diff --git a/CLAUDE.md b/CLAUDE.md index 4e9c2830..2ad07a8a 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -58,6 +58,7 @@ return render_template('page.html', settings=public_settings) ## Version Management +- Its important to update the version at the end of every plan - Version is stored in `config.py`: `VERSION = "X.XXX.XXX"` - When incrementing, only change the third segment (e.g., `0.238.024` -> `0.238.025`) - Include the current version in functional test file headers and documentation files @@ -83,7 +84,7 @@ return render_template('page.html', settings=public_settings) ## Release Notes -After completing code changes, offer to update `docs/explanation/release_notes.md`. +After completing plans and code changes, offer to update `docs/explanation/release_notes.md`. - Add entries under the current version from `config.py` - If the version was bumped, create a new section at the top: `### **(vX.XXX.XXX)**` diff --git a/application/single_app/app.py b/application/single_app/app.py index 2354b1b5..e0c8cfee 100644 --- a/application/single_app/app.py +++ b/application/single_app/app.py @@ -75,6 +75,7 @@ from route_backend_public_prompts import * from route_backend_user_agreement import register_route_backend_user_agreement from route_backend_conversation_export import register_route_backend_conversation_export +from route_backend_thoughts import register_route_backend_thoughts from route_backend_speech import register_route_backend_speech from route_backend_tts import register_route_backend_tts from route_enhanced_citations import register_enhanced_citations_routes @@ -641,6 +642,9 @@ def list_semantic_kernel_plugins(): # ------------------- API User Agreement Routes ---------- register_route_backend_user_agreement(app) +# ------------------- API Thoughts Routes ---------------- +register_route_backend_thoughts(app) + # ------------------- Extenral Health Routes ---------- register_route_external_health(app) diff --git a/application/single_app/config.py b/application/single_app/config.py index da63c230..bec5513b 100644 --- a/application/single_app/config.py +++ b/application/single_app/config.py @@ -94,7 +94,7 @@ EXECUTOR_TYPE = 'thread' EXECUTOR_MAX_WORKERS = 30 SESSION_TYPE = 'filesystem' -VERSION = "0.239.004" +VERSION = "0.239.013" SECRET_KEY = os.getenv('SECRET_KEY', 'dev-secret-key-change-in-production') @@ -257,6 +257,8 @@ def get_redis_cache_infrastructure_endpoint(redis_hostname: str) -> str: storage_account_user_documents_container_name = "user-documents" storage_account_group_documents_container_name = "group-documents" storage_account_public_documents_container_name = "public-documents" +storage_account_personal_chat_container_name = "personal-chat" +storage_account_group_chat_container_name = "group-chat" # Initialize Azure Cosmos DB client cosmos_endpoint = os.getenv("AZURE_COSMOS_ENDPOINT") @@ -459,6 +461,18 @@ def get_redis_cache_infrastructure_endpoint(redis_hostname: str) -> str: default_ttl=-1 # TTL disabled by default, enabled per-document for auto-cleanup ) +cosmos_thoughts_container_name = "thoughts" +cosmos_thoughts_container = cosmos_database.create_container_if_not_exists( + id=cosmos_thoughts_container_name, + partition_key=PartitionKey(path="/user_id") +) + +cosmos_archived_thoughts_container_name = "archive_thoughts" +cosmos_archived_thoughts_container = cosmos_database.create_container_if_not_exists( + id=cosmos_archived_thoughts_container_name, + partition_key=PartitionKey(path="/user_id") +) + def ensure_custom_logo_file_exists(app, settings): """ If custom_logo_base64 or custom_logo_dark_base64 is present in settings, ensure the appropriate @@ -745,9 +759,11 @@ def initialize_clients(settings): # This addresses the issue where the application assumes containers exist if blob_service_client: for container_name in [ - storage_account_user_documents_container_name, - storage_account_group_documents_container_name, - storage_account_public_documents_container_name + storage_account_user_documents_container_name, + storage_account_group_documents_container_name, + storage_account_public_documents_container_name, + storage_account_personal_chat_container_name, + storage_account_group_chat_container_name ]: try: container_client = blob_service_client.get_container_client(container_name) diff --git a/application/single_app/functions_activity_logging.py b/application/single_app/functions_activity_logging.py index 2a653a47..efb6e780 100644 --- a/application/single_app/functions_activity_logging.py +++ b/application/single_app/functions_activity_logging.py @@ -1393,3 +1393,332 @@ def log_retention_policy_force_push( level=logging.ERROR ) debug_print(f"⚠️ Warning: Failed to log retention policy force push: {str(e)}") + + +# === AGENT & ACTION ACTIVITY LOGGING === + +def log_agent_creation( + user_id: str, + agent_id: str, + agent_name: str, + agent_display_name: Optional[str] = None, + scope: str = 'personal', + group_id: Optional[str] = None +) -> None: + """ + Log an agent creation activity. + + Args: + user_id: The ID of the user who created the agent + agent_id: The unique ID of the new agent + agent_name: The name of the agent + agent_display_name: The display name of the agent + scope: 'personal', 'group', or 'global' + group_id: The group ID (only for group scope) + """ + try: + activity_record = { + 'id': str(uuid.uuid4()), + 'user_id': user_id, + 'activity_type': 'agent_creation', + 'timestamp': datetime.utcnow().isoformat(), + 'created_at': datetime.utcnow().isoformat(), + 'entity_type': 'agent', + 'operation': 'create', + 'entity': { + 'id': agent_id, + 'name': agent_name, + 'display_name': agent_display_name or agent_name + }, + 'workspace_type': scope, + 'workspace_context': {} + } + if scope == 'group' and group_id: + activity_record['workspace_context']['group_id'] = group_id + + cosmos_activity_logs_container.create_item(body=activity_record) + log_event( + message=f"Agent created: {agent_name} ({scope}) by user {user_id}", + extra=activity_record, + level=logging.INFO + ) + debug_print(f"✅ Agent creation logged: {agent_name} ({scope})") + except Exception as e: + log_event( + message=f"Error logging agent creation: {str(e)}", + extra={'user_id': user_id, 'agent_id': agent_id, 'scope': scope, 'error': str(e)}, + level=logging.ERROR + ) + debug_print(f"⚠️ Warning: Failed to log agent creation: {str(e)}") + + +def log_agent_update( + user_id: str, + agent_id: str, + agent_name: str, + agent_display_name: Optional[str] = None, + scope: str = 'personal', + group_id: Optional[str] = None +) -> None: + """ + Log an agent update activity. + + Args: + user_id: The ID of the user who updated the agent + agent_id: The unique ID of the agent + agent_name: The name of the agent + agent_display_name: The display name of the agent + scope: 'personal', 'group', or 'global' + group_id: The group ID (only for group scope) + """ + try: + activity_record = { + 'id': str(uuid.uuid4()), + 'user_id': user_id, + 'activity_type': 'agent_update', + 'timestamp': datetime.utcnow().isoformat(), + 'created_at': datetime.utcnow().isoformat(), + 'entity_type': 'agent', + 'operation': 'update', + 'entity': { + 'id': agent_id, + 'name': agent_name, + 'display_name': agent_display_name or agent_name + }, + 'workspace_type': scope, + 'workspace_context': {} + } + if scope == 'group' and group_id: + activity_record['workspace_context']['group_id'] = group_id + + cosmos_activity_logs_container.create_item(body=activity_record) + log_event( + message=f"Agent updated: {agent_name} ({scope}) by user {user_id}", + extra=activity_record, + level=logging.INFO + ) + debug_print(f"✅ Agent update logged: {agent_name} ({scope})") + except Exception as e: + log_event( + message=f"Error logging agent update: {str(e)}", + extra={'user_id': user_id, 'agent_id': agent_id, 'scope': scope, 'error': str(e)}, + level=logging.ERROR + ) + debug_print(f"⚠️ Warning: Failed to log agent update: {str(e)}") + + +def log_agent_deletion( + user_id: str, + agent_id: str, + agent_name: str, + scope: str = 'personal', + group_id: Optional[str] = None +) -> None: + """ + Log an agent deletion activity. + + Args: + user_id: The ID of the user who deleted the agent + agent_id: The unique ID of the agent + agent_name: The name of the agent + scope: 'personal', 'group', or 'global' + group_id: The group ID (only for group scope) + """ + try: + activity_record = { + 'id': str(uuid.uuid4()), + 'user_id': user_id, + 'activity_type': 'agent_deletion', + 'timestamp': datetime.utcnow().isoformat(), + 'created_at': datetime.utcnow().isoformat(), + 'entity_type': 'agent', + 'operation': 'delete', + 'entity': { + 'id': agent_id, + 'name': agent_name + }, + 'workspace_type': scope, + 'workspace_context': {} + } + if scope == 'group' and group_id: + activity_record['workspace_context']['group_id'] = group_id + + cosmos_activity_logs_container.create_item(body=activity_record) + log_event( + message=f"Agent deleted: {agent_name} ({scope}) by user {user_id}", + extra=activity_record, + level=logging.INFO + ) + debug_print(f"✅ Agent deletion logged: {agent_name} ({scope})") + except Exception as e: + log_event( + message=f"Error logging agent deletion: {str(e)}", + extra={'user_id': user_id, 'agent_id': agent_id, 'scope': scope, 'error': str(e)}, + level=logging.ERROR + ) + debug_print(f"⚠️ Warning: Failed to log agent deletion: {str(e)}") + + +def log_action_creation( + user_id: str, + action_id: str, + action_name: str, + action_type: Optional[str] = None, + scope: str = 'personal', + group_id: Optional[str] = None +) -> None: + """ + Log an action/plugin creation activity. + + Args: + user_id: The ID of the user who created the action + action_id: The unique ID of the new action + action_name: The name of the action + action_type: The type of the action (e.g., 'openapi', 'sql_query') + scope: 'personal', 'group', or 'global' + group_id: The group ID (only for group scope) + """ + try: + activity_record = { + 'id': str(uuid.uuid4()), + 'user_id': user_id, + 'activity_type': 'action_creation', + 'timestamp': datetime.utcnow().isoformat(), + 'created_at': datetime.utcnow().isoformat(), + 'entity_type': 'action', + 'operation': 'create', + 'entity': { + 'id': action_id, + 'name': action_name, + 'type': action_type + }, + 'workspace_type': scope, + 'workspace_context': {} + } + if scope == 'group' and group_id: + activity_record['workspace_context']['group_id'] = group_id + + cosmos_activity_logs_container.create_item(body=activity_record) + log_event( + message=f"Action created: {action_name} ({scope}) by user {user_id}", + extra=activity_record, + level=logging.INFO + ) + debug_print(f"✅ Action creation logged: {action_name} ({scope})") + except Exception as e: + log_event( + message=f"Error logging action creation: {str(e)}", + extra={'user_id': user_id, 'action_id': action_id, 'scope': scope, 'error': str(e)}, + level=logging.ERROR + ) + debug_print(f"⚠️ Warning: Failed to log action creation: {str(e)}") + + +def log_action_update( + user_id: str, + action_id: str, + action_name: str, + action_type: Optional[str] = None, + scope: str = 'personal', + group_id: Optional[str] = None +) -> None: + """ + Log an action/plugin update activity. + + Args: + user_id: The ID of the user who updated the action + action_id: The unique ID of the action + action_name: The name of the action + action_type: The type of the action + scope: 'personal', 'group', or 'global' + group_id: The group ID (only for group scope) + """ + try: + activity_record = { + 'id': str(uuid.uuid4()), + 'user_id': user_id, + 'activity_type': 'action_update', + 'timestamp': datetime.utcnow().isoformat(), + 'created_at': datetime.utcnow().isoformat(), + 'entity_type': 'action', + 'operation': 'update', + 'entity': { + 'id': action_id, + 'name': action_name, + 'type': action_type + }, + 'workspace_type': scope, + 'workspace_context': {} + } + if scope == 'group' and group_id: + activity_record['workspace_context']['group_id'] = group_id + + cosmos_activity_logs_container.create_item(body=activity_record) + log_event( + message=f"Action updated: {action_name} ({scope}) by user {user_id}", + extra=activity_record, + level=logging.INFO + ) + debug_print(f"✅ Action update logged: {action_name} ({scope})") + except Exception as e: + log_event( + message=f"Error logging action update: {str(e)}", + extra={'user_id': user_id, 'action_id': action_id, 'scope': scope, 'error': str(e)}, + level=logging.ERROR + ) + debug_print(f"⚠️ Warning: Failed to log action update: {str(e)}") + + +def log_action_deletion( + user_id: str, + action_id: str, + action_name: str, + action_type: Optional[str] = None, + scope: str = 'personal', + group_id: Optional[str] = None +) -> None: + """ + Log an action/plugin deletion activity. + + Args: + user_id: The ID of the user who deleted the action + action_id: The unique ID of the action + action_name: The name of the action + action_type: The type of the action + scope: 'personal', 'group', or 'global' + group_id: The group ID (only for group scope) + """ + try: + activity_record = { + 'id': str(uuid.uuid4()), + 'user_id': user_id, + 'activity_type': 'action_deletion', + 'timestamp': datetime.utcnow().isoformat(), + 'created_at': datetime.utcnow().isoformat(), + 'entity_type': 'action', + 'operation': 'delete', + 'entity': { + 'id': action_id, + 'name': action_name, + 'type': action_type + }, + 'workspace_type': scope, + 'workspace_context': {} + } + if scope == 'group' and group_id: + activity_record['workspace_context']['group_id'] = group_id + + cosmos_activity_logs_container.create_item(body=activity_record) + log_event( + message=f"Action deleted: {action_name} ({scope}) by user {user_id}", + extra=activity_record, + level=logging.INFO + ) + debug_print(f"✅ Action deletion logged: {action_name} ({scope})") + except Exception as e: + log_event( + message=f"Error logging action deletion: {str(e)}", + extra={'user_id': user_id, 'action_id': action_id, 'scope': scope, 'error': str(e)}, + level=logging.ERROR + ) + debug_print(f"⚠️ Warning: Failed to log action deletion: {str(e)}") diff --git a/application/single_app/functions_content.py b/application/single_app/functions_content.py index 376d23f4..3116ed82 100644 --- a/application/single_app/functions_content.py +++ b/application/single_app/functions_content.py @@ -352,7 +352,7 @@ def generate_embedding( embedding_model = selected_embedding_model['deploymentName'] while True: - random_delay = random.uniform(0.5, 2.0) + random_delay = random.uniform(0.05, 0.2) time.sleep(random_delay) try: @@ -385,3 +385,102 @@ def generate_embedding( except Exception as e: raise + +def generate_embeddings_batch( + texts, + batch_size=16, + max_retries=5, + initial_delay=1.0, + delay_multiplier=2.0 +): + """Generate embeddings for multiple texts in batches. + + Azure OpenAI embeddings API accepts a list of strings as input. + This reduces per-call overhead and delay significantly. + + Args: + texts: List of text strings to embed. + batch_size: Number of texts per API call (default 16). + max_retries: Max retries on rate limit errors. + initial_delay: Initial retry delay in seconds. + delay_multiplier: Multiplier for exponential backoff. + + Returns: + list of (embedding, token_usage) tuples, one per input text. + """ + settings = get_settings() + + enable_embedding_apim = settings.get('enable_embedding_apim', False) + + if enable_embedding_apim: + embedding_model = settings.get('azure_apim_embedding_deployment') + embedding_client = AzureOpenAI( + api_version=settings.get('azure_apim_embedding_api_version'), + azure_endpoint=settings.get('azure_apim_embedding_endpoint'), + api_key=settings.get('azure_apim_embedding_subscription_key')) + else: + if (settings.get('azure_openai_embedding_authentication_type') == 'managed_identity'): + token_provider = get_bearer_token_provider(DefaultAzureCredential(), cognitive_services_scope) + + embedding_client = AzureOpenAI( + api_version=settings.get('azure_openai_embedding_api_version'), + azure_endpoint=settings.get('azure_openai_embedding_endpoint'), + azure_ad_token_provider=token_provider + ) + + embedding_model_obj = settings.get('embedding_model', {}) + if embedding_model_obj and embedding_model_obj.get('selected'): + selected_embedding_model = embedding_model_obj['selected'][0] + embedding_model = selected_embedding_model['deploymentName'] + else: + embedding_client = AzureOpenAI( + api_version=settings.get('azure_openai_embedding_api_version'), + azure_endpoint=settings.get('azure_openai_embedding_endpoint'), + api_key=settings.get('azure_openai_embedding_key') + ) + + embedding_model_obj = settings.get('embedding_model', {}) + if embedding_model_obj and embedding_model_obj.get('selected'): + selected_embedding_model = embedding_model_obj['selected'][0] + embedding_model = selected_embedding_model['deploymentName'] + + results = [] + for i in range(0, len(texts), batch_size): + batch = texts[i:i + batch_size] + retries = 0 + current_delay = initial_delay + + while True: + random_delay = random.uniform(0.05, 0.2) + time.sleep(random_delay) + + try: + response = embedding_client.embeddings.create( + model=embedding_model, + input=batch + ) + + for item in response.data: + token_usage = None + if hasattr(response, 'usage') and response.usage: + token_usage = { + 'prompt_tokens': response.usage.prompt_tokens // len(batch), + 'total_tokens': response.usage.total_tokens // len(batch), + 'model_deployment_name': embedding_model + } + results.append((item.embedding, token_usage)) + break + + except RateLimitError as e: + retries += 1 + if retries > max_retries: + raise + + wait_time = current_delay * random.uniform(1.0, 1.5) + time.sleep(wait_time) + current_delay *= delay_multiplier + + except Exception as e: + raise + + return results diff --git a/application/single_app/functions_documents.py b/application/single_app/functions_documents.py index ce08066d..110afbd2 100644 --- a/application/single_app/functions_documents.py +++ b/application/single_app/functions_documents.py @@ -1646,6 +1646,191 @@ def save_chunks(page_text_content, page_number, file_name, user_id, document_id, # Return token usage information for accumulation return token_usage +def save_chunks_batch(chunks_data, user_id, document_id, group_id=None, public_workspace_id=None): + """ + Save multiple chunks at once using batch embedding and batch AI Search upload. + Significantly faster than calling save_chunks() per chunk. + + Args: + chunks_data: list of dicts with keys: page_text_content, page_number, file_name + user_id: The user ID + document_id: The document ID + group_id: Optional group ID for group documents + public_workspace_id: Optional public workspace ID for public documents + + Returns: + dict with 'total_tokens', 'prompt_tokens', 'model_deployment_name' + """ + from functions_content import generate_embeddings_batch + + current_time = datetime.now(timezone.utc).strftime('%Y-%m-%dT%H:%M:%SZ') + is_group = group_id is not None + is_public_workspace = public_workspace_id is not None + + # Retrieve metadata once for all chunks + try: + if is_public_workspace: + metadata = get_document_metadata( + document_id=document_id, + user_id=user_id, + public_workspace_id=public_workspace_id + ) + elif is_group: + metadata = get_document_metadata( + document_id=document_id, + user_id=user_id, + group_id=group_id + ) + else: + metadata = get_document_metadata( + document_id=document_id, + user_id=user_id + ) + + if not metadata: + raise ValueError(f"No metadata found for document {document_id}") + + version = metadata.get("version") if metadata.get("version") else 1 + except Exception as e: + log_event(f"[save_chunks_batch] Error retrieving metadata for document {document_id}: {repr(e)}", level=logging.ERROR) + raise + + # Generate all embeddings in batches + texts = [c['page_text_content'] for c in chunks_data] + try: + embedding_results = generate_embeddings_batch(texts) + except Exception as e: + log_event(f"[save_chunks_batch] Error generating batch embeddings for document {document_id}: {e}", level=logging.ERROR) + raise + + # Check for vision analysis once + vision_analysis = metadata.get('vision_analysis') + vision_text = "" + if vision_analysis: + vision_text_parts = [] + vision_text_parts.append("\n\n=== AI Vision Analysis ===") + vision_text_parts.append(f"Model: {vision_analysis.get('model', 'unknown')}") + if vision_analysis.get('description'): + vision_text_parts.append(f"\nDescription: {vision_analysis['description']}") + if vision_analysis.get('objects'): + objects_list = vision_analysis['objects'] + if isinstance(objects_list, list): + vision_text_parts.append(f"\nObjects Detected: {', '.join(objects_list)}") + else: + vision_text_parts.append(f"\nObjects Detected: {objects_list}") + if vision_analysis.get('text'): + vision_text_parts.append(f"\nVisible Text: {vision_analysis['text']}") + if vision_analysis.get('analysis'): + vision_text_parts.append(f"\nContextual Analysis: {vision_analysis['analysis']}") + vision_text = "\n".join(vision_text_parts) + + # Build all chunk documents + chunk_documents = [] + total_token_usage = {'total_tokens': 0, 'prompt_tokens': 0, 'model_deployment_name': None} + + for idx, chunk_info in enumerate(chunks_data): + embedding, token_usage = embedding_results[idx] + page_number = chunk_info['page_number'] + file_name = chunk_info['file_name'] + page_text_content = chunk_info['page_text_content'] + + if token_usage: + total_token_usage['total_tokens'] += token_usage.get('total_tokens', 0) + total_token_usage['prompt_tokens'] += token_usage.get('prompt_tokens', 0) + if not total_token_usage['model_deployment_name']: + total_token_usage['model_deployment_name'] = token_usage.get('model_deployment_name') + + chunk_id = f"{document_id}_{page_number}" + enhanced_chunk_text = page_text_content + vision_text if vision_text else page_text_content + + if is_public_workspace: + chunk_document = { + "id": chunk_id, + "document_id": document_id, + "chunk_id": str(page_number), + "chunk_text": enhanced_chunk_text, + "embedding": embedding, + "file_name": file_name, + "chunk_keywords": [], + "chunk_summary": "", + "page_number": page_number, + "author": [], + "title": "", + "document_classification": "None", + "document_tags": metadata.get('tags', []), + "chunk_sequence": page_number, + "upload_date": current_time, + "version": version, + "public_workspace_id": public_workspace_id + } + elif is_group: + shared_group_ids = metadata.get('shared_group_ids', []) if metadata else [] + chunk_document = { + "id": chunk_id, + "document_id": document_id, + "chunk_id": str(page_number), + "chunk_text": enhanced_chunk_text, + "embedding": embedding, + "file_name": file_name, + "chunk_keywords": [], + "chunk_summary": "", + "page_number": page_number, + "author": [], + "title": "", + "document_classification": "None", + "document_tags": metadata.get('tags', []), + "chunk_sequence": page_number, + "upload_date": current_time, + "version": version, + "group_id": group_id, + "shared_group_ids": shared_group_ids + } + else: + shared_user_ids = metadata.get('shared_user_ids', []) if metadata else [] + chunk_document = { + "id": chunk_id, + "document_id": document_id, + "chunk_id": str(page_number), + "chunk_text": enhanced_chunk_text, + "embedding": embedding, + "file_name": file_name, + "chunk_keywords": [], + "chunk_summary": "", + "page_number": page_number, + "author": [], + "title": "", + "document_classification": "None", + "document_tags": metadata.get('tags', []), + "chunk_sequence": page_number, + "upload_date": current_time, + "version": version, + "user_id": user_id, + "shared_user_ids": shared_user_ids + } + + chunk_documents.append(chunk_document) + + # Batch upload to AI Search + try: + if is_public_workspace: + search_client = CLIENTS["search_client_public"] + elif is_group: + search_client = CLIENTS["search_client_group"] + else: + search_client = CLIENTS["search_client_user"] + + # Upload in sub-batches of 32 to avoid request size limits + upload_batch_size = 32 + for i in range(0, len(chunk_documents), upload_batch_size): + sub_batch = chunk_documents[i:i + upload_batch_size] + search_client.upload_documents(documents=sub_batch) + + except Exception as e: + log_event(f"[save_chunks_batch] Error uploading batch to AI Search for document {document_id}: {e}", level=logging.ERROR) + raise + + return total_token_usage + def get_document_metadata_for_citations(document_id, user_id=None, group_id=None, public_workspace_id=None): """ Retrieve keywords and abstract from a document for creating metadata citations. @@ -4669,37 +4854,30 @@ def process_single_tabular_sheet(df, document_id, user_id, file_name, update_cal # Consider accumulating page count in the caller if needed. update_callback(number_of_pages=num_chunks_final) - # Save chunks, prepending the header to each + # Save chunks, prepending the header to each — use batch processing for speed + all_chunks = [] for idx, chunk_rows_content in enumerate(final_chunks_content, start=1): - # Prepend header - header length does not count towards chunk size limit chunk_with_header = header_string + chunk_rows_content - - update_callback( - current_file_chunk=idx, - status=f"Saving chunk {idx}/{num_chunks_final} from {file_name}..." - ) - - args = { + all_chunks.append({ "page_text_content": chunk_with_header, "page_number": idx, - "file_name": file_name, - "user_id": user_id, - "document_id": document_id - } + "file_name": file_name + }) - if is_public_workspace: - args["public_workspace_id"] = public_workspace_id - elif is_group: - args["group_id"] = group_id + if all_chunks: + update_callback( + current_file_chunk=1, + status=f"Batch processing {num_chunks_final} chunks from {file_name}..." + ) - token_usage = save_chunks(**args) - total_chunks_saved += 1 - - # Accumulate embedding tokens - if token_usage: - total_embedding_tokens += token_usage.get('total_tokens', 0) - if not embedding_model_name: - embedding_model_name = token_usage.get('model_deployment_name') + batch_token_usage = save_chunks_batch( + all_chunks, user_id, document_id, + group_id=group_id, public_workspace_id=public_workspace_id + ) + total_chunks_saved = len(all_chunks) + if batch_token_usage: + total_embedding_tokens = batch_token_usage.get('total_tokens', 0) + embedding_model_name = batch_token_usage.get('model_deployment_name') return total_chunks_saved, total_embedding_tokens, embedding_model_name @@ -4729,63 +4907,75 @@ def process_tabular(document_id, user_id, temp_file_path, original_filename, fil args["group_id"] = group_id upload_to_blob(**args) + update_callback(enhanced_citations=True, status=f"Enhanced citations enabled for {file_ext}") - try: - if file_ext == '.csv': - # Process CSV - # Read CSV, attempt to infer header, keep data as string initially - df = pandas.read_csv( - temp_file_path, - keep_default_na=False, - dtype=str + # When enhanced citations is on, index a single schema summary chunk + # instead of row-by-row chunking. The tabular processing plugin handles analysis. + if enable_enhanced_citations: + try: + if file_ext == '.csv': + df_preview = pandas.read_csv(temp_file_path, keep_default_na=False, dtype=str, nrows=5) + full_df = pandas.read_csv(temp_file_path, keep_default_na=False, dtype=str) + elif file_ext in ('.xlsx', '.xls', '.xlsm'): + engine = 'openpyxl' if file_ext in ('.xlsx', '.xlsm') else 'xlrd' + df_preview = pandas.read_excel(temp_file_path, engine=engine, keep_default_na=False, dtype=str, nrows=5) + full_df = pandas.read_excel(temp_file_path, engine=engine, keep_default_na=False, dtype=str) + else: + raise ValueError(f"Unsupported tabular file type: {file_ext}") + + row_count = len(full_df) + columns = list(df_preview.columns) + preview_rows = df_preview.head(5).to_string(index=False) + + schema_summary = ( + f"Tabular data file: {original_filename}\n" + f"Columns ({len(columns)}): {', '.join(columns)}\n" + f"Total rows: {row_count}\n" + f"Preview (first 5 rows):\n{preview_rows}\n\n" + f"This file is available for detailed analysis via the Tabular Processing plugin." ) - args = { - "df": df, - "document_id": document_id, - "user_id": user_id, + + update_callback(number_of_pages=1, status=f"Indexing schema summary for {original_filename}...") + + save_args = { + "page_text_content": schema_summary, + "page_number": 1, "file_name": original_filename, - "update_callback": update_callback + "user_id": user_id, + "document_id": document_id } - if is_public_workspace: - args["public_workspace_id"] = public_workspace_id + save_args["public_workspace_id"] = public_workspace_id elif is_group: - args["group_id"] = group_id + save_args["group_id"] = group_id - result = process_single_tabular_sheet(**args) - if isinstance(result, tuple) and len(result) == 3: - chunks, tokens, model = result - total_chunks_saved = chunks - total_embedding_tokens += tokens - if not embedding_model_name: - embedding_model_name = model - else: - total_chunks_saved = result - - elif file_ext in ('.xlsx', '.xls', '.xlsm'): - # Process