Azure Document Intelligence (formrecognizer) - 'InvalidContent' when passing pdf

I upload a pdf file to my streamlit application like this:

import streamlit as st

uploaded_file = st.file_uploader("Upload pdf file", type="pdf")
result = analyze_general_document(uploaded_file)

I want to analzye this pdf using the Azure Document Intelligence python package like this:

from io import BytesIO
from azure.core.credentials import AzureKeyCredential
from azure.ai.formrecognizer import DocumentAnalysisClient


def set_client(secrets: dict):
    endpoint = secrets["AI_DOCS_BASE"]
    key = secrets["AI_DOCS_KEY"]
    document_analysis_client = DocumentAnalysisClient(endpoint=endpoint, credential=AzureKeyCredential(key))
    return document_analysis_client


def analyze_general_document(uploaded_file, secrets: dict):
    print(f"File type: {uploaded_file.type}")
    print(f"File size: {uploaded_file.size} bytes")
    client = set_client(secrets)
    # poller = client.begin_analyze_document_from_url("prebuilt-document", formUrl)
    poller = client.begin_analyze_document("prebuilt-document", document=uploaded_file)

I can successfully print the file type and file size as you can see in the terminal output:

File type: application/pdf
File size: 6928426 bytes

Also opening the file with PyMuPDF works fine as well.

However the method begin_analyze_document throws the following exeception:

Traceback (most recent call last):
  File "C:\Users\myuser\AppData\Local\miniconda3\envs\projectai\Lib\site-packages\streamlit\runtime\scriptrunner\exec_code.py", line 88, in exec_func_with_error_handling
    result = func()
             ^^^^^^
  File "C:\Users\myuser\AppData\Local\miniconda3\envs\projectai\Lib\site-packages\streamlit\runtime\scriptrunner\script_runner.py", line 579, in code_to_exec
    exec(code, module.__dict__)
  File "C:\Users\myuser\Documents\visual-studio-code\project\project-ai-docs\webapp\app.py", line 79, in <module>
    main()
  File "C:\Users\myuser\Documents\visual-studio-code\project\project-ai-docs\webapp\app.py", line 61, in main
    zip_content = process_pdf(uploaded_file, secrets)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\myuser\Documents\visual-studio-code\project\project-ai-docs\webapp\app_backend.py", line 40, in process_pdf
    analyze_general_document(uploaded_file, secrets)
  File "C:\Users\myuser\Documents\visual-studio-code\project\project-ai-docs\webapp\az_document_intelligence.py", line 18, in analyze_general_document
    poller = client.begin_analyze_document("prebuilt-document", document=uploaded_file)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\myuser\AppData\Local\miniconda3\envs\projectai\Lib\site-packages\azure\core\tracing\decorator.py", line 105, in wrapper_use_tracer
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\myuser\AppData\Local\miniconda3\envs\projectai\Lib\site-packages\azure\ai\formrecognizer\_document_analysis_client.py", line 129, in begin_analyze_document
    return _client_op_path.begin_analyze_document(  # type: ignore
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\myuser\AppData\Local\miniconda3\envs\projectai\Lib\site-packages\azure\core\tracing\decorator.py", line 105, in wrapper_use_tracer
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\myuser\AppData\Local\miniconda3\envs\projectai\Lib\site-packages\azure\ai\formrecognizer\_generated\v2023_07_31\operations\_document_models_operations.py", line 518, in begin_analyze_document
    raw_result = self._analyze_document_initial(  # type: ignore
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\myuser\AppData\Local\miniconda3\envs\projectai\Lib\site-packages\azure\ai\formrecognizer\_generated\v2023_07_31\operations\_document_models_operations.py", line 443, in _analyze_document_initial
    raise HttpResponseError(response=response)
azure.core.exceptions.HttpResponseError: (InvalidRequest) Invalid request.
Code: InvalidRequest
Message: Invalid request.
Inner error: {
    "code": "InvalidContent",
    "message": "The file is corrupted or format is unsupported. Refer to documentation for the list of supported formats."
}

