SREDSimplify Docs

Appearance

File Handling & Indexing ​

The platform supports ingesting various document types (PDF, DOCX, CSV, TXT) to serve as context for the AI.

Upload Flow ​

  1. User drops a file into the ContextManagerPane.
  2. The frontend validates the file size and type against constants defined in the app.
  3. The file is uploaded to the backend via a multipart/form-data request.
  4. The backend stores the file as a BLOB in PostgreSQL.

Smart Indexing Polling ​

Because processing large PDFs or Docx files for AI embedding can take time, files have an indexStatus property:

  • PENDING: File metadata exists but indexing has not started yet.
  • INDEXING: Currently being processed by the backend/Python service.
  • READY: Ready to be used as context.
  • FAILED: Processing failed.

Polling Mechanism ​

The frontend utilizes a smart conditional polling strategy via React Query (refetchInterval).

  • It checks the list of files in the current project.
  • If any file has indexStatus === 'INDEXING', the hook polls the backend every 4 seconds.
  • Once all files are no longer INDEXING, polling automatically suspends to preserve network resources.