Excel (potentially multiple sheets) - excel_file = pandas.ExcelFile( - temp_file_path, - engine='openpyxl' if file_ext in ('.xlsx', '.xlsm') else 'xlrd' - ) - sheet_names = excel_file.sheet_names - base_name, ext = os.path.splitext(original_filename) - - accumulated_total_chunks = 0 - for sheet_name in sheet_names: - update_callback(status=f"Processing sheet '{sheet_name}'...") - # Read specific sheet, get values (not formulas), keep data as string - # Note: pandas typically reads values, not formulas by default. - df = excel_file.parse(sheet_name, keep_default_na=False, dtype=str) + token_usage = save_chunks(**save_args) + total_chunks_saved = 1 + if token_usage: + total_embedding_tokens = token_usage.get('total_tokens', 0) + embedding_model_name = token_usage.get('model_deployment_name') - # Create effective filename for this sheet - effective_filename = f"{base_name}-{sheet_name}{ext}" if len(sheet_names) > 1 else original_filename + # Don't return here — fall through to metadata extraction below + except Exception as e: + log_event(f"[process_tabular] Error creating schema summary, falling back to row-by-row: {e}", level=logging.WARNING) + # Fall through to existing row-by-row processing + # Only do row-by-row chunking if schema-only didn't produce chunks + if total_chunks_saved == 0: + try: + if file_ext == '.csv': + # Process CSV + # Read CSV, attempt to infer header, keep data as string initially + df = pandas.read_csv( + temp_file_path, + keep_default_na=False, + dtype=str + ) args = { "df": df, "document_id": document_id, "user_id": user_id, - "file_name": effective_filename, + "file_name": original_filename, "update_callback": update_callback } @@ -4797,21 +4987,62 @@ def process_tabular(document_id, user_id, temp_file_path, original_filename, fil result = process_single_tabular_sheet(**args) if isinstance(result, tuple) and len(result) == 3: chunks, tokens, model = result - accumulated_total_chunks += chunks + total_chunks_saved = chunks total_embedding_tokens += tokens if not embedding_model_name: embedding_model_name = model else: - accumulated_total_chunks += result + total_chunks_saved = result - total_chunks_saved = accumulated_total_chunks # Total across all sheets + elif file_ext in ('.xlsx', '.xls', '.xlsm'): + # Process Excel (potentially multiple sheets) + excel_file = pandas.ExcelFile( + temp_file_path, + engine='openpyxl' if file_ext in ('.xlsx', '.xlsm') else 'xlrd' + ) + sheet_names = excel_file.sheet_names + base_name, ext = os.path.splitext(original_filename) + accumulated_total_chunks = 0 + for sheet_name in sheet_names: + update_callback(status=f"Processing sheet '{sheet_name}'...") + # Read specific sheet, get values (not formulas), keep data as string + # Note: pandas typically reads values, not formulas by default. + df = excel_file.parse(sheet_name, keep_default_na=False, dtype=str) - except pandas.errors.EmptyDataError: - print(f"Warning: Tabular file or sheet is empty: {original_filename}") - update_callback(status=f"Warning: File/sheet is empty - {original_filename}", number_of_pages=0) - except Exception as e: - raise Exception(f"Failed processing Tabular file {original_filename}: {e}") + # Create effective filename for this sheet + effective_filename = f"{base_name}-{sheet_name}{ext}" if len(sheet_names) > 1 else original_filename + + args = { + "df": df, + "document_id": document_id, + "user_id": user_id, + "file_name": effective_filename, + "update_callback": update_callback + } + + if is_public_workspace: + args["public_workspace_id"] = public_workspace_id + elif is_group: + args["group_id"] = group_id + + result = process_single_tabular_sheet(**args) + if isinstance(result, tuple) and len(result) == 3: + chunks, tokens, model = result + accumulated_total_chunks += chunks + total_embedding_tokens += tokens + if not embedding_model_name: + embedding_model_name = model + else: + accumulated_total_chunks += result + + total_chunks_saved = accumulated_total_chunks # Total across all sheets + + except pandas.errors.EmptyDataError: + log_event(f"[process_tabular] Warning: Tabular file or sheet is empty: {original_filename}", level=logging.WARNING) + update_callback(status=f"Warning: File/sheet is empty - {original_filename}", number_of_pages=0) + except Exception as e: + raise Exception(f"Failed processing Tabular file {original_filename}: {e}") # Extract metadata if enabled and chunks were processed settings = get_settings() diff --git a/application/single_app/functions_global_actions.py b/application/single_app/functions_global_actions.py index 91f0d9f9..4d7293cd 100644 --- a/application/single_app/functions_global_actions.py +++ b/application/single_app/functions_global_actions.py @@ -60,12 +60,13 @@ def get_global_action(action_id, return_type=SecretReturnType.TRIGGER): return None -def save_global_action(action_data): +def save_global_action(action_data, user_id=None): """ Save or update a global action. Args: action_data (dict): Action data to save + user_id (str, optional): The user ID of the person performing the action Returns: dict: Saved action data or None if failed @@ -76,8 +77,27 @@ def save_global_action(action_data): action_data['id'] = str(uuid.uuid4()) # Add metadata action_data['is_global'] = True - action_data['created_at'] = datetime.utcnow().isoformat() - action_data['updated_at'] = datetime.utcnow().isoformat() + now = datetime.utcnow().isoformat() + + # Check if this is a new action or an update to preserve created_by/created_at + existing_action = None + try: + existing_action = cosmos_global_actions_container.read_item( + item=action_data['id'], + partition_key=action_data['id'] + ) + except Exception: + pass + + if existing_action: + action_data['created_by'] = existing_action.get('created_by', user_id) + action_data['created_at'] = existing_action.get('created_at', now) + else: + action_data['created_by'] = user_id + action_data['created_at'] = now + action_data['modified_by'] = user_id + action_data['modified_at'] = now + action_data['updated_at'] = now print(f"💾 Saving global action: {action_data.get('name', 'Unknown')}") # Store secrets in Key Vault before upsert action_data = keyvault_plugin_save_helper(action_data, scope_value=action_data.get('id'), scope="global") diff --git a/application/single_app/functions_global_agents.py b/application/single_app/functions_global_agents.py index 5cf6a3d4..87976510 100644 --- a/application/single_app/functions_global_agents.py +++ b/application/single_app/functions_global_agents.py @@ -163,25 +163,46 @@ def get_global_agent(agent_id): return None -def save_global_agent(agent_data): +def save_global_agent(agent_data, user_id=None): """ Save or update a global agent. Args: agent_data (dict): Agent data to save + user_id (str, optional): The user ID of the person performing the action Returns: dict: Saved agent data or None if failed """ try: - user_id = get_current_user_id() + if user_id is None: + user_id = get_current_user_id() cleaned_agent = sanitize_agent_payload(agent_data) if 'id' not in cleaned_agent: cleaned_agent['id'] = str(uuid.uuid4()) cleaned_agent['is_global'] = True cleaned_agent['is_group'] = False - cleaned_agent['created_at'] = datetime.utcnow().isoformat() - cleaned_agent['updated_at'] = datetime.utcnow().isoformat() + now = datetime.utcnow().isoformat() + + # Check if this is a new agent or an update to preserve created_by/created_at + existing_agent = None + try: + existing_agent = cosmos_global_agents_container.read_item( + item=cleaned_agent['id'], + partition_key=cleaned_agent['id'] + ) + except Exception: + pass + + if existing_agent: + cleaned_agent['created_by'] = existing_agent.get('created_by', user_id) + cleaned_agent['created_at'] = existing_agent.get('created_at', now) + else: + cleaned_agent['created_by'] = user_id + cleaned_agent['created_at'] = now + cleaned_agent['modified_by'] = user_id + cleaned_agent['modified_at'] = now + cleaned_agent['updated_at'] = now log_event( "Saving global agent.", extra={"agent_name": cleaned_agent.get('name', 'Unknown')}, diff --git a/application/single_app/functions_group_actions.py b/application/single_app/functions_group_actions.py index bc6aa4ea..c0d264b1 100644 --- a/application/single_app/functions_group_actions.py +++ b/application/single_app/functions_group_actions.py @@ -82,14 +82,36 @@ def get_group_action( return _clean_action(action, group_id, return_type) -def save_group_action(group_id: str, action_data: Dict[str, Any]) -> Dict[str, Any]: +def save_group_action(group_id: str, action_data: Dict[str, Any], user_id: Optional[str] = None) -> Dict[str, Any]: """Create or update a group action entry.""" payload = dict(action_data) action_id = payload.get("id") or str(uuid.uuid4()) payload["id"] = action_id payload["group_id"] = group_id - payload["last_updated"] = datetime.utcnow().isoformat() + now = datetime.utcnow().isoformat() + payload["last_updated"] = now + + # Track who created/modified this action + existing_action = None + try: + existing_action = cosmos_group_actions_container.read_item( + item=action_id, + partition_key=group_id, + ) + except exceptions.CosmosResourceNotFoundError: + pass + except Exception: + pass + + if existing_action: + payload["created_by"] = existing_action.get("created_by", user_id) + payload["created_at"] = existing_action.get("created_at", now) + else: + payload["created_by"] = user_id + payload["created_at"] = now + payload["modified_by"] = user_id + payload["modified_at"] = now payload.setdefault("name", "") payload.setdefault("displayName", payload.get("name", "")) diff --git a/application/single_app/functions_group_agents.py b/application/single_app/functions_group_agents.py index 8bf6f87c..7cbb8324 100644 --- a/application/single_app/functions_group_agents.py +++ b/application/single_app/functions_group_agents.py @@ -63,16 +63,38 @@ def get_group_agent(group_id: str, agent_id: str) -> Optional[Dict[str, Any]]: return None -def save_group_agent(group_id: str, agent_data: Dict[str, Any]) -> Dict[str, Any]: +def save_group_agent(group_id: str, agent_data: Dict[str, Any], user_id: Optional[str] = None) -> Dict[str, Any]: """Create or update a group agent entry.""" payload = sanitize_agent_payload(agent_data) agent_id = payload.get("id") or str(uuid.uuid4()) payload["id"] = agent_id payload["group_id"] = group_id - payload["last_updated"] = datetime.utcnow().isoformat() + now = datetime.utcnow().isoformat() + payload["last_updated"] = now payload["is_global"] = False payload["is_group"] = True + # Track who created/modified this agent + existing_agent = None + try: + existing_agent = cosmos_group_agents_container.read_item( + item=agent_id, + partition_key=group_id, + ) + except exceptions.CosmosResourceNotFoundError: + pass + except Exception: + pass + + if existing_agent: + payload["created_by"] = existing_agent.get("created_by", user_id) + payload["created_at"] = existing_agent.get("created_at", now) + else: + payload["created_by"] = user_id + payload["created_at"] = now + payload["modified_by"] = user_id + payload["modified_at"] = now + # Required/defaulted fields payload.setdefault("name", "") payload.setdefault("display_name", payload.get("name", "")) diff --git a/application/single_app/functions_personal_actions.py b/application/single_app/functions_personal_actions.py index 6345438e..91d849f3 100644 --- a/application/single_app/functions_personal_actions.py +++ b/application/single_app/functions_personal_actions.py @@ -113,15 +113,26 @@ def save_personal_action(user_id, action_data): existing_action = get_personal_action(user_id, action_data['name']) # Preserve existing ID if updating, or generate new ID if creating + now = datetime.utcnow().isoformat() if existing_action: - # Update existing action - preserve the original ID + # Update existing action - preserve the original ID and creation tracking action_data['id'] = existing_action['id'] + action_data['created_by'] = existing_action.get('created_by', user_id) + action_data['created_at'] = existing_action.get('created_at', now) elif 'id' not in action_data or not action_data['id']: # New action - generate UUID for ID action_data['id'] = str(uuid.uuid4()) - + action_data['created_by'] = user_id + action_data['created_at'] = now + else: + # Has an ID but no existing action found - treat as new + action_data['created_by'] = user_id + action_data['created_at'] = now + action_data['modified_by'] = user_id + action_data['modified_at'] = now + action_data['user_id'] = user_id - action_data['last_updated'] = datetime.utcnow().isoformat() + action_data['last_updated'] = now # Validate required fields required_fields = ['name', 'displayName', 'type', 'description'] diff --git a/application/single_app/functions_personal_agents.py b/application/single_app/functions_personal_agents.py index a4a5e47d..3c6c275e 100644 --- a/application/single_app/functions_personal_agents.py +++ b/application/single_app/functions_personal_agents.py @@ -128,9 +128,33 @@ def save_personal_agent(user_id, agent_data): cleaned_agent.setdefault(field, '') if 'id' not in cleaned_agent: cleaned_agent['id'] = str(f"{user_id}_{cleaned_agent.get('name', 'default')}") - + + # Check if this is a new agent or an update to preserve created_by/created_at + existing_agent = None + try: + existing_agent = cosmos_personal_agents_container.read_item( + item=cleaned_agent['id'], + partition_key=user_id + ) + except exceptions.CosmosResourceNotFoundError: + pass + except Exception: + pass + + now = datetime.utcnow().isoformat() + if existing_agent: + # Preserve original creation tracking + cleaned_agent['created_by'] = existing_agent.get('created_by', user_id) + cleaned_agent['created_at'] = existing_agent.get('created_at', now) + else: + # New agent + cleaned_agent['created_by'] = user_id + cleaned_agent['created_at'] = now + cleaned_agent['modified_by'] = user_id + cleaned_agent['modified_at'] = now + cleaned_agent['user_id'] = user_id - cleaned_agent['last_updated'] = datetime.utcnow().isoformat() + cleaned_agent['last_updated'] = now cleaned_agent['is_global'] = False cleaned_agent['is_group'] = False diff --git a/application/single_app/functions_settings.py b/application/single_app/functions_settings.py index 8176939d..f3dc59de 100644 --- a/application/single_app/functions_settings.py +++ b/application/single_app/functions_settings.py @@ -25,6 +25,7 @@ def get_settings(use_cosmos=False): 'enable_text_plugin': True, 'enable_default_embedding_model_plugin': False, 'enable_fact_memory_plugin': True, + 'enable_tabular_processing_plugin': False, 'enable_multi_agent_orchestration': False, 'max_rounds_per_agent': 1, 'enable_semantic_kernel': False, @@ -205,6 +206,9 @@ def get_settings(use_cosmos=False): 'require_member_of_feedback_admin': False, 'enable_conversation_archiving': False, + # Processing Thoughts + 'enable_thoughts': False, + # Search and Extract 'azure_ai_search_endpoint': '', 'azure_ai_search_key': '', @@ -391,6 +395,9 @@ def update_settings(new_settings): # always fetch the latest settings doc, which includes your merges settings_item = get_settings() settings_item.update(new_settings) + # Dependency enforcement: tabular processing requires enhanced citations + if not settings_item.get('enable_enhanced_citations', False): + settings_item['enable_tabular_processing_plugin'] = False cosmos_settings_container.upsert_item(settings_item) cache_updater = getattr(app_settings_cache, "update_settings_cache", None) if callable(cache_updater): diff --git a/application/single_app/functions_thoughts.py b/application/single_app/functions_thoughts.py new file mode 100644 index 00000000..c6ffe9dd --- /dev/null +++ b/application/single_app/functions_thoughts.py @@ -0,0 +1,256 @@ +# functions_thoughts.py + +import uuid +import time +from datetime import datetime, timezone +from config import cosmos_thoughts_container, cosmos_archived_thoughts_container, cosmos_messages_container +from functions_appinsights import log_event +from functions_settings import get_settings + + +class ThoughtTracker: + """Stateful per-request tracker that writes processing step records to Cosmos DB. + + Each add_thought() call immediately upserts a document so that polling + clients can see partial progress before the final response is sent. + + All Cosmos writes are wrapped in try/except so thought errors never + interrupt the chat processing flow. + """ + + def __init__(self, conversation_id, message_id, thread_id, user_id): + self.conversation_id = conversation_id + self.message_id = message_id + self.thread_id = thread_id + self.user_id = user_id + self.current_index = 0 + settings = get_settings() + self.enabled = settings.get('enable_thoughts', False) + + def add_thought(self, step_type, content, detail=None): + """Write a thought step to Cosmos immediately. + + Args: + step_type: One of search, tabular_analysis, web_search, + agent_tool_call, generation, content_safety. + content: Short human-readable description of the step. + detail: Optional technical detail (function names, params, etc.). + + Returns: + The thought document id, or None if disabled/failed. + """ + if not self.enabled: + return None + + thought_id = str(uuid.uuid4()) + thought_doc = { + 'id': thought_id, + 'conversation_id': self.conversation_id, + 'message_id': self.message_id, + 'thread_id': self.thread_id, + 'user_id': self.user_id, + 'step_index': self.current_index, + 'step_type': step_type, + 'content': content, + 'detail': detail, + 'duration_ms': None, + 'timestamp': datetime.now(timezone.utc).isoformat() + } + self.current_index += 1 + + try: + cosmos_thoughts_container.upsert_item(thought_doc) + except Exception as e: + log_event(f"ThoughtTracker.add_thought failed: {e}", level="WARNING") + return None + + return thought_id + + def complete_thought(self, thought_id, duration_ms): + """Patch an existing thought with its duration after the step finishes.""" + if not self.enabled or not thought_id: + return + + try: + thought_doc = cosmos_thoughts_container.read_item( + item=thought_id, + partition_key=self.user_id + ) + thought_doc['duration_ms'] = duration_ms + cosmos_thoughts_container.upsert_item(thought_doc) + except Exception as e: + log_event(f"ThoughtTracker.complete_thought failed: {e}", level="WARNING") + + def timed_thought(self, step_type, content, detail=None): + """Convenience: add a thought and return a timer helper. + + Usage: + timer = tracker.timed_thought('search', 'Searching documents...') + # ... do work ... + timer.stop() + """ + start = time.time() + thought_id = self.add_thought(step_type, content, detail) + return _ThoughtTimer(self, thought_id, start) + + +class _ThoughtTimer: + """Helper returned by ThoughtTracker.timed_thought() for auto-duration capture.""" + + def __init__(self, tracker, thought_id, start_time): + self._tracker = tracker + self._thought_id = thought_id + self._start = start_time + + def stop(self): + elapsed_ms = int((time.time() - self._start) * 1000) + self._tracker.complete_thought(self._thought_id, elapsed_ms) + return elapsed_ms + + +# --------------------------------------------------------------------------- +# CRUD helpers +# --------------------------------------------------------------------------- + +def get_thoughts_for_message(conversation_id, message_id, user_id): + """Return all thoughts for a specific assistant message, ordered by step_index.""" + try: + query = ( + "SELECT * FROM c " + "WHERE c.conversation_id = @conv_id " + "AND c.message_id = @msg_id " + "ORDER BY c.step_index ASC" + ) + params = [ + {"name": "@conv_id", "value": conversation_id}, + {"name": "@msg_id", "value": message_id}, + ] + results = list(cosmos_thoughts_container.query_items( + query=query, + parameters=params, + partition_key=user_id + )) + return results + except Exception as e: + log_event(f"get_thoughts_for_message failed: {e}", level="WARNING") + return [] + + +def get_pending_thoughts(conversation_id, user_id): + """Return the latest thoughts for a conversation that are still in-progress. + + Used by the polling endpoint. Retrieves thoughts created within the last + 5 minutes for the conversation, grouped by the most recent message_id. + """ + try: + five_minutes_ago = datetime.now(timezone.utc) + from datetime import timedelta + five_minutes_ago = (five_minutes_ago - timedelta(minutes=5)).isoformat() + + query = ( + "SELECT * FROM c " + "WHERE c.conversation_id = @conv_id " + "AND c.timestamp >= @since " + "ORDER BY c.timestamp DESC" + ) + params = [ + {"name": "@conv_id", "value": conversation_id}, + {"name": "@since", "value": five_minutes_ago}, + ] + results = list(cosmos_thoughts_container.query_items( + query=query, + parameters=params, + partition_key=user_id + )) + + if not results: + return [] + + # Group by the most recent message_id + latest_message_id = results[0].get('message_id') + latest_thoughts = [ + t for t in results if t.get('message_id') == latest_message_id + ] + # Return in ascending step_index order + latest_thoughts.sort(key=lambda t: t.get('step_index', 0)) + return latest_thoughts + except Exception as e: + log_event(f"get_pending_thoughts failed: {e}", level="WARNING") + return [] + + +def get_thoughts_for_conversation(conversation_id, user_id): + """Return all thoughts for a conversation.""" + try: + query = ( + "SELECT * FROM c " + "WHERE c.conversation_id = @conv_id " + "ORDER BY c.timestamp ASC" + ) + params = [ + {"name": "@conv_id", "value": conversation_id}, + ] + results = list(cosmos_thoughts_container.query_items( + query=query, + parameters=params, + partition_key=user_id + )) + return results + except Exception as e: + log_event(f"get_thoughts_for_conversation failed: {e}", level="WARNING") + return [] + + +def archive_thoughts_for_conversation(conversation_id, user_id): + """Copy all thoughts for a conversation to the archive container, then delete originals.""" + try: + thoughts = get_thoughts_for_conversation(conversation_id, user_id) + for thought in thoughts: + archived = dict(thought) + archived['archived_at'] = datetime.now(timezone.utc).isoformat() + cosmos_archived_thoughts_container.upsert_item(archived) + + for thought in thoughts: + cosmos_thoughts_container.delete_item( + item=thought['id'], + partition_key=user_id + ) + except Exception as e: + log_event(f"archive_thoughts_for_conversation failed: {e}", level="WARNING") + + +def delete_thoughts_for_conversation(conversation_id, user_id): + """Delete all thoughts for a conversation.""" + try: + thoughts = get_thoughts_for_conversation(conversation_id, user_id) + for thought in thoughts: + cosmos_thoughts_container.delete_item( + item=thought['id'], + partition_key=user_id + ) + except Exception as e: + log_event(f"delete_thoughts_for_conversation failed: {e}", level="WARNING") + + +def delete_thoughts_for_message(message_id, user_id): + """Delete all thoughts associated with a specific assistant message.""" + try: + query = ( + "SELECT * FROM c " + "WHERE c.message_id = @msg_id" + ) + params = [ + {"name": "@msg_id", "value": message_id}, + ] + results = list(cosmos_thoughts_container.query_items( + query=query, + parameters=params, + partition_key=user_id + )) + for thought in results: + cosmos_thoughts_container.delete_item( + item=thought['id'], + partition_key=user_id + ) + except Exception as e: + log_event(f"delete_thoughts_for_message failed: {e}", level="WARNING") diff --git a/application/single_app/route_backend_agents.py b/application/single_app/route_backend_agents.py index 57097ee5..2f631af7 100644 --- a/application/single_app/route_backend_agents.py +++ b/application/single_app/route_backend_agents.py @@ -23,6 +23,11 @@ from functions_appinsights import log_event from json_schema_validation import validate_agent from swagger_wrapper import swagger_route, get_auth_security +from functions_activity_logging import ( + log_agent_creation, + log_agent_update, + log_agent_deletion, +) bpa = Blueprint('admin_agents', __name__) @@ -147,6 +152,18 @@ def set_user_agents(): for agent_name in agents_to_delete: delete_personal_agent(user_id, agent_name) + # Log individual agent activities + for agent in filtered_agents: + a_name = agent.get('name', '') + a_id = agent.get('id', '') + a_display = agent.get('display_name', a_name) + if a_name in current_agent_names: + log_agent_update(user_id=user_id, agent_id=a_id, agent_name=a_name, agent_display_name=a_display, scope='personal') + else: + log_agent_creation(user_id=user_id, agent_id=a_id, agent_name=a_name, agent_display_name=a_display, scope='personal') + for agent_name in agents_to_delete: + log_agent_deletion(user_id=user_id, agent_id=agent_name, agent_name=agent_name, scope='personal') + log_event("User agents updated", extra={"user_id": user_id, "agents_count": len(filtered_agents)}) return jsonify({'success': True}) @@ -175,6 +192,9 @@ def delete_user_agent(agent_name): # Delete from personal_agents container delete_personal_agent(user_id, agent_name) + # Log agent deletion activity + log_agent_deletion(user_id=user_id, agent_id=agent_to_delete.get('id', agent_name), agent_name=agent_name, scope='personal') + # Check if there are any agents left and if they match global_selected_agent remaining_agents = get_personal_agents(user_id) if len(remaining_agents) > 0: @@ -270,11 +290,12 @@ def create_group_agent_route(): cleaned_payload.pop(key, None) try: - saved = save_group_agent(active_group, cleaned_payload) + saved = save_group_agent(active_group, cleaned_payload, user_id=user_id) except Exception as exc: debug_print('Failed to save group agent: %s', exc) return jsonify({'error': 'Unable to save agent'}), 500 + log_agent_creation(user_id=user_id, agent_id=saved.get('id', ''), agent_name=saved.get('name', ''), agent_display_name=saved.get('display_name', ''), scope='group', group_id=active_group) return jsonify(saved), 201 @@ -325,11 +346,12 @@ def update_group_agent_route(agent_id): return jsonify({'error': str(exc)}), 400 try: - saved = save_group_agent(active_group, cleaned_payload) + saved = save_group_agent(active_group, cleaned_payload, user_id=user_id) except Exception as exc: debug_print('Failed to update group agent %s: %s', agent_id, exc) return jsonify({'error': 'Unable to update agent'}), 500 + log_agent_update(user_id=user_id, agent_id=agent_id, agent_name=saved.get('name', ''), agent_display_name=saved.get('display_name', ''), scope='group', group_id=active_group) return jsonify(saved), 200 @@ -360,6 +382,7 @@ def delete_group_agent_route(agent_id): if not removed: return jsonify({'error': 'Agent not found'}), 404 + log_agent_deletion(user_id=user_id, agent_id=agent_id, agent_name=agent_id, scope='group', group_id=active_group) return jsonify({'message': 'Agent deleted'}), 200 # User endpoint to set selected agent (new model, not legacy default_agent) @@ -504,10 +527,11 @@ def add_agent(): cleaned_agent['id'] = '15b0c92a-741d-42ff-ba0b-367c7ee0c848' # Save to global agents container - result = save_global_agent(cleaned_agent) + result = save_global_agent(cleaned_agent, user_id=str(get_current_user_id())) if not result: return jsonify({'error': 'Failed to save agent.'}), 500 + log_agent_creation(user_id=str(get_current_user_id()), agent_id=cleaned_agent.get('id', ''), agent_name=cleaned_agent.get('name', ''), agent_display_name=cleaned_agent.get('display_name', ''), scope='global') log_event("Agent added", extra={"action": "add", "agent": {k: v for k, v in cleaned_agent.items() if k != 'id'}, "user": str(get_current_user_id())}) # --- HOT RELOAD TRIGGER --- setattr(builtins, "kernel_reload_needed", True) @@ -615,10 +639,11 @@ def edit_agent(agent_name): return jsonify({'error': 'Agent not found.'}), 404 # Save the updated agent - result = save_global_agent(cleaned_agent) + result = save_global_agent(cleaned_agent, user_id=str(get_current_user_id())) if not result: return jsonify({'error': 'Failed to save agent.'}), 500 + log_agent_update(user_id=str(get_current_user_id()), agent_id=cleaned_agent.get('id', ''), agent_name=agent_name, agent_display_name=cleaned_agent.get('display_name', ''), scope='global') log_event( f"Agent {agent_name} edited", extra={ @@ -660,6 +685,7 @@ def delete_agent(agent_name): if not success: return jsonify({'error': 'Failed to delete agent.'