Address lack of support for office formats

Author: pamelafox

app/backend/prepdocslib/pdfparser.py

@@ -14,6 +14,7 @@
 )
 from azure.core.credentials import AzureKeyCredential
 from azure.core.credentials_async import AsyncTokenCredential
+from azure.core.exceptions import HttpResponseError
 from PIL import Image
 from pypdf import PdfReader
 
@@ -68,6 +69,7 @@ async def parse(self, content: IO) -> AsyncGenerator[Page, None]:
         async with DocumentIntelligenceClient(
             endpoint=self.endpoint, credential=self.credential
         ) as document_intelligence_client:
+            file_analyzed = False
             if self.use_content_understanding:
                 if self.content_understanding_endpoint is None:
                     raise ValueError("Content Understanding is enabled but no endpoint was provided")
@@ -77,15 +79,29 @@ async def parse(self, content: IO) -> AsyncGenerator[Page, None]:
                     )
                 cu_describer = ContentUnderstandingDescriber(self.content_understanding_endpoint, self.credential)
                 content_bytes = content.read()
-                poller = await document_intelligence_client.begin_analyze_document(
-                    model_id="prebuilt-layout",
-                    analyze_request=AnalyzeDocumentRequest(bytes_source=content_bytes),
-                    output=["figures"],
-                    features=["ocrHighResolution"],
-                    output_content_format="markdown",
-                )
-                doc_for_pymupdf = pymupdf.open(stream=io.BytesIO(content_bytes))
-            else:
+                try:
+                    poller = await document_intelligence_client.begin_analyze_document(
+                        model_id="prebuilt-layout",
+                        analyze_request=AnalyzeDocumentRequest(bytes_source=content_bytes),
+                        output=["figures"],
+                        features=["ocrHighResolution"],
+                        output_content_format="markdown",
+                    )
+                    doc_for_pymupdf = pymupdf.open(stream=io.BytesIO(content_bytes))
+                    file_analyzed = True
+                except HttpResponseError as e:
+                    content.seek(0)
+                    if e.error.code == "InvalidArgument":
+                        logger.warning(
+                            "This document type does not support media description. Proceeding with standard analysis."
+                        )
+                    else:
+                        logger.warning(
+                            "Unexpected error analyzing document for media description: %s. Proceeding with standard analysis.",
+                            e,
+                        )
+
+            if file_analyzed is False:
                 poller = await document_intelligence_client.begin_analyze_document(
                     model_id=self.model_id, analyze_request=content, content_type="application/octet-stream"
                 )

docs/deploy_features.md

@@ -163,7 +163,6 @@ By default, if your documents contain image-like figures, the data ingestion pro
 so users will not be able to ask questions about them.
 
 You can optionably enable the description of media content using Azure Content Understanding. When enabled, the data ingestion process will send figures to Azure Content Understanding and replace the figure with the description in the indexed document.
-To learn more about this process and compare it to the gpt-4 vision integration, see [this guide](./data_ingestion.md#media-description).
 
 To enable media description with Azure Content Understanding, run:
 
@@ -175,6 +174,9 @@ If you have already run `azd up`, you will need to run `azd provision` to create
 If you have already indexed your documents and want to re-index them with the media descriptions,
 first [remove the existing documents](./data_ingestion.md#removing-documents) and then [re-ingest the data](./data_ingestion.md#indexing-additional-documents).
 
+⚠️ This feature does not yet support DOCX, PPTX, or XLSX formats. If you have figures in those formats, they will be ignored.
+Convert them first to PDF or image formats to enable media description.
+
 ## Enabling client-side chat history
 
 This feature allows users to view the chat history of their conversation, stored in the browser using [IndexedDB](https://developer.mozilla.org/docs/Web/API/IndexedDB_API). That means the chat history will be available only on the device where the chat was initiated. To enable browser-stored chat history, run:

tests/test_pdfparser.py

@@ -4,6 +4,7 @@
 import pathlib
 from unittest.mock import AsyncMock, MagicMock
 
+import azure.core.exceptions
 import pymupdf
 import pytest
 from azure.ai.documentintelligence.aio import DocumentIntelligenceClient
@@ -308,3 +309,48 @@ async def mock_describe_image(self, image_bytes):
         pages[0].text
         == "# Simple Figure\n\nThis text is before the figure and NOT part of it.\n\n\n<figure><figcaption>Figure 1<br>Pie chart</figcaption></figure>\n\n\nThis is text after the figure that's not part of it."
     )
+
+
+@pytest.mark.asyncio
+async def test_parse_unsupportedformat(monkeypatch, caplog):
+    mock_poller = MagicMock()
+
+    async def mock_begin_analyze_document(self, model_id, analyze_request, **kwargs):
+        class MockErrorResponse:
+            def __init__(self):
+                self.reason = "InvalidArgument"
+                self.status_code = 400
+                self.error = {"code": "InvalidArgument"}
+
+        if kwargs.get("features") == ["ocrHighResolution"]:
+            raise azure.core.exceptions.HttpResponseError(message="InvalidArgument", response=MockErrorResponse())
+        else:
+            return mock_poller
+
+    async def mock_poller_result():
+        return AnalyzeResult(
+            content="Page content",
+            pages=[DocumentPage(page_number=1, spans=[DocumentSpan(offset=0, length=12)])],
+            tables=[],
+            figures=[],
+        )
+
+    monkeypatch.setattr(DocumentIntelligenceClient, "begin_analyze_document", mock_begin_analyze_document)
+    monkeypatch.setattr(mock_poller, "result", mock_poller_result)
+
+    parser = DocumentAnalysisParser(
+        endpoint="https://example.com",
+        credential=MockAzureCredential(),
+        use_content_understanding=True,
+        content_understanding_endpoint="https://example.com",
+    )
+    content = io.BytesIO(b"pdf content bytes")
+    content.name = "test.docx"
+    with caplog.at_level(logging.WARNING):
+        pages = [page async for page in parser.parse(content)]
+        assert "This document type does not support media description." in caplog.text
+
+    assert len(pages) == 1
+    assert pages[0].page_num == 0
+    assert pages[0].offset == 0
+    assert pages[0].text == "Page content"