n\t\t\t\treturn {\n\t\t\t\t\ttype: 'text',\n\t\t\t\t\ttext: JSON.stringify({ ...doc }),\n\t\t\t\t};\n\t\t\t} else {\n\t\t\t\treturn {\n\t\t\t\t\ttype: 'text',\n\t\t\t\t\tpageContent: JSON.stringify({ pageContent: doc.pageContent }),\n\t\t\t\t};\n\t\t\t}\n\t\t});\n\n\t\t// Log the AI event for analytics\n\t\tlogAiEvent(context, 'ai-vector-store-searched', { input: query });\n\n\t\treturn [\n\t\t\t{\n\t\t\t\tjson: {\n\t\t\t\t\tresponse: serializedDocs,\n\t\t\t\t},\n\t\t\t\tpairedItem: {\n\t\t\t\t\titem: itemIndex,\n\t\t\t\t},\n\t\t\t},\n\t\t];\n\t} finally {\n\t\t// Release the vector store client if a release method was provided\n\t\targs.releaseVectorStoreClient?.(vectorStore);\n\t}\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,0BAMO;AAEP,qBAAqD;AASrD,eAAsB,qCACrB,SACA,MACA,YACA,WACgC;AAChC,QAAM,aAAS,yCAAyB,SAAS,SAAS;AAC1D,QAAM,cAAc,MAAM,KAAK;AAAA,IAC9B;AAAA;AAAA,IAEA;AAAA,IACA;AAAA,IACA;AAAA,EACD;AAEA,MAAI;AAEH,UAAM,YAAY,QAAQ,aAAa;AACvC,UAAM,OAAO,UAAU,SAAS;AAChC,UAAM,QAAQ,OAAO,KAAK,KAAK,UAAU,WAAW,KAAK,KAAK,QAAQ;AAEtE,QAAI,CAAC,SAAS,OAAO,UAAU,UAAU;AACxC,YAAM,IAAI,MAAM,+DAA+D;AAAA,IAChF;AAEA,UAAM,OAAO,QAAQ,iBAAiB,QAAQ,WAAW,CAAC;AAC1D,iDAAoB,QAAQ,MAAM,QAAQ,QAAQ,CAAC;AACnD,UAAM,cAAc,QAAQ,iBAAiB,eAAe,WAAW,KAAK;AAC5E,kDAAqB,eAAe,aAAa,QAAQ,QAAQ,CAAC;AAElE,UAAM,0BAA0B,QAAQ;AAAA,MACvC;AAAA,MACA;AAAA,MACA;AAAA,IACD;AACA,kDAAqB,2BAA2B,yBAAyB,QAAQ,QAAQ,CAAC;AAG1F,UAAM,gBAAgB,MAAM,WAAW,WAAW,KAAK;AAGvD,QAAI,OAAO,MAAM,YAAY,gCAAgC,eAAe,MAAM,MAAM;AAGxF,QAAI,eAAe,KAAK,SAAS,GAAG;AACnC,YAAM,WAAY,MAAM,QAAQ;AAAA,QAC/B,wCAAoB;AAAA,QACpB;AAAA,MACD;AACA,YAAM,YAAY,KAAK,IAAI,CAAC,CAAC,GAAG,MAAM,GAAG;AAEzC,YAAM,oBAAoB,MAAM,SAAS,kBAAkB,WAAW,KAAK;AAC3E,aAAO,kBAAkB,IAAI,CAAC,QAAQ;AACrC,cAAM,EAAE,gBAAgB,GAAG,SAAS,IAAI,IAAI,YAAY,CAAC;AACzD,eAAO,CAAC,EAAE,GAAG,KAAK,SAAS,GAAG,kBAAkB,CAAC;AAAA,MAClD,CAAC;AAAA,IACF;AAGA,UAAM,iBAAiB,KAAK,IAAI,CAAC,CAAC,GAAG,MAAM;AAC1C,UAAI,yBAAyB;AAC5B,eAAO;AAAA,UACN,MAAM;AAAA,UACN,MAAM,KAAK,UAAU,EAAE,GAAG,IAAI,CAAC;AAAA,QAChC;AAAA,MACD,OAAO;AACN,eAAO;AAAA,UACN,MAAM;AAAA,UACN,aAAa,KAAK,UAAU,EAAE,aAAa,IAAI,YAAY,CAAC;AAAA,QAC7D;AAAA,MACD;AAAA,IACD,CAAC;AAGD,mCAAW,SAAS,4BAA4B,EAAE,OAAO,MAAM,CAAC;AAEhE,WAAO;AAAA,MACN;AAAA,QACC,MAAM;AAAA,UACL,UAAU;AAAA,QACX;AAAA,QACA,YAAY;AAAA,UACX,MAAM;AAAA,QACP;AAAA,MACD;AAAA,IACD;AAAA,EACD,UAAE;AAED,SAAK,2BAA2B,WAAW;AAAA,EAC5C;AACD;","names":[]}

