298 lines
12 KiB
TypeScript
298 lines
12 KiB
TypeScript
import { flatten } from 'lodash'
|
|
import { v4 as uuidv4 } from 'uuid'
|
|
import { createClient } from '@supabase/supabase-js'
|
|
import { Document } from '@langchain/core/documents'
|
|
import { Embeddings } from '@langchain/core/embeddings'
|
|
import { SupabaseVectorStore, SupabaseLibArgs } from '@langchain/community/vectorstores/supabase'
|
|
import { ICommonObject, INode, INodeData, INodeOutputsValue, INodeParams, IndexingResult } from '../../../src/Interface'
|
|
import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src/utils'
|
|
import { addMMRInputParams, resolveVectorStoreOrRetriever } from '../VectorStoreUtils'
|
|
import { index } from '../../../src/indexing'
|
|
import { FilterParser } from './filterParser'
|
|
|
|
class Supabase_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 = 'Supabase'
|
|
this.name = 'supabase'
|
|
this.version = 4.0
|
|
this.type = 'Supabase'
|
|
this.icon = 'supabase.svg'
|
|
this.category = 'Vector Stores'
|
|
this.description = 'Upsert embedded data and perform similarity or mmr search upon query using Supabase via pgvector extension'
|
|
this.baseClasses = [this.type, 'VectorStoreRetriever', 'BaseRetriever']
|
|
this.credential = {
|
|
label: 'Connect Credential',
|
|
name: 'credential',
|
|
type: 'credential',
|
|
credentialNames: ['supabaseApi']
|
|
}
|
|
this.inputs = [
|
|
{
|
|
label: 'Document',
|
|
name: 'document',
|
|
type: 'Document',
|
|
list: true,
|
|
optional: true
|
|
},
|
|
{
|
|
label: 'Embeddings',
|
|
name: 'embeddings',
|
|
type: 'Embeddings'
|
|
},
|
|
{
|
|
label: 'Record Manager',
|
|
name: 'recordManager',
|
|
type: 'RecordManager',
|
|
description: 'Keep track of the record to prevent duplication',
|
|
optional: true
|
|
},
|
|
{
|
|
label: 'Supabase Project URL',
|
|
name: 'supabaseProjUrl',
|
|
type: 'string'
|
|
},
|
|
{
|
|
label: 'Table Name',
|
|
name: 'tableName',
|
|
type: 'string'
|
|
},
|
|
{
|
|
label: 'Query Name',
|
|
name: 'queryName',
|
|
type: 'string'
|
|
},
|
|
{
|
|
label: 'Supabase Metadata Filter',
|
|
name: 'supabaseMetadataFilter',
|
|
type: 'json',
|
|
optional: true,
|
|
additionalParams: true,
|
|
acceptVariable: true
|
|
},
|
|
{
|
|
label: 'Supabase RPC Filter',
|
|
name: 'supabaseRPCFilter',
|
|
type: 'string',
|
|
rows: 4,
|
|
placeholder: `filter("metadata->a::int", "gt", 5)
|
|
.filter("metadata->c::int", "gt", 7)
|
|
.filter("metadata->>stuff", "eq", "right");`,
|
|
description:
|
|
'Query builder-style filtering. If this is set, will override the metadata filter. Refer <a href="https://js.langchain.com/v0.1/docs/integrations/vectorstores/supabase/#metadata-query-builder-filtering" target="_blank">here</a> for more information',
|
|
optional: true,
|
|
additionalParams: true
|
|
},
|
|
{
|
|
label: 'Top K',
|
|
name: 'topK',
|
|
description: 'Number of top results to fetch. Default to 4',
|
|
placeholder: '4',
|
|
type: 'number',
|
|
additionalParams: true,
|
|
optional: true
|
|
}
|
|
]
|
|
addMMRInputParams(this.inputs)
|
|
this.outputs = [
|
|
