mirror of https://github.com/FlowiseAI/Flowise.git
Updates to Elasticsearch VectoreStore Functionality.
vinodkiran
parent
f108c62acf
commit
57760dc633
|
|
@ -14,14 +14,19 @@ class ElasticSearchUserPassword implements INodeCredential {
|
|||
this.description =
|
||||
'Refer to <a target="_blank" href="https://www.elastic.co/guide/en/kibana/current/tutorial-secure-access-to-kibana.html">official guide</a> on how to get User Password from ElasticSearch'
|
||||
this.inputs = [
|
||||
{
|
||||
label: 'Cloud ID',
|
||||
name: 'cloudId',
|
||||
type: 'string'
|
||||
},
|
||||
{
|
||||
label: 'ElasticSearch User',
|
||||
name: 'elasticSearchUser',
|
||||
name: 'username',
|
||||
type: 'string'
|
||||
},
|
||||
{
|
||||
label: 'ElasticSearch Password',
|
||||
name: 'elasticSearchPassword',
|
||||
name: 'password',
|
||||
type: 'password'
|
||||
}
|
||||
]
|
||||
|
|
|
|||
|
|
@ -0,0 +1,193 @@
|
|||
import {
|
||||
getBaseClasses,
|
||||
getCredentialData,
|
||||
getCredentialParam,
|
||||
ICommonObject,
|
||||
INodeData,
|
||||
INodeOutputsValue,
|
||||
INodeParams
|
||||
} from '../../../src'
|
||||
import { Client, ClientOptions } from '@elastic/elasticsearch'
|
||||
import { ElasticClientArgs, ElasticVectorSearch } from 'langchain/vectorstores/elasticsearch'
|
||||
import { Embeddings } from 'langchain/embeddings/base'
|
||||
import { VectorStore } from 'langchain/vectorstores/base'
|
||||
import { Document } from 'langchain/document'
|
||||
|
||||
export abstract class ElasticSearchBase {
|
||||
label: string
|
||||
name: string
|
||||
version: number
|
||||
description: string
|
||||
type: string
|
||||
icon: string
|
||||
category: string
|
||||
baseClasses: string[]
|
||||
inputs: INodeParams[]
|
||||
credential: INodeParams
|
||||
outputs: INodeOutputsValue[]
|
||||
|
||||
protected constructor() {
|
||||
this.type = 'Elasticsearch'
|
||||
this.icon = 'elasticsearch.png'
|
||||
this.category = 'Vector Stores'
|
||||
this.baseClasses = [this.type, 'VectorStoreRetriever', 'BaseRetriever']
|
||||
this.credential = {
|
||||
label: 'Connect Credential',
|
||||
name: 'credential',
|
||||
type: 'credential',
|
||||
credentialNames: ['elasticsearchApi', 'elasticSearchUserPassword']
|
||||
}
|
||||
this.inputs = [
|
||||
{
|
||||
label: 'Embeddings',
|
||||
name: 'embeddings',
|
||||
type: 'Embeddings'
|
||||
},
|
||||
{
|
||||
label: 'Index Name',
|
||||
name: 'indexName',
|
||||
placeholder: '<INDEX_NAME>',
|
||||
type: 'string'
|
||||
},
|
||||
{
|
||||
label: 'Top K',
|
||||
name: 'topK',
|
||||
description: 'Number of top results to fetch. Default to 4',
|
||||
placeholder: '4',
|
||||
type: 'number',
|
||||
additionalParams: true,
|
||||
optional: true
|
||||
},
|
||||
{
|
||||
label: 'Similarity',
|
||||
name: 'similarity',
|
||||
description: 'Similarity measure used in Elasticsearch.',
|
||||
type: 'options',
|
||||
default: 'l2_norm',
|
||||
options: [
|
||||
{
|
||||
label: 'l2_norm',
|
||||
name: 'l2_norm'
|
||||
},
|
||||
{
|
||||
label: 'dot_product',
|
||||
name: 'dot_product'
|
||||
},
|
||||
{
|
||||
label: 'cosine',
|
||||
name: 'cosine'
|
||||
}
|
||||
],
|
||||
additionalParams: true,
|
||||
optional: true
|
||||
}
|
||||
]
|
||||
this.outputs = [
|
||||
{
|
||||
label: 'Elasticsearch Retriever',
|
||||
name: 'retriever',
|
||||
baseClasses: this.baseClasses
|
||||
},
|
||||
{
|
||||
label: 'Elasticsearch Vector Store',
|
||||
name: 'vectorStore',
|
||||
baseClasses: [this.type, ...getBaseClasses(ElasticVectorSearch)]
|
||||
}
|
||||
]
|
||||
}
|
||||
|
||||
abstract constructVectorStore(
|
||||
embeddings: Embeddings,
|