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/retrieveAsToolOperation.js

@@ -0,0 +1,95 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var retrieveAsToolOperation_exports = {};
+__export(retrieveAsToolOperation_exports, {
+  handleRetrieveAsToolOperation: () => handleRetrieveAsToolOperation
+});
+module.exports = __toCommonJS(retrieveAsToolOperation_exports);
+var import_fromAIToolFactory = require("../../../../../utils/fromAIToolFactory");
+var import_helpers = require("../../../../../utils/helpers");
+var import_logWrapper = require("../../../../../utils/logWrapper");
+var import_n8n_workflow = require("n8n-workflow");
+async function handleRetrieveAsToolOperation(context, args, embeddings, itemIndex) {
+  const toolDescription = context.getNodeParameter("toolDescription", itemIndex);
+  const node = context.getNode();
+  const { typeVersion } = node;
+  const toolName = typeVersion < 1.3 ? context.getNodeParameter("toolName", itemIndex) : (0, import_n8n_workflow.nodeNameToToolName)(node);
+  const vectorStoreTool = (0, import_fromAIToolFactory.createToolFromNode)(node, {
+    name: toolName,
+    description: toolDescription,
+    extraArgs: [{ key: "input", description: "Query to search for. Required" }],
+    func: async (query) => {
+      const topK = context.getNodeParameter("topK", itemIndex, 4);
+      const useReranker = context.getNodeParameter("useReranker", itemIndex, false);
+      const includeDocumentMetadata = context.getNodeParameter(
+        "includeDocumentMetadata",
+        itemIndex,
+        true
+      );
+      const filter = (0, import_helpers.getMetadataFiltersValues)(context, itemIndex);
+      const queryString = typeof query === "string" ? query : query.input;
+      (0, import_n8n_workflow.assert)(typeof queryString === "string", "Query must be of type string");
+      const vectorStore = await args.getVectorStoreClient(
+        context,
+        void 0,
+        embeddings,
+        itemIndex
+      );
+      try {
+        const embeddedPrompt = await embeddings.embedQuery(queryString);
+        let documents = await vectorStore.similaritySearchVectorWithScore(
+          embeddedPrompt,
+          topK,
+          filter
+        );
+        if (useReranker && documents.length > 0) {
+          const reranker = await context.getInputConnectionData(
+            import_n8n_workflow.NodeConnectionTypes.AiReranker,
+            0
+          );
+          const docs = documents.map(([doc]) => doc);
+          const rerankedDocuments = await reranker.compressDocuments(docs, queryString);
+          documents = rerankedDocuments.map((doc) => {
+            const { relevanceScore, ...metadata } = doc.metadata;
+            return [{ ...doc, metadata }, relevanceScore];
+          });
+        }
+        return documents.map((document) => {
+          if (includeDocumentMetadata) {
+            return { type: "text", text: JSON.stringify(document[0]) };
+          }
+          return {
+            type: "text",
+            text: JSON.stringify({ pageContent: document[0].pageContent })
+          };
+        }).filter((document) => !!document);
+      } finally {
+        args.releaseVectorStoreClient?.(vectorStore);
+      }
+    }
+  });
+  return {
+    response: (0, import_logWrapper.logWrapper)(vectorStoreTool, context)
+  };
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  handleRetrieveAsToolOperation
+});
+//# sourceMappingURL=retrieveAsToolOperation.js.map