{
|
|
label: 'Supabase Retriever',
|
|
name: 'retriever',
|
|
baseClasses: this.baseClasses
|
|
},
|
|
{
|
|
label: 'Supabase Vector Store',
|
|
name: 'vectorStore',
|
|
baseClasses: [this.type, ...getBaseClasses(SupabaseVectorStore)]
|
|
}
|
|
]
|
|
}
|
|
|
|
//@ts-ignore
|
|
vectorStoreMethods = {
|
|
async upsert(nodeData: INodeData, options: ICommonObject): Promise<Partial<IndexingResult>> {
|
|
const supabaseProjUrl = nodeData.inputs?.supabaseProjUrl as string
|
|
const tableName = nodeData.inputs?.tableName as string
|
|
const queryName = nodeData.inputs?.queryName as string
|
|
const docs = nodeData.inputs?.document as Document[]
|
|
const embeddings = nodeData.inputs?.embeddings as Embeddings
|
|
const recordManager = nodeData.inputs?.recordManager
|
|
|
|
const credentialData = await getCredentialData(nodeData.credential ?? '', options)
|
|
const supabaseApiKey = getCredentialParam('supabaseApiKey', credentialData, nodeData)
|
|
|
|
const client = createClient(supabaseProjUrl, supabaseApiKey)
|
|
|
|
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 {
|
|
if (recordManager) {
|
|
const vectorStore = await SupabaseUpsertVectorStore.fromExistingIndex(embeddings, {
|
|
client,
|
|
tableName: tableName,
|
|
queryName: queryName
|
|
})
|
|
await recordManager.createSchema()
|
|
const res = await index({
|
|
docsSource: finalDocs,
|
|
recordManager,
|
|
vectorStore,
|
|
options: {
|
|
cleanup: recordManager?.cleanup,
|
|
sourceIdKey: recordManager?.sourceIdKey ?? 'source',
|
|
vectorStoreName: tableName + '_' + queryName
|
|
}
|
|
})
|
|
return res
|
|
} else {
|
|
await SupabaseUpsertVectorStore.fromDocuments(finalDocs, embeddings, {
|
|
client,
|
|
tableName: tableName,
|
|
queryName: queryName
|
|
})
|
|
return { numAdded: finalDocs.length, addedDocs: finalDocs }
|
|
}
|
|
} catch (e) {
|
|
throw new Error(e)
|
|
}
|
|
},
|
|
async delete(nodeData: INodeData, ids: string[], options: ICommonObject): Promise<void> {
|
|
const supabaseProjUrl = nodeData.inputs?.supabaseProjUrl as string
|
|
const tableName = nodeData.inputs?.tableName as string
|
|
const queryName = nodeData.inputs?.queryName as string
|
|
const embeddings = nodeData.inputs?.embeddings as Embeddings
|
|
const recordManager = nodeData.inputs?.recordManager
|
|
|
|
const credentialData = await getCredentialData(nodeData.credential ?? '', options)
|
|
const supabaseApiKey = getCredentialParam('supabaseApiKey', credentialData, nodeData)
|
|
|
|
const client = createClient(supabaseProjUrl, supabaseApiKey)
|
|
|
|
const supabaseStore = new SupabaseVectorStore(embeddings, {
|
|
client,
|
|
tableName: tableName,
|
|
queryName: queryName
|
|
})
|
|
|
|
try {
|
|
if (recordManager) {
|
|
const vectorStoreName = tableName + '_' + queryName
|
|
await recordManager.createSchema()
|
|
;(recordManager as any).namespace = (recordManager as any).namespace + '_' + vectorStoreName
|
|
const keys: string[] = await recordManager.listKeys({})
|
|
|
|
await supabaseStore.delete({ ids: keys })
|
|
await recordManager.deleteKeys(keys)
|
|
} else {
|
|