||||
elasticSearchClientArgs: ElasticClientArgs,
|
||||
docs: Document<Record<string, any>>[] | undefined
|
||||
): Promise<VectorStore>
|
||||
|
||||
async init(nodeData: INodeData, _: string, options: ICommonObject, docs: Document<Record<string, any>>[] | undefined): Promise<any> {
|
||||
const credentialData = await getCredentialData(nodeData.credential ?? '', options)
|
||||
const endPoint = getCredentialParam('endpoint', credentialData, nodeData)
|
||||
const cloudId = getCredentialParam('cloudId', credentialData, nodeData)
|
||||
const indexName = nodeData.inputs?.indexName as string
|
||||
const embeddings = nodeData.inputs?.embeddings as Embeddings
|
||||
const topK = nodeData.inputs?.topK as string
|
||||
const similarityMeasure = nodeData.inputs?.similarityMeasure as string
|
||||
const k = topK ? parseFloat(topK) : 4
|
||||
const output = nodeData.outputs?.output as string
|
||||
|
||||
const elasticSearchClientArgs = this.prepareClientArgs(endPoint, cloudId, credentialData, nodeData, similarityMeasure, indexName)
|
||||
|
||||
const vectorStore = await this.constructVectorStore(embeddings, elasticSearchClientArgs, docs)
|
||||
|
||||
if (output === 'retriever') {
|
||||
return vectorStore.asRetriever(k)
|
||||
} else if (output === 'vectorStore') {
|
||||
;(vectorStore as any).k = k
|
||||
return vectorStore
|
||||
}
|
||||
return vectorStore
|
||||
}
|
||||
|
||||
protected prepareConnectionOptions(
|
||||
endPoint: string | undefined,
|
||||
cloudId: string | undefined,
|
||||
credentialData: ICommonObject,
|
||||
nodeData: INodeData
|
||||
) {
|
||||
let elasticSearchClientOptions: ClientOptions = {}
|
||||
if (endPoint) {
|
||||
let apiKey = getCredentialParam('apiKey', credentialData, nodeData)
|
||||
elasticSearchClientOptions = {
|
||||
node: endPoint,
|
||||
auth: {
|
||||
apiKey: apiKey
|
||||
}
|
||||
}
|
||||
} else if (cloudId) {
|
||||
let username = getCredentialParam('username', credentialData, nodeData)
|
||||
let password = getCredentialParam('password', credentialData, nodeData)
|
||||
elasticSearchClientOptions = {
|
||||
cloud: {
|
||||
id: cloudId
|
||||
},
|
||||
auth: {
|
||||
username: username,
|
||||
password: password
|
||||
}
|
||||
}
|
||||
}
|
||||
return elasticSearchClientOptions
|
||||
}
|
||||
|
||||
protected prepareClientArgs(
|
||||
endPoint: string | undefined,
|
||||
cloudId: string | undefined,
|
||||
credentialData: ICommonObject,
|
||||
nodeData: INodeData,
|
||||
similarityMeasure: string,
|
||||
indexName: string
|
||||
) {
|
||||
let elasticSearchClientOptions = this.prepareConnectionOptions(endPoint, cloudId, credentialData, nodeData)
|
||||
let vectorSearchOptions = {}
|
||||
switch (similarityMeasure) {
|
||||
case 'dot_product':
|
||||
vectorSearchOptions = {
|
||||
similarity: 'dot_product'
|
||||
}
|
||||
break
|
||||
case 'cosine':
|
||||
vectorSearchOptions = {
|
||||
similarity: 'cosine'
|
||||
}
|
||||
break
|
||||
default:
|
||||
vectorSearchOptions = {
|
||||
similarity: 'l2_norm'
|
||||
}
|
||||
}
|
||||
const elasticSearchClientArgs: ElasticClientArgs = {
|
||||
client: new Client(elasticSearchClientOptions),
|
||||
indexName: indexName,
|
||||
vectorSearchOptions: vectorSearchOptions
|
||||
}
|
||||
return elasticSearchClientArgs
|
||||
}
|
||||
}
|
||||
|
|
@ -1,110 +1,30 @@
|
|||
import { ICommonObject, INode, INodeData, INodeOutputsValue, INodeParams } from '../../../src/Interface'
|
||||
import { ICommonObject, INode, INodeData } from '../../../src/Interface'
|
||||
import { Embeddings } from 'langchain/embeddings/base'
|
||||