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/retrieveAsToolOperation.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../../nodes/vector_store/shared/createVectorStoreNode/operations/retrieveAsToolOperation.ts"],"sourcesContent":["import type { Embeddings } from '@langchain/core/embeddings';\nimport type { BaseDocumentCompressor } from '@langchain/core/retrievers/document_compressors';\nimport type { VectorStore } from '@langchain/core/vectorstores';\nimport { createToolFromNode } from '@utils/fromAIToolFactory';\nimport { getMetadataFiltersValues } from '@utils/helpers';\nimport { logWrapper } from '@utils/logWrapper';\nimport type { ISupplyDataFunctions, SupplyData } from 'n8n-workflow';\nimport { assert, NodeConnectionTypes, nodeNameToToolName } from 'n8n-workflow';\n\nimport type { VectorStoreNodeConstructorArgs } from '../types';\n\n/**\n * Handles the 'retrieve-as-tool' operation mode\n * Returns a tool that can be used with AI Agent nodes\n */\nexport async function handleRetrieveAsToolOperation<T extends VectorStore = VectorStore>(\n\tcontext: ISupplyDataFunctions,\n\targs: VectorStoreNodeConstructorArgs<T>,\n\tembeddings: Embeddings,\n\titemIndex: number,\n): Promise<SupplyData> {\n\t// Get the tool configuration parameters\n\tconst toolDescription = context.getNodeParameter('toolDescription', itemIndex) as string;\n\n\tconst node = context.getNode();\n\tconst { typeVersion } = node;\n\tconst toolName =\n\t\ttypeVersion < 1.3\n\t\t\t? (context.getNodeParameter('toolName', itemIndex) as string)\n\t\t\t: nodeNameToToolName(node);\n\n\t// Create a Dynamic Tool that wraps vector store search functionality\n\tconst vectorStoreTool = createToolFromNode(node, {\n\t\tname: toolName,\n\t\tdescription: toolDescription,\n\t\textraArgs: [{ key: 'input', description: 'Query to search for. Required' }],\n\t\tfunc: async (query) => {\n\t\t\tconst topK = context.getNodeParameter('topK', itemIndex, 4) as number;\n\t\t\tconst useReranker = context.getNodeParameter('useReranker', itemIndex, false) as boolean;\n\t\t\tconst includeDocumentMetadata = context.getNodeParameter(\n\t\t\t\t'includeDocumentMetadata',\n\t\t\t\titemIndex,\n\t\t\t\ttrue,\n\t\t\t) as boolean;\n\n\t\t\t// Get metadata filters\n\t\t\tconst filter = getMetadataFiltersValues(context, itemIndex);\n\n\t\t\t// Extract the query string - it can be either a string (DynamicTool) or an object with 'input' key (DynamicStructuredTool)\n\t\t\tconst queryString = typeof query === 'string' ? query : query.input;\n\t\t\tassert(typeof queryString === 'string', 'Query must be of type string');\n\n\t\t\t// For each tool use, get a fresh vector store client.\n\t\t\t// We don't pass in a filter here only later in the similaritySearchVectorWithScore\n\t\t\t// method to avoid an exception with some vector stores like Supabase or Pinecone(#AI-740)\n\t\t\tconst vectorStore = await args.getVectorStoreClient(\n\t\t\t\tcontext,\n\t\t\t\tundefined,\n\t\t\t\tembeddings,\n\t\t\t\titemIndex,\n\t\t\t);\n\n\t\t\ttry {\n\t\t\t\t// Embed the input query\n\t\t\t\tconst embeddedPrompt = await embeddings.embedQuery(queryString);\n\n\t\t\t\t// Search for similar documents\n\t\t\t\tlet documents = await vectorStore.similaritySearchVectorWithScore(\n\t\t\t\t\tembeddedPrompt,\n\t\t\t\t\ttopK,\n\t\t\t\t\tfilter,\n\t\t\t\t);\n\n\t\t\t\t// If reranker is used, rerank the documents\n\t\t\t\tif (useReranker && documents.length > 0) {\n\t\t\t\t\tconst reranker = (await context.getInputConnectionData(\n\t\t\t\t\t\tNodeConnectionTypes.AiReranker,\n\t\t\t\t\t\t0,\n\t\t\t\t\t)) as BaseDocumentCompressor;\n\n\t\t\t\t\tconst docs = documents.map(([doc]) => doc);\n\t\t\t\t\tconst rerankedDocuments = await reranker.compressDocuments(docs, queryString);\n\t\t\t\t\tdocuments = rerankedDocuments.map((doc) => {\n\t\t\t\t\t\tconst { relevanceScore, ...metadata } = doc.metadata;\n\t\t\t\t\t\treturn [{ ...doc, metadata }, relevanceScore];\n\t\t\t\t\t});\n\t\t\t\t}\n\n\t\t\t\t// Format the documents for the tool output\n\t\t\t\treturn documents\n\t\t\t\t\t.map((document) => {\n\t\t\t\t\t\tif (includeDocumentMetadata) {\n\t\t\t\t\t\t\treturn { type: 'text', text: JSON.stringify(document[0]) };\n\t\t\t\t\t\t}\n\t\t\t\t\t\treturn {\n\t\t\t\t\t\t\ttype: 'text',\n\t\t\t\t\t\t\ttext: JSON.stringify({ pageContent: document[0].pageContent }),\n\t\t\t\t\t\t};\n\t\t\t\t\t})\n\t\t\t\t\t.filter((document) => !!document);\n\t\t\t} finally {\n\t\t\t\t// Release the vector store client if a release method was provided\n\t\t\t\targs.releaseVectorStoreClient?.(vectorStore);\n\t\t\t}\n\t\t},\n\t});\n\n\t// Return the vector store tool with logging wrapper\n\treturn {\n\t\tresponse: logWrapper(vectorStoreTool, context),\n\t};\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,+BAAmC;AACnC,qBAAyC;AACzC,wBAA2B;AAE3B,0BAAgE;AAQhE,eAAsB,8BACrB,SACA,MACA,YACA,WACsB;AAEtB,QAAM,kBAAkB,QAAQ,iBAAiB,mBAAmB,SAAS;AAE7E,QAAM,OAAO,QAAQ,QAAQ;AAC7B,QAAM,EAAE,YAAY,IAAI;AACxB,QAAM,WACL,cAAc,MACV,QAAQ,iBAAiB,YAAY,SAAS,QAC/C,wCAAmB,IAAI;AAG3B,QAAM,sBAAkB,6CAAmB,MAAM;AAAA,IAChD,MAAM;AAAA,IACN,aAAa;AAAA,IACb,WAAW,CAAC,EAAE,KAAK,SAAS,aAAa,gCAAgC,CAAC;AAAA,IAC1E,MAAM,OAAO,UAAU;AACtB,YAAM,OAAO,QAAQ,iBAAiB,QAAQ,WAAW,CAAC;AAC1D,YAAM,cAAc,QAAQ,iBAAiB,eAAe,WAAW,KAAK;AAC5E,YAAM,0BAA0B,QAAQ;AAAA,QACvC;AAAA,QACA;AAAA,QACA;AAAA,MACD;AAGA,YAAM,aAAS,yCAAyB,SAAS,SAAS;AAG1D,YAAM,cAAc,OAAO,UAAU,WAAW,QAAQ,MAAM;AAC9D,sCAAO,OAAO,gBAAgB,UAAU,8BAA8B;AAKtE,YAAM,cAAc,MAAM,KAAK;AAAA,QAC9B;AAAA,QACA;AAAA,QACA;AAAA,QACA;AAAA,MACD;AAEA,UAAI;AAEH,cAAM,iBAAiB,MAAM,WAAW,WAAW,WAAW;AAG9D,YAAI,YAAY,MAAM,YAAY;AAAA,UACjC;AAAA,UACA;AAAA,UACA;AAAA,QACD;AAGA,YAAI,eAAe,UAAU,SAAS,GAAG;AACxC,gBAAM,WAAY,MAAM,QAAQ;AAAA,YAC/B,wCAAoB;AAAA,YACpB;AAAA,UACD;AAEA,gBAAM,OAAO,UAAU,IAAI,CAAC,CAAC,GAAG,MAAM,GAAG;AACzC,gBAAM,oBAAoB,MAAM,SAAS,kBAAkB,MAAM,WAAW;AAC5E,sBAAY,kBAAkB,IAAI,CAAC,QAAQ;AAC1C,kBAAM,EAAE,gBAAgB,GAAG,SAAS,IAAI,IAAI;AAC5C,mBAAO,CAAC,EAAE,GAAG,KAAK,SAAS,GAAG,cAAc;AAAA,UAC7C,CAAC;AAAA,QACF;AAGA,eAAO,UACL,IAAI,CAAC,aAAa;AAClB,cAAI,yBAAyB;AAC5B,mBAAO,EAAE,MAAM,QAAQ,MAAM,KAAK,UAAU,SAAS,CAAC,CAAC,EAAE;AAAA,UAC1D;AACA,iBAAO;AAAA,YACN,MAAM;AAAA,YACN,MAAM,KAAK,UAAU,EAAE,aAAa,SAAS,CAAC,EAAE,YAAY,CAAC;AAAA,UAC9D;AAAA,QACD,CAAC,EACA,OAAO,CAAC,aAAa,CAAC,CAAC,QAAQ;AAAA,MAClC,UAAE;AAED,aAAK,2BAA2B,WAAW;AAAA,MAC5C;AAAA,IACD;AAAA,EACD,CAAC;AAGD,SAAO;AAAA,IACN,cAAU,8BAAW,iBAAiB,OAAO;AAAA,EAC9C;AACD;","names":[]}