await supabaseStore.delete({ ids })
|
|
}
|
|
} catch (e) {
|
|
throw new Error(e)
|
|
}
|
|
}
|
|
}
|
|
|
|
async init(nodeData: INodeData, _: string, options: ICommonObject): Promise<any> {
|
|
const supabaseProjUrl = nodeData.inputs?.supabaseProjUrl as string
|
|
const tableName = nodeData.inputs?.tableName as string
|
|
const queryName = nodeData.inputs?.queryName as string
|
|
const embeddings = nodeData.inputs?.embeddings as Embeddings
|
|
const supabaseMetadataFilter = nodeData.inputs?.supabaseMetadataFilter
|
|
const supabaseRPCFilter = nodeData.inputs?.supabaseRPCFilter
|
|
|
|
const credentialData = await getCredentialData(nodeData.credential ?? '', options)
|
|
const supabaseApiKey = getCredentialParam('supabaseApiKey', credentialData, nodeData)
|
|
|
|
const client = createClient(supabaseProjUrl, supabaseApiKey)
|
|
|
|
const obj: SupabaseLibArgs = {
|
|
client,
|
|
tableName,
|
|
queryName
|
|
}
|
|
|
|
if (supabaseMetadataFilter) {
|
|
const metadatafilter = typeof supabaseMetadataFilter === 'object' ? supabaseMetadataFilter : JSON.parse(supabaseMetadataFilter)
|
|
obj.filter = metadatafilter
|
|
}
|
|
|
|
if (supabaseRPCFilter) {
|
|
obj.filter = FilterParser.parseFilterString(supabaseRPCFilter)
|
|
}
|
|
|
|
const vectorStore = await SupabaseVectorStore.fromExistingIndex(embeddings, obj)
|
|
|
|
return resolveVectorStoreOrRetriever(nodeData, vectorStore, obj.filter)
|
|
}
|
|
}
|
|
|
|
class SupabaseUpsertVectorStore extends SupabaseVectorStore {
|
|
async addVectors(vectors: number[][], documents: Document[], options?: { ids?: string[] | number[] }): Promise<string[]> {
|
|
if (vectors.length === 0) {
|
|
return []
|
|
}
|
|
const rows = vectors.map((embedding, idx) => ({
|
|
content: documents[idx].pageContent,
|
|
embedding,
|
|
metadata: documents[idx].metadata
|
|
}))
|
|
|
|
let returnedIds: string[] = []
|
|
for (let i = 0; i < rows.length; i += this.upsertBatchSize) {
|
|
const chunk = rows.slice(i, i + this.upsertBatchSize).map((row, j) => {
|
|
if (options?.ids) {
|
|
return { id: options.ids[i + j], ...row }
|
|
}
|
|
return row
|
|
})
|
|
|
|
let res = await this.client.from(this.tableName).upsert(chunk).select()
|
|
|
|
if (res.error) {
|
|
// If the error is due to null value in column "id", we will generate a new id for the row
|
|
if (res.error.message.includes(`null value in column "id"`)) {
|
|
const chunk = rows.slice(i, i + this.upsertBatchSize).map((row, y) => {
|
|
if (options?.ids) {
|
|
return { id: options.ids[i + y], ...row }
|
|
}
|
|
return { id: uuidv4(), ...row }
|
|
})
|
|
res = await this.client.from(this.tableName).upsert(chunk).select()
|
|
|
|
if (res.error) {
|
|
throw new Error(`Error inserting: ${res.error.message} ${res.status} ${res.statusText}`)
|
|
}
|
|
} else {
|
|
throw new Error(`Error inserting: ${res.error.message} ${res.status} ${res.statusText}`)
|
|
}
|
|
}
|
|
|
|
if (res.data) {
|
|
returnedIds = returnedIds.concat(res.data.map((row) => row.id))
|
|
}
|
|
}
|
|
|
|
return returnedIds
|
|
}
|
|
}
|
|
|
|
module.exports = { nodeClass: Supabase_VectorStores }
|