}), 500 + log_agent_deletion(user_id=str(get_current_user_id()), agent_id=agent_to_delete.get('id', ''), agent_name=agent_name, scope='global') log_event("Agent deleted", extra={"action": "delete", "agent_name": agent_name, "user": str(get_current_user_id())}) # --- HOT RELOAD TRIGGER --- setattr(builtins, "kernel_reload_needed", True) diff --git a/application/single_app/route_backend_chats.py b/application/single_app/route_backend_chats.py index e452fed4..8154b7d4 100644 --- a/application/single_app/route_backend_chats.py +++ b/application/single_app/route_backend_chats.py @@ -28,6 +28,7 @@ from functions_activity_logging import log_chat_activity, log_conversation_creation, log_token_usage from flask import current_app from swagger_wrapper import swagger_route, get_auth_security +from functions_thoughts import ThoughtTracker def get_kernel(): @@ -39,6 +40,185 @@ def get_kernel_agents(): log_event(f"[SKChat] get_kernel_agents - g.kernel_agents: {type(g_agents)} ({len(g_agents) if g_agents else 0} agents), builtins.kernel_agents: {type(builtins_agents)} ({len(builtins_agents) if builtins_agents else 0} agents)", level=logging.INFO) return g_agents or builtins_agents +async def run_tabular_sk_analysis(user_question, tabular_filenames, user_id, + conversation_id, gpt_model, settings, + source_hint="workspace", group_id=None, + public_workspace_id=None): + """Run lightweight SK with TabularProcessingPlugin to analyze tabular data. + + Creates a temporary Kernel with only the TabularProcessingPlugin, uses the + same chat model as the user's session, and returns computed analysis results. + Returns None on failure for graceful degradation. + """ + from semantic_kernel import Kernel as SKKernel + from semantic_kernel.connectors.ai.open_ai import AzureChatCompletion + from semantic_kernel.connectors.ai.function_choice_behavior import FunctionChoiceBehavior + from semantic_kernel.connectors.ai.open_ai.prompt_execution_settings.azure_chat_prompt_execution_settings import AzureChatPromptExecutionSettings + from semantic_kernel.contents.chat_history import ChatHistory as SKChatHistory + from semantic_kernel_plugins.tabular_processing_plugin import TabularProcessingPlugin + + try: + log_event(f"[Tabular SK Analysis] Starting analysis for files: {tabular_filenames}", level=logging.INFO) + + # 1. Create lightweight kernel with only tabular plugin + kernel = SKKernel() + tabular_plugin = TabularProcessingPlugin() + kernel.add_plugin(tabular_plugin, plugin_name="tabular_processing") + + # 2. Create chat service using same config as main chat + enable_gpt_apim = settings.get('enable_gpt_apim', False) + if enable_gpt_apim: + chat_service = AzureChatCompletion( + service_id="tabular-analysis", + deployment_name=gpt_model, + endpoint=settings.get('azure_apim_gpt_endpoint'), + api_key=settings.get('azure_apim_gpt_subscription_key'), + api_version=settings.get('azure_apim_gpt_api_version'), + ) + else: + auth_type = settings.get('azure_openai_gpt_authentication_type') + if auth_type == 'managed_identity': + token_provider = get_bearer_token_provider(DefaultAzureCredential(), cognitive_services_scope) + chat_service = AzureChatCompletion( + service_id="tabular-analysis", + deployment_name=gpt_model, + endpoint=settings.get('azure_openai_gpt_endpoint'), + api_version=settings.get('azure_openai_gpt_api_version'), + ad_token_provider=token_provider, + ) + else: + chat_service = AzureChatCompletion( + service_id="tabular-analysis", + deployment_name=gpt_model, + endpoint=settings.get('azure_openai_gpt_endpoint'), + api_key=settings.get('azure_openai_gpt_key'), + api_version=settings.get('azure_openai_gpt_api_version'), + ) + kernel.add_service(chat_service) + + # 3. Pre-dispatch: load file schemas to eliminate discovery LLM rounds + source_context = f"source='{source_hint}'" + if group_id: + source_context += f", group_id='{group_id}'" + if public_workspace_id: + source_context += f", public_workspace_id='{public_workspace_id}'" + + schema_parts = [] + for fname in tabular_filenames: + try: + container, blob_path = tabular_plugin._resolve_blob_location_with_fallback( + user_id, conversation_id, fname, source_hint, + group_id=group_id, public_workspace_id=public_workspace_id + ) + df = tabular_plugin._read_tabular_blob_to_dataframe(container, blob_path) + df_numeric = tabular_plugin._try_numeric_conversion(df.copy()) + schema_info = { + "filename": fname, + "row_count": len(df), + "columns": list(df.columns), + "dtypes": {col: str(dtype) for col, dtype in df_numeric.dtypes.items()}, + "preview": df.head(3).to_dict(orient='records') + } + schema_parts.append(json.dumps(schema_info, indent=2, default=str)) + log_event(f"[Tabular SK Analysis] Pre-loaded schema for {fname} ({len(df)} rows)", level=logging.DEBUG) + except Exception as e: + log_event(f"[Tabular SK Analysis] Failed to pre-load schema for {fname}: {e}", level=logging.WARNING) + schema_parts.append(json.dumps({"filename": fname, "error": f"Could not pre-load: {str(e)}"})) + + schema_context = "\n".join(schema_parts) + + # 4. Build chat history with pre-loaded schemas + chat_history = SKChatHistory() + chat_history.add_system_message( + "You are a data analyst. Use the tabular_processing plugin functions to " + "analyze the data and answer the user's question.\n\n" + f"FILE SCHEMAS (pre-loaded — do NOT call list_tabular_files or describe_tabular_file):\n" + f"{schema_context}\n\n" + "IMPORTANT: Batch multiple independent function calls in a SINGLE response. " + "For example, call multiple aggregate_column or group_by_aggregate functions " + "at once rather than one at a time.\n\n" + "Return the computed results clearly." + ) + + chat_history.add_user_message( + f"Analyze the tabular data to answer: {user_question}\n" + f"Use user_id='{user_id}', conversation_id='{conversation_id}', {source_context}." + ) + + # 5. Execute with auto function calling + execution_settings = AzureChatPromptExecutionSettings( + service_id="tabular-analysis", + function_choice_behavior=FunctionChoiceBehavior.Auto( + maximum_auto_invoke_attempts=5 + ), + ) + + result = await chat_service.get_chat_message_contents( + chat_history, execution_settings, kernel=kernel + ) + + if result and result[0].content: + analysis = result[0].content + # Cap at 20k characters to stay within token budget + if len(analysis) > 20000: + analysis = analysis[:20000] + "\n[Analysis truncated]" + log_event(f"[Tabular SK Analysis] Analysis complete, {len(analysis)} chars", level=logging.INFO) + return analysis + log_event("[Tabular SK Analysis] No content in SK response", level=logging.WARNING) + return None + + except Exception as e: + log_event(f"[Tabular SK Analysis] Error: {e}", level=logging.WARNING, exceptionTraceback=True) + return None + +def collect_tabular_sk_citations(user_id, conversation_id): + """Collect plugin invocations from the tabular SK analysis and convert to citation format.""" + from semantic_kernel_plugins.plugin_invocation_logger import get_plugin_logger + + plugin_logger = get_plugin_logger() + plugin_invocations = plugin_logger.get_invocations_for_conversation(user_id, conversation_id) + + if not plugin_invocations: + return [] + + def make_json_serializable(obj): + if obj is None: + return None + elif isinstance(obj, (str, int, float, bool)): + return obj + elif isinstance(obj, dict): + return {str(k): make_json_serializable(v) for k, v in obj.items()} + elif isinstance(obj, (list, tuple)): + return [make_json_serializable(item) for item in obj] + else: + return str(obj) + + citations = [] + for inv in plugin_invocations: + timestamp_str = None + if inv.timestamp: + if hasattr(inv.timestamp, 'isoformat'): + timestamp_str = inv.timestamp.isoformat() + else: + timestamp_str = str(inv.timestamp) + + citation = { + 'tool_name': f"{inv.plugin_name}.{inv.function_name}", + 'function_name': inv.function_name, + 'plugin_name': inv.plugin_name, + 'function_arguments': make_json_serializable(inv.parameters), + 'function_result': make_json_serializable(inv.result), + 'duration_ms': inv.duration_ms, + 'timestamp': timestamp_str, + 'success': inv.success, + 'error_message': make_json_serializable(inv.error_message), + 'user_id': inv.user_id + } + citations.append(citation) + + log_event(f"[Tabular SK Citations] Collected {len(citations)} tool execution citations", level=logging.INFO) + return citations + def register_route_backend_chats(app): @app.route('/api/chat', methods=['POST']) @swagger_route(security=get_auth_security()) @@ -668,6 +848,18 @@ def result_requires_message_reload(result: Any) -> bool: conversation_item['last_updated'] = datetime.utcnow().isoformat() cosmos_conversations_container.upsert_item(conversation_item) # Update timestamp and potentially title + + # Generate assistant_message_id early for thought tracking + assistant_message_id = f"{conversation_id}_assistant_{int(time.time())}_{random.randint(1000,9999)}" + + # Initialize thought tracker + thought_tracker = ThoughtTracker( + conversation_id=conversation_id, + message_id=assistant_message_id, + thread_id=current_user_thread_id, + user_id=user_id + ) + # region 3 - Content Safety # --------------------------------------------------------------------- # 3) Check Content Safety (but DO NOT return 403). @@ -679,6 +871,7 @@ def result_requires_message_reload(result: Any) -> bool: blocklist_matches = [] if settings.get('enable_content_safety') and "content_safety_client" in CLIENTS: + thought_tracker.add_thought('content_safety', 'Checking content safety...') try: content_safety_client = CLIENTS["content_safety_client"] request_obj = AnalyzeTextOptions(text=user_message) @@ -836,6 +1029,7 @@ def result_requires_message_reload(result: Any) -> bool: # Perform the search + thought_tracker.add_thought('search', f"Searching {document_scope or 'personal'} workspace documents for '{(search_query or user_message)[:50]}'") try: # Prepare search arguments # Set default and maximum values for top_n @@ -899,6 +1093,8 @@ def result_requires_message_reload(result: Any) -> bool: }), 500 if search_results: + unique_doc_names = set(doc.get('file_name', 'Unknown') for doc in search_results) + thought_tracker.add_thought('search', f"Found {len(search_results)} results from {len(unique_doc_names)} documents") retrieved_texts = [] combined_documents = [] classifications_found = set(conversation_item.get('classification', [])) # Load existing @@ -953,6 +1149,70 @@ def result_requires_message_reload(result: Any) -> bool: 'documents': combined_documents # Keep track of docs used }) + # Auto-detect tabular files in search results and prompt the LLM to use the plugin + if settings.get('enable_tabular_processing_plugin', False) and settings.get('enable_enhanced_citations', False): + tabular_files_in_results = set() + for source_doc in combined_documents: + fname = source_doc.get('file_name', '') + if fname and any(fname.lower().endswith(ext) for ext in TABULAR_EXTENSIONS): + tabular_files_in_results.add(fname) + + if tabular_files_in_results: + # Determine source based on document_scope, not just active IDs + if document_scope == 'group' and active_group_id: + tabular_source_hint = "group" + elif document_scope == 'public' and active_public_workspace_id: + tabular_source_hint = "public" + else: + tabular_source_hint = "workspace" + + tabular_filenames_str = ", ".join(tabular_files_in_results) + + # Run SK tabular analysis to pre-compute results + tabular_analysis = asyncio.run(run_tabular_sk_analysis( + user_question=user_message, + tabular_filenames=tabular_files_in_results, + user_id=user_id, + conversation_id=conversation_id, + gpt_model=gpt_model, + settings=settings, + source_hint=tabular_source_hint, + group_id=active_group_id if tabular_source_hint == "group" else None, + public_workspace_id=active_public_workspace_id if tabular_source_hint == "public" else None, + )) + + if tabular_analysis: + # Inject pre-computed analysis results as context + tabular_system_msg = ( + f"The following analysis was computed from the tabular file(s) " + f"{tabular_filenames_str} using data analysis functions:\n\n" + f"{tabular_analysis}\n\n" + f"Use these computed results to answer the user's question accurately." + ) + else: + # Fallback: instruct LLM to use plugin functions (for agent mode) + tabular_system_msg = ( + f"IMPORTANT: The search results include data from tabular file(s): {tabular_filenames_str}. " + f"The search results contain only a schema summary (column names and a few sample rows), NOT the full data. " + f"You MUST use the tabular_processing plugin functions to answer ANY question about these files. " + f"Do NOT attempt to answer using the schema summary alone — it is incomplete. " + f"Available functions: describe_tabular_file, aggregate_column, filter_rows, query_tabular_data, group_by_aggregate. " + f"Use source='{tabular_source_hint}'" + + (f" and group_id='{active_group_id}'" if tabular_source_hint == "group" else "") + + (f" and public_workspace_id='{active_public_workspace_id}'" if tabular_source_hint == "public" else "") + + "." + ) + system_messages_for_augmentation.append({ + 'role': 'system', + 'content': tabular_system_msg + }) + + # Collect tool execution citations from SK tabular analysis + if tabular_analysis: + tabular_sk_citations = collect_tabular_sk_citations(user_id, conversation_id) + if tabular_sk_citations: + agent_citations_list.extend(tabular_sk_citations) + # Loop through each source document/chunk used for this message for source_doc in combined_documents: # 4. Create a citation dictionary, selecting the desired fields @@ -1138,8 +1398,8 @@ def result_requires_message_reload(result: Any) -> bool: """ # Update the system message with enhanced content and updated documents array if system_messages_for_augmentation: - system_messages_for_augmentation[-1]['content'] = system_prompt_search - system_messages_for_augmentation[-1]['documents'] = combined_documents + system_messages_for_augmentation[0]['content'] = system_prompt_search + system_messages_for_augmentation[0]['documents'] = combined_documents # --- END NEW METADATA CITATIONS --- # Update conversation classifications if new ones were found @@ -1489,6 +1749,7 @@ def result_requires_message_reload(result: Any) -> bool: }), status_code if web_search_enabled: + thought_tracker.add_thought('web_search', f"Searching the web for '{(search_query or user_message)[:50]}'") perform_web_search( settings=settings, conversation_id=conversation_id, @@ -1504,7 +1765,9 @@ def result_requires_message_reload(result: Any) -> bool: agent_citations_list=agent_citations_list, web_search_citations_list=web_search_citations_list, ) - + if web_search_citations_list: + thought_tracker.add_thought('web_search', f"Got {len(web_search_citations_list)} web search results") + # region 5 - FINAL conversation history preparation # --------------------------------------------------------------------- # 5) Prepare FINAL conversation history for GPT (including summarization) @@ -1650,6 +1913,7 @@ def result_requires_message_reload(result: Any) -> bool: allowed_roles_in_history = ['user', 'assistant'] # Add 'system' if you PERSIST general system messages not related to augmentation max_file_content_length_in_history = 50000 # Increased limit for all file content in history max_tabular_content_length_in_history = 50000 # Same limit for tabular data consistency + chat_tabular_files = set() # Track tabular files uploaded directly to chat for message in recent_messages: role = message.get('role') @@ -1685,25 +1949,38 @@ def result_requires_message_reload(result: Any) -> bool: filename = message.get('filename', 'uploaded_file') file_content = message.get('file_content', '') # Assuming file content is stored is_table = message.get('is_table', False) - - # Use higher limit for tabular data that needs complete analysis - content_limit = max_tabular_content_length_in_history if is_table else max_file_content_length_in_history - - display_content = file_content[:content_limit] - if len(file_content) > content_limit: - display_content += "..." - - # Enhanced message for tabular data - if is_table: + file_content_source = message.get('file_content_source', '') + + # Tabular files stored in blob (enhanced citations enabled) - reference plugin + if is_table and file_content_source == 'blob': + chat_tabular_files.add(filename) # Track for mini SK analysis conversation_history_for_api.append({ - 'role': 'system', # Represent file as system info - 'content': f"[User uploaded a tabular data file named '{filename}'. This is CSV format data for analysis:\n{display_content}]\nThis is complete tabular data in CSV format. You can perform calculations, analysis, and data operations on this dataset." + 'role': 'system', + 'content': f"[User uploaded a tabular data file named '{filename}'. " + f"The file is stored in blob storage and available for analysis. " + f"Use the tabular_processing plugin functions (list_tabular_files, describe_tabular_file, " + f"aggregate_column, filter_rows, query_tabular_data, group_by_aggregate) to analyze this data. " + f"The file source is 'chat'.]" }) else: - conversation_history_for_api.append({ - 'role': 'system', # Represent file as system info - 'content': f"[User uploaded a file named '{filename}'. Content preview:\n{display_content}]\nUse this file context if relevant." - }) + # Use higher limit for tabular data that needs complete analysis + content_limit = max_tabular_content_length_in_history if is_table else max_file_content_length_in_history + + display_content = file_content[:content_limit] + if len(file_content) > content_limit: + display_content += "..." + + # Enhanced message for tabular data + if is_table: + conversation_history_for_api.append({ + 'role': 'system', # Represent file as system info + 'content': f"[User uploaded a tabular data file named '{filename}'. This is CSV format data for analysis:\n{display_content}]\nThis is complete tabular data in CSV format. You can perform calculations, analysis, and data operations on this dataset." + }) + else: + conversation_history_for_api.append({ + 'role': 'system', # Represent file as system info + 'content': f"[User uploaded a file named '{filename}'. Content preview:\n{display_content}]\nUse this file context if relevant." + }) elif role == 'image': # Handle image uploads with extracted text and vision analysis filename = message.get('filename', 'uploaded_image') is_user_upload = message.get('metadata', {}).get('is_user_upload', False) @@ -1767,6 +2044,45 @@ def result_requires_message_reload(result: Any) -> bool: # Ignored roles: 'safety', 'blocked', 'system' (if they are only for augmentation/summary) + # --- Mini SK analysis for tabular files uploaded directly to chat --- + if chat_tabular_files and settings.get('enable_tabular_processing_plugin', False) and settings.get('enable_enhanced_citations', False): + chat_tabular_filenames_str = ", ".join(chat_tabular_files) + log_event( + f"[Chat Tabular SK] Detected {len(chat_tabular_files)} tabular file(s) uploaded to chat: {chat_tabular_filenames_str}", + level=logging.INFO + ) + + chat_tabular_analysis = asyncio.run(run_tabular_sk_analysis( + user_question=user_message, + tabular_filenames=chat_tabular_files, + user_id=user_id, + conversation_id=conversation_id, + gpt_model=gpt_model, + settings=settings, + source_hint="chat", + )) + + if chat_tabular_analysis: + # Inject pre-computed analysis results as context + conversation_history_for_api.append({ + 'role': 'system', + 'content': ( + f"The following analysis was computed from the chat-uploaded tabular file(s) " + f"{chat_tabular_filenames_str} using data analysis functions:\n\n" + f"{chat_tabular_analysis}\n\n" + f"Use these computed results to answer the user's question accurately." + ) + }) + + # Collect tool execution citations from SK tabular analysis + chat_tabular_sk_citations = collect_tabular_sk_citations(user_id, conversation_id) + if chat_tabular_sk_citations: + agent_citations_list.extend(chat_tabular_sk_citations) + + debug_print(f"[Chat Tabular SK] Analysis injected, {len(chat_tabular_analysis)} chars") + else: + debug_print("[Chat Tabular SK] Analysis returned None, relying on existing file context messages") + # Ensure the very last message is the current user's message (it should be if fetched correctly) if not conversation_history_for_api or conversation_history_for_api[-1]['role'] != 'user': debug_print("Warning: Last message in history is not the user's current message. Appending.") @@ -2110,6 +2426,23 @@ def orchestrator_error(e): }) if selected_agent: + thought_tracker.add_thought('agent_tool_call', f"Sending to agent '{getattr(selected_agent, 'display_name', getattr(selected_agent, 'name', 'unknown'))}'") + + # Register callback to write plugin thoughts to Cosmos in real-time + callback_key = f"{user_id}:{conversation_id}" + plugin_logger = get_plugin_logger() + + def on_plugin_invocation(inv): + duration_str = f" ({int(inv.duration_ms)}ms)" if inv.duration_ms else "" + tool_name = f"{inv.plugin_name}.{inv.function_name}" + thought_tracker.add_thought( + 'agent_tool_call', + f"Agent called {tool_name}{duration_str}", + detail=f"success={inv.success}" + ) + + plugin_logger.register_callback(callback_key, on_plugin_invocation) + def invoke_selected_agent(): return asyncio.run(run_sk_call( selected_agent.invoke, @@ -2120,16 +2453,18 @@ def agent_success(result): msg = str(result) notice = None agent_used = getattr(selected_agent, 'name', 'All Plugins') - + + # Deregister real-time thought callback + plugin_logger.deregister_callbacks(callback_key) + # Get the actual model deployment used by the agent actual_model_deployment = getattr(selected_agent, 'deployment_name', None) or agent_used debug_print(f"Agent '{agent_used}' using deployment: {actual_model_deployment}") - + # Extract detailed plugin invocations for enhanced agent citations - plugin_logger = get_plugin_logger() - # CRITICAL FIX: Filter by user_id and conversation_id to prevent cross-conversation contamination + # (Thoughts already written to Cosmos in real-time by callback) plugin_invocations = plugin_logger.get_invocations_for_conversation(user_id, conversation_id) - + # Convert plugin invocations to citation format with detailed information detailed_citations = [] for inv in plugin_invocations: @@ -2204,6 +2539,7 @@ def make_json_serializable(obj): ) return (msg, actual_model_deployment, "agent", notice) def agent_error(e): + plugin_logger.deregister_callbacks(callback_key) debug_print(f"Error during Semantic Kernel Agent invocation: {str(e)}") log_event( f"Error during Semantic Kernel Agent invocation: {str(e)}", @@ -2244,8 +2580,17 @@ def foundry_agent_success(result): or agent_used ) + # Deregister real-time thought callback + plugin_logger.deregister_callbacks(callback_key) + foundry_citations = getattr(selected_agent, 'last_run_citations', []) or [] if foundry_citations: + # Emit thoughts for Foundry agent citations/tool calls + for citation in foundry_citations: + thought_tracker.add_thought( + 'agent_tool_call', + f"Agent retrieved citation from Azure AI Foundry" + ) for citation in foundry_citations: try: serializable = json.loads(json.dumps(citation, default=str)) @@ -2282,6 +2627,7 @@ def foundry_agent_success(result): return (msg, actual_model_deployment, 'agent', notice) def foundry_agent_error(e): + plugin_logger.deregister_callbacks(callback_key) log_event( f"Error during Azure AI Foundry agent invocation: {str(e)}", extra={ @@ -2360,6 +2706,7 @@ def kernel_error(e): 'on_error': kernel_error }) + thought_tracker.add_thought('generation', 'Generating response...') def invoke_gpt_fallback(): if not conversation_history_for_api: raise Exception('Cannot generate response: No conversation history available.') @@ -2510,8 +2857,8 @@ def gpt_error(e): if hasattr(selected_agent, 'name'): agent_name = selected_agent.name - assistant_message_id = f"{conversation_id}_assistant_{int(time.time())}_{random.randint(1000,9999)}" - + # assistant_message_id was generated earlier for thought tracking + # Get user_info and thread_id from the user message for ownership tracking and threading user_info_for_assistant = None user_thread_id = None @@ -2672,7 +3019,8 @@ def gpt_error(e): 'web_search_citations': web_search_citations_list, 'agent_citations': agent_citations_list, 'reload_messages': reload_messages_required, - 'kernel_fallback_notice': kernel_fallback_notice + 'kernel_fallback_notice': kernel_fallback_notice, + 'thoughts_enabled': thought_tracker.enabled }), 200 except Exception as e: @@ -3111,10 +3459,27 @@ def generate(): conversation_item['last_updated'] = datetime.utcnow().isoformat() cosmos_conversations_container.upsert_item(conversation_item) - + + # Generate assistant_message_id early for thought tracking + assistant_message_id = f"{conversation_id}_assistant_{int(time.time())}_{random.randint(1000,9999)}" + + # Initialize thought tracker for streaming path + thought_tracker = ThoughtTracker( + conversation_id=conversation_id, + message_id=assistant_message_id, + thread_id=current_user_thread_id, + user_id=user_id + ) + + def emit_thought(step_type, content, detail=None): + """Add a thought to Cosmos and return an SSE event string.""" + thought_tracker.add_thought(step_type, content, detail) + return f"data: {json.dumps({'type': 'thought', 'step_index': thought_tracker.current_index - 1, 'step_type': step_type, 'content': content})}\n\n" + # Hybrid search (if enabled) combined_documents = [] if hybrid_search_enabled: + yield emit_thought('search', f"Searching {document_scope or 'personal'} workspace documents for '{(search_query or user_message)[:50]}'") try: search_args = { "query": search_query, @@ -3144,8 +3509,10 @@ def generate(): search_results = hybrid_search(**search_args) except Exception as e: debug_print(f"Error during hybrid search: {e}") - + if search_results: + unique_doc_names_stream = set(doc.get('file_name', 'Unknown') for doc in search_results) + yield emit_thought('search', f"Found {len(search_results)} results from {len(unique_doc_names_stream)} documents") retrieved_texts = [] for doc in search_results: @@ -3319,11 +3686,60 @@ def generate(): 'content': system_prompt_search, 'documents': combined_documents }) - + + # Auto-detect tabular files in search results and run SK analysis + if settings.get('enable_tabular_processing_plugin', False) and settings.get('enable_enhanced_citations', False): + tabular_files_in_results = set() + for source_doc in combined_documents: + fname = source_doc.get('file_name', '') + if fname and any(fname.lower().endswith(ext) for ext in TABULAR_EXTENSIONS): + tabular_files_in_results.add(fname) + + if tabular_files_in_results: + # Determine source based on document_scope, not just active IDs + if document_scope == 'group' and active_group_id: + tabular_source_hint = "group" + elif document_scope == 'public' and active_public_workspace_id: + tabular_source_hint = "public" + else: + tabular_source_hint = "workspace" + + tabular_filenames_str = ", ".join(tabular_files_in_results) + + # Run SK tabular analysis to pre-compute results + tabular_analysis = asyncio.run(run_tabular_sk_analysis( + user_question=user_message, + tabular_filenames=tabular_files_in_results, + user_id=user_id, + conversation_id=conversation_id, + gpt_model=gpt_model, + settings=settings, + source_hint=tabular_source_hint, + group_id=active_group_id if tabular_source_hint == "group" else None, + public_workspace_id=active_public_workspace_id if tabular_source_hint == "public" else None, + )) + + if tabular_analysis: + system_messages_for_augmentation.append({ + 'role': 'system', + 'content': ( + f"The following analysis was computed from the tabular file(s) " + f"{tabular_filenames_str} using data analysis functions:\n\n" + f"{tabular_analysis}\n\n" + f"Use these computed results to answer the user's question accurately." + ) + }) + + # Collect tool execution citations from SK tabular analysis + tabular_sk_citations = collect_tabular_sk_citations(user_id, conversation_id) + if tabular_sk_citations: + agent_citations_list.extend(tabular_sk_citations) + # Reorder hybrid citations list in descending order based on page_number hybrid_citations_list.sort(key=lambda x: x.get('page_number', 0), reverse=True) if web_search_enabled: + yield emit_thought('web_search', f"Searching the web for '{(search_query or user_message)[:50]}'") perform_web_search( settings=settings, conversation_id=conversation_id, @@ -3339,6 +3755,8 @@ def generate(): agent_citations_list=agent_citations_list, web_search_citations_list=web_search_citations_list, ) + if web_search_citations_list: + yield emit_thought('web_search', f"Got {len(web_search_citations_list)} web search results") # Update message chat type message_chat_type = None @@ -3381,15 +3799,108 @@ def generate(): 'content': aug_msg['content'] }) - # Add recent messages + # Add recent messages (with file role handling) allowed_roles_in_history = ['user', 'assistant'] + max_file_content_length_in_history = 50000 + max_tabular_content_length_in_history = 50000 + chat_tabular_files = set() # Track tabular files uploaded directly to chat + for message in recent_messages: - if message.get('role') in allowed_roles_in_history: + role = message.get('role') + content = message.get('content', '') + + if role in allowed_roles_in_history: conversation_history_for_api.append({ - 'role': message['role'], - 'content': message.get('content', '') + 'role': role, + 'content': content }) - + elif role == 'file': + filename = message.get('filename', 'uploaded_file') + file_content = message.get('file_content', '') + is_table = message.get('is_table', False) + file_content_source = message.get('file_content_source', '') + + # Tabular files stored in blob - track for mini SK analysis + if is_table and file_content_source == 'blob': + chat_tabular_files.add(filename) + conversation_history_for_api.append({ + 'role': 'system', + 'content': ( + f"[User uploaded a tabular data file named '{filename}'. " + f"The file is stored in blob storage and available for analysis. " + f"Use the tabular_processing plugin functions (list_tabular_files, " + f"describe_tabular_file, aggregate_column, filter_rows, " + f"query_tabular_data, group_by_aggregate) to analyze this data. " + f"The file source is 'chat'.]" + ) + }) + else: + content_limit = ( + max_tabular_content_length_in_history if is_table + else max_file_content_length_in_history + ) + display_content = file_content[:content_limit] + if len(file_content) > content_limit: + display_content += "..." + + if is_table: + conversation_history_for_api.append({ + 'role': 'system', + 'content': ( + f"[User uploaded a tabular data file named '{filename}'. " + f"This is CSV format data for analysis:\n{display_content}]\n" + f"This is complete tabular data in CSV format. You can perform " + f"calculations, analysis, and data operations on this dataset." + ) + }) + else: + conversation_history_for_api.append({ + 'role': 'system', + 'content': ( + f"[User uploaded a file named '{filename}'. " + f"Content preview:\n{display_content}]\n" + f"Use this file context if relevant." + ) + }) + + # --- Mini SK analysis for tabular files uploaded directly to chat --- + if chat_tabular_files and settings.get('enable_tabular_processing_plugin', False) and settings.get('enable_enhanced_citations', False): + chat_tabular_filenames_str = ", ".join(chat_tabular_files) + log_event( + f"[Chat Tabular SK] Streaming: Detected {len(chat_tabular_files)} tabular file(s) uploaded to chat: {chat_tabular_filenames_str}", + level=logging.INFO + ) + + chat_tabular_analysis = asyncio.run(run_tabular_sk_analysis( + user_question=user_message, + tabular_filenames=chat_tabular_files, + user_id=user_id, + conversation_id=conversation_id, + gpt_model=gpt_model, + settings=settings, + source_hint="chat", + )) + + if chat_tabular_analysis: + conversation_history_for_api.append({ + 'role': 'system', + 'content': ( + f"The following analysis was computed from the chat-uploaded tabular file(s) " + f"{chat_tabular_filenames_str} using data analysis functions:\n\n" + f"{chat_tabular_analysis}\n\n" + f"Use these computed results to answer the user's question accurately." + ) + }) + + # Collect tool execution citations + chat_tabular_sk_citations = collect_tabular_sk_citations(user_id, conversation_id) + if chat_tabular_sk_citations: + agent_citations_list.extend(chat_tabular_sk_citations) + + debug_print(f"[Chat Tabular SK] Streaming: Analysis injected, {len(chat_tabular_analysis)} chars") + else: + debug_print("[Chat Tabular SK] Streaming: Analysis returned None, relying on existing file context") + except Exception as e: yield f"data: {json.dumps({'error': f'History error: {str(e)}'})}\n\n" return @@ -3472,7 +3983,7 @@ def generate(): # Stream the response accumulated_content = "" token_usage_data = None # Will be populated from final stream chunk - assistant_message_id = f"{conversation_id}_assistant_{int(time.time())}_{random.randint(1000,9999)}" + # assistant_message_id was generated earlier for thought tracking final_model_used = gpt_model # Default to gpt_model, will be overridden if agent is used # DEBUG: Check agent streaming decision @@ -3482,8 +3993,23 @@ def generate(): try: if use_agent_streaming and selected_agent: # Stream from agent using invoke_stream + yield emit_thought('agent_tool_call', f"Sending to agent '{agent_display_name_used or agent_name_used}'") debug_print(f"--- Streaming from Agent: {agent_name_used} ---") - + + # Register callback to persist plugin thoughts to Cosmos in real-time + callback_key = f"{user_id}:{conversation_id}" + plugin_logger_cb = get_plugin_logger() + + def on_plugin_invocation_streaming(inv): + duration_str = f" ({int(inv.duration_ms)}ms)" if inv.duration_ms else "" + tool_name = f"{inv.plugin_name}.