Document AI

• Client: google-cloud-documentai (documentai_v1.DocumentProcessorServiceClient)
• Transport: gRPC (the client default), forced to REST during a mock scope. Document AI has no emulator env var, so drongo forces the client onto its REST transport and serves it from the HTTP layer.
• Backend: per-project.

Use the normal client with no transport argument. Document AI uses regional endpoints, so pass the region the same way you would against the real API.

Covers processors (create / get / list / delete, enable / disable), processor types, batch processing, and - the headline - document processing driven by a handler that runs your code, so a test exercises real extraction logic rather than a stub.

Create a processor and process a document

Register a processor handler and process_document runs it (like the Cloud Run / Vertex executable handlers). The handler receives (content, mime_type) and returns the resulting Document:

from drongo import documentai, mock_gcp
from google.api_core.client_options import ClientOptions


@mock_gcp
def test_process():
    from google.cloud import documentai_v1

    location = "us"
    parent = f"projects/my-project/locations/{location}"
    opts = ClientOptions(api_endpoint=f"{location}-documentai.googleapis.com")
    client = documentai_v1.DocumentProcessorServiceClient(client_options=opts)

    processor = client.create_processor(
        parent=parent,
        processor=documentai_v1.Processor(type_="OCR_PROCESSOR", display_name="my-ocr"),
    )

    @documentai.processor_handler(processor.name)
    def extract(content, mime_type):
        return {
            "text": content.decode(),
            "entities": [{"type": "invoice_id", "mentionText": "INV-42"}],
        }

    response = client.process_document(
        request=documentai_v1.ProcessRequest(
            name=processor.name,
            raw_document=documentai_v1.RawDocument(
                content=b"INV-42 total $10", mime_type="text/plain"
            ),
        )
    )
    assert response.document.text == "INV-42 total $10"
    assert response.document.entities[0].type_ == "invoice_id"

With no handler registered, process_document returns the decoded input as the document text for text mime types (empty otherwise). Missing processors raise google.api_core.exceptions.NotFound.

Coverage

Operation	Status
Processors: create / get / list / delete (LRO)	Supported
Enable / disable processor (LRO)	Supported
Process document (process_document, via a handler)	Supported
Batch process (batch_process_documents, LRO)	Supported
Processor types: fetch / list / get	Supported
Processor versions (deploy / train / evaluate)	Planned
Human review, evaluations	Planned