UNPKG

dtamind-components

Version:

DTAmindai Components

198 lines (184 loc) 7.94 kB
import { flatten } from 'lodash' import { Embeddings } from '@langchain/core/embeddings' import { SingleStoreVectorStore, SingleStoreVectorStoreConfig } from '@langchain/community/vectorstores/singlestore' import { Document } from '@langchain/core/documents' import { ICommonObject, INode, INodeData, INodeOutputsValue, INodeParams, IndexingResult } from '../../../src/Interface' import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src/utils' class SingleStore_VectorStores implements INode { label: string name: string version: number description: string type: string icon: string category: string badge: string baseClasses: string[] inputs: INodeParams[] credential: INodeParams outputs: INodeOutputsValue[] constructor() { this.label = 'SingleStore' this.name = 'singlestore' this.version = 1.0 this.type = 'SingleStore' this.icon = 'singlestore.svg' this.category = 'Vector Stores' this.description = 'Upsert embedded data and perform similarity search upon query using SingleStore, a fast and distributed cloud relational database' this.baseClasses = [this.type, 'VectorStoreRetriever', 'BaseRetriever'] this.credential = { label: 'Connect Credential', name: 'credential', type: 'credential', description: 'Needed when using SingleStore cloud hosted', optional: true, credentialNames: ['singleStoreApi'] } this.inputs = [ { label: 'Document', name: 'document', type: 'Document', list: true, optional: true }, { label: 'Embeddings', name: 'embeddings', type: 'Embeddings' }, { label: 'Host', name: 'host', type: 'string' }, { label: 'Database', name: 'database', type: 'string' }, { label: 'Table Name', name: 'tableName', type: 'string', placeholder: 'embeddings', additionalParams: true, optional: true }, { label: 'Content Column Name', name: 'contentColumnName', type: 'string', placeholder: 'content', additionalParams: true, optional: true }, { label: 'Vector Column Name', name: 'vectorColumnName', type: 'string', placeholder: 'vector', additionalParams: true, optional: true }, { label: 'Metadata Column Name', name: 'metadataColumnName', type: 'string', placeholder: 'metadata', additionalParams: true, optional: true }, { label: 'Top K', name: 'topK', placeholder: '4', type: 'number', additionalParams: true, optional: true } ] this.outputs = [ { label: 'SingleStore Retriever', name: 'retriever', baseClasses: this.baseClasses }, { label: 'SingleStore Vector Store', name: 'vectorStore', baseClasses: [this.type, ...getBaseClasses(SingleStoreVectorStore)] } ] } //@ts-ignore vectorStoreMethods = { async upsert(nodeData: INodeData, options: ICommonObject): Promise<Partial<IndexingResult>> { const credentialData = await getCredentialData(nodeData.credential ?? '', options) const user = getCredentialParam('user', credentialData, nodeData) const password = getCredentialParam('password', credentialData, nodeData) const singleStoreConnectionConfig = { connectionOptions: { host: nodeData.inputs?.host as string, port: 3306, user, password, database: nodeData.inputs?.database as string }, ...(nodeData.inputs?.tableName ? { tableName: nodeData.inputs.tableName as string } : {}), ...(nodeData.inputs?.contentColumnName ? { contentColumnName: nodeData.inputs.contentColumnName as string } : {}), ...(nodeData.inputs?.vectorColumnName ? { vectorColumnName: nodeData.inputs.vectorColumnName as string } : {}), ...(nodeData.inputs?.metadataColumnName ? { metadataColumnName: nodeData.inputs.metadataColumnName as string } : {}) } as SingleStoreVectorStoreConfig const docs = nodeData.inputs?.document as Document[] const embeddings = nodeData.inputs?.embeddings as Embeddings const flattenDocs = docs && docs.length ? flatten(docs) : [] const finalDocs = [] for (let i = 0; i < flattenDocs.length; i += 1) { if (flattenDocs[i] && flattenDocs[i].pageContent) { finalDocs.push(new Document(flattenDocs[i])) } } try { const vectorStore = new SingleStoreVectorStore(embeddings, singleStoreConnectionConfig) vectorStore.addDocuments.bind(vectorStore)(finalDocs) return { numAdded: finalDocs.length, addedDocs: finalDocs } } catch (e) { throw new Error(e) } } } async init(nodeData: INodeData, _: string, options: ICommonObject): Promise<any> { const credentialData = await getCredentialData(nodeData.credential ?? '', options) const user = getCredentialParam('user', credentialData, nodeData) const password = getCredentialParam('password', credentialData, nodeData) const singleStoreConnectionConfig = { connectionOptions: { host: nodeData.inputs?.host as string, port: 3306, user, password, database: nodeData.inputs?.database as string }, ...(nodeData.inputs?.tableName ? { tableName: nodeData.inputs.tableName as string } : {}), ...(nodeData.inputs?.contentColumnName ? { contentColumnName: nodeData.inputs.contentColumnName as string } : {}), ...(nodeData.inputs?.vectorColumnName ? { vectorColumnName: nodeData.inputs.vectorColumnName as string } : {}), ...(nodeData.inputs?.metadataColumnName ? { metadataColumnName: nodeData.inputs.metadataColumnName as string } : {}) } as SingleStoreVectorStoreConfig const embeddings = nodeData.inputs?.embeddings as Embeddings const output = nodeData.outputs?.output as string const topK = nodeData.inputs?.topK as string const k = topK ? parseFloat(topK) : 4 const vectorStore = new SingleStoreVectorStore(embeddings, singleStoreConnectionConfig) if (output === 'retriever') { const retriever = vectorStore.asRetriever(k) return retriever } else if (output === 'vectorStore') { ;(vectorStore as any).k = k return vectorStore } return vectorStore } } module.exports = { nodeClass: SingleStore_VectorStores }