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/retrieveOperation.js

@@ -0,0 +1,55 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var retrieveOperation_exports = {};
+__export(retrieveOperation_exports, {
+  handleRetrieveOperation: () => handleRetrieveOperation
+});
+module.exports = __toCommonJS(retrieveOperation_exports);
+var import_n8n_workflow = require("n8n-workflow");
+var import_helpers = require("../../../../../utils/helpers");
+var import_logWrapper = require("../../../../../utils/logWrapper");
+async function handleRetrieveOperation(context, args, embeddings, itemIndex) {
+  const filter = (0, import_helpers.getMetadataFiltersValues)(context, itemIndex);
+  const useReranker = context.getNodeParameter("useReranker", itemIndex, false);
+  const vectorStore = await args.getVectorStoreClient(context, filter, embeddings, itemIndex);
+  let response = vectorStore;
+  if (useReranker) {
+    const reranker = await context.getInputConnectionData(
+      import_n8n_workflow.NodeConnectionTypes.AiReranker,
+      0
+    );
+    response = {
+      reranker,
+      vectorStore: (0, import_logWrapper.logWrapper)(vectorStore, context)
+    };
+  } else {
+    response = (0, import_logWrapper.logWrapper)(vectorStore, context);
+  }
+  return {
+    response,
+    closeFunction: async () => {
+      args.releaseVectorStoreClient?.(vectorStore);
+    }
+  };
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  handleRetrieveOperation
+});
+//# sourceMappingURL=retrieveOperation.js.map