import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src'
|
||||
|
||||
import { Client, ClientOptions } from '@elastic/elasticsearch'
|
||||
import { ElasticClientArgs, ElasticVectorSearch } from 'langchain/vectorstores/elasticsearch'
|
||||
import { ElasticSearchBase } from './ElasticSearchBase'
|
||||
import { VectorStore } from 'langchain/vectorstores/base'
|
||||
import { Document } from 'langchain/document'
|
||||
|
||||
class ElasicsearchExisting_VectorStores implements INode {
|
||||
label: string
|
||||
name: string
|
||||
version: number
|
||||
description: string
|
||||
type: string
|
||||
icon: string
|
||||
category: string
|
||||
baseClasses: string[]
|
||||
inputs: INodeParams[]
|
||||
credential: INodeParams
|
||||
outputs: INodeOutputsValue[]
|
||||
|
||||
class ElasicsearchExisting_VectorStores extends ElasticSearchBase implements INode {
|
||||
constructor() {
|
||||
super()
|
||||
this.label = 'Elasticsearch Load Existing Index'
|
||||
this.name = 'ElasticsearchIndex'
|
||||
this.version = 1.0
|
||||
this.type = 'Elasticsearch'
|
||||
this.icon = 'elasticsearch.png'
|
||||
this.category = 'Vector Stores'
|
||||
this.description = 'Load existing index from Elasticsearch (i.e: Document has been upserted)'
|
||||
this.baseClasses = [this.type, 'VectorStoreRetriever', 'BaseRetriever']
|
||||
this.credential = {
|
||||
label: 'Connect Credential',
|
||||
name: 'credential',
|
||||
type: 'credential',
|
||||
credentialNames: ['elasticsearchApi', 'elasticSearchUserPassword']
|
||||
}
|
||||
this.inputs = [
|
||||
{
|
||||
label: 'Embeddings',
|
||||
name: 'embeddings',
|
||||
type: 'Embeddings'
|
||||
},
|
||||
{
|
||||
label: 'Index Name',
|
||||
name: 'indexName',
|
||||
placeholder: '<INDEX_NAME>',
|
||||
type: 'string'
|
||||
},
|
||||
{
|
||||
label: 'Top K',
|
||||
name: 'topK',
|
||||
description: 'Number of top results to fetch. Default to 4',
|
||||
placeholder: '4',
|
||||
type: 'number',
|
||||
additionalParams: true,
|
||||
optional: true
|
||||
}
|
||||
]
|
||||
this.outputs = [
|
||||
{
|
||||
label: 'Elasticsearch Retriever',
|
||||
name: 'retriever',
|
||||
baseClasses: this.baseClasses
|
||||
},
|
||||
{
|
||||
label: 'Elasticsearch Vector Store',
|
||||
name: 'vectorStore',
|
||||
baseClasses: [this.type, ...getBaseClasses(ElasticVectorSearch)]
|
||||
}
|
||||
]
|
||||
this.description = 'Load existing index from Elasticsearch (i.e: Document has been upserted)'
|
||||
}
|
||||
|
||||
async constructVectorStore(
|
||||
embeddings: Embeddings,
|
||||
elasticSearchClientArgs: ElasticClientArgs,
|
||||
docs: Document<Record<string, any>>[] | undefined
|
||||
): Promise<VectorStore> {
|
||||
return await ElasticVectorSearch.fromExistingIndex(embeddings, elasticSearchClientArgs)
|
||||
}
|
||||
|
||||
async init(nodeData: INodeData, _: string, options: ICommonObject): Promise<any> {
|
||||
const credentialData = await getCredentialData(nodeData.credential ?? '', options)
|
||||
const endPoint = getCredentialParam('endpoint', credentialData, nodeData)
|
||||
const apiKey = getCredentialParam('apiKey', credentialData, nodeData)
|
||||
const indexName = nodeData.inputs?.indexName as string
|
||||
const embeddings = nodeData.inputs?.embeddings as Embeddings
|
||||
const topK = nodeData.inputs?.topK as string
|
||||
|
||||
const k = topK ? parseFloat(topK) : 4
|
||||
const output = nodeData.outputs?.output as string
|
||||
|
||||
// eslint-disable-next-line no-console
|
||||
console.log('EndPoint:: ' + endPoint + ', APIKey:: ' + apiKey + ', Index:: ' + indexName)
|
||||
|
||||
const elasticSearchClientOptions: ClientOptions = {
|
||||