{inv.function_name}" + thought_tracker.add_thought( + 'agent_tool_call', + f"Agent called {tool_name}{duration_str}" + ) + + plugin_logger_cb.register_callback(callback_key, on_plugin_invocation_streaming) + # Import required classes from semantic_kernel.contents.chat_message_content import ChatMessageContent @@ -3524,7 +4050,6 @@ async def stream_agent_async(): return chunks, usage_data # Execute async streaming - import asyncio try: # Try to get existing event loop loop = asyncio.get_event_loop() @@ -3539,36 +4064,49 @@ async def stream_agent_async(): try: # Run streaming and collect chunks and usage chunks, stream_usage = loop.run_until_complete(stream_agent_async()) - - # Yield chunks to frontend - for chunk_content in chunks: - accumulated_content += chunk_content - yield f"data: {json.dumps({'content': chunk_content})}\n\n" - - # Try to capture token usage from stream metadata - if stream_usage: - # stream_usage is a CompletionUsage object, not a dict - prompt_tokens = getattr(stream_usage, 'prompt_tokens', 0) - completion_tokens = getattr(stream_usage, 'completion_tokens', 0) - total_tokens = getattr(stream_usage, 'total_tokens', None) - - # Calculate total if not provided - if total_tokens is None or total_tokens == 0: - total_tokens = prompt_tokens + completion_tokens - - token_usage_data = { - 'prompt_tokens': prompt_tokens, - 'completion_tokens': completion_tokens, - 'total_tokens': total_tokens, - 'captured_at': datetime.utcnow().isoformat() - } - debug_print(f"[Agent Streaming Tokens] From metadata - prompt: {prompt_tokens}, completion: {completion_tokens}, total: {total_tokens}") except Exception as stream_error: + plugin_logger_cb.deregister_callbacks(callback_key) debug_print(f"❌ Agent streaming error: {stream_error}") import traceback traceback.print_exc() yield f"data: {json.dumps({'error': f'Agent streaming failed: {str(stream_error)}'})}\n\n" return + + # Deregister callback (agent completed successfully) + plugin_logger_cb.deregister_callbacks(callback_key) + + # Emit SSE-only events for streaming UI (Cosmos writes already done by callback) + agent_plugin_invocations = plugin_logger_cb.get_invocations_for_conversation(user_id, conversation_id) + for inv in agent_plugin_invocations: + duration_str = f" ({int(inv.duration_ms)}ms)" if inv.duration_ms else "" + tool_name = f"{inv.plugin_name}.{inv.function_name}" + content = f"Agent called {tool_name}{duration_str}" + yield f"data: {json.dumps({'type': 'thought', 'step_index': thought_tracker.current_index, 'step_type': 'agent_tool_call', 'content': content})}\n\n" + thought_tracker.current_index += 1 + + # Yield chunks to frontend + for chunk_content in chunks: + accumulated_content += chunk_content + yield f"data: {json.dumps({'content': chunk_content})}\n\n" + + # Try to capture token usage from stream metadata + if stream_usage: + # stream_usage is a CompletionUsage object, not a dict + prompt_tokens = getattr(stream_usage, 'prompt_tokens', 0) + completion_tokens = getattr(stream_usage, 'completion_tokens', 0) + total_tokens = getattr(stream_usage, 'total_tokens', None) + + # Calculate total if not provided + if total_tokens is None or total_tokens == 0: + total_tokens = prompt_tokens + completion_tokens + + token_usage_data = { + 'prompt_tokens': prompt_tokens, + 'completion_tokens': completion_tokens, + 'total_tokens': total_tokens, + 'captured_at': datetime.utcnow().isoformat() + } + debug_print(f"[Agent Streaming Tokens] From metadata - prompt: {prompt_tokens}, completion: {completion_tokens}, total: {total_tokens}") # Collect token usage from kernel services if not captured from stream if not token_usage_data: @@ -3650,6 +4188,7 @@ def make_json_serializable(obj): else: # Stream from regular GPT model (non-agent) + yield emit_thought('generation', 'Generating response...') debug_print(f"--- Streaming from GPT ({gpt_model}) ---") # Prepare stream parameters @@ -3818,7 +4357,8 @@ def make_json_serializable(obj): 'agent_citations': agent_citations_list, 'agent_display_name': agent_display_name_used if use_agent_streaming else None, 'agent_name': agent_name_used if use_agent_streaming else None, - 'full_content': accumulated_content + 'full_content': accumulated_content, + 'thoughts_enabled': thought_tracker.enabled } yield f"data: {json.dumps(final_data)}\n\n" diff --git a/application/single_app/route_backend_conversations.py b/application/single_app/route_backend_conversations.py index f267d729..ac25bd8c 100644 --- a/application/single_app/route_backend_conversations.py +++ b/application/single_app/route_backend_conversations.py @@ -8,6 +8,7 @@ from functions_debug import debug_print from swagger_wrapper import swagger_route, get_auth_security from functions_activity_logging import log_conversation_creation, log_conversation_deletion, log_conversation_archival +from functions_thoughts import archive_thoughts_for_conversation, delete_thoughts_for_conversation def register_route_backend_conversations(app): @@ -430,7 +431,14 @@ def delete_conversation(conversation_id): cosmos_archived_messages_container.upsert_item(archived_doc) cosmos_messages_container.delete_item(doc['id'], partition_key=conversation_id) - + + # Archive/delete thoughts for conversation + user_id_for_thoughts = conversation_item.get('user_id') + if archiving_enabled: + archive_thoughts_for_conversation(conversation_id, user_id_for_thoughts) + else: + delete_thoughts_for_conversation(conversation_id, user_id_for_thoughts) + # Log conversation deletion before actual deletion log_conversation_deletion( user_id=conversation_item.get('user_id'), @@ -530,7 +538,13 @@ def delete_multiple_conversations(): cosmos_archived_messages_container.upsert_item(archived_message) cosmos_messages_container.delete_item(message['id'], partition_key=conversation_id) - + + # Archive/delete thoughts for conversation + if archiving_enabled: + archive_thoughts_for_conversation(conversation_id, user_id) + else: + delete_thoughts_for_conversation(conversation_id, user_id) + # Log conversation deletion before actual deletion log_conversation_deletion( user_id=user_id, diff --git a/application/single_app/route_backend_documents.py b/application/single_app/route_backend_documents.py index 28d4ef69..0e9d490b 100644 --- a/application/single_app/route_backend_documents.py +++ b/application/single_app/route_backend_documents.py @@ -7,6 +7,7 @@ from utils_cache import invalidate_personal_search_cache from functions_debug import * from functions_activity_logging import log_document_upload, log_document_metadata_update_transaction +import io import os import requests from flask import current_app @@ -72,7 +73,58 @@ def get_file_content(): filename = items_sorted[0].get('filename', 'Untitled') is_table = items_sorted[0].get('is_table', False) - debug_print(f"[GET_FILE_CONTENT] Filename: {filename}, is_table: {is_table}") + file_content_source = items_sorted[0].get('file_content_source', '') + debug_print(f"[GET_FILE_CONTENT] Filename: {filename}, is_table: {is_table}, source: {file_content_source}") + + # Handle blob-stored tabular files (enhanced citations enabled) + if file_content_source == 'blob': + blob_container = items_sorted[0].get('blob_container', '') + blob_path = items_sorted[0].get('blob_path', '') + debug_print(f"[GET_FILE_CONTENT] Blob-stored file: container={blob_container}, path={blob_path}") + + if not blob_container or not blob_path: + return jsonify({'error': 'Blob storage reference is incomplete'}), 500 + + try: + blob_service_client = CLIENTS.get("storage_account_office_docs_client") + if not blob_service_client: + return jsonify({'error': 'Blob storage client not available'}), 500 + + blob_client = blob_service_client.get_blob_client( + container=blob_container, + blob=blob_path + ) + stream = blob_client.download_blob() + blob_data = stream.readall() + + # Convert to CSV using pandas for display + file_ext = os.path.splitext(filename)[1].lower() + if file_ext == '.csv': + import pandas + df = pandas.read_csv(io.BytesIO(blob_data)) + combined_content = df.to_csv(index=False) + elif file_ext in ['.xlsx', '.xlsm']: + import pandas + df = pandas.read_excel(io.BytesIO(blob_data), engine='openpyxl') + combined_content = df.to_csv(index=False) + elif file_ext == '.xls': + import pandas + df = pandas.read_excel(io.BytesIO(blob_data), engine='xlrd') + combined_content = df.to_csv(index=False) + else: + combined_content = blob_data.decode('utf-8', errors='replace') + + debug_print(f"[GET_FILE_CONTENT] Successfully read blob content, length: {len(combined_content)}") + return jsonify({ + 'file_content': combined_content, + 'filename': filename, + 'is_table': is_table, + 'file_content_source': 'blob' + }), 200 + + except Exception as blob_err: + debug_print(f"[GET_FILE_CONTENT] Error reading from blob: {blob_err}") + return jsonify({'error': f'Error reading file from storage: {str(blob_err)}'}), 500 add_file_task_to_file_processing_log(document_id=file_id, user_id=user_id, content="Combining file content from chunks, filename: " + filename + ", is_table: " + str(is_table)) combined_parts = [] diff --git a/application/single_app/route_backend_plugins.py b/application/single_app/route_backend_plugins.py index 77aab866..f4d4dca0 100644 --- a/application/single_app/route_backend_plugins.py +++ b/application/single_app/route_backend_plugins.py @@ -32,6 +32,11 @@ from functions_debug import debug_print from json_schema_validation import validate_plugin +from functions_activity_logging import ( + log_action_creation, + log_action_update, + log_action_deletion, +) def discover_plugin_types(): # Dynamically discover allowed plugin types from available plugin classes. @@ -345,6 +350,19 @@ def set_user_plugins(): except Exception as e: debug_print(f"Error saving personal actions for user {user_id}: {e}") return jsonify({'error': 'Failed to save plugins'}), 500 + + # Log individual action activities + for plugin in filtered_plugins: + p_name = plugin.get('name', '') + p_id = plugin.get('id', '') + p_type = plugin.get('type', '') + if p_name in current_action_names: + log_action_update(user_id=user_id, action_id=p_id, action_name=p_name, action_type=p_type, scope='personal') + else: + log_action_creation(user_id=user_id, action_id=p_id, action_name=p_name, action_type=p_type, scope='personal') + for plugin_name in (current_action_names - new_plugin_names): + log_action_deletion(user_id=user_id, action_id=plugin_name, action_name=plugin_name, scope='personal') + log_event("User plugins updated", extra={"user_id": user_id, "plugins_count": len(filtered_plugins)}) return jsonify({'success': True}) @@ -360,6 +378,7 @@ def delete_user_plugin(plugin_name): if not deleted: return jsonify({'error': 'Plugin not found.'}), 404 + log_action_deletion(user_id=user_id, action_id=plugin_name, action_name=plugin_name, scope='personal') log_event("User plugin deleted", extra={"user_id": user_id, "plugin_name": plugin_name}) return jsonify({'success': True}) @@ -460,6 +479,13 @@ def create_group_action_route(): for key in ('group_id', 'last_updated', 'user_id', 'is_global', 'is_group', 'scope'): payload.pop(key, None) + # Handle endpoint based on plugin type (same logic as personal plugins) + plugin_type = payload.get('type', '') + if plugin_type in ['sql_schema', 'sql_query']: + payload.setdefault('endpoint', f'sql://{plugin_type}') + elif plugin_type == 'msgraph': + payload.setdefault('endpoint', 'https://graph.microsoft.com') + # Merge with schema to ensure all required fields are present (same as global actions) schema_dir = os.path.join(current_app.root_path, 'static', 'json', 'schemas') merged = get_merged_plugin_settings(payload.get('type'), payload, schema_dir) @@ -467,11 +493,12 @@ def create_group_action_route(): payload['additionalFields'] = merged.get('additionalFields', payload.get('additionalFields', {})) try: - saved = save_group_action(active_group, payload) + saved = save_group_action(active_group, payload, user_id=user_id) except Exception as exc: debug_print('Failed to save group action: %s', exc) return jsonify({'error': 'Unable to save action'}), 500 + log_action_creation(user_id=user_id, action_id=saved.get('id', ''), action_name=saved.get('name', ''), action_type=saved.get('type', ''), scope='group', group_id=active_group) return jsonify(saved), 201 @@ -516,6 +543,13 @@ def update_group_action_route(action_id): merged['is_group'] = True merged['id'] = existing.get('id', action_id) + # Handle endpoint based on plugin type (same logic as personal plugins) + plugin_type = merged.get('type', '') + if plugin_type in ['sql_schema', 'sql_query']: + merged.setdefault('endpoint', f'sql://{plugin_type}') + elif plugin_type == 'msgraph': + merged.setdefault('endpoint', 'https://graph.microsoft.com') + try: validate_group_action_payload(merged, partial=False) except ValueError as exc: @@ -528,11 +562,12 @@ def update_group_action_route(action_id): merged['additionalFields'] = schema_merged.get('additionalFields', merged.get('additionalFields', {})) try: - saved = save_group_action(active_group, merged) + saved = save_group_action(active_group, merged, user_id=user_id) except Exception as exc: debug_print('Failed to update group action %s: %s', action_id, exc) return jsonify({'error': 'Unable to update action'}), 500 + log_action_update(user_id=user_id, action_id=action_id, action_name=saved.get('name', ''), action_type=saved.get('type', ''), scope='group', group_id=active_group) return jsonify(saved), 200 @@ -563,6 +598,7 @@ def delete_group_action_route(action_id): if not removed: return jsonify({'error': 'Action not found'}), 404 + log_action_deletion(user_id=user_id, action_id=action_id, action_name=action_id, scope='group', group_id=active_group) return jsonify({'message': 'Action deleted'}), 200 @bpap.route('/api/user/plugins/types', methods=['GET']) @@ -588,6 +624,8 @@ def get_core_plugin_settings(): 'enable_text_plugin': bool(settings.get('enable_text_plugin', True)), 'enable_default_embedding_model_plugin': bool(settings.get('enable_default_embedding_model_plugin', True)), 'enable_fact_memory_plugin': bool(settings.get('enable_fact_memory_plugin', True)), + 'enable_tabular_processing_plugin': bool(settings.get('enable_tabular_processing_plugin', False)), + 'enable_enhanced_citations': bool(settings.get('enable_enhanced_citations', False)), 'enable_semantic_kernel': bool(settings.get('enable_semantic_kernel', False)), 'allow_user_plugins': bool(settings.get('allow_user_plugins', True)), 'allow_group_plugins': bool(settings.get('allow_group_plugins', True)), @@ -610,6 +648,7 @@ def update_core_plugin_settings(): 'enable_text_plugin', 'enable_default_embedding_model_plugin', 'enable_fact_memory_plugin', + 'enable_tabular_processing_plugin', 'allow_user_plugins', 'allow_group_plugins' ] @@ -627,6 +666,11 @@ def update_core_plugin_settings(): return jsonify({'error': f"Field '{key}' must be a boolean."}), 400 updates[key] = data[key] logging.info("Validated plugin settings: %s", updates) + # Dependency: tabular processing requires enhanced citations + if updates.get('enable_tabular_processing_plugin', False): + full_settings = get_settings() + if not full_settings.get('enable_enhanced_citations', False): + return jsonify({'error': 'Tabular Processing requires Enhanced Citations to be enabled.'}), 400 # Update settings success = update_settings(updates) if success: @@ -692,9 +736,10 @@ def add_plugin(): new_plugin['id'] = plugin_id # Save to global actions container - save_global_action(new_plugin) + save_global_action(new_plugin, user_id=str(get_current_user_id())) - log_event("Plugin added", extra={"action": "add", "plugin": new_plugin, "user": str(getattr(request, 'user', 'unknown'))}) + log_action_creation(user_id=str(get_current_user_id()), action_id=plugin_id, action_name=new_plugin.get('name', ''), action_type=new_plugin.get('type', ''), scope='global') + log_event("Plugin added", extra={"action": "add", "plugin": new_plugin, "user": str(get_current_user_id())}) # --- HOT RELOAD TRIGGER --- setattr(builtins, "kernel_reload_needed", True) @@ -753,9 +798,10 @@ def edit_plugin(plugin_name): # Delete old and save updated if 'id' in found_plugin: delete_global_action(found_plugin['id']) - save_global_action(updated_plugin) + save_global_action(updated_plugin, user_id=str(get_current_user_id())) - log_event("Plugin edited", extra={"action": "edit", "plugin": updated_plugin, "user": str(getattr(request, 'user', 'unknown'))}) + log_action_update(user_id=str(get_current_user_id()), action_id=updated_plugin.get('id', ''), action_name=plugin_name, action_type=updated_plugin.get('type', ''), scope='global') + log_event("Plugin edited", extra={"action": "edit", "plugin": updated_plugin, "user": str(get_current_user_id())}) # --- HOT RELOAD TRIGGER --- setattr(builtins, "kernel_reload_needed", True) return jsonify({'success': True}) @@ -796,7 +842,8 @@ def delete_plugin(plugin_name): if 'id' in plugin_to_delete: delete_global_action(plugin_to_delete['id']) - log_event("Plugin deleted", extra={"action": "delete", "plugin_name": plugin_name, "user": str(getattr(request, 'user', 'unknown'))}) + log_action_deletion(user_id=str(get_current_user_id()), action_id=plugin_to_delete.get('id', ''), action_name=plugin_name, action_type=plugin_to_delete.get('type', ''), scope='global') + log_event("Plugin deleted", extra={"action": "delete", "plugin_name": plugin_name, "user": str(get_current_user_id())}) # --- HOT RELOAD TRIGGER --- setattr(builtins, "kernel_reload_needed", True) return jsonify({'success': True}) @@ -928,4 +975,116 @@ def _merge_group_and_global_actions(group_actions, global_actions): return normalized_actions +@bpap.route('/api/plugins/test-sql-connection', methods=['POST']) +@swagger_route(security=get_auth_security()) +@login_required +@user_required +def test_sql_connection(): + """Test a SQL database connection using provided configuration.""" + data = request.get_json(silent=True) or {} + database_type = (data.get('database_type') or 'sqlserver').lower() + connection_method = data.get('connection_method', 'parameters') + connection_string = data.get('connection_string', '') + server = data.get('server', '') + database = data.get('database', '') + port = data.get('port', '') + driver = data.get('driver', '') + username = data.get('username', '') + password = data.get('password', '') + auth_type = data.get('auth_type', 'username_password') + timeout = min(int(data.get('timeout', 10)), 15) # Cap at 15 seconds for test + + # Map azure_sql to sqlserver + if database_type in ('azure_sql', 'azuresql'): + database_type = 'sqlserver' + + try: + if database_type == 'sqlserver': + import pyodbc + if connection_method == 'connection_string' and connection_string: + conn = pyodbc.connect(connection_string, timeout=timeout) + else: + if not server or not database: + return jsonify({'success': False, 'error': 'Server and database are required for individual parameters connection.'}), 400 + drv = driver or 'ODBC Driver 17 for SQL Server' + conn_str = f"DRIVER={{{drv}}};SERVER={server};DATABASE={database}" + if port: + conn_str += f",{port}" + if auth_type == 'username_password' and username and password: + conn_str += f";UID={username};PWD={password}" + elif auth_type == 'managed_identity': + conn_str += ";Authentication=ActiveDirectoryMsi" + elif auth_type == 'integrated': + conn_str += ";Trusted_Connection=yes" + conn = pyodbc.connect(conn_str, timeout=timeout) + cursor = conn.cursor() + cursor.execute("SELECT 1") + cursor.close() + conn.close() + return jsonify({'success': True, 'message': f'Successfully connected to {data.get("database", "database")} on {data.get("server", "server")}.'}) + + elif database_type == 'postgresql': + import psycopg2 + if connection_method == 'connection_string' and connection_string: + conn = psycopg2.connect(connection_string, connect_timeout=timeout) + else: + if not server or not database: + return jsonify({'success': False, 'error': 'Server and database are required.'}), 400 + conn_params = {'host': server, 'database': database, 'connect_timeout': timeout} + if port: + conn_params['port'] = int(port) + if username: + conn_params['user'] = username + if password: + conn_params['password'] = password + conn = psycopg2.connect(**conn_params) + cursor = conn.cursor() + cursor.execute("SELECT 1") + cursor.close() + conn.close() + return jsonify({'success': True, 'message': f'Successfully connected to PostgreSQL database {data.get("database", "")}.'}) + + elif database_type == 'mysql': + import pymysql + if connection_method == 'connection_string' and connection_string: + # pymysql doesn't natively parse connection strings, so use params + return jsonify({'success': False, 'error': 'MySQL test connection requires individual parameters, not a connection string.'}), 400 + if not server or not database: + return jsonify({'success': False, 'error': 'Server and database are required.'}), 400 + conn_params = {'host': server, 'database': database, 'connect_timeout': timeout} + if port: + conn_params['port'] = int(port) + if username: + conn_params['user'] = username + if password: + conn_params['password'] = password + conn = pymysql.connect(**conn_params) + cursor = conn.cursor() + cursor.execute("SELECT 1") + cursor.close() + conn.close() + return jsonify({'success': True, 'message': f'Successfully connected to MySQL database {data.get("database", "")}.'}) + + elif database_type == 'sqlite': + import sqlite3 + db_path = connection_string or database + if not db_path: + return jsonify({'success': False, 'error': 'Database path is required for SQLite.'}), 400 + conn = sqlite3.connect(db_path, timeout=timeout) + cursor = conn.cursor() + cursor.execute("SELECT 1") + cursor.close() + conn.close() + return jsonify({'success': True, 'message': f'Successfully connected to SQLite database.'