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/retrieveOperation.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../../nodes/vector_store/shared/createVectorStoreNode/operations/retrieveOperation.ts"],"sourcesContent":["import type { Embeddings } from '@langchain/core/embeddings';\nimport type { BaseDocumentCompressor } from '@langchain/core/retrievers/document_compressors';\nimport type { VectorStore } from '@langchain/core/vectorstores';\nimport { NodeConnectionTypes, type ISupplyDataFunctions, type SupplyData } from 'n8n-workflow';\n\nimport { getMetadataFiltersValues } from '@utils/helpers';\nimport { logWrapper } from '@utils/logWrapper';\n\nimport type { VectorStoreNodeConstructorArgs } from '../types';\n\n/**\n * Handles the 'retrieve' operation mode\n * Returns the vector store to be used with AI nodes\n */\nexport async function handleRetrieveOperation<T extends VectorStore = VectorStore>(\n\tcontext: ISupplyDataFunctions,\n\targs: VectorStoreNodeConstructorArgs<T>,\n\tembeddings: Embeddings,\n\titemIndex: number,\n): Promise<SupplyData> {\n\t// Get metadata filters\n\tconst filter = getMetadataFiltersValues(context, itemIndex);\n\tconst useReranker = context.getNodeParameter('useReranker', itemIndex, false) as boolean;\n\n\t// Get the vector store client\n\tconst vectorStore = await args.getVectorStoreClient(context, filter, embeddings, itemIndex);\n\tlet response: VectorStore | { reranker: BaseDocumentCompressor; vectorStore: VectorStore } =\n\t\tvectorStore;\n\n\tif (useReranker) {\n\t\tconst reranker = (await context.getInputConnectionData(\n\t\t\tNodeConnectionTypes.AiReranker,\n\t\t\t0,\n\t\t)) as BaseDocumentCompressor;\n\n\t\t// Return reranker and vector store with log wrapper\n\t\tresponse = {\n\t\t\treranker,\n\t\t\tvectorStore: logWrapper(vectorStore, context),\n\t\t};\n\t} else {\n\t\t// Return the vector store with logging wrapper\n\t\tresponse = logWrapper(vectorStore, context);\n\t}\n\n\treturn {\n\t\tresponse,\n\t\tcloseFunction: async () => {\n\t\t\t// Release the vector store client if a release method was provided\n\t\t\targs.releaseVectorStoreClient?.(vectorStore);\n\t\t},\n\t};\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,0BAAgF;AAEhF,qBAAyC;AACzC,wBAA2B;AAQ3B,eAAsB,wBACrB,SACA,MACA,YACA,WACsB;AAEtB,QAAM,aAAS,yCAAyB,SAAS,SAAS;AAC1D,QAAM,cAAc,QAAQ,iBAAiB,eAAe,WAAW,KAAK;AAG5E,QAAM,cAAc,MAAM,KAAK,qBAAqB,SAAS,QAAQ,YAAY,SAAS;AAC1F,MAAI,WACH;AAED,MAAI,aAAa;AAChB,UAAM,WAAY,MAAM,QAAQ;AAAA,MAC/B,wCAAoB;AAAA,MACpB;AAAA,IACD;AAGA,eAAW;AAAA,MACV;AAAA,MACA,iBAAa,8BAAW,aAAa,OAAO;AAAA,IAC7C;AAAA,EACD,OAAO;AAEN,mBAAW,8BAAW,aAAa,OAAO;AAAA,EAC3C;AAEA,SAAO;AAAA,IACN;AAAA,IACA,eAAe,YAAY;AAE1B,WAAK,2BAA2B,WAAW;AAAA,IAC5C;AAAA,EACD;AACD;","names":[]}