node: endPoint,
|
||||
auth: {
|
||||
apiKey: apiKey
|
||||
}
|
||||
}
|
||||
|
||||
const elasticSearchClientArgs: ElasticClientArgs = {
|
||||
client: new Client(elasticSearchClientOptions),
|
||||
indexName: indexName
|
||||
}
|
||||
|
||||
const vectorStore = await ElasticVectorSearch.fromExistingIndex(embeddings, elasticSearchClientArgs)
|
||||
// eslint-disable-next-line no-console
|
||||
console.log('vectorStore ::' + vectorStore._vectorstoreType())
|
||||
if (output === 'retriever') {
|
||||
return vectorStore.asRetriever(k)
|
||||
} else if (output === 'vectorStore') {
|
||||
;(vectorStore as any).k = k
|
||||
return vectorStore
|
||||
}
|
||||
return vectorStore
|
||||
return super.init(nodeData, _, options, undefined)
|
||||
}
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -1,148 +1,39 @@
|
|||
import { ICommonObject, INode, INodeData, INodeOutputsValue, INodeParams } from '../../../src/Interface'
|
||||
import { ICommonObject, INode, INodeData } from '../../../src/Interface'
|
||||
import { Embeddings } from 'langchain/embeddings/base'
|
||||
import { Document } from 'langchain/document'
|
||||
import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src'
|
||||
|
||||
import { Client, ClientOptions } from '@elastic/elasticsearch'
|
||||
import { ElasticClientArgs, ElasticVectorSearch } from 'langchain/vectorstores/elasticsearch'
|
||||
import { flatten } from 'lodash'
|
||||
import { ElasticSearchBase } from './ElasticSearchBase'
|
||||
import { VectorStore } from 'langchain/vectorstores/base'
|
||||
|
||||
class ElasicsearchUpsert_VectorStores implements INode {
|
||||
label: string
|
||||
name: string
|
||||
version: number
|
||||
description: string
|
||||
type: string
|
||||
icon: string
|
||||
category: string
|
||||
baseClasses: string[]
|
||||
inputs: INodeParams[]
|
||||
credential: INodeParams
|
||||
outputs: INodeOutputsValue[]
|
||||
|
||||
class ElasicsearchUpsert_VectorStores extends ElasticSearchBase implements INode {
|
||||
constructor() {
|
||||
super()
|
||||
this.label = 'Elasticsearch Upsert Document'
|
||||
this.name = 'ElasticsearchUpsert'
|
||||
this.version = 1.0
|
||||
this.type = 'Elasticsearch'
|
||||
this.icon = 'elasticsearch.png'
|
||||
this.category = 'Vector Stores'
|
||||
this.description = 'Upsert documents to Elasticsearch'
|
||||
this.baseClasses = [this.type, 'VectorStoreRetriever', 'BaseRetriever']
|
||||
this.credential = {
|
||||
label: 'Connect Credential',
|
||||
name: 'credential',
|
||||
type: 'credential',
|
||||
credentialNames: ['elasticsearchApi', 'elasticSearchUserPassword']
|
||||
}
|
||||
this.inputs = [
|
||||
{
|
||||
label: 'Document',
|
||||
name: 'document',
|
||||
type: 'Document',
|
||||
list: true
|
||||
},
|
||||
{
|
||||
label: 'Embeddings',
|
||||
name: 'embeddings',
|
||||
type: 'Embeddings'
|
||||
},
|
||||
{
|
||||
label: 'Index Name',
|
||||
name: 'indexName',
|
||||
placeholder: '<INDEX_NAME>',
|
||||
type: 'string'
|
||||
},
|
||||
{
|
||||
label: 'Top K',
|
||||
name: 'topK',
|
||||
description: 'Number of top results to fetch. Default to 4',
|
||||
placeholder: '4',
|
||||
type: 'number',
|
||||
additionalParams: true,
|
||||
optional: true
|
||||
},
|
||||
{
|
||||
label: 'Similarity',
|
||||
name: 'similarity',
|
||||
description: 'Similarity measure used in Elasticsearch.',
|
||||
type: 'options',
|
||||
default: 'l2_norm',
|
||||
options: [
|
||||
{
|
||||
label: 'l2_norm',
|
||||
name: 'l2_norm'
|
||||
},
|
||||
{
|
||||
label: 'dot_product',
|
||||
name: 'dot_product'
|
||||
},
|
||||
{
|
||||
label: 'cosine',
|
||||
name: 'cosine'
|
||||