}) + + else: + return jsonify({'success': False, 'error': f'Unsupported database type: {database_type}'}), 400 + except ImportError as e: + return jsonify({'success': False, 'error': f'Database driver not installed: {str(e)}'}), 400 + except Exception as e: + error_msg = str(e) + # Sanitize error message to avoid leaking sensitive details + if 'password' in error_msg.lower() or 'pwd' in error_msg.lower(): + error_msg = 'Authentication failed. Please check your credentials.' + return jsonify({'success': False, 'error': f'Connection failed: {error_msg}'}), 400 diff --git a/application/single_app/route_backend_thoughts.py b/application/single_app/route_backend_thoughts.py new file mode 100644 index 00000000..a7624a3f --- /dev/null +++ b/application/single_app/route_backend_thoughts.py @@ -0,0 +1,80 @@ +# route_backend_thoughts.py + +from flask import request, jsonify +from functions_authentication import login_required, user_required, get_current_user_id +from functions_settings import get_settings +from functions_thoughts import get_thoughts_for_message, get_pending_thoughts +from swagger_wrapper import swagger_route, get_auth_security +from functions_appinsights import log_event + + +def register_route_backend_thoughts(app): + + @app.route('/api/conversations//messages//thoughts', methods=['GET']) + @swagger_route(security=get_auth_security()) + @login_required + @user_required + def api_get_message_thoughts(conversation_id, message_id): + """Return persisted thoughts for a specific assistant message.""" + user_id = get_current_user_id() + if not user_id: + return jsonify({'error': 'User not authenticated'}), 401 + + settings = get_settings() + if not settings.get('enable_thoughts', False): + return jsonify({'thoughts': [], 'enabled': False}), 200 + + try: + thoughts = get_thoughts_for_message(conversation_id, message_id, user_id) + # Strip internal Cosmos fields before returning + sanitized = [] + for t in thoughts: + sanitized.append({ + 'id': t.get('id'), + 'step_index': t.get('step_index'), + 'step_type': t.get('step_type'), + 'content': t.get('content'), + 'detail': t.get('detail'), + 'duration_ms': t.get('duration_ms'), + 'timestamp': t.get('timestamp') + }) + return jsonify({'thoughts': sanitized, 'enabled': True}), 200 + except Exception as e: + log_event(f"api_get_message_thoughts error: {e}", level="WARNING") + return jsonify({'error': 'Failed to retrieve thoughts'}), 500 + + @app.route('/api/conversations//thoughts/pending', methods=['GET']) + @swagger_route(security=get_auth_security()) + @login_required + @user_required + def api_get_pending_thoughts(conversation_id): + """Return the latest in-progress thoughts for a conversation. + + Used by the non-streaming frontend to poll for thought updates + while waiting for the chat response. + """ + user_id = get_current_user_id() + if not user_id: + return jsonify({'error': 'User not authenticated'}), 401 + + settings = get_settings() + if not settings.get('enable_thoughts', False): + return jsonify({'thoughts': [], 'enabled': False}), 200 + + try: + thoughts = get_pending_thoughts(conversation_id, user_id) + sanitized = [] + for t in thoughts: + sanitized.append({ + 'id': t.get('id'), + 'step_index': t.get('step_index'), + 'step_type': t.get('step_type'), + 'content': t.get('content'), + 'detail': t.get('detail'), + 'duration_ms': t.get('duration_ms'), + 'timestamp': t.get('timestamp') + }) + return jsonify({'thoughts': sanitized, 'enabled': True}), 200 + except Exception as e: + log_event(f"api_get_pending_thoughts error: {e}", level="WARNING") + return jsonify({'error': 'Failed to retrieve pending thoughts'}), 500 diff --git a/application/single_app/route_backend_user_agreement.py b/application/single_app/route_backend_user_agreement.py index f46559ff..b76213b3 100644 --- a/application/single_app/route_backend_user_agreement.py +++ b/application/single_app/route_backend_user_agreement.py @@ -130,7 +130,7 @@ def api_accept_user_agreement(): return jsonify({"error": "workspace_id and workspace_type are required"}), 400 # Validate workspace type - valid_types = ["personal", "group", "public"] + valid_types = ["personal", "group", "public", "chat"] if workspace_type not in valid_types: return jsonify({"error": f"Invalid workspace_type. Must be one of: {', '.join(valid_types)}"}), 400 diff --git a/application/single_app/route_enhanced_citations.py b/application/single_app/route_enhanced_citations.py index c81ef225..44a35223 100644 --- a/application/single_app/route_enhanced_citations.py +++ b/application/single_app/route_enhanced_citations.py @@ -8,6 +8,7 @@ import requests import mimetypes import io +import pandas from functions_authentication import login_required, user_required, get_current_user_id from functions_settings import get_settings, enabled_required @@ -15,7 +16,7 @@ from functions_group import get_user_groups from functions_public_workspaces import get_user_visible_public_workspace_ids_from_settings from swagger_wrapper import swagger_route, get_auth_security -from config import CLIENTS, storage_account_user_documents_container_name, storage_account_group_documents_container_name, storage_account_public_documents_container_name, IMAGE_EXTENSIONS, VIDEO_EXTENSIONS, AUDIO_EXTENSIONS +from config import CLIENTS, storage_account_user_documents_container_name, storage_account_group_documents_container_name, storage_account_public_documents_container_name, storage_account_personal_chat_container_name, IMAGE_EXTENSIONS, VIDEO_EXTENSIONS, AUDIO_EXTENSIONS, TABULAR_EXTENSIONS, cosmos_messages_container from functions_debug import debug_print def register_enhanced_citations_routes(app): @@ -183,6 +184,189 @@ def get_enhanced_citation_pdf(): except Exception as e: return jsonify({"error": str(e)}), 500 + @app.route("/api/enhanced_citations/tabular", methods=["GET"]) + @swagger_route(security=get_auth_security()) + @login_required + @user_required + @enabled_required("enable_enhanced_citations") + def get_enhanced_citation_tabular(): + """ + Serve original tabular file (CSV, XLSX, etc.) from blob storage for download. + Used for chat-uploaded tabular files stored in blob storage. + """ + conversation_id = request.args.get("conversation_id") + file_id = request.args.get("file_id") + + if not conversation_id or not file_id: + return jsonify({"error": "conversation_id and file_id are required"}), 400 + + user_id = get_current_user_id() + if not user_id: + return jsonify({"error": "User not authenticated"}), 401 + + try: + # Look up the file message in Cosmos to get blob reference + query_str = """ + SELECT * FROM c + WHERE c.conversation_id = @conversation_id + AND c.id = @file_id + """ + items = list(cosmos_messages_container.query_items( + query=query_str, + parameters=[ + {'name': '@conversation_id', 'value': conversation_id}, + {'name': '@file_id', 'value': file_id} + ], + partition_key=conversation_id + )) + + if not items: + return jsonify({"error": "File not found"}), 404 + + file_msg = items[0] + file_content_source = file_msg.get('file_content_source', '') + + if file_content_source != 'blob': + return jsonify({"error": "File is not stored in blob storage"}), 400 + + blob_container = file_msg.get('blob_container', '') + blob_path = file_msg.get('blob_path', '') + filename = file_msg.get('filename', 'download') + + if not blob_container or not blob_path: + return jsonify({"error": "Blob reference is incomplete"}), 500 + + blob_service_client = CLIENTS.get("storage_account_office_docs_client") + if not blob_service_client: + return jsonify({"error": "Storage not available"}), 500 + + blob_client = blob_service_client.get_blob_client( + container=blob_container, + blob=blob_path + ) + stream = blob_client.download_blob() + content = stream.readall() + + # Determine content type + content_type, _ = mimetypes.guess_type(filename) + if not content_type: + content_type = 'application/octet-stream' + + return Response( + content, + content_type=content_type, + headers={ + 'Content-Length': str(len(content)), + 'Content-Disposition': f'attachment; filename="{filename}"', + 'Cache-Control': 'private, max-age=300', + } + ) + + except Exception as e: + debug_print(f"Error serving tabular citation: {e}") + return jsonify({"error": str(e)}), 500 + + @app.route("/api/enhanced_citations/tabular_workspace", methods=["GET"]) + @swagger_route(security=get_auth_security()) + @login_required + @user_required + @enabled_required("enable_enhanced_citations") + def get_enhanced_citation_tabular_workspace(): + """ + Serve tabular file (CSV, XLSX, etc.) from blob storage for workspace documents. + Uses doc_id to look up the document across personal, group, and public workspaces. + """ + doc_id = request.args.get("doc_id") + if not doc_id: + return jsonify({"error": "doc_id is required"}), 400 + + user_id = get_current_user_id() + if not user_id: + return jsonify({"error": "User not authenticated"}), 401 + + try: + doc_response, status_code = get_document(user_id, doc_id) + if status_code != 200: + return doc_response, status_code + + raw_doc = doc_response.get_json() + file_name = raw_doc.get('file_name', '') + ext = file_name.lower().split('.')[-1] if '.' in file_name else '' + + if ext not in ('csv', 'xlsx', 'xls', 'xlsm'): + return jsonify({"error": "File is not a tabular file"}), 400 + + return serve_enhanced_citation_content(raw_doc, force_download=True) + + except Exception as e: + debug_print(f"Error serving tabular workspace citation: {e}") + return jsonify({"error": str(e)}), 500 + + @app.route("/api/enhanced_citations/tabular_preview", methods=["GET"]) + @swagger_route(security=get_auth_security()) + @login_required + @user_required + @enabled_required("enable_enhanced_citations") + def get_enhanced_citation_tabular_preview(): + """ + Return JSON preview of a tabular file for rendering as an HTML table. + Reads the file into a pandas DataFrame and returns columns + rows as JSON. + """ + doc_id = request.args.get("doc_id") + max_rows = min(int(request.args.get("max_rows", 200)), 500) + if not doc_id: + return jsonify({"error": "doc_id is required"}), 400 + + user_id = get_current_user_id() + if not user_id: + return jsonify({"error": "User not authenticated"}), 401 + + try: + doc_response, status_code = get_document(user_id, doc_id) + if status_code != 200: + return doc_response, status_code + + raw_doc = doc_response.get_json() + file_name = raw_doc.get('file_name', '') + ext = file_name.lower().rsplit('.', 1)[-1] if '.' in file_name else '' + if ext not in ('csv', 'xlsx', 'xls', 'xlsm'): + return jsonify({"error": "File is not a tabular file"}), 400 + + # Download blob + workspace_type, container_name = determine_workspace_type_and_container(raw_doc) + blob_name = get_blob_name(raw_doc, workspace_type) + blob_service_client = CLIENTS.get("storage_account_office_docs_client") + if not blob_service_client: + return jsonify({"error": "Blob storage client not available"}), 500 + blob_client = blob_service_client.get_blob_client(container=container_name, blob=blob_name) + data = blob_client.download_blob().readall() + + # Read into DataFrame + if ext == 'csv': + df = pandas.read_csv(io.BytesIO(data), keep_default_na=False, dtype=str) + elif ext in ('xlsx', 'xlsm'): + df = pandas.read_excel(io.BytesIO(data), engine='openpyxl', keep_default_na=False, dtype=str) + elif ext == 'xls': + df = pandas.read_excel(io.BytesIO(data), engine='xlrd', keep_default_na=False, dtype=str) + else: + return jsonify({"error": f"Unsupported file type: {ext}"}), 400 + + total_rows = len(df) + preview = df.head(max_rows) + + return jsonify({ + "filename": file_name, + "total_rows": total_rows, + "total_columns": len(df.columns), + "columns": list(df.columns), + "rows": preview.values.tolist(), + "truncated": total_rows > max_rows + }) + + except Exception as e: + debug_print(f"Error generating tabular preview: {e}") + return jsonify({"error": str(e)}), 500 + def get_document(user_id, doc_id): """ Get document metadata - searches across all enabled workspace types diff --git a/application/single_app/route_frontend_admin_settings.py b/application/single_app/route_frontend_admin_settings.py index 578e1545..fdd77ce5 100644 --- a/application/single_app/route_frontend_admin_settings.py +++ b/application/single_app/route_frontend_admin_settings.py @@ -98,6 +98,8 @@ def admin_settings(): settings['enable_text_plugin'] = False if 'enable_fact_memory_plugin' not in settings: settings['enable_fact_memory_plugin'] = False + if 'enable_tabular_processing_plugin' not in settings: + settings['enable_tabular_processing_plugin'] = False if 'enable_default_embedding_model_plugin' not in settings: settings['enable_default_embedding_model_plugin'] = False if 'enable_multi_agent_orchestration' not in settings: @@ -809,9 +811,10 @@ def is_valid_url(url): 'require_member_of_safety_violation_admin': require_member_of_safety_violation_admin, # ADDED 'require_member_of_feedback_admin': require_member_of_feedback_admin, # ADDED - # Feedback & Archiving + # Feedback, Archiving & Thoughts 'enable_user_feedback': form_data.get('enable_user_feedback') == 'on', 'enable_conversation_archiving': form_data.get('enable_conversation_archiving') == 'on', + 'enable_thoughts': form_data.get('enable_thoughts') == 'on', # Search (Web Search via Azure AI Foundry agent) 'enable_web_search': enable_web_search, diff --git a/application/single_app/route_frontend_chats.py b/application/single_app/route_frontend_chats.py index ca0feb1a..67a41879 100644 --- a/application/single_app/route_frontend_chats.py +++ b/application/single_app/route_frontend_chats.py @@ -237,8 +237,33 @@ def upload_file(): # Handle XML, YAML, and LOG files as text for inline chat extracted_content = extract_text_file(temp_file_path) elif file_ext_nodot in TABULAR_EXTENSIONS: - extracted_content = extract_table_file(temp_file_path, file_ext) is_table = True + + # Upload tabular file to blob storage for tabular processing plugin access + if settings.get('enable_enhanced_citations', False): + try: + blob_service_client = CLIENTS.get("storage_account_office_docs_client") + if blob_service_client: + blob_path = f"{user_id}/{conversation_id}/{filename}" + blob_client = blob_service_client.get_blob_client( + container=storage_account_personal_chat_container_name, + blob=blob_path + ) + metadata = { + "conversation_id": str(conversation_id), + "user_id": str(user_id) + } + with open(temp_file_path, "rb") as blob_f: + blob_client.upload_blob(blob_f, overwrite=True, metadata=metadata) + log_event(f"Uploaded chat tabular file to blob storage: {blob_path}") + except Exception as blob_err: + log_event( + f"Warning: Failed to upload chat tabular file to blob storage: {blob_err}", + level=logging.WARNING + ) + else: + # Only extract content for Cosmos storage when enhanced citations is disabled + extracted_content = extract_table_file(temp_file_path, file_ext) else: return jsonify({'error': 'Unsupported file type'}), 400 @@ -395,25 +420,50 @@ def upload_file(): current_thread_id = str(uuid.uuid4()) - file_message = { - 'id': file_message_id, - 'conversation_id': conversation_id, - 'role': 'file', - 'filename': filename, - 'file_content': extracted_content, - 'is_table': is_table, - 'timestamp': datetime.utcnow().isoformat(), - 'model_deployment_name': None, - 'metadata': { - 'thread_info': { - 'thread_id': current_thread_id, - 'previous_thread_id': previous_thread_id, - 'active_thread': True, - 'thread_attempt': 1 + # When enhanced citations is enabled and file is tabular, store a lightweight + # reference without file_content to avoid Cosmos DB size limits. + # The tabular data lives in blob storage and is served from there. + if is_table and settings.get('enable_enhanced_citations', False): + file_message = { + 'id': file_message_id, + 'conversation_id': conversation_id, + 'role': 'file', + 'filename': filename, + 'is_table': is_table, + 'file_content_source': 'blob', + 'blob_container': storage_account_personal_chat_container_name, + 'blob_path': f"{user_id}/{conversation_id}/{filename}", + 'timestamp': datetime.utcnow().isoformat(), + 'model_deployment_name': None, + 'metadata': { + 'thread_info': { + 'thread_id': current_thread_id, + 'previous_thread_id': previous_thread_id, + 'active_thread': True, + 'thread_attempt': 1 + } } } - } - + else: + file_message = { + 'id': file_message_id, + 'conversation_id': conversation_id, + 'role': 'file', + 'filename': filename, + 'file_content': extracted_content, + 'is_table': is_table, + 'timestamp': datetime.utcnow().isoformat(), + 'model_deployment_name': None, + 'metadata': { + 'thread_info': { + 'thread_id': current_thread_id, + 'previous_thread_id': previous_thread_id, + 'active_thread': True, + 'thread_attempt': 1 + } + } + } + # Add vision analysis if available if vision_analysis: file_message['vision_analysis'] = vision_analysis diff --git a/application/single_app/semantic_kernel_loader.py b/application/single_app/semantic_kernel_loader.py index 78f54203..eb9685bc 100644 --- a/application/single_app/semantic_kernel_loader.py +++ b/application/single_app/semantic_kernel_loader.py @@ -19,6 +19,7 @@ from semantic_kernel.functions.kernel_plugin import KernelPlugin from semantic_kernel_plugins.embedding_model_plugin import EmbeddingModelPlugin from semantic_kernel_plugins.fact_memory_plugin import FactMemoryPlugin +from semantic_kernel_plugins.tabular_processing_plugin import TabularProcessingPlugin from functions_settings import get_settings, get_user_settings from foundry_agent_runtime import AzureAIFoundryChatCompletionAgent from functions_appinsights import log_event, get_appinsights_logger @@ -408,6 +409,13 @@ def load_embedding_model_plugin(kernel: Kernel, settings): description="Provides text embedding functions using the configured embedding model." ) +def load_tabular_processing_plugin(kernel: Kernel): + kernel.add_plugin( + TabularProcessingPlugin(), + plugin_name="tabular_processing", + description="Provides data analysis on tabular files (CSV, XLSX) stored in blob storage. Can list files, describe schemas, aggregate columns, filter rows, run queries, and perform group-by operations." + ) + def load_core_plugins_only(kernel: Kernel, settings): """Load only core plugins for model-only conversations without agents.""" debug_print(f"[SK Loader] Loading core plugins only for model-only mode...") @@ -429,6 +437,10 @@ def load_core_plugins_only(kernel: Kernel, settings): load_text_plugin(kernel) log_event("[SK Loader] Loaded Text plugin.", level=logging.INFO) + if settings.get('enable_tabular_processing_plugin', False) and settings.get('enable_enhanced_citations', False): + load_tabular_processing_plugin(kernel) + log_event("[SK Loader] Loaded Tabular Processing plugin.", level=logging.INFO) + # =================== Semantic Kernel Initialization =================== def initialize_semantic_kernel(user_id: str=None, redis_client=None): debug_print(f"[SK Loader] Initializing Semantic Kernel and plugins...") @@ -1013,6 +1025,14 @@ def load_plugins_for_kernel(kernel, plugin_manifests, settings, mode_label="glob except Exception as e: log_event(f"[SK Loader] Failed to load Fact Memory Plugin: {e}", level=logging.WARNING) + # Register Tabular Processing Plugin if enabled (requires enhanced citations) + if settings.get('enable_tabular_processing_plugin', False) and settings.get('enable_enhanced_citations', False): + try: + load_tabular_processing_plugin(kernel) + log_event("[SK Loader] Loaded Tabular Processing plugin.", level=logging.INFO) + except Exception as e: + log_event(f"[SK Loader] Failed to load Tabular Processing plugin: {e}", level=logging.WARNING) + # Conditionally load static embedding model plugin if settings.get('enable_default_embedding_model_plugin', True): try: @@ -1357,7 +1377,11 @@ def load_user_semantic_kernel(kernel: Kernel, settings, user_id: str, redis_clie load_embedding_model_plugin(kernel, settings) print(f"[SK Loader] Loaded Default Embedding Model plugin.") log_event("[SK Loader] Loaded Default Embedding Model plugin.", level=logging.INFO) - + + if settings.get('enable_tabular_processing_plugin', False) and settings.get('enable_enhanced_citations', False): + load_tabular_processing_plugin(kernel) + log_event("[SK Loader] Loaded Tabular Processing plugin.", level=logging.INFO) + # Get selected agent from user settings (this still needs to be in user settings for UI state) user_settings = get_user_settings(user_id).get('settings', {}) selected_agent = user_settings.get('selected_agent') diff --git a/application/single_app/semantic_kernel_plugins/plugin_invocation_logger.py b/application/single_app/semantic_kernel_plugins/plugin_invocation_logger.py index f982f0a4..bddf9cda 100644 --- a/application/single_app/semantic_kernel_plugins/plugin_invocation_logger.py +++ b/application/single_app/semantic_kernel_plugins/plugin_invocation_logger.py @@ -11,6 +11,7 @@ import logging import functools import inspect +import threading from typing import Any, Dict, List, Optional, Callable from datetime import datetime from dataclasses import dataclass, asdict @@ -51,24 +52,29 @@ def __init__(self): self.invocations: List[PluginInvocation] = [] self.max_history = 1000 # Keep last 1000 invocations in memory self.logger = get_appinsights_logger() or logging.getLogger(__name__) + self._callbacks: Dict[str, List[Callable[[PluginInvocation], None]]] = {} + self._callback_lock = threading.Lock() def log_invocation(self, invocation: PluginInvocation): """Log a plugin invocation to Application Insights and local history.""" # Add to local history self.invocations.append(invocation) - + # Trim history if needed if len(self.invocations) > self.max_history: self.invocations = self.invocations[-self.max_history:] - + # Enhanced terminal logging self._log_to_terminal(invocation) - + # Log to Application Insights self._log_to_appinsights(invocation) - + # Log to standard logging self._log_to_standard(invocation) + + # Fire registered thought callbacks + self._fire_callbacks(invocation) def _log_to_terminal(self, invocation: PluginInvocation): """Log detailed invocation information to terminal.""" @@ -277,6 +283,34 @@ def clear_history(self): """Clear the invocation history.""" self.invocations.clear() + def register_callback(self, key, callback): + """Register a callback fired on each plugin invocation for the given key. + + Args: + key: A string key, typically f"{user_id}:{conversation_id}". + callback: Called with the PluginInvocation after it is logged. + """ + with self._callback_lock: + if key not in self._callbacks: + self._callbacks[key] = [] + self._callbacks[key].append(callback) + + def deregister_callbacks(self, key): + """Remove all callbacks for the given key.""" + with self._callback_lock: + self._callbacks.pop(key, None) + + def _fire_callbacks(self, invocation): + """Fire matching callbacks for this invocation's user+conversation.""" + key = f"{invocation.user_id}:{invocation.conversation_id}" + with self._callback_lock: + callbacks = list(self._callbacks.get(key, [])) + for cb in callbacks: + try: + cb(invocation) + except Exception as e: + log_event(f"Plugin invocation callback error: {e}", level="WARNING") + # Global instance _plugin_logger = PluginInvocationLogger() diff --git a/application/single_app/semantic_kernel_plugins/tabular_processing_plugin.py b/application/single_app/semantic_kernel_plugins/tabular_processing_plugin.py new file mode 100644 index 00000000..a525250b --- /dev/null +++ b/application/single_app/semantic_kernel_plugins/tabular_processing_plugin.py @@ -0,0 +1,515 @@ +# tabular_processing_plugin.py +""" +TabularProcessingPlugin for Semantic Kernel: provides data analysis operations +on tabular files (CSV, XLSX, XLS, XLSM) stored in Azure Blob Storage. + +Works with workspace documents (user-documents, group-documents, public-documents) +and chat-uploaded documents (personal-chat container). +""" +import asyncio +import io +import json +import logging +import pandas +from typing import Annotated, Optional, List +from semantic_kernel.functions import kernel_function +from semantic_kernel_plugins.plugin_invocation_logger import plugin_function_logger +from functions_appinsights import log_event +from config import ( + CLIENTS, + TABULAR_EXTENSIONS, + storage_account_user_documents_container_name, + storage_account_personal_chat_container_name, + storage_account_group_documents_container_name, + storage_account_public_documents_container_name, +) + + +class TabularProcessingPlugin: + """Provides data analysis functions on tabular files stored in blob storage.""" + + SUPPORTED_EXTENSIONS = {'.csv', '.xlsx', '.xls', '.xlsm'} + + def __init__(self): + self._df_cache = {} # Per-instance cache: (container, blob_name) -> DataFrame + + def _get_blob_service_client(self): + """Get the blob service client from CLIENTS cache.""" + client = CLIENTS.get("storage_account_office_docs_client") + if not client: + raise RuntimeError("Blob storage client not available. Enhanced citations must be enabled.") + return client + + def _list_tabular_blobs(self, container_name: str, prefix: str) -> List[str]: + """List all tabular file blobs under a given prefix.""" + client = self._get_blob_service_client() + container_client = client.get_container_client(container_name) + blobs = [] + for blob in container_client.list_blobs(name_starts_with=prefix): + name_lower = blob['name'].lower() + if any(name_lower.endswith(ext) for ext in self.SUPPORTED_EXTENSIONS): + blobs.append(blob['name']) + return blobs + + def _read_tabular_blob_to_dataframe(self, container_name: str, blob_name: str) -> pandas.DataFrame: + """Download a blob and read it into a pandas DataFrame. Uses per-instance cache.""" + cache_key = (container_name, blob_name) + if cache_key in self._df_cache: + log_event(f"[TabularProcessingPlugin] Cache hit for {blob_name}", level=logging.DEBUG) + return self._df_cache[cache_key].copy() + + client = self._get_blob_service_client() + blob_client = client.get_blob_client(container=container_name, blob=blob_name) + stream = blob_client.download_blob() + data = stream.readall() + + name_lower = blob_name.lower() + if name_lower.endswith('.csv'): + df = pandas.read_csv(io.BytesIO(data), keep_default_na=False, dtype=str) + elif name_lower.endswith('.xlsx') or name_lower.endswith('.xlsm'): + df = pandas.read_excel(io.BytesIO(data), engine='openpyxl', keep_default_na=False, dtype=str) + elif name_lower.endswith('.xls'): + df = pandas.read_excel(io.BytesIO(data), engine='xlrd', keep_default_na=False, dtype=str) + else: + raise ValueError(f"Unsupported tabular file type: {blob_name}") + + self._df_cache[cache_key] = df + log_event(f"[TabularProcessingPlugin] Cached DataFrame for {blob_name} ({len(df)} rows)", level=logging.DEBUG) + return df.copy() + + def _try_numeric_conversion(self, df: pandas.DataFrame) -> pandas.DataFrame: + """Attempt to convert string columns to numeric where possible.""" + for col in df.columns: + try: + df[col] = pandas.to_numeric(df[col]) + except (ValueError, TypeError): + pass + return df + + def _resolve_blob_location(self, user_id: str, conversation_id: str, filename: str, source: str, + group_id: str = None, public_workspace_id: str = None) -> tuple: + """Resolve container name and blob path from source type.""" + source = source.lower().strip() + if source == 'chat': + container = storage_account_personal_chat_container_name + blob_path = f"{user_id}/{conversation_id}/{filename}" + elif source == 'workspace': + container = storage_account_user_documents_container_name + blob_path = f"{user_id}/{filename}" + elif source == 'group': + if not group_id: + raise ValueError("group_id is required for source='group'") + container = storage_account_group_documents_container_name + blob_path = f"{group_id}/{filename}" + elif source == 'public': + if not public_workspace_id: + raise ValueError("public_workspace_id is required for source='public'") + container = storage_account_public_documents_container_name + blob_path = f"{public_workspace_id}/{filename}" + else: + raise ValueError(f"Unknown source '{source}'. Use 'workspace', 'chat', 'group', or 'public'.") + return container, blob_path + + def _resolve_blob_location_with_fallback(self, user_id: str, conversation_id: str, filename: str, source: str, + group_id: str = None, public_workspace_id: str = None) -> tuple: + """Try primary source first, then fall back to other containers if blob not found.""" + source = source.lower().strip() + attempts = [] + + # Primary attempt based on specified source + try: + primary = self._resolve_blob_location(user_id, conversation_id, filename, source, group_id, public_workspace_id) + attempts.append(primary) + except ValueError: + pass + + # Fallback attempts in priority order (skip the primary source) + if source != 'workspace': + attempts.append((storage_account_user_documents_container_name, f"{user_id}/{filename}")) + if source != 'group' and group_id: + attempts.append((storage_account_group_documents_container_name, f"{group_id}/{filename}")) + if source != 'public' and public_workspace_id: + attempts.append((storage_account_public_documents_container_name, f"{public_workspace_id}/{filename}")) + if source != 'chat': + attempts.append((storage_account_personal_chat_container_name, f"{user_id}/{conversation_id}/{filename}")) + + client = self._get_blob_service_client() + for container, blob_path in attempts: + try: + blob_client = client.get_blob_client(container=container, blob=blob_path) + if blob_client.exists(): + log_event(f"[TabularProcessingPlugin] Found blob at {container}/{blob_path}", level=logging.DEBUG) + return container, blob_path + except Exception: + continue + + # If nothing found, return primary for the original error message + if attempts: + return attempts[0] + raise ValueError(f"Could not resolve blob location for {filename}") + + @kernel_function( + description=( + "List all tabular data files available for a user. Checks workspace documents " + "(user-documents container), chat-uploaded documents (personal-chat container), " + "and optionally group or public workspace documents. " + "Returns a JSON list of available files with their source." + ), + name="list_tabular_files" + ) + @plugin_function_logger("TabularProcessingPlugin") + async def list_tabular_files( + self, + user_id: Annotated[str, "The user ID (from Scope ID in Conversation Metadata)"], + conversation_id: Annotated[str, "The conversation ID (from Conversation Metadata)"], + group_id: Annotated[Optional[str], "Group ID (for group workspace documents)"] = None, + public_workspace_id: Annotated[Optional[str], "Public workspace ID (for public workspace documents)"] = None, + ) -> Annotated[str, "JSON list of available tabular files"]: + """List all tabular files available for the user across all accessible containers.""" + def _sync_work(): + results = [] + try: + workspace_prefix = f"{user_id}/" + workspace_blobs = self._list_tabular_blobs( + storage_account_user_documents_container_name, workspace_prefix + ) + for blob in workspace_blobs: + filename = blob.split('/')[-1] + results.append({ + "filename": filename, + "blob_path": blob, + "source": "workspace", + "container": storage_account_user_documents_container_name + }) + except Exception as e: + log_event(f"[TabularProcessingPlugin] Error listing workspace blobs: {e}", level=logging.WARNING) + + try: + chat_prefix = f"{user_id}/{conversation_id}/" + chat_blobs = self._list_tabular_blobs( + storage_account_personal_chat_container_name, chat_prefix + ) + for blob in chat_blobs: + filename = blob.split('/')[-1] + results.append({ + "filename": filename, + "blob_path": blob, + "source": "chat", + "container": storage_account_personal_chat_container_name + }) + except Exception as e: + log_event(f"[TabularProcessingPlugin] Error listing chat blobs: {e}", level=logging.WARNING) + + if group_id: + try: + group_prefix = f"{group_id}/" + group_blobs = self._list_tabular_blobs( + storage_account_group_documents_container_name, group_prefix + ) + for blob in group_blobs: + filename = blob.split('/')[-1] + results.append({ + "filename": filename, + "blob_path": blob, + "source": "group", + "container": storage_account_group_documents_container_name + }) + except Exception as e: + log_event(f"[TabularProcessingPlugin] Error listing group blobs: {e}", level=logging.WARNING) + + if public_workspace_id: + try: + public_prefix = f"{public_workspace_id}/" + public_blobs = self._list_tabular_blobs( + storage_account_public_documents_container_name, public_prefix + ) + for blob in public_blobs: + filename = blob.split('/')[-1] + results.append({ + "filename": filename, + "blob_path": blob, + "source": "public", + "container": storage_account_public_documents_container_name + }) + except Exception as e: + log_event(f"[TabularProcessingPlugin] Error listing public blobs: {e}", level=logging.WARNING) + + return json.dumps(results, indent=2) + return await asyncio.to_thread(_sync_work) + + @kernel_function( + description=( + "Get a summary of a tabular file including column names, row count, data types, " + "and a preview of the first few rows." + ), + name="describe_tabular_file" + ) + @plugin_function_logger("TabularProcessingPlugin") + async def describe_tabular_file( + self, + user_id: Annotated[str, "The user ID (from Scope ID in Conversation Metadata)"], + conversation_id: Annotated[str, "The conversation ID (from Conversation Metadata)"], + filename: Annotated[str, "The filename of the tabular file"], + source: Annotated[str, "Source: 'workspace', 'chat', 'group', or 'public'"] = "chat", + group_id: Annotated[Optional[str], "Group ID (for group workspace documents)"] = None, + public_workspace_id: Annotated[Optional[str], "Public workspace ID (for public workspace documents)"] = None, + ) -> Annotated[str, "JSON summary of the tabular file"]: + """Get schema and preview of a tabular file.""" + def _sync_work(): + try: + container, blob_path = self._resolve_blob_location( + user_id, conversation_id, filename, source, + group_id=group_id, public_workspace_id=public_workspace_id + ) + df = self._read_tabular_blob_to_dataframe(container, blob_path) + df_numeric = self._try_numeric_conversion(df.copy()) + + summary = { + "filename": filename, + "row_count": len(df), + "column_count": len(df.columns), + "columns": list(df.columns), + "dtypes": {col: str(dtype) for col, dtype in df_numeric.dtypes.items()}, + "preview": df.head(5).to_dict(orient='records'), + "null_counts": df.isnull().sum().to_dict() + } + return json.dumps(summary, indent=2, default=str) + except Exception as e: + log_event(f"[TabularProcessingPlugin] Error describing file: {e}", level=logging.WARNING) + return json.dumps({"error": str(e)}) + return await asyncio.to_thread(_sync_work) + + @kernel_function( + description=( + "Execute an aggregation operation on a column of a tabular file. " + "Supported operations: sum, mean, count, min, max, median, std, nunique, value_counts." + ), + name="aggregate_column" + ) + @plugin_function_logger("TabularProcessingPlugin") + async def aggregate_column( + self, + user_id: Annotated[str, "The user ID (from Scope ID in Conversation Metadata)"], + conversation_id: Annotated[str, "The conversation ID (from Conversation Metadata)"], + filename: Annotated[str, "The filename of the tabular file"], + column: Annotated[str, "The column name to aggregate"], + operation: Annotated[str, "Aggregation: sum, mean, count, min, max, median, std, nunique, value_counts"], + source: Annotated[str, "Source: 'workspace', 'chat', 'group', or 'public'"] = "chat", + group_id: Annotated[Optional[str], "Group ID (for group workspace documents)"] = None, + public_workspace_id: Annotated[Optional[str], "Public workspace ID (for public workspace documents)"] = None, + ) -> Annotated[str, "JSON result of the aggregation"]: + """Execute an aggregation operation on a column.""" + def _sync_work(): + try: + container, blob_path = self._resolve_blob_location( + user_id, conversation_id, filename, source, + group_id=group_id, public_workspace_id=public_workspace_id + ) + df = self._read_tabular_blob_to_dataframe(container, blob_path) + df = self._try_numeric_conversion(df) + + if column not in df.columns: + return json.dumps({"error": f"Column '{column}' not found. Available: {list(df.columns)}"}) + + series = df[column] + op = operation.lower().strip() + + if op == 'sum': + result = series.sum() + elif op == 'mean': + result = series.mean() + elif op == 'count': + result = series.count() + elif op == 'min': + result = series.min() + elif op == 'max': + result = series.max() + elif op == 'median': + result = series.median() + elif op == 'std': + result = series.std() + elif op == 'nunique': + result = series.nunique() + elif op == 'value_counts': + result = series.value_counts().to_dict() + else: + return json.dumps({"error": f"Unsupported operation: {operation}. Use sum, mean, count, min, max, median, std, nunique, value_counts."}) + + return json.dumps({"column": column, "operation": op, "result": result}, indent=2, default=str) + except Exception as e: + log_event(f"[TabularProcessingPlugin] Error aggregating column: {e}", level=logging.WARNING) + return json.dumps({"error": str(e)}) + return await asyncio.to_thread(_sync_work) + + @kernel_function( + description=( + "Filter rows in a tabular file based on conditions and return matching rows. " + "Supports operators: ==, !=, >, <, >=, <=, contains, startswith, endswith." + ), + name="filter_rows" + ) + @plugin_function_logger("TabularProcessingPlugin") + async def filter_rows( + self, + user_id: Annotated[str, "The user ID (from Scope ID in Conversation Metadata)"], + conversation_id: Annotated[str, "The conversation ID (from Conversation Metadata)"], + filename: Annotated[str, "The filename of the tabular file"], + column: Annotated[str, "The column to filter on"], + operator: Annotated[str, "Operator: ==, !=, >, <, >=, <=, contains, startswith, endswith"], + value: Annotated[str, "The value to compare against"], + source: Annotated[str, "Source: 'workspace', 'chat', 'group', or 'public'"] = "chat", + max_rows: Annotated[str, "Maximum rows to return"] = "100", + group_id: Annotated[Optional[str], "Group ID (for group workspace documents)"] = None, + public_workspace_id: Annotated[Optional[str], "Public workspace ID (for public workspace documents)"] = None, + ) -> Annotated[str, "JSON list of matching rows"]: + """Filter rows based on a condition.""" + def _sync_work(): + try: + container, blob_path = self._resolve_blob_location( + user_id, conversation_id, filename, source, + group_id=group_id, public_workspace_id=public_workspace_id + ) + df = self._read_tabular_blob_to_dataframe(container, blob_path) + df = self._try_numeric_conversion(df) + + if column not in df.columns: + return json.dumps({"error": f"Column '{column}' not found. Available: {list(df.columns)}"}) + + series = df[column] + op = operator.strip().lower() + + numeric_value = None + try: + numeric_value = float(value) + except (ValueError, TypeError): + pass + + if op == '==' or op == 'equals': + if numeric_value is not None and pandas.api.types.is_numeric_dtype(series): + mask = series == numeric_value + else: + mask = series.astype(str).str.lower() == value.lower() + elif op == '!=': + if numeric_value is not None and pandas.api.types.is_numeric_dtype(series): + mask = series != numeric_value + else: + mask = series.astype(str).str.lower() != value.lower() + elif op == '>': + mask = series > numeric_value + elif op == '<': + mask = series < numeric_value + elif op == '>=': + mask = series >= numeric_value + elif op == '<=': + mask = series <= numeric_value + elif op == 'contains': + mask = series.astype(str).str.contains(value, case=False, na=False) + elif op == 'startswith': + mask = series.astype(str).str.lower().str.startswith(value.lower()) + elif op == 'endswith': + mask = series.astype(str).str.lower().str.endswith(value.lower()) + else: + return json.dumps({"error": f"Unsupported operator: {operator}"}) + + limit = int(max_rows) + filtered = df[mask].head(limit) + return json.dumps({ + "total_matches": int(mask.sum()), + "returned_rows": len(filtered), + "data": filtered.to_dict(orient='records') + }, indent=2, default=str) + except Exception as e: + log_event(f"[TabularProcessingPlugin] Error filtering rows: {e}", level=logging.WARNING) + return json.dumps({"error": str(e)}) + return await asyncio.to_thread(_sync_work) + + @kernel_function( + description=( + "Execute a pandas query expression against a tabular file for advanced analysis. " + "The query string uses pandas DataFrame.query() syntax. " + "Examples: 'Age > 30 and State == \"CA\"', 'Price < 100'" + ), + name="query_tabular_data" + ) + @plugin_function_logger("TabularProcessingPlugin") + async def query_tabular_data( + self, + user_id: Annotated[str, "The user ID (from Scope ID in Conversation Metadata)"], + conversation_id: Annotated[str, "The conversation ID (from Conversation Metadata)"], + filename: Annotated[str, "The filename of the tabular file"], + query_expression: Annotated[str, "Pandas query expression (e.g. 'Age > 30 and State == \"CA\"')"], + source: Annotated[str, "Source: 'workspace', 'chat', 'group', or 'public'"] = "chat", + max_rows: Annotated[str, "Maximum rows to return"] = "100", + group_id: Annotated[Optional[str], "Group ID (for group workspace documents)"] = None, + public_workspace_id: Annotated[Optional[str], "Public workspace ID (for public workspace documents)"] = None, + ) -> Annotated[str, "JSON result of the query"]: + """Execute a pandas query expression against a tabular file.""" + def _sync_work(): + try: + container, blob_path = self._resolve_blob_location( + user_id, conversation_id, filename, source, + group_id=group_id, public_workspace_id=public_workspace_id + ) + df = self._read_tabular_blob_to_dataframe(container, blob_path) + df = self._try_numeric_conversion(df) + + result_df = df.query(query_expression) + limit = int(max_rows) + return json.dumps({ + "total_matches": len(result_df), + "returned_rows": min(len(result_df), limit), + "data": result_df.head(limit).to_dict(orient='records') + }, indent=2, default=str) + except Exception as e: + log_event(f"[TabularProcessingPlugin] Error querying data: {e}", level=logging.WARNING) + return json.dumps({"error": f"Query error: {str(e)}. Ensure column names and values are correct."}) + return await asyncio.to_thread(_sync_work) + + @kernel_function( + description=( + "Perform a group-by aggregation on a tabular file. " + "Groups data by one column and aggregates another column. " + "Supported operations: sum, mean, count, min, max." + ), + name="group_by_aggregate" + ) + @plugin_function_logger("TabularProcessingPlugin") + async def group_by_aggregate( + self, + user_id: Annotated[str, "The user ID (from Scope ID in Conversation Metadata)"], + conversation_id: Annotated[str, "The conversation ID (from Conversation Metadata)"], + filename: Annotated[str, "The filename of the tabular file"], + group_by_column: Annotated[str, "The column to group by"], + aggregate_column: Annotated[str, "The column to aggregate"], + operation: Annotated[str, "Aggregation operation: sum, mean, count, min, max"], + source: Annotated[str, "Source: 'workspace', 'chat', 'group', or 'public'"] = "chat", + group_id: Annotated[Optional[str], "Group ID (for group workspace documents)"] = None, + public_workspace_id: Annotated[Optional[str], "Public workspace ID (for public workspace documents)"] = None, + ) -> Annotated[str, "JSON result of the group-by aggregation"]: + """Group by one column and aggregate another.""" + def _sync_work(): + try: + container, blob_path = self._resolve_blob_location( + user_id, conversation_id, filename, source, + group_id=group_id, public_workspace_id=public_workspace_id + ) + df = self._read_tabular_blob_to_dataframe(container, blob_path) + df = self._try_numeric_conversion(df) + + for col in [group_by_column, aggregate_column]: + if col not in df.columns: + return json.dumps({"error": f"Column '{col}' not found. Available: {list(df.columns)}"}) + + op = operation.lower().strip() + grouped = df.groupby(group_by_column)[aggregate_column].agg(op) + return json.dumps({ + "group_by": group_by_column, + "aggregate_column": aggregate_column, + "operation": op, + "groups": len(grouped), + "result": grouped.to_dict() + }, indent=2, default=str) + except Exception as e: + log_event(f"[TabularProcessingPlugin] Error in group-by: {e}", level=logging.WARNING) + return json.dumps({"error": str(e)}) + return await asyncio.to_thread(_sync_work) diff --git a/application/single_app/static/css/chats.css b/application/single_app/static/css/chats.css index 38e11c3a..6a64bbc1 100644 --- a/application/single_app/static/css/chats.css +++ b/application/single_app/static/css/chats.css @@ -1676,4 +1676,160 @@ mark.search-highlight { 100% { transform: scale(1.05); } +} + +/* ============================================= + Processing Thoughts + ============================================= */ + +/* Loading indicator thought text */ +.thought-live-text { + font-style: italic; + white-space: nowrap; + overflow: hidden; + text-overflow: ellipsis; + max-width: 300px; +} + +/* Toggle button in message footer */ +.thoughts-toggle-btn { + font-size: 0.9rem; + color: #6c757d; + padding: 0 0.25rem; + border: none; + background: none; + cursor: pointer; + transition: color 0.15s ease-in-out; +} + +.thoughts-toggle-btn:hover { + color: #ffc107; +} + +/* Collapsible container inside message bubble */ +.thoughts-container { + max-height: 300px; + overflow-y: auto; + font-size: 0.85rem; +} + +/* Timeline wrapper */ +.thoughts-list { + position: relative; + padding-left: 1.25rem; +} + +/* Vertical timeline line */ +.thoughts-list::before { + content: ''; + position: absolute; + left: 0.5rem; + top: 0.25rem; + bottom: 0.25rem; + width: 2px; + background: linear-gradient(to bottom, #0d6efd, #6ea8fe); + border-radius: 1px; +} + +/* Individual thought step */ +.thought-step { + display: flex; + align-items: flex-start; + padding-left: 0.75rem; + padding-top: 0.25rem; + padding-bottom: 0.25rem; + position: relative; +} + +/* Timeline node dot */ +.thought-step::before { + content: ''; + position: absolute; + left: -1rem; + top: 0.55rem; + width: 8px; + height: 8px; + border-radius: 50%; + background-color: #0d6efd; + border: 2px solid #fff; + box-shadow: 0 0 0 1px #0d6efd; + z-index: 1; +} + +/* Last thought step gets a slightly different dot */ +.thought-step:last-child::before { + background-color: #198754; + box-shadow: 0 0 0 1px #198754; +} + +.thought-step i { + flex-shrink: 0; + margin-top: 2px; +} + +/* Streaming cursor thought badge pulse animation */ +.animate-pulse { + animation: thought-pulse 1.5s ease-in-out infinite; +} + +/* Streaming thought display (before content arrives) */ +.streaming-thought-display { + display: flex; + align-items: center; + padding: 0.5rem 0; +} + +/* Light mode: use darker, more readable colors */ +.streaming-thought-display .badge { + background-color: rgba(13, 110, 253, 0.08) !important; + color: #0a58ca !important; + border-color: rgba(13, 110, 253, 0.25) !important; +} + +/* Dark mode: lighter accent colors */ +[data-bs-theme="dark"] .streaming-thought-display .badge { + background-color: rgba(13, 202, 240, 0.15) !important; + color: #6edff6 !important; + border-color: rgba(13, 202, 240, 0.3) !important; +} + +@keyframes thought-pulse { + 0%, 100% { + opacity: 1; + } + 50% { + opacity: 0.6; + } +} + +/* Dark mode overrides */ +[data-bs-theme="dark"] .thoughts-toggle-btn { + color: #adb5bd; +} + +[data-bs-theme="dark"] .thoughts-toggle-btn:hover { + color: #ffc107; +} + +[data-bs-theme="dark"] .thought-step { + /* Dark mode dot border matches dark background */ +} + +[data-bs-theme="dark"] .thought-step::before { + border-color: #212529; + background-color: #6ea8fe; + box-shadow: 0 0 0 1px #6ea8fe; +} + +[data-bs-theme="dark"] .thought-step:last-child::before { + background-color: #75b798; + box-shadow: 0 0 0 1px #75b798; +} + +[data-bs-theme="dark"] .thoughts-list::before { + background: linear-gradient(to bottom, #6ea8fe, #9ec5fe); +} + +[data-bs-theme="dark"] .thoughts-container { + border-top-color: #495057 !important; } \ No newline at end of file diff --git a/application/single_app/static/css/styles.css b/application/single_app/static/css/styles.css index e537590d..eacc8859 100644 --- a/application/single_app/static/css/styles.css +++ b/application/single_app/static/css/styles.css @@ -502,6 +502,95 @@ main { flex-grow: 1; } +/* ============================================ + Item cards (agents/actions grid view) + ============================================ */ +.item-card { + cursor: default; + transition: all 0.3s ease; + border: 1px solid #dee2e6; + border-radius: 0.375rem; + background-color: #ffffff; +} + +.item-card:hover { + border-color: #adb5bd; + transform: translateY(-2px); + box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1); +} + +.item-card .card-title { + font-weight: 600; + font-size: 0.9rem; + color: #212529; +} + +.item-card .card-text { + color: #6c757d; + font-size: 0.8rem; + line-height: 1.4; +} + +.item-card .item-card-icon { + color: #0d6efd; +} + +.item-card .item-card-buttons { + border-top: 1px solid #f0f0f0; + padding-top: 0.5rem; +} + +/* Dark mode for item cards */ +[data-bs-theme="dark"] .item-card { + background-color: #343a40; + border: 1px solid #495057; + color: #e9ecef; +} + +[data-bs-theme="dark"] .item-card:hover { + background-color: #3d444b; + border-color: #6c757d; +} + +[data-bs-theme="dark"] .item-card .card-title { + color: #e9ecef; +} + +[data-bs-theme="dark"] .item-card .card-text { + color: #adb5bd; +} + +[data-bs-theme="dark"] .item-card .item-card-icon { + color: #6ea8fe; +} + +[data-bs-theme="dark"] .item-card .item-card-buttons { + border-top-color: #495057; +} + +/* Improved table column layout for agents and actions */ +.item-list-table th:nth-child(1), +.item-list-table td:nth-child(1) { + width: 28%; + min-width: 140px; +} + +.item-list-table th:nth-child(2), +.item-list-table td:nth-child(2) { + width: 47%; + max-width: 0; + overflow: hidden; + text-overflow: ellipsis; + white-space: nowrap; +} + +.item-list-table th:nth-child(3), +.item-list-table td:nth-child(3) { + width: 25%; + min-width: 160px; + white-space: nowrap; +} + /* Connection type buttons */ .connection-type-btn { border: 2px solid #dee2e6; @@ -854,3 +943,171 @@ main { [data-bs-theme="dark"] .message-content a:visited { color: #b399ff !important; /* Purple-ish for visited links */ } + +/* ============================================ + Rendered Markdown — table & code block styles + Shared by agent detail view, template preview, + and any non-chat area that renders Markdown. + ============================================ */ + +/* --- Tables --- */ +.rendered-markdown table { + width: 100%; + max-width: 100%; + margin: 0.75rem 0; + border-collapse: collapse; + border-spacing: 0; + border: 1px solid #dee2e6; + border-radius: 0.375rem; + overflow: hidden; + background-color: var(--bs-body-bg); + box-shadow: 0 0.125rem 0.25rem rgba(0, 0, 0, 0.075); + font-size: 0.875rem; + display: block; + overflow-x: auto; + white-space: nowrap; + -webkit-overflow-scrolling: touch; +} + +@media (min-width: 768px) { + .rendered-markdown table { + display: table; + white-space: normal; + } +} + +.rendered-markdown table th, +.rendered-markdown table td { + padding: 0.5rem 0.75rem; + border-bottom: 1px solid #dee2e6; + border-right: 1px solid #dee2e6; + text-align: left; + vertical-align: top; + word-wrap: break-word; + line-height: 1.4; +} + +.rendered-markdown table th:last-child, +.rendered-markdown table td:last-child { + border-right: none; +} + +.rendered-markdown table thead th { + background-color: #f8f9fa; + font-weight: 600; + color: #495057; + border-bottom: 2px solid #dee2e6; +} + +.rendered-markdown table tbody tr:nth-child(even) { + background-color: rgba(0, 0, 0, 0.02); +} + +.rendered-markdown table tbody tr:hover { + background-color: rgba(0, 0, 0, 0.04); + transition: background-color 0.15s ease-in-out; +} + +.rendered-markdown table th[align="center"], +.rendered-markdown table td[align="center"] { + text-align: center; +} + +.rendered-markdown table th[align="right"], +.rendered-markdown table td[align="right"] { + text-align: right; +} + +/* Dark mode tables */ +[data-bs-theme="dark"] .rendered-markdown table { + border-color: #495057; + background-color: var(--bs-dark); + color: #e9ecef; +} + +[data-bs-theme="dark"] .rendered-markdown table th, +[data-bs-theme="dark"] .rendered-markdown table td { + border-color: #495057; +} + +[data-bs-theme="dark"] .rendered-markdown table thead th { + background-color: #343a40; + color: #e9ecef; + border-bottom-color: #495057; +} + +[data-bs-theme="dark"] .rendered-markdown table tbody tr:nth-child(even) { + background-color: rgba(255, 255, 255, 0.05); +} + +[data-bs-theme="dark"] .rendered-markdown table tbody tr:hover { + background-color: rgba(255, 255, 255, 0.1); +} + +.rendered-markdown table code { + background-color: rgba(0, 0, 0, 0.1); + padding: 0.125rem 0.25rem; + border-radius: 0.25rem; + font-size: 0.8em; +} + +[data-bs-theme="dark"] .rendered-markdown table code { + background-color: rgba(255, 255, 255, 0.1); +} + +/* --- Code blocks --- */ +.rendered-markdown pre, +.rendered-markdown pre[class*="language-"] { + overflow-x: auto; + max-width: 100%; + width: 100%; + box-sizing: border-box; + display: block; + white-space: pre; + background-color: #1e1e1e; + color: #d4d4d4; + border-radius: 0.375rem; + padding: 1rem; + margin: 0.75rem 0; + font-size: 0.85rem; + line-height: 1.5; +} + +.rendered-markdown pre code { + display: block; + min-width: 0; + max-width: 100%; + overflow-x: auto; + white-space: pre; + background: transparent; + color: inherit; + padding: 0; + font-size: inherit; +} + +/* Inline code */ +.rendered-markdown code:not(pre code) { + background-color: rgba(0, 0, 0, 0.06); + padding: 0.15rem 0.35rem; + border-radius: 0.25rem; + font-size: 0.85em; + color: #d63384; +} + +[data-bs-theme="dark"] .rendered-markdown code:not(pre code) { + background-color: rgba(255, 255, 255, 0.1); + color: #e685b5; +} + +/* Blockquotes */ +.rendered-markdown blockquote { + border-left: 4px solid #dee2e6; + padding-left: 1em; + color: #6c757d; + margin: 0.75rem 0; +} + +[data-bs-theme="dark"] .rendered-markdown blockquote { + border-left-color: #495057; + color: #adb5bd; +} diff --git a/application/single_app/static/images/custom_logo.png b/application/single_app/static/images/custom_logo.png new file mode 100644 index 00000000..ecf6e652 Binary files /dev/null and b/application/single_app/static/images/custom_logo.png differ diff --git a/application/single_app/static/images/custom_logo_dark.png b/application/single_app/static/images/custom_logo_dark.png new file mode 100644 index 00000000..4f281945 Binary files /dev/null and b/application/single_app/static/images/custom_logo_dark.png differ diff --git a/application/single_app/static/js/admin/admin_settings.js b/application/single_app/static/js/admin/admin_settings.js index 85719128..9cf4580d 100644 --- a/application/single_app/static/js/admin/admin_settings.js +++ b/application/single_app/static/js/admin/admin_settings.js @@ -1237,10 +1237,11 @@ function setupToggles() { const mathToggle = document.getElementById('toggle-math-plugin'); const textToggle = document.getElementById('toggle-text-plugin'); const factMemoryToggle = document.getElementById('toggle-fact-memory-plugin'); + const tabularProcessingToggle = document.getElementById('toggle-tabular-processing-plugin'); const embeddingToggle = document.getElementById('toggle-default-embedding-model-plugin'); const allowUserPluginsToggle = document.getElementById('toggle-allow-user-plugins'); const allowGroupPluginsToggle = document.getElementById('toggle-allow-group-plugins'); - const toggles = [timeToggle, httpToggle, waitToggle, mathToggle, textToggle, factMemoryToggle, embeddingToggle, allowUserPluginsToggle, allowGroupPluginsToggle]; + const toggles = [timeToggle, httpToggle, waitToggle, mathToggle, textToggle, factMemoryToggle, tabularProcessingToggle, embeddingToggle, allowUserPluginsToggle, allowGroupPluginsToggle]; // Feedback area let feedbackDiv = document.getElementById('core-plugin-toggles-feedback'); if (!feedbackDiv) { @@ -1270,6 +1271,16 @@ function setupToggles() { if (textToggle) textToggle.checked = !!settings.enable_text_plugin; if (embeddingToggle) embeddingToggle.checked = !!settings.enable_default_embedding_model_plugin; if (factMemoryToggle) factMemoryToggle.checked = !!settings.enable_fact_memory_plugin; + if (tabularProcessingToggle) { + tabularProcessingToggle.checked = !!settings.enable_tabular_processing_plugin; + const ecEnabled = !!settings.enable_enhanced_citations; + tabularProcessingToggle.disabled = !ecEnabled; + const depNote = document.getElementById('tabular-processing-dependency-note'); + if (depNote) { + depNote.textContent = ecEnabled ? 'Requires Enhanced Citations' : 'Requires Enhanced Citations (currently disabled)'; + depNote.className = ecEnabled ? 'text-muted d-block ms-4' : 'text-danger d-block ms-4'; + } + } if (allowUserPluginsToggle) allowUserPluginsToggle.checked = !!settings.allow_user_plugins; if (allowGroupPluginsToggle) allowGroupPluginsToggle.checked = !!settings.allow_group_plugins; } catch (err) { @@ -1291,6 +1302,7 @@ function setupToggles() { enable_text_plugin: textToggle ? textToggle.checked : false, enable_default_embedding_model_plugin: embeddingToggle ? embeddingToggle.checked : false, enable_fact_memory_plugin: factMemoryToggle ? factMemoryToggle.checked : false, + enable_tabular_processing_plugin: tabularProcessingToggle ? tabularProcessingToggle.checked : false, allow_user_plugins: allowUserPluginsToggle ? allowUserPluginsToggle.checked : false, allow_group_plugins: allowGroupPluginsToggle ? allowGroupPluginsToggle.checked : false }; @@ -3827,11 +3839,12 @@ function checkOptionalFeaturesEnabled(stepNumber) { return endpoint && key; } - case 11: // User feedback and archiving - // Check if feedback is enabled + case 11: // User feedback, archiving, and thoughts + // Check if feedback, archiving, or thoughts is enabled const feedbackEnabled = document.getElementById('enable_user_feedback')?.checked; const archivingEnabled = document.getElementById('enable_conversation_archiving')?.checked; - return feedbackEnabled || archivingEnabled; + const thoughtsEnabled = document.getElementById('enable_thoughts')?.checked; + return feedbackEnabled || archivingEnabled || thoughtsEnabled; case 12: // Enhanced citations and image generation // Check if enhanced citations or image generation is enabled diff --git a/application/single_app/static/js/chat/chat-enhanced-citations.js b/application/single_app/static/js/chat/chat-enhanced-citations.js index dcda708b..9d4344bb 100644 --- a/application/single_app/static/js/chat/chat-enhanced-citations.js +++ b/application/single_app/static/js/chat/chat-enhanced-citations.js @@ -18,11 +18,13 @@ export function getFileType(fileName) { const imageExtensions = ['jpg', 'jpeg', 'png', 'bmp', 'tiff', 'tif']; const videoExtensions = ['mp4', 'mov', 'avi', 'mkv', 'flv', 'webm', 'wmv', 'm4v', '3gp']; const audioExtensions = ['mp3', 'wav', 'ogg', 'aac', 'flac', 'm4a']; - + const tabularExtensions = ['csv', 'xlsx', 'xls', 'xlsm']; + if (imageExtensions.includes(ext)) return 'image'; if (ext === 'pdf') return 'pdf'; if (videoExtensions.includes(ext)) return 'video'; if (audioExtensions.includes(ext)) return 'audio'; + if (tabularExtensions.includes(ext)) return 'tabular'; return 'other'; } @@ -66,6 +68,9 @@ export function showEnhancedCitationModal(docId, pageNumberOrTimestamp, citation const audioTimestamp = convertTimestampToSeconds(pageNumberOrTimestamp); showAudioModal(docId, audioTimestamp, docMetadata.file_name); break; + case 'tabular': + showTabularDownloadModal(docId, docMetadata.file_name); + break; default: // Fall back to text citation for unsupported types import('./chat-citations.js').then(module => { @@ -291,6 +296,119 @@ export function showAudioModal(docId, timestamp, fileName) { modalInstance.show(); } +/** + * Show tabular file preview modal with data table + * @param {string} docId - Document ID + * @param {string} fileName - File name + */ +export function showTabularDownloadModal(docId, fileName) { + console.log(`Showing tabular preview modal for docId: ${docId}, fileName: ${fileName}`); + showLoadingIndicator(); + + // Create or get tabular modal + let tabularModal = document.getElementById("enhanced-tabular-modal"); + if (!tabularModal) { + tabularModal = createTabularModal(); + } + + const title = tabularModal.querySelector(".modal-title"); + const tableContainer = tabularModal.querySelector("#enhanced-tabular-table-container"); + const rowInfo = tabularModal.querySelector("#enhanced-tabular-row-info"); + const downloadBtn = tabularModal.querySelector("#enhanced-tabular-download"); + const errorContainer = tabularModal.querySelector("#enhanced-tabular-error"); + + title.textContent = `Tabular Data: ${fileName}`; + tableContainer.innerHTML = '

Loading data preview...

