feat: wire PDF and text file content into AI chat messages

PDF uploads were stored in S3 and had text extracted during upload, but
fetch_upload_images() filtered exclusively for image MIME types, so
document content never reached the AI.

- Add fetch_upload_documents() in storage_service.py to retrieve
  extracted_content for PDFs and text files
- Update ai_sessions.py chat endpoint to call both fetch_upload_images
  and fetch_upload_documents, injecting document text as context
- Add PDF text extraction in _generate_ai_description (pypdf)
- Add pypdf>=4.0.0 to requirements.txt
- Fix test_db teardown to avoid connection pool issues
- Add 5 tests for fetch_upload_documents

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

backend/app/api/endpoints/ai_sessions.py

@@ -280,18 +280,28 @@ async def send_chat_message(
     user_id = current_user.id
     account_id = current_user.account_id
 
-    # Fetch attached images from S3 (if any)
+    # Fetch attached uploads from S3 (if any)
     images = None
+    message = data.message
     if data.upload_ids:
-        from app.services.storage_service import fetch_upload_images
+        from app.services.storage_service import fetch_upload_images, fetch_upload_documents
         images = await fetch_upload_images(data.upload_ids, account_id, db) or None
 
+        # Inject document text (PDFs, text files) as context in the message
+        documents = await fetch_upload_documents(data.upload_ids, account_id, db)
+        if documents:
+            doc_parts = []
+            for doc in documents:
+                doc_parts.append(f"--- Attached file: {doc['filename']} ---\n{doc['text']}")
+            doc_context = "\n\n".join(doc_parts)
+            message = f"{message}\n\n[Attached document content]\n{doc_context}"
+
     try:
         ai_content, suggested_flows, session, fork_metadata, actions_data, questions_data = await unified_chat_service.send_chat_message(
             session_id=session_id,
             user_id=user_id,
             account_id=account_id,
-            message=data.message,
+            message=message,
             db=db,
             images=images,
         )

backend/app/api/endpoints/uploads.py

@@ -61,6 +61,40 @@ async def _generate_ai_description(upload_id: UUID, file_data: bytes, content_ty
                     max_tokens=100,
                 )
                 upload.ai_description = description
+            elif content_type == "application/pdf":
+                try:
+                    from pypdf import PdfReader
+                    import io as _io
+
+                    reader = PdfReader(_io.BytesIO(file_data))
+                    pages_text = []
+                    for page in reader.pages:
+                        page_text = page.extract_text()
+                        if page_text:
+                            pages_text.append(page_text)
+                    text_content = "\n\n".join(pages_text)
+                except Exception:
+                    logger.warning("PDF text extraction failed for upload %s", upload_id)
+                    text_content = ""
+
+                if text_content:
+                    upload.extracted_content = text_content[:10000]
+
+                    if len(text_content) > 2000:
+                        summary, _, _ = await _call_ai(
+                            system_base="You are a technical document analyst for IT troubleshooting.",
+                            rag_context="",
+                            history=[],
+                            new_message=f"Summarize this PDF content in 2-3 sentences:\n\n{text_content[:5000]}",
+                            max_tokens=200,
+                        )
+                        upload.content_summary = summary
+                        upload.ai_description = summary
+                    else:
+                        upload.ai_description = f"PDF document: {upload.filename}"
+                else:
+                    upload.ai_description = f"PDF document (no extractable text): {upload.filename}"
+
             elif content_type.startswith("text/") or content_type in (
                 "application/json", "application/xml", "application/yaml",
             ):