}
|
||||
],
|
||||
additionalParams: true,
|
||||
optional: true
|
||||
}
|
||||
]
|
||||
this.outputs = [
|
||||
{
|
||||
label: 'Elasticsearch Retriever',
|
||||
name: 'retriever',
|
||||
baseClasses: this.baseClasses
|
||||
},
|
||||
{
|
||||
label: 'Elasticsearch Vector Store',
|
||||
name: 'vectorStore',
|
||||
baseClasses: [this.type, ...getBaseClasses(ElasticVectorSearch)]
|
||||
}
|
||||
]
|
||||
this.inputs.unshift({
|
||||
label: 'Document',
|
||||
name: 'document',
|
||||
type: 'Document',
|
||||
list: true
|
||||
})
|
||||
}
|
||||
|
||||
async constructVectorStore(
|
||||
embeddings: Embeddings,
|
||||
elasticSearchClientArgs: ElasticClientArgs,
|
||||
docs: Document<Record<string, any>>[]
|
||||
): Promise<VectorStore> {
|
||||
const vectorStore = new ElasticVectorSearch(embeddings, elasticSearchClientArgs)
|
||||
await vectorStore.addDocuments(docs)
|
||||
return vectorStore
|
||||
}
|
||||
|
||||
async init(nodeData: INodeData, _: string, options: ICommonObject): Promise<any> {
|
||||
const credentialData = await getCredentialData(nodeData.credential ?? '', options)
|
||||
const endPoint = getCredentialParam('endpoint', credentialData, nodeData)
|
||||
const apiKey = getCredentialParam('apiKey', credentialData, nodeData)
|
||||
const docs = nodeData.inputs?.document as Document[]
|
||||
const indexName = nodeData.inputs?.indexName as string
|
||||
const embeddings = nodeData.inputs?.embeddings as Embeddings
|
||||
const topK = nodeData.inputs?.topK as string
|
||||
const k = topK ? parseFloat(topK) : 4
|
||||
const output = nodeData.outputs?.output as string
|
||||
const similarityMeasure = nodeData.inputs?.similarityMeasure as string
|
||||
|
||||
// eslint-disable-next-line no-console
|
||||
console.log('EndPoint:: ' + endPoint + ', APIKey:: ' + apiKey + ', Index:: ' + indexName)
|
||||
|
||||
const elasticSearchClientOptions: ClientOptions = {
|
||||
node: endPoint,
|
||||
auth: {
|
||||
apiKey: apiKey
|
||||
}
|
||||
}
|
||||
let vectorSearchOptions = {}
|
||||
switch (similarityMeasure) {
|
||||
case 'dot_product':
|
||||
vectorSearchOptions = {
|
||||
similarity: 'dot_product'
|
||||
}
|
||||
break
|
||||
case 'cosine':
|
||||
vectorSearchOptions = {
|
||||
similarity: 'cosine'
|
||||
}
|
||||
break
|
||||
default:
|
||||
vectorSearchOptions = {
|
||||
similarity: 'l2_norm'
|
||||
}
|
||||
}
|
||||
const elasticSearchClientArgs: ElasticClientArgs = {
|
||||
client: new Client(elasticSearchClientOptions),
|
||||
indexName: indexName,
|
||||
vectorSearchOptions: vectorSearchOptions
|
||||
}
|
||||
|
||||
const flattenDocs = docs && docs.length ? flatten(docs) : []
|
||||
const finalDocs = []
|
||||
|
|
@ -150,15 +41,14 @@ class ElasicsearchUpsert_VectorStores implements INode {
|
|||
finalDocs.push(new Document(flattenDocs[i]))
|
||||
}
|
||||
|
||||
const vectorStore = await ElasticVectorSearch.fromDocuments(finalDocs, embeddings, elasticSearchClientArgs)
|
||||
|
||||
if (output === 'retriever') {
|
||||
return vectorStore.asRetriever(k)
|
||||
} else if (output === 'vectorStore') {
|
||||
;(vectorStore as any).k = k
|
||||
return vectorStore
|
||||
}
|
||||
return vectorStore
|
||||
// The following code is a workaround for a bug (Langchain Issue #1589) in the underlying library.
|
||||
// Store does not support object in metadata and fail silently
|
||||
finalDocs.forEach((d) => {
|
||||
delete d.metadata.pdf
|
||||
delete d.metadata.loc
|
||||
})
|
||||
// end of workaround
|
||||
return super.init(nodeData, _, options, flattenDocs)
|
||||
}
|
||||
}
|
||||
|
||||
|
|
|
|||
Loading…
Reference in New Issue