upload_router

upload_router

Upload / Paste router for ingesting documents into the knowledge store.

Classes

Functions

ingest_paste async

ingest_paste(body: PasteRequest) -> IngestResponse

Ingest pasted text into the knowledge store.

Source code in src/openjarvis/server/upload_router.py

@router.post("/ingest", response_model=IngestResponse)
async def ingest_paste(body: PasteRequest) -> IngestResponse:
    """Ingest pasted text into the knowledge store."""
    text = body.content.strip()
    if not text:
        raise HTTPException(status_code=400, detail="Content is empty")

    store = _get_store()
    doc_id = str(uuid.uuid4())
    chunks = _chunk_text(text)

    for idx, chunk in enumerate(chunks):
        store.store(
            chunk,
            source="upload",
            doc_type="paste",
            doc_id=doc_id,
            title=body.title or "Pasted text",
            chunk_index=idx,
        )

    logger.info("Ingested %d chunks from pasted text (doc_id=%s)", len(chunks), doc_id)
    return IngestResponse(chunks_added=len(chunks))

ingest_files async

ingest_files(files: List[UploadFile] = File(...), title: Optional[str] = Form(None)) -> IngestResponse

Ingest uploaded files into the knowledge store.

Source code in src/openjarvis/server/upload_router.py

@router.post("/ingest/files", response_model=IngestResponse)
async def ingest_files(
    files: List[UploadFile] = File(...),
    title: Optional[str] = Form(None),
) -> IngestResponse:
    """Ingest uploaded files into the knowledge store."""
    store = _get_store()
    total_chunks = 0

    for upload in files:
        filename = upload.filename or "untitled"
        ext = ""
        if "." in filename:
            ext = "." + filename.rsplit(".", 1)[-1].lower()

        if ext not in _ALLOWED_EXTENSIONS:
            allowed = ", ".join(sorted(_ALLOWED_EXTENSIONS))
            raise HTTPException(
                status_code=400,
                detail=(
                    f"Unsupported file type: {ext}. "
                    f"Allowed: {allowed}"
                ),
            )

        data = await upload.read()

        # Parse content based on extension
        if ext in (".txt", ".md", ".csv"):
            try:
                text = data.decode("utf-8")
            except UnicodeDecodeError:
                text = data.decode("latin-1")
        elif ext == ".pdf":
            text = _extract_text_from_pdf(data)
        elif ext == ".docx":
            text = _extract_text_from_docx(data)
        else:
            continue

        text = text.strip()
        if not text:
            continue

        doc_id = str(uuid.uuid4())
        doc_title = title or filename
        chunks = _chunk_text(text)

        for idx, chunk in enumerate(chunks):
            store.store(
                chunk,
                source="upload",
                doc_type=ext.lstrip("."),
                doc_id=doc_id,
                title=doc_title,
                chunk_index=idx,
            )

        total_chunks += len(chunks)
        logger.info(
            "Ingested %d chunks from file %s (doc_id=%s)",
            len(chunks),
            filename,
            doc_id,
        )

    return IngestResponse(chunks_added=total_chunks)