Refactor AzureDIService to enhance document analysis logging and update table extraction logic. Temporarily disable table extraction and improve content type logging. Update TableAgent to reflect changes in context handling.

src/agents/table_agent.py

@@ -35,9 +35,9 @@ class TableAgent(BaseAgent):
             
             # Store both text and tables in context
             ctx["text"] = result["text"]
-            ctx["tables"] = result["tables"]
+            #ctx["tables"] = result["tables"]
             
-            self.logger.info(f"Extracted {len(result['text'])} characters of text and {len(result['tables'])} tables")
+            self.logger.info(f"Extracted {len(result['text'])} characters of text including tables")
             return result
             
         except Exception as e:

src/services/azure_di_service.py

@@ -1,6 +1,8 @@
 """Real wrapper goes here – currently not used by stub agents."""
 import logging
+import json
 from azure.ai.documentintelligence import DocumentIntelligenceClient
+from azure.ai.documentintelligence.models import DocumentContentFormat
 from azure.core.credentials import AzureKeyCredential
 from azure.core.exceptions import HttpResponseError
 
@@ -14,38 +16,35 @@ class AzureDIService:
             self.logger.info("Starting document analysis with Azure Document Intelligence")
             
             # Analyze the entire document at once
-            poller = self.client.begin_analyze_document("prebuilt-layout", body=pdf_bytes)
+            #poller = self.client.begin_analyze_document("prebuilt-layout", body=pdf_bytes)
+
+            poller = self.client.begin_analyze_document(
+                "prebuilt-layout", 
+                body=pdf_bytes, 
+                content_type="application/octet-stream", 
+                output_content_format=DocumentContentFormat.MARKDOWN
+            )
             result = poller.result()
             
-            # Extract text content
-            text_content = ""
+            # Log the raw result structure
+            self.logger.info("Inspecting Azure DI result structure:")
+            self.logger.info(f"Result type: {type(result)}")
+            self.logger.info(f"Result attributes: {dir(result)}")
+            
+            # Check if content exists and log its type
             if hasattr(result, "content"):
-                text_content = result.content
-                self.logger.info(f"Extracted {len(text_content)} characters of text")
+                self.logger.info(f"Content type: {type(result.content)}")
+                self.logger.info(f"Content preview: {result.content[:500]}")
             
-            # Extract tables as HTML strings
-            tables = []
+            # Check if tables exist and log their structure
             if hasattr(result, "tables"):
-                self.logger.info(f"Found {len(result.tables)} tables in the document")
-                for table in result.tables:
-                    # Simple HTML rendering for demo; you can improve this
-                    html = "<table>"
-                    for row_idx in range(table.row_count):
-                        html += "<tr>"
-                        for col_idx in range(table.column_count):
-                            cell = next((c for c in table.cells if c.row_index == row_idx and c.column_index == col_idx), None)
-                            html += f"<td>{cell.content if cell else ''}</td>"
-                        html += "</tr>"
-                    html += "</table>"
-                    tables.append(html)
-            else:
-                self.logger.warning("No tables found in the document")
+                self.logger.info(f"Number of tables: {len(result.tables)}")
+                if result.tables:
+                    self.logger.info(f"First table structure: {dir(result.tables[0])}")
+                    self.logger.info(f"First table cells: {[cell.content for cell in result.tables[0].cells]}")
             
-            # Return both text and tables
-            return {
-                "text": text_content,
-                "tables": tables
-            }
+            # For now, return empty result until we understand the structure
+            return {"text": result.content}
             
         except HttpResponseError as e:
             self.logger.error(f"Azure Document Intelligence API error: {str(e)}")