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/updateOperation.js

@@ -0,0 +1,69 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var updateOperation_exports = {};
+__export(updateOperation_exports, {
+  handleUpdateOperation: () => handleUpdateOperation
+});
+module.exports = __toCommonJS(updateOperation_exports);
+var import_n8n_workflow = require("n8n-workflow");
+var import_helpers = require("../../../../../utils/helpers");
+var import_N8nJsonLoader = require("../../../../../utils/N8nJsonLoader");
+var import_processDocuments = require("../../processDocuments");
+var import_utils = require("../utils");
+async function handleUpdateOperation(context, args, embeddings) {
+  if (!(0, import_utils.isUpdateSupported)(args)) {
+    throw new import_n8n_workflow.NodeOperationError(
+      context.getNode(),
+      "Update operation is not implemented for this Vector Store"
+    );
+  }
+  const items = context.getInputData();
+  const loader = new import_N8nJsonLoader.N8nJsonLoader(context);
+  const resultData = [];
+  for (let itemIndex = 0; itemIndex < items.length; itemIndex++) {
+    const itemData = items[itemIndex];
+    const documentId = context.getNodeParameter("id", itemIndex, "", {
+      extractValue: true
+    });
+    const vectorStore = await args.getVectorStoreClient(context, void 0, embeddings, itemIndex);
+    try {
+      const { processedDocuments, serializedDocuments } = await (0, import_processDocuments.processDocument)(
+        loader,
+        itemData,
+        itemIndex
+      );
+      if (processedDocuments?.length !== 1) {
+        throw new import_n8n_workflow.NodeOperationError(context.getNode(), "Single document per item expected");
+      }
+      resultData.push(...serializedDocuments);
+      await vectorStore.addDocuments(processedDocuments, {
+        ids: [documentId]
+      });
+      (0, import_helpers.logAiEvent)(context, "ai-vector-store-updated");
+    } finally {
+      args.releaseVectorStoreClient?.(vectorStore);
+    }
+  }
+  return resultData;
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  handleUpdateOperation
+});
+//# sourceMappingURL=updateOperation.js.map

package/dist/nodes/vector_store/shared/createVectorStoreNode/operations/updateOperation.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../../nodes/vector_store/shared/createVectorStoreNode/operations/updateOperation.ts"],"sourcesContent":["import type { Embeddings } from '@langchain/core/embeddings';\nimport type { VectorStore } from '@langchain/core/vectorstores';\nimport type { IExecuteFunctions, INodeExecutionData } from 'n8n-workflow';\nimport { NodeOperationError } from 'n8n-workflow';\n\nimport { logAiEvent } from '@utils/helpers';\nimport { N8nJsonLoader } from '@utils/N8nJsonLoader';\n\nimport { processDocument } from '../../processDocuments';\nimport type { VectorStoreNodeConstructorArgs } from '../types';\nimport { isUpdateSupported } from '../utils';\n\n/**\n * Handles the 'update' operation mode\n * Updates existing documents in the vector store by ID\n */\nexport async function handleUpdateOperation<T extends VectorStore = VectorStore>(\n\tcontext: IExecuteFunctions,\n\targs: VectorStoreNodeConstructorArgs<T>,\n\tembeddings: Embeddings,\n): Promise<INodeExecutionData[]> {\n\t// First check if update operation is supported by this vector store\n\tif (!isUpdateSupported(args)) {\n\t\tthrow new NodeOperationError(\n\t\t\tcontext.getNode(),\n\t\t\t'Update operation is not implemented for this Vector Store',\n\t\t);\n\t}\n\n\t// Get input items\n\tconst items = context.getInputData();\n\t// Create a loader for processing document data\n\tconst loader = new N8nJsonLoader(context);\n\n\tconst resultData: INodeExecutionData[] = [];\n\n\t// Process each input item\n\tfor (let itemIndex = 0; itemIndex < items.length; itemIndex++) {\n\t\tconst itemData = items[itemIndex];\n\n\t\t// Get the document ID to update\n\t\tconst documentId = context.getNodeParameter('id', itemIndex, '', {\n\t\t\textractValue: true,\n\t\t}) as string;\n\n\t\t// Get the vector store client\n\t\tconst vectorStore = await args.getVectorStoreClient(context, undefined, embeddings, itemIndex);\n\n\t\ttry {\n\t\t\t// Process the document from the input\n\t\t\tconst { processedDocuments, serializedDocuments } = await processDocument(\n\t\t\t\tloader,\n\t\t\t\titemData,\n\t\t\t\titemIndex,\n\t\t\t);\n\n\t\t\t// Validate that we have exactly one document to update\n\t\t\tif (processedDocuments?.length !== 1) {\n\t\t\t\tthrow new NodeOperationError(context.getNode(), 'Single document per item expected');\n\t\t\t}\n\n\t\t\t// Add the serialized document to the result\n\t\t\tresultData.push(...serializedDocuments);\n\n\t\t\t// Use document ID to update the existing document\n\t\t\tawait vectorStore.addDocuments(processedDocuments, {\n\t\t\t\tids: [documentId],\n\t\t\t});\n\n\t\t\t// Log the AI event for analytics\n\t\t\tlogAiEvent(context, 'ai-vector-store-updated');\n\t\t} finally {\n\t\t\t// Release the vector store client if a release method was provided\n\t\t\targs.releaseVectorStoreClient?.(vectorStore);\n\t\t}\n\t}\n\n\treturn resultData;\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,0BAAmC;AAEnC,qBAA2B;AAC3B,2BAA8B;AAE9B,8BAAgC;AAEhC,mBAAkC;AAMlC,eAAsB,sBACrB,SACA,MACA,YACgC;AAEhC,MAAI,KAAC,gCAAkB,IAAI,GAAG;AAC7B,UAAM,IAAI;AAAA,MACT,QAAQ,QAAQ;AAAA,MAChB;AAAA,IACD;AAAA,EACD;AAGA,QAAM,QAAQ,QAAQ,aAAa;AAEnC,QAAM,SAAS,IAAI,mCAAc,OAAO;AAExC,QAAM,aAAmC,CAAC;AAG1C,WAAS,YAAY,GAAG,YAAY,MAAM,QAAQ,aAAa;AAC9D,UAAM,WAAW,MAAM,SAAS;AAGhC,UAAM,aAAa,QAAQ,iBAAiB,MAAM,WAAW,IAAI;AAAA,MAChE,cAAc;AAAA,IACf,CAAC;AAGD,UAAM,cAAc,MAAM,KAAK,qBAAqB,SAAS,QAAW,YAAY,SAAS;AAE7F,QAAI;AAEH,YAAM,EAAE,oBAAoB,oBAAoB,IAAI,UAAM;AAAA,QACzD;AAAA,QACA;AAAA,QACA;AAAA,MACD;AAGA,UAAI,oBAAoB,WAAW,GAAG;AACrC,cAAM,IAAI,uCAAmB,QAAQ,QAAQ,GAAG,mCAAmC;AAAA,MACpF;AAGA,iBAAW,KAAK,GAAG,mBAAmB;AAGtC,YAAM,YAAY,aAAa,oBAAoB;AAAA,QAClD,KAAK,CAAC,UAAU;AAAA,MACjB,CAAC;AAGD,qCAAW,SAAS,yBAAyB;AAAA,IAC9C,UAAE;AAED,WAAK,2BAA2B,WAAW;AAAA,IAC5C;AAAA,EACD;AAEA,SAAO;AACR;","names":[]}

