farcasclaudiu/Flowise
1
0
Fork 0
mirror of https://github.com/farcasclaudiu/Flowise.git synced 2026-06-28 19:00:59 +03:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #110 from FlowiseAI/bugfix/MultipleDocuments

Browse Source 
Bugfix/Add fixes to multiple documents
This commit is contained in:
Henry Heng
2023-05-16 12:55:39 +01:00
committed by GitHub
parent 1d9181f099 daf7fd3a50
commit 5d73189179
7 changed files with 136 additions and 85 deletions
Download Patch File Download Diff File Expand all files Collapse all files
packages/components/nodes/documentloaders/Cheerio/Cheerio.ts
+8 -7
View File
@@ -3,6 +3,7 @@ import { TextSplitter } from 'langchain/text_splitter'
import { CheerioWebBaseLoader } from 'langchain/document_loaders/web/cheerio'
import { test } from 'linkifyjs'
import { getAvailableURLs } from '../../../src'
class Cheerio_DocumentLoaders implements INode {
label: string
name: string
@@ -33,13 +34,6 @@ class Cheerio_DocumentLoaders implements INode {
type: 'TextSplitter',
optional: true
},
{
label: 'Metadata',
name: 'metadata',
type: 'json',
optional: true,
additionalParams: true
},
{
label: 'Web Scrap for Relative Links',
name: 'webScrap',
@@ -54,6 +48,13 @@ class Cheerio_DocumentLoaders implements INode {
default: 10,
optional: true,
additionalParams: true
},
{
label: 'Metadata',
name: 'metadata',
type: 'json',
optional: true,
additionalParams: true
}
]
}
packages/components/nodes/documentloaders/Csv/Csv.ts
+23 -9
View File
@@ -1,7 +1,6 @@
import { INode, INodeData, INodeParams } from '../../../src/Interface'
import { TextSplitter } from 'langchain/text_splitter'
import { CSVLoader } from 'langchain/document_loaders/fs/csv'
import { getBlob } from '../../../src/utils'
class Csv_DocumentLoaders implements INode {
label: string
@@ -58,20 +57,35 @@ class Csv_DocumentLoaders implements INode {
const columnName = nodeData.inputs?.columnName as string
const metadata = nodeData.inputs?.metadata
const blob = new Blob(getBlob(csvFileBase64))
const loader = new CSVLoader(blob, columnName.trim().length === 0 ? undefined : columnName.trim())
let docs = []
let alldocs = []
let files: string[] = []
if (textSplitter) {
docs = await loader.loadAndSplit(textSplitter)
if (csvFileBase64.startsWith('[') && csvFileBase64.endsWith(']')) {
files = JSON.parse(csvFileBase64)
} else {
docs = await loader.load()
files = [csvFileBase64]
}
for (const file of files) {
const splitDataURI = file.split(',')
splitDataURI.pop()
const bf = Buffer.from(splitDataURI.pop() || '', 'base64')
const blob = new Blob([bf])
const loader = new CSVLoader(blob, columnName.trim().length === 0 ? undefined : columnName.trim())
if (textSplitter) {
const docs = await loader.loadAndSplit(textSplitter)
alldocs.push(...docs)
} else {
const docs = await loader.load()
alldocs.push(...docs)
}
}
if (metadata) {
const parsedMetadata = typeof metadata === 'object' ? metadata : JSON.parse(metadata)
let finaldocs = []
for (const doc of docs) {
for (const doc of alldocs) {
const newdoc = {
...doc,
metadata: {
@@ -84,7 +98,7 @@ class Csv_DocumentLoaders implements INode {
return finaldocs
}
return docs
return alldocs
}
}
packages/components/nodes/documentloaders/Docx/Docx.ts
+23 -9
View File
@@ -1,7 +1,6 @@
import { INode, INodeData, INodeParams } from '../../../src/Interface'
import { TextSplitter } from 'langchain/text_splitter'
import { DocxLoader } from 'langchain/document_loaders/fs/docx'
import { getBlob } from '../../../src/utils'
class Docx_DocumentLoaders implements INode {
label: string
@@ -49,20 +48,35 @@ class Docx_DocumentLoaders implements INode {
const docxFileBase64 = nodeData.inputs?.docxFile as string
const metadata = nodeData.inputs?.metadata
const blob = new Blob(getBlob(docxFileBase64))