'; + rowInfo.textContent = ''; + errorContainer.classList.add('d-none'); + + const downloadUrl = `/api/enhanced_citations/tabular_workspace?doc_id=${encodeURIComponent(docId)}`; + downloadBtn.href = downloadUrl; + downloadBtn.download = fileName; + + // Show modal immediately with loading state + const modalInstance = new bootstrap.Modal(tabularModal); + modalInstance.show(); + + // Fetch preview data + const previewUrl = `/api/enhanced_citations/tabular_preview?doc_id=${encodeURIComponent(docId)}`; + fetch(previewUrl) + .then(response => { + if (!response.ok) throw new Error(`HTTP ${response.status}`); + return response.json(); + }) + .then(data => { + hideLoadingIndicator(); + if (data.error) { + showTabularError(tableContainer, errorContainer, data.error); + return; + } + renderTabularPreview(tableContainer, rowInfo, data); + }) + .catch(error => { + hideLoadingIndicator(); + console.error('Error loading tabular preview:', error); + showTabularError(tableContainer, errorContainer, 'Could not load data preview.'); + }); +} + +/** + * Render tabular data as an HTML table + * @param {HTMLElement} container - Table container element + * @param {HTMLElement} rowInfo - Row info display element + * @param {Object} data - Preview data from API + */ +function renderTabularPreview(container, rowInfo, data) { + const { columns, rows, total_rows, truncated } = data; + + // Build table HTML + let html = ''; + + // Header + html += ''; + for (const col of columns) { + const escaped = col.replace(/&/g, '&').replace(//g, '>'); + html += ``; + } + html += ''; + + // Body + html += ''; + for (const row of rows) { + html += ''; + for (const cell of row) { + const val = cell === null || cell === undefined ? '' : String(cell); + const escaped = val.replace(/&/g, '&').replace(//g, '>'); + html += ``; + } + html += ''; + } + html += '
${escaped}
${escaped}
'; + + container.innerHTML = html; + + // Row info + const displayedRows = rows.length; + const totalFormatted = total_rows.toLocaleString(); + if (truncated) { + rowInfo.textContent = `Showing ${displayedRows.toLocaleString()} of ${totalFormatted} rows`; + } else { + rowInfo.textContent = `${totalFormatted} rows, ${columns.length} columns`; + } +} + +/** + * Show error state in tabular modal with download fallback + * @param {HTMLElement} tableContainer - Table container element + * @param {HTMLElement} errorContainer - Error display element + * @param {string} message - Error message + */ +function showTabularError(tableContainer, errorContainer, message) { + tableContainer.innerHTML = '
'; + errorContainer.textContent = message + ' You can still download the file below.'; + errorContainer.classList.remove('d-none'); +} + /** * Convert timestamp string to seconds * @param {string|number} timestamp - Timestamp in various formats @@ -445,3 +563,36 @@ function createPdfModal() { document.body.appendChild(modal); return modal; } + +/** + * Create tabular file preview modal HTML structure + * @returns {HTMLElement} - Modal element + */ +function createTabularModal() { + const modal = document.createElement("div"); + modal.id = "enhanced-tabular-modal"; + modal.classList.add("modal", "fade"); + modal.tabIndex = -1; + modal.innerHTML = ` +
+
+
+
Tabular Data Citation
+ +
+
+
+
+
+
+ + + Download File + +
+
+
+ `; + document.body.appendChild(modal); + return modal; +} diff --git a/application/single_app/static/js/chat/chat-input-actions.js b/application/single_app/static/js/chat/chat-input-actions.js index 77851319..c0c7832b 100644 --- a/application/single_app/static/js/chat/chat-input-actions.js +++ b/application/single_app/static/js/chat/chat-input-actions.js @@ -127,11 +127,11 @@ export function fetchFileContent(conversationId, fileId) { hideLoadingIndicator(); if (data.file_content && data.filename) { - showFileContentPopup(data.file_content, data.filename, data.is_table); + showFileContentPopup(data.file_content, data.filename, data.is_table, data.file_content_source, conversationId, fileId); } else if (data.error) { showToast(data.error, "danger"); } else { - ashowToastlert("Unexpected response from server.", "danger"); + showToast("Unexpected response from server.", "danger"); } }) .catch((error) => { @@ -141,7 +141,7 @@ export function fetchFileContent(conversationId, fileId) { }); } -export function showFileContentPopup(fileContent, filename, isTable) { +export function showFileContentPopup(fileContent, filename, isTable, fileContentSource, conversationId, fileId) { let modalContainer = document.getElementById("file-modal"); if (!modalContainer) { modalContainer = document.createElement("div"); @@ -155,6 +155,7 @@ export function showFileContentPopup(fileContent, filename, isTable) {
Uploaded File: ${filename}
+
`; @@ -760,6 +763,7 @@ export function appendMessage(
${senderLabel}
${mainMessageHtml} ${citationContentContainerHtml} + ${thoughtsHtml.containerHtml} ${metadataContainerHtml} ${footerContentHtml} @@ -816,6 +820,9 @@ export function appendMessage( } }); } + + // Attach thoughts toggle listener + attachThoughtsToggleListener(messageDiv, messageId, currentConversationId); const maskBtn = messageDiv.querySelector(".mask-btn"); if (maskBtn) { @@ -1516,6 +1523,7 @@ export function actuallySendMessage(finalMessageToSend) { } // Regular non-streaming fetch + startThoughtPolling(currentConversationId); fetch("/api/chat", { method: "POST", headers: { @@ -1547,6 +1555,7 @@ export function actuallySendMessage(finalMessageToSend) { }) .then((data) => { // Only successful responses reach here + stopThoughtPolling(); hideLoadingIndicatorInChatbox(); console.log("--- Data received from /api/chat ---"); @@ -1688,6 +1697,7 @@ export function actuallySendMessage(finalMessageToSend) { } }) .catch((error) => { + stopThoughtPolling(); hideLoadingIndicatorInChatbox(); console.error("Error sending message:", error); diff --git a/application/single_app/static/js/chat/chat-streaming.js b/application/single_app/static/js/chat/chat-streaming.js index faf6f59e..d2b5b218 100644 --- a/application/single_app/static/js/chat/chat-streaming.js +++ b/application/single_app/static/js/chat/chat-streaming.js @@ -5,6 +5,7 @@ import { loadUserSettings, saveUserSetting } from './chat-layout.js'; import { showToast } from './chat-toast.js'; import { updateSidebarConversationTitle } from './chat-sidebar-conversations.js'; import { applyScopeLock } from './chat-documents.js'; +import { handleStreamingThought } from './chat-thoughts.js'; let streamingEnabled = false; let currentEventSource = null; @@ -207,8 +208,11 @@ export function sendMessageWithStreaming(messageData, tempUserMessageId, current handleStreamError(tempAiMessageId, data.partial_content || accumulatedContent, data.error); return; } - - if (data.content) { + + if (data.type === 'thought') { + handleStreamingThought(data); + // Continue reading — don't fall through to content handling + } else if (data.content) { // Append chunk to accumulated content accumulatedContent += data.content; updateStreamingMessage(tempAiMessageId, accumulatedContent); diff --git a/application/single_app/static/js/chat/chat-thoughts.js b/application/single_app/static/js/chat/chat-thoughts.js new file mode 100644 index 00000000..a780bd3f --- /dev/null +++ b/application/single_app/static/js/chat/chat-thoughts.js @@ -0,0 +1,215 @@ +// chat-thoughts.js + +import { updateLoadingIndicatorText } from './chat-loading-indicator.js'; +import { escapeHtml } from './chat-utils.js'; + +let thoughtPollingInterval = null; +let lastSeenThoughtIndex = -1; + +// --------------------------------------------------------------------------- +// Icon map: step_type → Bootstrap Icon class +// --------------------------------------------------------------------------- +function getThoughtIcon(stepType) { + const iconMap = { + 'search': 'bi-search', + 'tabular_analysis': 'bi-table', + 'web_search': 'bi-globe', + 'agent_tool_call': 'bi-robot', + 'generation': 'bi-lightning', + 'content_safety': 'bi-shield-check' + }; + return iconMap[stepType] || 'bi-stars'; +} + +// --------------------------------------------------------------------------- +// Polling (non-streaming mode) +// --------------------------------------------------------------------------- + +/** + * Start polling for pending thoughts while waiting for a non-streaming response. + * @param {string} conversationId - The current conversation ID. + */ +export function startThoughtPolling(conversationId) { + if (!conversationId) return; + if (!window.appSettings?.enable_thoughts) return; + + stopThoughtPolling(); // clear any previous interval + lastSeenThoughtIndex = -1; + + thoughtPollingInterval = setInterval(() => { + fetch(`/api/conversations/${conversationId}/thoughts/pending`, { + credentials: 'same-origin' + }) + .then(r => r.json()) + .then(data => { + if (data.thoughts && data.thoughts.length > 0) { + const latest = data.thoughts[data.thoughts.length - 1]; + if (latest.step_index > lastSeenThoughtIndex) { + lastSeenThoughtIndex = latest.step_index; + const icon = getThoughtIcon(latest.step_type); + updateLoadingIndicatorText(latest.content, icon); + } + } + }) + .catch(() => { /* ignore polling errors */ }); + }, 2000); +} + +/** + * Stop the thought polling interval. + */ +export function stopThoughtPolling() { + if (thoughtPollingInterval) { + clearInterval(thoughtPollingInterval); + thoughtPollingInterval = null; + } + lastSeenThoughtIndex = -1; +} + +// --------------------------------------------------------------------------- +// Streaming handler +// --------------------------------------------------------------------------- + +/** + * Handle a streaming thought event received via SSE. + * Updates the streaming message placeholder with a styled thought indicator. + * When actual content starts streaming, updateStreamingMessage() will overwrite this. + * @param {object} thoughtData - { step_index, step_type, content } + */ +export function handleStreamingThought(thoughtData) { + // Find the streaming message's content area + const messageElement = document.querySelector('[data-message-id^="temp_ai_"]'); + if (!messageElement) return; + + const contentElement = messageElement.querySelector('.message-text'); + if (!contentElement) return; + + const icon = getThoughtIcon(thoughtData.step_type); + // Replace entire content with styled thought indicator (visually distinct from AI response) + contentElement.innerHTML = `
+ + ${escapeHtml(thoughtData.content)} + +
`; +} + +// --------------------------------------------------------------------------- +// Per-message collapsible: toggle button + container HTML +// --------------------------------------------------------------------------- + +/** + * Create HTML for the thoughts toggle button and hidden container. + * Returns an object with { toggleHtml, containerHtml }. + * @param {string} messageId + */ +export function createThoughtsToggleHtml(messageId) { + if (!window.appSettings?.enable_thoughts) { + return { toggleHtml: '', containerHtml: '' }; + } + + const containerId = `thoughts-${messageId || Date.now()}`; + const toggleHtml = ``; + const containerHtml = `
Loading thoughts...
`; + + return { toggleHtml, containerHtml }; +} + +/** + * Attach event listener for the thoughts toggle button inside a message div. + * @param {HTMLElement} messageDiv + * @param {string} messageId + * @param {string} conversationId + */ +export function attachThoughtsToggleListener(messageDiv, messageId, conversationId) { + const toggleBtn = messageDiv.querySelector('.thoughts-toggle-btn'); + if (!toggleBtn) return; + + toggleBtn.addEventListener('click', () => { + const targetId = toggleBtn.getAttribute('aria-controls'); + const container = messageDiv.querySelector(`#${targetId}`); + if (!container) return; + + // Store scroll position + const scrollContainer = document.getElementById('chat-messages-container'); + const currentScroll = scrollContainer?.scrollTop || window.pageYOffset; + + const isExpanded = !container.classList.contains('d-none'); + if (isExpanded) { + container.classList.add('d-none'); + toggleBtn.setAttribute('aria-expanded', 'false'); + toggleBtn.title = 'Show processing thoughts'; + toggleBtn.innerHTML = ''; + } else { + container.classList.remove('d-none'); + toggleBtn.setAttribute('aria-expanded', 'true'); + toggleBtn.title = 'Hide processing thoughts'; + toggleBtn.innerHTML = ''; + + // Lazy-load thoughts on first expand + if (container.innerHTML.includes('Loading thoughts')) { + loadThoughtsForMessage(conversationId, messageId, container); + } + } + + // Restore scroll position + setTimeout(() => { + if (scrollContainer) { + scrollContainer.scrollTop = currentScroll; + } else { + window.scrollTo(0, currentScroll); + } + }, 10); + }); +} + +// --------------------------------------------------------------------------- +// Fetch + render thoughts for a message +// --------------------------------------------------------------------------- + +/** + * Fetch thoughts for a specific message from the API and render them. + * @param {string} conversationId + * @param {string} messageId + * @param {HTMLElement} container + */ +function loadThoughtsForMessage(conversationId, messageId, container) { + fetch(`/api/conversations/${conversationId}/messages/${messageId}/thoughts`, { + credentials: 'same-origin' + }) + .then(r => r.json()) + .then(data => { + if (!data.enabled) { + container.innerHTML = '
Processing thoughts are disabled.
'; + return; + } + if (!data.thoughts || data.thoughts.length === 0) { + container.innerHTML = '
No processing thoughts recorded for this message.
'; + return; + } + container.innerHTML = renderThoughtsList(data.thoughts); + }) + .catch(err => { + console.error('Error loading thoughts:', err); + container.innerHTML = '
Failed to load processing thoughts.
'; + }); +} + +/** + * Render a list of thought steps as HTML. + * @param {Array} thoughts + * @returns {string} HTML string + */ +function renderThoughtsList(thoughts) { + let html = '
'; + thoughts.forEach(t => { + const icon = getThoughtIcon(t.step_type); + const durationStr = t.duration_ms != null ? `(${t.duration_ms}ms)` : ''; + html += `
+ + ${escapeHtml(t.content || '')} + ${durationStr} +
`; + }); + html += '
'; + return html; +} diff --git a/application/single_app/static/js/plugin_common.js b/application/single_app/static/js/plugin_common.js index e40158b9..29a88a24 100644 --- a/application/single_app/static/js/plugin_common.js +++ b/application/single_app/static/js/plugin_common.js @@ -2,6 +2,10 @@ // Shared logic for admin_plugins.js and workspace_plugins.js // Exports: functions for modal field handling, validation, label toggling, table rendering, and plugin CRUD import { showToast } from "./chat/chat-toast.js" +import { + humanizeName, truncateDescription, + openViewModal, createActionCard +} from './workspace/view-utils.js'; // Fetch merged plugin settings from backend given type and current settings export async function fetchAndMergePluginSettings(pluginType, currentSettings = {}) { @@ -60,8 +64,7 @@ export function escapeHtml(str) { } // Render plugins table (parameterized for tbody selector and button handlers) -export function renderPluginsTable({plugins, tbodySelector, onEdit, onDelete, ensureTable = true, isAdmin = false}) { - console.log('Rendering plugins table with %d plugins', plugins.length); +export function renderPluginsTable({plugins, tbodySelector, onEdit, onDelete, onView, ensureTable = true, isAdmin = false}) { // Optionally ensure the table is present before rendering if (ensureTable) { ensurePluginsTableInRoot(); @@ -75,29 +78,33 @@ export function renderPluginsTable({plugins, tbodySelector, onEdit, onDelete, en plugins.forEach(plugin => { const tr = document.createElement('tr'); const safeName = escapeHtml(plugin.name); - const safeDisplayName = escapeHtml(plugin.display_name || plugin.name); - const safeDesc = escapeHtml(plugin.description || 'No description available'); + const displayName = humanizeName(plugin.display_name || plugin.name); + const safeDisplayName = escapeHtml(displayName); + const description = plugin.description || 'No description available'; + const truncatedDesc = escapeHtml(truncateDescription(description, 90)); let actionButtons = ''; let globalBadge = plugin.is_global ? ' Global' : ''; - // Show action buttons for: - // - Admin context: all actions (global and personal) - // - User context: only personal actions (not global) + // View button always shown + let viewButton = ``; + + // Edit/Delete buttons based on context + let editDeleteButtons = ''; if (isAdmin || !plugin.is_global) { - actionButtons = ` -
+ editDeleteButtons = ` -
- `; + `; } + actionButtons = `
${viewButton}${editDeleteButtons}
`; tr.innerHTML = ` - ${safeDisplayName}${globalBadge} - ${safeDesc} + ${safeDisplayName}${globalBadge} + ${truncatedDesc} ${actionButtons} `; tbody.appendChild(tr); @@ -109,6 +116,34 @@ export function renderPluginsTable({plugins, tbodySelector, onEdit, onDelete, en tbody.querySelectorAll('.delete-plugin-btn').forEach(btn => { btn.onclick = () => onDelete(btn.getAttribute('data-plugin-name')); }); + tbody.querySelectorAll('.view-plugin-btn').forEach(btn => { + btn.onclick = () => { + if (onView) { + onView(btn.getAttribute('data-plugin-name')); + } + }; + }); +} + +// Render plugins grid (card-based view) +export function renderPluginsGrid({plugins, containerSelector, onEdit, onDelete, onView, isAdmin = false}) { + const container = document.querySelector(containerSelector); + if (!container) return; + container.innerHTML = ''; + if (!plugins.length) { + container.innerHTML = '
No actions found.