Why is the pdf considered invalid? I also tried wrapping it in a BytesIO object like this but it didn't work either:

def analyze_general_document(uploaded_file, secrets: dict):
    print(f"File type: {uploaded_file.type}")
    print(f"File size: {uploaded_file.size} bytes")
    # Read the file as bytes
    file_bytes = uploaded_file.read()
    client = set_client(secrets)
    # poller = client.begin_analyze_document_from_url("prebuilt-document", formUrl)
    poller = client.begin_analyze_document("prebuilt-document", document=BytesIO(file_bytes))

Azure Document Intelligence (formrecognizer) - 'InvalidContent' when passing pdf

You can use the below code that Analyze the pdf file with Azure Document Intelligence by uploaded with streamlit using python,

Code:

import streamlit as st
from io import BytesIO
from azure.core.credentials import AzureKeyCredential
from azure.ai.formrecognizer import DocumentAnalysisClient
from PyPDF2 import PdfReader

def set_client(secrets: dict):
    endpoint = secrets["AI_DOCS_BASE"]
    key = secrets["AI_DOCS_KEY"]
    return DocumentAnalysisClient(endpoint=endpoint, credential=AzureKeyCredential(key))

def validate_pdf(file):
    try:
        file.seek(0)  # Reset pointer
        reader = PdfReader(file)
        if len(reader.pages) == 0:
            raise ValueError("The PDF has no pages.")
        print(f"PDF is valid with {len(reader.pages)} pages.")
    except Exception as e:
        raise ValueError(f"PDF validation failed: {e}")

def check_pdf_metadata(file):
    file.seek(0)  # Reset pointer
    reader = PdfReader(file)
    if reader.is_encrypted:
        raise ValueError("Encrypted PDFs are not supported.")
    print(f"PDF has {len(reader.pages)} pages and is not encrypted.")

def analyze_general_document(uploaded_file, secrets: dict):

    validate_pdf(uploaded_file)
    check_pdf_metadata(uploaded_file)

    uploaded_file.seek(0)  # Reset pointer
    file_bytes = uploaded_file.read()
    client = set_client(secrets)
    try:
        print("Sending file to Azure Document Intelligence...")
        poller = client.begin_analyze_document("prebuilt-document", document=BytesIO(file_bytes))
        result = poller.result()
        print("Analysis successful!")
        return result
    except Exception as e:
        raise RuntimeError(f"Azure Document Intelligence error: {e}")

# Streamlit Application
def main():
    st.title("Azure Document Intelligence PDF Analyzer")
    uploaded_file = st.file_uploader("Upload PDF file", type="pdf")

    if uploaded_file is not None:
        # Display file details
        st.write(f"File Name: {uploaded_file.name}")
        st.write(f"File Type: {uploaded_file.type}")
        st.write(f"File Size: {uploaded_file.size} bytes")

        # Secrets for Azure setup
        secrets = {
            "AI_DOCS_BASE": "https://xxxxxxx.cognitiveservices.azure.com/",
            "AI_DOCS_KEY": "xxxxx"
        }

        try:
            # Analyze PDF
            result = analyze_general_document(uploaded_file, secrets)
            st.success("PDF analysis completed!")
            st.json(result.to_dict())  # Display results in Streamlit
        except Exception as e:
            st.error(f"Error: {e}")

if __name__ == "__main__":
    main()

Output:

PS C:\Users\xxxx> streamlit run set.py

  You can now view your Streamlit app in your browser.

  Local URL: http://localhost:8xxx1
  Network URL: http://192.168.1.8:8xxx1

PDF is valid with 20 pages.
PDF has 20 pages and is not encrypted.
Sending file to Azure Document Intelligence...
Analysis successful!

Browser:

Reference: azure.ai.formrecognizer.DocumentAnalysisClient class | Microsoft Learn