const loader = new DocxLoader(blob)
let docs = []
let alldocs = []
let files: string[] = []
if (textSplitter) {
docs = await loader.loadAndSplit(textSplitter)
if (docxFileBase64.startsWith('[') && docxFileBase64.endsWith(']')) {
files = JSON.parse(docxFileBase64)
} else {
docs = await loader.load()
files = [docxFileBase64]
}
for (const file of files) {
const splitDataURI = file.split(',')
splitDataURI.pop()
const bf = Buffer.from(splitDataURI.pop() || '', 'base64')
const blob = new Blob([bf])
const loader = new DocxLoader(blob)
if (textSplitter) {
const docs = await loader.loadAndSplit(textSplitter)
alldocs.push(...docs)
} else {
const docs = await loader.load()
alldocs.push(...docs)
}
}
if (metadata) {
const parsedMetadata = typeof metadata === 'object' ? metadata : JSON.parse(metadata)
let finaldocs = []
for (const doc of docs) {
for (const doc of alldocs) {
const newdoc = {
...doc,
metadata: {
@@ -75,7 +89,7 @@ class Docx_DocumentLoaders implements INode {
return finaldocs
}
return docs
return alldocs
}
}
packages/components/nodes/documentloaders/Json/Json.ts
+23 -9
View File
@@ -1,7 +1,6 @@
import { INode, INodeData, INodeParams } from '../../../src/Interface'
import { TextSplitter } from 'langchain/text_splitter'
import { JSONLoader } from 'langchain/document_loaders/fs/json'
import { getBlob } from '../../../src/utils'
class Json_DocumentLoaders implements INode {
label: string
@@ -64,20 +63,35 @@ class Json_DocumentLoaders implements INode {
pointers = outputString.split(',').map((pointer) => '/' + pointer.trim())
}
const blob = new Blob(getBlob(jsonFileBase64))
const loader = new JSONLoader(blob, pointers.length != 0 ? pointers : undefined)
let docs = []
let alldocs = []
let files: string[] = []
if (textSplitter) {
docs = await loader.loadAndSplit(textSplitter)
if (jsonFileBase64.startsWith('[') && jsonFileBase64.endsWith(']')) {
files = JSON.parse(jsonFileBase64)
} else {
docs = await loader.load()
files = [jsonFileBase64]
}
for (const file of files) {
const splitDataURI = file.split(',')
splitDataURI.pop()
const bf = Buffer.from(splitDataURI.pop() || '', 'base64')
const blob = new Blob([bf])
const loader = new JSONLoader(blob, pointers.length != 0 ? pointers : undefined)
if (textSplitter) {
const docs = await loader.loadAndSplit(textSplitter)
alldocs.push(...docs)
} else {
const docs = await loader.load()
alldocs.push(...docs)
}
}
if (metadata) {
const parsedMetadata = typeof metadata === 'object' ? metadata : JSON.parse(metadata)
let finaldocs = []
for (const doc of docs) {
for (const doc of alldocs) {
const newdoc = {
...doc,
metadata: {
@@ -90,7 +104,7 @@ class Json_DocumentLoaders implements INode {
return finaldocs
}
return docs
return alldocs
}
}
packages/components/nodes/documentloaders/Pdf/Pdf.ts
+36 -18
View File
@@ -1,7 +1,6 @@
import { INode, INodeData, INodeParams } from '../../../src/Interface'
import { TextSplitter } from 'langchain/text_splitter'
import { PDFLoader } from 'langchain/document_loaders/fs/pdf'
import { getBlob } from '../../../src/utils'
class Pdf_DocumentLoaders implements INode {
label: string
@@ -66,30 +65,49 @@ class Pdf_DocumentLoaders implements INode {
const usage = nodeData.inputs?.usage as string
const metadata = nodeData.inputs?.metadata
const blob = new Blob(getBlob(pdfFileBase64))
let docs = []
if (usage === 'perFile') {
// @ts-ignore
const loader = new PDFLoader(blob, { splitPages: false, pdfjs: () => import('pdf-parse/lib/pdf.js/v1.10.100/build/pdf.js') })
if (textSplitter) {
docs = await loader.loadAndSplit(textSplitter)
} else {
docs = await loader.load()
}
let alldocs = []
let files: string[] = []
if (pdfFileBase64.startsWith('[') && pdfFileBase64.endsWith(']')) {
files = JSON.parse(pdfFileBase64)
} else {
// @ts-ignore
