Using document_intelligence_client.begin_analyze_document returns Session.request() got an unexpected keyword argument 'body'

Question

When I'm trying to use the following script:

  # Open local file in binary mode
    with open(file_path, "rb") as f:
        poller = document_intelligence_client.begin_analyze_document(
            model_id="prebuilt-layout",
            body=f
        )

I get the error:
Session.request() got an unexpected keyword argument 'body'

I've checked the version: Version: 1.0.0
and in case tried updating it with pip. It still returns the same error.
The key and endpoint are correct.

In essence, I want to use local PDF files and use document intelligence to retrieve the text and tables, maybe even figures later if I can figure this out first. I have a custom extraction model, but I'm just trying to test this first to see if everything is in order.

I've tried changing

body=f -> document=f

but the error just changes to

Session.request() got an unexpected keyword argument 'document'

Answer 1

Hi @Maxime Jack Raymond Jens Cara,

Apologies for the delay in response. The error occurs because begin_analyze_document() does not accept a raw file stream (body=f or document=f). Instead, it requires either a URL (urlSource) or a Base64-encoded file (base64Source).

Since you're working with local PDF files, you'll need to encode the file in Base64 before passing it to the API. Here's the correct approach:

import os
import base64
from azure.core.credentials import AzureKeyCredential
from azure.ai.documentintelligence import DocumentIntelligenceClient
from azure.ai.documentintelligence.models import AnalyzeResult

# Set your Azure endpoint and key
endpoint = "<END_POINT>"  # Replace with your endpoint
key = "YOUR_KEY"  # Replace with your key

def analyze_layout(local_file_path):
    document_intelligence_client = DocumentIntelligenceClient(
        endpoint=endpoint, credential=AzureKeyCredential(key)
    )

    # Read the local PDF and encode it to Base64
    with open(local_file_path, "rb") as file_stream:
        base64_data = base64.b64encode(file_stream.read()).decode("utf-8")

    # Send request with base64Source
    poller = document_intelligence_client.begin_analyze_document(
        model_id="prebuilt-layout", analyze_request={"base64Source": base64_data}
    )

    result: AnalyzeResult = poller.result()

    for page in result.pages:
        print(f"----Analyzing layout from page #{page.page_number}----")
        print(f"Page dimensions: {page.width} x {page.height} {page.unit}")

        if page.lines:
            for line in page.lines:
                print(f"Line: '{line.content}'")

    if result.tables:
        for table in result.tables:
            print(f"Table with {table.row_count} rows and {table.column_count} columns")
            for cell in table.cells:
                print(f"Cell[{cell.row_index}][{cell.column_index}]: '{cell.content}'")

if __name__ == "__main__":
    local_file_path = r"FILE_PATH"  # Change this to your local file path
    analyze_layout(local_file_path)

Here is the output:

For more info: Get started with Document Intelligence.

This should resolve the error and allow you to process local PDFs correctly. Let me know if you need any further assistance!

---

If this answers your query, do click Accept Answer and Yes for was this answer helpful.