GCP node implementation

[LakshmireddyP]

how to create custom node for adding files to the GCP bucket which are coming from document loaders.

[Ganesh-403]

@LakshmireddyP
There isn't a built-in Google Cloud Storage (GCS) upload node in Flowise today. Document Loaders are designed to read documents (from PDFs, APIs, Google Drive, S3, etc.) and output Flowise Document objects for downstream processing, not to write files to external storage. :contentReference[oaicite:0]{index=0}

If your goal is:

Document Loader → Upload original file(s) to a GCS bucket

then a custom node is the right approach.

High-level implementation

Your custom node would:

1. Accept the output from a Document Loader (or the file path/binary if available).
2. Use the official @google-cloud/storage SDK.
3. Upload the file to your target bucket.
4. Return either:
   • the original documents (so the flow can continue), or
   • the uploaded GCS URI (gs://bucket/path/file.pdf) as metadata.

The upload itself is straightforward with the GCS SDK:

import { Storage } from '@google-cloud/storage'

const storage = new Storage()

await storage
    .bucket(bucketName)
    .upload(localFilePath, {
        destination: destinationPath
    })

Google's Storage SDK is the recommended way to upload objects to Cloud Storage. :contentReference[oaicite:1]{index=1}

Where should the custom node live?

From what I can tell, Flowise doesn't currently have documentation specifically for creating a "GCS Upload" node.

However, existing integrations are implemented as node packages under the Flowise source, so the general approach would be to:

• create a new custom node
• define its inputs (bucket, destination path, credentials, documents/files)
• call the GCS SDK
• return the documents so downstream nodes (embeddings, vector stores, etc.) can continue processing

If you don't need a reusable node, another option is to use the Custom Document Loader, which supports custom JavaScript and async operations in a sandbox. Depending on your workflow, that may be sufficient without modifying the Flowise codebase. :contentReference[oaicite:2]{index=2}

One clarification

Are you trying to upload:

1. the original files (PDF, DOCX, etc.) to GCS, or
2. the processed Document[] objects after the loader has extracted text?

That distinction matters because most Document Loaders emit LangChain Document objects (text + metadata), not the original file bytes. If your goal is to preserve the original files, you'll likely want to intercept them before or during the loading stage rather than after the documents have been parsed.

If this solves your problem, feel free to mark it as the accepted answer so others can find it easily.