const loader = new PDFLoader(blob, { pdfjs: () => import('pdf-parse/lib/pdf.js/v1.10.100/build/pdf.js') })
if (textSplitter) {
docs = await loader.loadAndSplit(textSplitter)
files = [pdfFileBase64]
}
for (const file of files) {
const splitDataURI = file.split(',')
splitDataURI.pop()
const bf = Buffer.from(splitDataURI.pop() || '', 'base64')
if (usage === 'perFile') {
const loader = new PDFLoader(new Blob([bf]), {
splitPages: false,
// @ts-ignore
pdfjs: () => import('pdf-parse/lib/pdf.js/v1.10.100/build/pdf.js')
})
if (textSplitter) {
const docs = await loader.loadAndSplit(textSplitter)
alldocs.push(...docs)
} else {
const docs = await loader.load()
alldocs.push(...docs)
}
} else {
docs = await loader.load()
// @ts-ignore
const loader = new PDFLoader(new Blob([bf]), { pdfjs: () => import('pdf-parse/lib/pdf.js/v1.10.100/build/pdf.js') })
if (textSplitter) {
const docs = await loader.loadAndSplit(textSplitter)
alldocs.push(...docs)
} else {
const docs = await loader.load()
alldocs.push(...docs)
}
}
}
if (metadata) {
const parsedMetadata = typeof metadata === 'object' ? metadata : JSON.parse(metadata)
let finaldocs = []
for (const doc of docs) {
for (const doc of alldocs) {
const newdoc = {
...doc,
metadata: {
@@ -102,7 +120,7 @@ class Pdf_DocumentLoaders implements INode {
return finaldocs
}
return docs
return alldocs
}
}
packages/components/nodes/documentloaders/Text/Text.ts
+23 -9
View File
@@ -1,7 +1,6 @@
import { INode, INodeData, INodeParams } from '../../../src/Interface'
import { TextSplitter } from 'langchain/text_splitter'
import { TextLoader } from 'langchain/document_loaders/fs/text'
import { getBlob } from '../../../src/utils'
class Text_DocumentLoaders implements INode {
label: string
@@ -49,20 +48,35 @@ class Text_DocumentLoaders implements INode {
const txtFileBase64 = nodeData.inputs?.txtFile as string
const metadata = nodeData.inputs?.metadata
const blob = new Blob(getBlob(txtFileBase64))
const loader = new TextLoader(blob)
let docs = []
let alldocs = []
let files: string[] = []
if (textSplitter) {
docs = await loader.loadAndSplit(textSplitter)
if (txtFileBase64.startsWith('[') && txtFileBase64.endsWith(']')) {
files = JSON.parse(txtFileBase64)
} else {
docs = await loader.load()
files = [txtFileBase64]
}
for (const file of files) {
const splitDataURI = file.split(',')
splitDataURI.pop()
const bf = Buffer.from(splitDataURI.pop() || '', 'base64')
const blob = new Blob([bf])
const loader = new TextLoader(blob)
if (textSplitter) {
const docs = await loader.loadAndSplit(textSplitter)
alldocs.push(...docs)
} else {
const docs = await loader.load()
alldocs.push(...docs)
}
}
if (metadata) {
const parsedMetadata = typeof metadata === 'object' ? metadata : JSON.parse(metadata)
let finaldocs = []
for (const doc of docs) {
for (const doc of alldocs) {
const newdoc = {
...doc,
metadata: {
@@ -74,7 +88,7 @@ class Text_DocumentLoaders implements INode {
}
return finaldocs
}
return docs
return alldocs
}
}
packages/components/src/utils.ts
-24
View File
@@ -152,30 +152,6 @@ export const getInputVariables = (paramValue: string): string[] => {
return inputVariables
}
/**
* Get blob
* @param {string} fileBase64Str
* @returns {Buffer[]}
*/
export const getBlob = (fileBase64Str: string) => {
let bufferArray: Buffer[] = []
let files: string[] = []
if (fileBase64Str.startsWith('[') && fileBase64Str.endsWith(']')) {
files = JSON.parse(fileBase64Str)
} else {
files = [fileBase64Str]
}
for (const file of files) {
const splitDataURI = file.split(',')
splitDataURI.pop()
const bf = Buffer.from(splitDataURI.pop() || '', 'base64')
bufferArray.push(bf)
}
return bufferArray
}
export const getAvailableURLs = async (url: string, limit: number) => {
try {
const availableUrls: string[] = []