package/dist/nodes/vector_store/shared/createVectorStoreNode/types.js

@@ -0,0 +1,17 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var types_exports = {};
+module.exports = __toCommonJS(types_exports);
+//# sourceMappingURL=types.js.map

package/dist/nodes/vector_store/shared/createVectorStoreNode/types.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../nodes/vector_store/shared/createVectorStoreNode/types.ts"],"sourcesContent":["import type { Document } from '@langchain/core/documents';\nimport type { Embeddings } from '@langchain/core/embeddings';\nimport type { VectorStore } from '@langchain/core/vectorstores';\nimport type {\n\tIExecuteFunctions,\n\tINodeCredentialDescription,\n\tINodeProperties,\n\tILoadOptionsFunctions,\n\tINodeListSearchResult,\n\tIcon,\n\tISupplyDataFunctions,\n\tThemeIconColor,\n\tIDataObject,\n\tNodeParameterValueType,\n} from 'n8n-workflow';\n\nexport type NodeOperationMode = 'insert' | 'load' | 'retrieve' | 'update' | 'retrieve-as-tool';\n\nexport interface NodeMeta {\n\tdisplayName: string;\n\tname: string;\n\thidden?: boolean;\n\tdescription: string;\n\tdocsUrl: string;\n\ticon: Icon;\n\ticonColor?: ThemeIconColor;\n\tcredentials?: INodeCredentialDescription[];\n\toperationModes?: NodeOperationMode[];\n\tcategories?: string[];\n\tsubcategories?: Record<string, string[]>;\n}\n\nexport interface VectorStoreNodeConstructorArgs<T extends VectorStore = VectorStore> {\n\tmeta: NodeMeta;\n\tmethods?: {\n\t\tlistSearch?: {\n\t\t\t[key: string]: (\n\t\t\t\tthis: ILoadOptionsFunctions,\n\t\t\t\tfilter?: string,\n\t\t\t\tpaginationToken?: string,\n\t\t\t) => Promise<INodeListSearchResult>;\n\t\t};\n\t\tactionHandler?: {\n\t\t\t[functionName: string]: (\n\t\t\t\tthis: ILoadOptionsFunctions,\n\t\t\t\tpayload: IDataObject | string | undefined,\n\t\t\t) => Promise<NodeParameterValueType>;\n\t\t};\n\t};\n\n\tsharedFields: INodeProperties[];\n\tinsertFields?: INodeProperties[];\n\tloadFields?: INodeProperties[];\n\tretrieveFields?: INodeProperties[];\n\tupdateFields?: INodeProperties[];\n\n\t/**\n\t * Function to populate the vector store with documents\n\t * Used during the 'insert' operation mode\n\t */\n\tpopulateVectorStore: (\n\t\tcontext: IExecuteFunctions | ISupplyDataFunctions,\n\t\tembeddings: Embeddings,\n\t\tdocuments: Array<Document<Record<string, unknown>>>,\n\t\titemIndex: number,\n\t) => Promise<void>;\n\n\t/**\n\t * Function to get the vector store client\n\t * This function is called for all operation modes\n\t */\n\tgetVectorStoreClient: (\n\t\tcontext: IExecuteFunctions | ISupplyDataFunctions,\n\t\tfilter: Record<string, never> | undefined,\n\t\tembeddings: Embeddings,\n\t\titemIndex: number,\n\t) => Promise<T>;\n\n\t/**\n\t * Optional function to release resources associated with the vector store client\n\t * Called after the vector store operations are complete\n\t */\n\treleaseVectorStoreClient?: (vectorStore: T) => void;\n}\n"],"mappings":";;;;;;;;;;;;;;AAAA;AAAA;","names":[]}

package/dist/nodes/vector_store/shared/createVectorStoreNode/utils.js

@@ -0,0 +1,48 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var utils_exports = {};
+__export(utils_exports, {
+  getOperationModeOptions: () => getOperationModeOptions,
+  isUpdateSupported: () => isUpdateSupported,
+  transformDescriptionForOperationMode: () => transformDescriptionForOperationMode
+});
+module.exports = __toCommonJS(utils_exports);
+var import_constants = require("./constants");
+function transformDescriptionForOperationMode(fields, mode) {
+  return fields.map((field) => ({
+    ...field,
+    displayOptions: { show: { mode: Array.isArray(mode) ? mode : [mode] } }
+  }));
+}
+function isUpdateSupported(args) {
+  return args.meta.operationModes?.includes("update") ?? false;
+}
+function getOperationModeOptions(args) {
+  const enabledOperationModes = args.meta.operationModes ?? import_constants.DEFAULT_OPERATION_MODES;
+  return import_constants.OPERATION_MODE_DESCRIPTIONS.filter(
+    ({ value }) => enabledOperationModes.includes(value)
+  );
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  getOperationModeOptions,
+  isUpdateSupported,
+  transformDescriptionForOperationMode
+});
+//# sourceMappingURL=utils.js.map

package/dist/nodes/vector_store/shared/createVectorStoreNode/utils.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../nodes/vector_store/shared/createVectorStoreNode/utils.ts"],"sourcesContent":["import type { VectorStore } from '@langchain/core/vectorstores';\nimport type { INodeProperties, INodePropertyOptions } from 'n8n-workflow';\n\nimport { DEFAULT_OPERATION_MODES, OPERATION_MODE_DESCRIPTIONS } from './constants';\nimport type { NodeOperationMode, VectorStoreNodeConstructorArgs } from './types';\n\n/**\n * Transforms field descriptions to show only for specific operation modes\n * This function adds displayOptions to each field to make it appear only for specified modes\n */\nexport function transformDescriptionForOperationMode(\n\tfields: INodeProperties[],\n\tmode: NodeOperationMode | NodeOperationMode[],\n): INodeProperties[] {\n\treturn fields.map((field) => ({\n\t\t...field,\n\t\tdisplayOptions: { show: { mode: Array.isArray(mode) ? mode : [mode] } },\n\t}));\n}\n\n/**\n * Checks if the update operation is supported for a specific vector store\n * A vector store supports updates if it explicitly includes 'update' in its operationModes\n */\nexport function isUpdateSupported<T extends VectorStore>(\n\targs: VectorStoreNodeConstructorArgs<T>,\n): boolean {\n\treturn args.meta.operationModes?.includes('update') ?? false;\n}\n\n/**\n * Returns the operation mode options enabled for a specific vector store\n * Filters the full list of operation modes based on what's enabled for this vector store\n */\nexport function getOperationModeOptions<T extends VectorStore>(\n\targs: VectorStoreNodeConstructorArgs<T>,\n): INodePropertyOptions[] {\n\tconst enabledOperationModes = args.meta.operationModes ?? DEFAULT_OPERATION_MODES;\n\n\treturn OPERATION_MODE_DESCRIPTIONS.filter(({ value }) =>\n\t\tenabledOperationModes.includes(value as NodeOperationMode),\n\t);\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAGA,uBAAqE;AAO9D,SAAS,qCACf,QACA,MACoB;AACpB,SAAO,OAAO,IAAI,CAAC,WAAW;AAAA,IAC7B,GAAG;AAAA,IACH,gBAAgB,EAAE,MAAM,EAAE,MAAM,MAAM,QAAQ,IAAI,IAAI,OAAO,CAAC,IAAI,EAAE,EAAE;AAAA,EACvE,EAAE;AACH;AAMO,SAAS,kBACf,MACU;AACV,SAAO,KAAK,KAAK,gBAAgB,SAAS,QAAQ,KAAK;AACxD;AAMO,SAAS,wBACf,MACyB;AACzB,QAAM,wBAAwB,KAAK,KAAK,kBAAkB;AAE1D,SAAO,6CAA4B;AAAA,IAAO,CAAC,EAAE,MAAM,MAClD,sBAAsB,SAAS,KAA0B;AAAA,EAC1D;AACD;","names":[]}