'; + return; + } + plugins.forEach(plugin => { + const card = createActionCard(plugin, { + onView: (p) => { if (onView) onView(p.name); }, + onEdit: (p) => onEdit(p.name), + onDelete: (p) => onDelete(p.name), + canManage: isAdmin || !plugin.is_global, + isAdmin + }); + container.appendChild(card); + }); } // Toggle auth fields and labels (parameterized for DOM elements) diff --git a/application/single_app/static/js/plugin_modal_stepper.js b/application/single_app/static/js/plugin_modal_stepper.js index 89076076..aa5b4e01 100644 --- a/application/single_app/static/js/plugin_modal_stepper.js +++ b/application/single_app/static/js/plugin_modal_stepper.js @@ -1,6 +1,10 @@ // plugin_modal_stepper.js // Multi-step modal functionality for action/plugin creation import { showToast } from "./chat/chat-toast.js"; +import { getTypeIcon } from "./workspace/view-utils.js"; + +// Action types hidden from the creation UI (backend plugins remain intact) +const HIDDEN_ACTION_TYPES = ['sql_schema', 'ui_test', 'queue_storage', 'blob_storage', 'embedding_model']; export class PluginModalStepper { @@ -129,6 +133,12 @@ export class PluginModalStepper { document.getElementById('sql-auth-type').addEventListener('change', () => this.handleSqlAuthTypeChange()); + // Test SQL connection button + const testConnBtn = document.getElementById('sql-test-connection-btn'); + if (testConnBtn) { + testConnBtn.addEventListener('click', () => this.testSqlConnection()); + } + // Set up display name to generated name conversion this.setupNameGeneration(); @@ -193,6 +203,8 @@ export class PluginModalStepper { if (!res.ok) throw new Error('Failed to load action types'); this.availableTypes = await res.json(); + // Hide deprecated/internal action types from the creation UI + this.availableTypes = this.availableTypes.filter(t => !HIDDEN_ACTION_TYPES.includes(t.type)); // Sort action types alphabetically by display name this.availableTypes.sort((a, b) => { const nameA = (a.display || a.displayName || a.type || a.name || '').toLowerCase(); @@ -271,10 +283,15 @@ export class PluginModalStepper { description.substring(0, maxLength) + '...' : description; const needsTruncation = description.length > maxLength; + const iconClass = getTypeIcon(type.type || type.name); + col.innerHTML = `
-
${this.escapeHtml(displayName)}
+
+ +
${this.escapeHtml(displayName)}
+

${this.escapeHtml(truncatedDescription)} ${needsTruncation ? ` @@ -538,43 +555,52 @@ export class PluginModalStepper { } if (stepNumber === 4) { - // Load additional settings schema for selected type - let options = {forceReload: true}; - this.getAdditionalSettingsSchema(this.selectedType, options); + const isSqlType = this.selectedType === 'sql_query' || this.selectedType === 'sql_schema'; const additionalFieldsDiv = document.getElementById('plugin-additional-fields-div'); - if (additionalFieldsDiv) { - // Only clear and rebuild if type changes - if (this.selectedType !== this.lastAdditionalFieldsType) { - additionalFieldsDiv.innerHTML = ''; - additionalFieldsDiv.classList.remove('d-none'); - if (this.selectedType) { - this.getAdditionalSettingsSchema(this.selectedType) - .then(schema => { - if (schema) { - this.buildAdditionalFieldsUI(schema, additionalFieldsDiv); - try { - if (this.isEditMode && this.originalPlugin && this.originalPlugin.additionalFields) { - this.populateDynamicAdditionalFields(this.originalPlugin.additionalFields); + + // For SQL types, hide additional fields entirely since Step 3 covers all SQL config + if (isSqlType && additionalFieldsDiv) { + additionalFieldsDiv.innerHTML = ''; + additionalFieldsDiv.classList.add('d-none'); + this.lastAdditionalFieldsType = this.selectedType; + } else { + // Load additional settings schema for selected type + let options = {forceReload: true}; + this.getAdditionalSettingsSchema(this.selectedType, options); + if (additionalFieldsDiv) { + // Only clear and rebuild if type changes + if (this.selectedType !== this.lastAdditionalFieldsType) { + additionalFieldsDiv.innerHTML = ''; + additionalFieldsDiv.classList.remove('d-none'); + if (this.selectedType) { + this.getAdditionalSettingsSchema(this.selectedType) + .then(schema => { + if (schema) { + this.buildAdditionalFieldsUI(schema, additionalFieldsDiv); + try { + if (this.isEditMode && this.originalPlugin && this.originalPlugin.additionalFields) { + this.populateDynamicAdditionalFields(this.originalPlugin.additionalFields); + } + } catch (error) { + console.error('Error populating dynamic additional fields:', error); } - } catch (error) { - console.error('Error populating dynamic additional fields:', error); + } else { + console.log('No additional settings schema found'); + additionalFieldsDiv.classList.add('d-none'); } - } else { - console.log('No additional settings schema found'); + }) + .catch(error => { + console.error(`Error fetching additional settings schema for type: ${this.selectedType} -- ${error}`); additionalFieldsDiv.classList.add('d-none'); - } - }) - .catch(error => { - console.error(`Error fetching additional settings schema for type: ${this.selectedType} -- ${error}`); - additionalFieldsDiv.classList.add('d-none'); - }); - } else { - console.warn('No plugin type selected'); - additionalFieldsDiv.classList.add('d-none'); + }); + } else { + console.warn('No plugin type selected'); + additionalFieldsDiv.classList.add('d-none'); + } + this.lastAdditionalFieldsType = this.selectedType; } - this.lastAdditionalFieldsType = this.selectedType; + // Otherwise, preserve user data and do not redraw } - // Otherwise, preserve user data and do not redraw } if (!this.isEditMode) { @@ -1230,6 +1256,80 @@ export class PluginModalStepper { this.updateSqlAuthInfo(); } + async testSqlConnection() { + const btn = document.getElementById('sql-test-connection-btn'); + const resultDiv = document.getElementById('sql-test-connection-result'); + const alertDiv = document.getElementById('sql-test-connection-alert'); + if (!btn || !resultDiv || !alertDiv) return; + + // Collect current SQL config from Step 3 + const databaseType = document.querySelector('input[name="sql-database-type"]:checked')?.value; + const connectionMethod = document.querySelector('input[name="sql-connection-method"]:checked')?.value || 'parameters'; + const authType = document.getElementById('sql-auth-type')?.value || 'username_password'; + + if (!databaseType) { + resultDiv.classList.remove('d-none'); + alertDiv.className = 'alert alert-warning mb-0 py-2 px-3 small'; + alertDiv.textContent = 'Please select a database type first.'; + return; + } + + const payload = { + database_type: databaseType, + connection_method: connectionMethod, + auth_type: authType + }; + + if (connectionMethod === 'connection_string') { + payload.connection_string = document.getElementById('sql-connection-string')?.value?.trim() || ''; + } else { + payload.server = document.getElementById('sql-server')?.value?.trim() || ''; + payload.database = document.getElementById('sql-database')?.value?.trim() || ''; + payload.port = document.getElementById('sql-port')?.value?.trim() || ''; + if (databaseType === 'sqlserver' || databaseType === 'azure_sql') { + payload.driver = document.getElementById('sql-driver')?.value || ''; + } + } + + if (authType === 'username_password') { + payload.username = document.getElementById('sql-username')?.value?.trim() || ''; + payload.password = document.getElementById('sql-password')?.value?.trim() || ''; + } + + payload.timeout = parseInt(document.getElementById('sql-timeout')?.value) || 10; + + // Show loading state + const originalText = btn.innerHTML; + btn.innerHTML = 'Testing...'; + btn.disabled = true; + resultDiv.classList.add('d-none'); + + try { + const response = await fetch('/api/plugins/test-sql-connection', { + method: 'POST', + headers: { 'Content-Type': 'application/json' }, + body: JSON.stringify(payload) + }); + const data = await response.json(); + + resultDiv.classList.remove('d-none'); + if (data.success) { + alertDiv.className = 'alert alert-success mb-0 py-2 px-3 small'; + alertDiv.innerHTML = '' + (data.message || 'Connection successful!'); + } else { + alertDiv.className = 'alert alert-danger mb-0 py-2 px-3 small'; + alertDiv.innerHTML = '' + (data.error || 'Connection failed.'); + } + } catch (error) { + resultDiv.classList.remove('d-none'); + alertDiv.className = 'alert alert-danger mb-0 py-2 px-3 small'; + alertDiv.innerHTML = 'Test failed: ' + (error.message || 'Network error'); + } finally { + btn.innerHTML = originalText; + btn.disabled = false; + } + } + updateSqlConnectionExamples() { const selectedType = document.querySelector('input[name="sql-database-type"]:checked')?.value; const examplesDiv = document.getElementById('sql-connection-examples'); @@ -1720,12 +1820,17 @@ export class PluginModalStepper { // Collect additional fields from the dynamic UI and MERGE with existing additionalFields // This preserves OpenAPI spec content and other auto-populated fields - try { - const dynamicFields = this.collectAdditionalFields(); - // Merge dynamicFields into additionalFields (preserving existing values) - additionalFields = { ...additionalFields, ...dynamicFields }; - } catch (e) { - throw new Error('Invalid additional fields input'); + // For SQL types, Step 3 already provides all necessary config — skip dynamic field merge + // to prevent empty Step 4 fields from overwriting populated Step 3 values + const isSqlType = this.selectedType === 'sql_query' || this.selectedType === 'sql_schema'; + if (!isSqlType) { + try { + const dynamicFields = this.collectAdditionalFields(); + // Merge dynamicFields into additionalFields (preserving existing values) + additionalFields = { ...additionalFields, ...dynamicFields }; + } catch (e) { + throw new Error('Invalid additional fields input'); + } } let metadata = {}; @@ -2106,6 +2211,7 @@ export class PluginModalStepper { populateAdvancedSummary() { const advancedSection = document.getElementById('summary-advanced-section'); + const isSqlType = this.selectedType === 'sql_query' || this.selectedType === 'sql_schema'; // Check if there's any metadata or additional fields const metadata = document.getElementById('plugin-metadata').value.trim(); @@ -2123,9 +2229,33 @@ export class PluginModalStepper { hasMetadata = metadata.length > 0 && metadata !== '{}'; } - // DRY: Use private helper to collect additional fields - let additionalFieldsObj = this.collectAdditionalFields(); - hasAdditionalFields = Object.keys(additionalFieldsObj).length > 0; + // For SQL types, additional fields are already shown in the SQL Database Configuration + // summary section, so skip showing them again in Advanced to avoid redundancy + if (!isSqlType) { + // DRY: Use private helper to collect additional fields + let additionalFieldsObj = this.collectAdditionalFields(); + hasAdditionalFields = Object.keys(additionalFieldsObj).length > 0; + + // Show/hide additional fields preview + const additionalFieldsPreview = document.getElementById('summary-additional-fields-preview'); + if (hasAdditionalFields) { + let previewContent = ''; + if (typeof additionalFieldsObj === 'object' && additionalFieldsObj !== null) { + previewContent = JSON.stringify(additionalFieldsObj, null, 2); + } else { + previewContent = ''; + } + document.getElementById('summary-additional-fields-content').textContent = previewContent; + additionalFieldsPreview.style.display = ''; + } else { + additionalFieldsPreview.style.display = 'none'; + } + } else { + // Hide additional fields for SQL types + const additionalFieldsPreview = document.getElementById('summary-additional-fields-preview'); + if (additionalFieldsPreview) additionalFieldsPreview.style.display = 'none'; + hasAdditionalFields = false; + } // Update has metadata/additional fields indicators document.getElementById('summary-has-metadata').textContent = hasMetadata ? 'Yes' : 'No'; @@ -2140,21 +2270,6 @@ export class PluginModalStepper { metadataPreview.style.display = 'none'; } - // Show/hide additional fields preview - const additionalFieldsPreview = document.getElementById('summary-additional-fields-preview'); - if (hasAdditionalFields) { - let previewContent = ''; - if (typeof additionalFieldsObj === 'object' && additionalFieldsObj !== null) { - previewContent = JSON.stringify(additionalFieldsObj, null, 2); - } else { - previewContent = ''; - } - document.getElementById('summary-additional-fields-content').textContent = previewContent; - additionalFieldsPreview.style.display = ''; - } else { - additionalFieldsPreview.style.display = 'none'; - } - // Show advanced section if there's any advanced content if (hasMetadata || hasAdditionalFields) { advancedSection.style.display = ''; diff --git a/application/single_app/static/js/workspace/group_agents.js b/application/single_app/static/js/workspace/group_agents.js index f97dbd07..608f029e 100644 --- a/application/single_app/static/js/workspace/group_agents.js +++ b/application/single_app/static/js/workspace/group_agents.js @@ -4,16 +4,23 @@ import { showToast } from "../chat/chat-toast.js"; import * as agentsCommon from "../agents_common.js"; import { AgentModalStepper } from "../agent_modal_stepper.js"; +import { + humanizeName, truncateDescription, escapeHtml as escapeHtmlUtil, + setupViewToggle, switchViewContainers, openViewModal, createAgentCard +} from './view-utils.js'; const tableBody = document.getElementById("group-agents-table-body"); const errorContainer = document.getElementById("group-agents-error"); const searchInput = document.getElementById("group-agents-search"); const createButton = document.getElementById("create-group-agent-btn"); const permissionWarning = document.getElementById("group-agents-permission-warning"); +const agentsListView = document.getElementById("group-agents-list-view"); +const agentsGridView = document.getElementById("group-agents-grid-view"); let agents = []; let filteredAgents = []; let agentStepper = null; +let currentViewMode = 'list'; let currentContext = window.groupWorkspaceContext || { activeGroupId: null, activeGroupName: "", @@ -21,14 +28,7 @@ let currentContext = window.groupWorkspaceContext || { }; function escapeHtml(value) { - if (!value) return ""; - return value.replace(/[&<>"']/g, (char) => ({ - "&": "&", - "<": "<", - ">": ">", - '"': """, - "'": "'" - }[char] || char)); + return escapeHtmlUtil(value); } function canManageAgents() { @@ -46,6 +46,7 @@ function groupAllowsModifications() { } function truncateName(name, maxLength = 18) { + // Kept for backward compat; prefer humanizeName for display if (!name || name.length <= maxLength) return name || ""; return `${name.substring(0, maxLength)}…`; } @@ -114,29 +115,61 @@ function renderAgentsTable(list) { list.forEach((agent) => { const tr = document.createElement("tr"); - const displayName = truncateName(agent.display_name || agent.displayName || agent.name || ""); - const description = escapeHtml(agent.description || "No description available."); - - let actionsHtml = "—"; + const rawName = agent.display_name || agent.displayName || agent.name || ""; + const displayName = humanizeName(rawName); + const fullDesc = agent.description || "No description available."; + const shortDesc = truncateDescription(fullDesc, 90); + + let actionsHtml = ` + + `; if (canManage) { - actionsHtml = ` - - `; } tr.innerHTML = ` - ${escapeHtml(displayName)} - ${description} + ${escapeHtml(displayName)} + ${escapeHtml(shortDesc)} ${actionsHtml}`; tableBody.appendChild(tr); }); } +function renderAgentsGrid(list) { + if (!agentsGridView) return; + agentsGridView.innerHTML = ''; + + if (!list.length) { + agentsGridView.innerHTML = '

No group agents found.
'; + return; + } + + const canManage = canManageAgents() && groupAllowsModifications(); + list.forEach(agent => { + const col = createAgentCard(agent, { + onChat: a => chatWithGroupAgent(a.name || a), + onView: a => openGroupAgentViewModal(a), + onEdit: canManage ? a => { + const found = agents.find(x => x.id === (a.id || a.name || a) || x.name === (a.name || a)); + openAgentModal(found || null); + } : null, + onDelete: canManage ? a => deleteGroupAgent(a.id || a.name || a) : null + }); + agentsGridView.appendChild(col); + }); +} + function filterAgents(term) { if (!term) { filteredAgents = agents.slice(); @@ -149,6 +182,23 @@ function filterAgents(term) { }); } renderAgentsTable(filteredAgents); + renderAgentsGrid(filteredAgents); +} + +// Open the view modal for a group agent with Chat/Edit/Delete actions +function openGroupAgentViewModal(agent) { + const canManage = canManageAgents() && groupAllowsModifications(); + const callbacks = { + onChat: (a) => chatWithGroupAgent(a.name) + }; + if (canManage) { + callbacks.onEdit = (a) => { + const found = agents.find(x => x.id === a.id || x.name === a.name); + openAgentModal(found || a); + }; + callbacks.onDelete = (a) => deleteGroupAgent(a.id || a.name); + } + openViewModal(agent, 'agent', callbacks); } function overrideAgentStepper(stepper) { @@ -343,7 +393,57 @@ async function fetchGroupAgents() { } } +async function chatWithGroupAgent(agentName) { + try { + const agent = agents.find(a => a.name === agentName); + if (!agent) { + throw new Error("Agent not found"); + } + + const payloadData = { + selected_agent: { + name: agentName, + display_name: agent.display_name || agent.displayName || agentName, + is_global: !!agent.is_global, + is_group: true, + group_id: currentContext.activeGroupId, + group_name: currentContext.activeGroupName + } + }; + + const resp = await fetch("/api/user/settings/selected_agent", { + method: "POST", + headers: { "Content-Type": "application/json" }, + body: JSON.stringify(payloadData) + }); + + if (!resp.ok) { + throw new Error("Failed to select agent"); + } + + window.location.href = "/chats"; + } catch (err) { + console.error("Error selecting group agent for chat:", err); + showToast("Error selecting agent for chat. Please try again.", "danger"); + } +} + function handleTableClick(event) { + const viewBtn = event.target.closest(".view-group-agent-btn"); + if (viewBtn) { + const agentName = viewBtn.dataset.agentName; + const agent = agents.find(a => a.name === agentName); + if (agent) openGroupAgentViewModal(agent); + return; + } + + const chatBtn = event.target.closest(".chat-group-agent-btn"); + if (chatBtn) { + const agentName = chatBtn.dataset.agentName; + chatWithGroupAgent(agentName); + return; + } + const editBtn = event.target.closest(".edit-group-agent-btn"); if (editBtn) { const agentId = editBtn.dataset.agentId; @@ -384,6 +484,11 @@ function initialize() { updatePermissionUI(); bindEventHandlers(); + setupViewToggle('groupAgents', 'groupAgentsViewPreference', (mode) => { + currentViewMode = mode; + switchViewContainers(mode, agentsListView, agentsGridView); + }); + if (document.getElementById("group-agents-tab-btn")?.classList.contains("active")) { fetchGroupAgents(); } diff --git a/application/single_app/static/js/workspace/group_plugins.js b/application/single_app/static/js/workspace/group_plugins.js index 60a7f42e..8acdf5bd 100644 --- a/application/single_app/static/js/workspace/group_plugins.js +++ b/application/single_app/static/js/workspace/group_plugins.js @@ -3,6 +3,10 @@ import { ensurePluginsTableInRoot, validatePluginManifest } from "../plugin_common.js"; import { showToast } from "../chat/chat-toast.js"; +import { + humanizeName, truncateDescription, escapeHtml as escapeHtmlUtil, + setupViewToggle, switchViewContainers, openViewModal, createActionCard +} from './view-utils.js'; const root = document.getElementById("group-plugins-root"); const permissionWarning = document.getElementById("group-plugins-permission-warning"); @@ -11,6 +15,7 @@ let plugins = []; let filteredPlugins = []; let templateReady = false; let listenersBound = false; +let currentViewMode = 'list'; let currentContext = window.groupWorkspaceContext || { activeGroupId: null, activeGroupName: "", @@ -18,14 +23,7 @@ let currentContext = window.groupWorkspaceContext || { }; function escapeHtml(value) { - if (!value) return ""; - return value.replace(/[&<>"']/g, (char) => ({ - "&": "&", - "<": "<", - ">": ">", - '"': """, - "'": "'" - }[char] || char)); + return escapeHtmlUtil(value); } function canManagePlugins() { @@ -66,6 +64,14 @@ function bindRootEvents() { }); root.addEventListener("click", async (event) => { + const viewBtn = event.target.closest(".view-group-plugin-btn"); + if (viewBtn) { + const pluginId = viewBtn.dataset.pluginId; + const plugin = plugins.find(x => x.id === pluginId || x.name === pluginId); + if (plugin) openGroupPluginViewModal(plugin); + return; + } + const createBtn = event.target.closest("#create-group-plugin-btn"); if (createBtn) { event.preventDefault(); @@ -148,23 +154,28 @@ function renderPluginsTable(list) { const canManage = canManagePlugins() && groupAllowsModifications(); list.forEach((plugin) => { const tr = document.createElement("tr"); - const displayName = plugin.displayName || plugin.display_name || plugin.name || ""; - const description = plugin.description || "No description available."; + const rawName = plugin.displayName || plugin.display_name || plugin.name || ""; + const displayName = humanizeName(rawName); + const fullDesc = plugin.description || "No description available."; + const shortDesc = truncateDescription(fullDesc, 90); const isGlobal = Boolean(plugin.is_global); - let actionsHtml = "—"; + // View button always visible + let actionsHtml = ` + `; + if (canManage && !isGlobal) { - actionsHtml = ` -
- - -
`; + actionsHtml += ` + + `; } else if (canManage && isGlobal) { - actionsHtml = "Managed globally"; + actionsHtml += `Managed globally`; } const titleHtml = isGlobal @@ -172,14 +183,36 @@ function renderPluginsTable(list) { : escapeHtml(displayName); tr.innerHTML = ` - ${titleHtml} - ${escapeHtml(description)} + ${titleHtml} + ${escapeHtml(shortDesc)} ${actionsHtml}`; tbody.appendChild(tr); }); } +function renderPluginsGrid(list) { + const gridView = document.getElementById('group-plugins-grid-view'); + if (!gridView) return; + gridView.innerHTML = ''; + + if (!list.length) { + gridView.innerHTML = '
No group actions found.
'; + return; + } + + const canManage = canManagePlugins() && groupAllowsModifications(); + list.forEach(plugin => { + const isGlobal = Boolean(plugin.is_global); + const col = createActionCard(plugin, { + onView: p => openGroupPluginViewModal(p), + onEdit: (canManage && !isGlobal) ? p => openPluginModal(p.id || p.name) : null, + onDelete: (canManage && !isGlobal) ? p => deleteGroupPlugin(p.id || p.name) : null + }); + gridView.appendChild(col); + }); +} + function filterPlugins(term) { if (!term) { filteredPlugins = plugins.slice(); @@ -192,6 +225,19 @@ function filterPlugins(term) { }); } renderPluginsTable(filteredPlugins); + renderPluginsGrid(filteredPlugins); +} + +// Open the view modal for a group action with Edit/Delete actions +function openGroupPluginViewModal(plugin) { + const canManage = canManagePlugins() && groupAllowsModifications(); + const isGlobal = Boolean(plugin.is_global); + const callbacks = {}; + if (canManage && !isGlobal) { + callbacks.onEdit = (p) => openPluginModal(p.id || p.name); + callbacks.onDelete = (p) => deleteGroupPlugin(p.id || p.name); + } + openViewModal(plugin, 'action', callbacks); } async function fetchGroupPlugins() { @@ -220,7 +266,17 @@ async function fetchGroupPlugins() { filteredPlugins = plugins.slice(); renderPluginsTable(filteredPlugins); + renderPluginsGrid(filteredPlugins); updatePermissionUI(); + + // Set up view toggle (only once after template is in DOM) + setupViewToggle('groupPlugins', 'groupPluginsViewPreference', (mode) => { + currentViewMode = mode; + switchViewContainers(mode, + document.getElementById('group-plugins-list-view'), + document.getElementById('group-plugins-grid-view') + ); + }); } catch (error) { console.error("Error loading group actions:", error); renderError(error.message || "Unable to load group actions."); diff --git a/application/single_app/static/js/workspace/view-utils.js b/application/single_app/static/js/workspace/view-utils.js new file mode 100644 index 00000000..3b78bc15 --- /dev/null +++ b/application/single_app/static/js/workspace/view-utils.js @@ -0,0 +1,523 @@ +// view-utils.js +// Shared utilities for list/grid view toggle, name humanization, and view modal +// Used by personal and group agents/actions workspace modules + +/** + * Convert a technical name to a human-readable display name. + * Handles underscores, camelCase, PascalCase, and consecutive uppercase. + * Examples: + * "sql_query" → "Sql Query" + * "myAgentName" → "My Agent Name" + * "OpenAPIPlugin" → "Open API Plugin" + * "log_analytics" → "Log Analytics" + */ +export function humanizeName(name) { + if (!name) return ""; + // Replace underscores and hyphens with spaces + let result = name.replace(/[_-]/g, " "); + // Insert space before uppercase letters that follow lowercase letters (camelCase) + result = result.replace(/([a-z])([A-Z])/g, "$1 $2"); + // Insert space between consecutive uppercase followed by lowercase (e.g., "APIPlugin" → "API Plugin") + result = result.replace(/([A-Z]+)([A-Z][a-z])/g, "$1 $2"); + // Capitalize first letter of each word + result = result.replace(/\b\w/g, (c) => c.toUpperCase()); + // Collapse multiple spaces + result = result.replace(/\s+/g, " ").trim(); + return result; +} + +/** + * Truncate a description string to maxLen characters, appending "…" if truncated. + */ +export function truncateDescription(text, maxLen = 100) { + if (!text) return ""; + if (text.length <= maxLen) return text; + return text.substring(0, maxLen).trimEnd() + "…"; +} + +/** + * Escape HTML entities to prevent XSS. + */ +export function escapeHtml(str) { + if (!str) return ""; + return str.replace(/[&<>"']/g, (c) => + ({ "&": "&", "<": "<", ">": ">", '"': """, "'": "'" }[c]) + ); +} + +/** + * Get an appropriate Bootstrap icon class for an action/plugin type. + */ +export function getTypeIcon(type) { + if (!type) return "bi-lightning-charge"; + const t = type.toLowerCase(); + if (t.includes("sql")) return "bi-database"; + if (t.includes("openapi")) return "bi-globe"; + if (t.includes("log_analytics")) return "bi-graph-up"; + if (t.includes("msgraph")) return "bi-microsoft"; + if (t.includes("databricks")) return "bi-bricks"; + if (t.includes("http") || t.includes("smart_http")) return "bi-cloud-arrow-up"; + if (t.includes("azure_function")) return "bi-lightning"; + if (t.includes("blob")) return "bi-file-earmark"; + if (t.includes("queue")) return "bi-inbox"; + if (t.includes("embedding")) return "bi-vector-pen"; + if (t.includes("fact_memory")) return "bi-brain"; + if (t.includes("math")) return "bi-calculator"; + if (t.includes("text")) return "bi-fonts"; + if (t.includes("time")) return "bi-clock"; + return "bi-lightning-charge"; +} + +/** + * Create the HTML string for a list/grid view toggle button group. + * @param {string} prefix - Unique prefix for element IDs (e.g., "agents", "plugins", "group-agents") + * @returns {string} HTML string + */ +export function createViewToggleHtml(prefix) { + return ` +
+ + + + +
`; +} + +/** + * Set up view toggle event listeners and restore saved preference. + * @param {string} prefix - Unique prefix matching createViewToggleHtml + * @param {string} storageKey - localStorage key for persistence + * @param {function} onSwitch - Callback receiving 'list' or 'grid' + */ +export function setupViewToggle(prefix, storageKey, onSwitch) { + const listRadio = document.getElementById(`${prefix}-view-list`); + const gridRadio = document.getElementById(`${prefix}-view-grid`); + if (!listRadio || !gridRadio) return; + + listRadio.addEventListener("change", () => { + if (listRadio.checked) { + localStorage.setItem(storageKey, "list"); + onSwitch("list"); + } + }); + + gridRadio.addEventListener("change", () => { + if (gridRadio.checked) { + localStorage.setItem(storageKey, "grid"); + onSwitch("grid"); + } + }); + + // Restore saved preference + const saved = localStorage.getItem(storageKey); + if (saved === "grid") { + gridRadio.checked = true; + listRadio.checked = false; + onSwitch("grid"); + } else { + onSwitch("list"); + } +} + +/** + * Toggle visibility of list and grid containers. + * @param {string} mode - 'list' or 'grid' + * @param {HTMLElement} listContainer - The list/table container element + * @param {HTMLElement} gridContainer - The grid container element + */ +export function switchViewContainers(mode, listContainer, gridContainer) { + if (listContainer) { + listContainer.classList.toggle("d-none", mode !== "list"); + } + if (gridContainer) { + gridContainer.classList.toggle("d-none", mode !== "grid"); + } +} + +// ============================================================================ +// VIEW MODAL — Lightweight read-only detail view +// ============================================================================ + +/** + * Open a read-only view modal for an agent or action. + * @param {object} item - The agent or action data object + * @param {'agent'|'action'} type - What kind of item this is + * @param {object} [callbacks] - Optional action callbacks { onChat, onEdit, onDelete } + */ +export function openViewModal(item, type, callbacks = {}) { + const modalEl = document.getElementById("item-view-modal"); + if (!modalEl) return; + + const titleEl = modalEl.querySelector(".modal-title"); + const bodyEl = modalEl.querySelector(".modal-body"); + const footerEl = modalEl.querySelector(".modal-footer"); + if (!titleEl || !bodyEl || !footerEl) return; + + if (type === "agent") { + titleEl.textContent = "Agent Details"; + bodyEl.innerHTML = buildAgentViewHtml(item); + } else { + titleEl.textContent = "Action Details"; + bodyEl.innerHTML = buildActionViewHtml(item); + } + + // Build footer buttons dynamically + footerEl.innerHTML = ''; + const { onChat, onEdit, onDelete } = callbacks; + + if (onChat && typeof onChat === 'function') { + const chatBtn = document.createElement('button'); + chatBtn.type = 'button'; + chatBtn.className = 'btn btn-primary'; + chatBtn.innerHTML = 'Chat'; + chatBtn.addEventListener('click', () => { + bootstrap.Modal.getInstance(modalEl)?.hide(); + onChat(item); + }); + footerEl.appendChild(chatBtn); + } + + if (onEdit && typeof onEdit === 'function') { + const editBtn = document.createElement('button'); + editBtn.type = 'button'; + editBtn.className = 'btn btn-outline-secondary'; + editBtn.innerHTML = 'Edit'; + editBtn.addEventListener('click', () => { + bootstrap.Modal.getInstance(modalEl)?.hide(); + onEdit(item); + }); + footerEl.appendChild(editBtn); + } + + if (onDelete && typeof onDelete === 'function') { + const delBtn = document.createElement('button'); + delBtn.type = 'button'; + delBtn.className = 'btn btn-outline-danger'; + delBtn.innerHTML = 'Delete'; + delBtn.addEventListener('click', () => { + bootstrap.Modal.getInstance(modalEl)?.hide(); + onDelete(item); + }); + footerEl.appendChild(delBtn); + } + + const closeBtn = document.createElement('button'); + closeBtn.type = 'button'; + closeBtn.className = 'btn btn-secondary'; + closeBtn.textContent = 'Close'; + closeBtn.setAttribute('data-bs-dismiss', 'modal'); + footerEl.appendChild(closeBtn); + + const modal = new bootstrap.Modal(modalEl); + modal.show(); +} + +function buildAgentViewHtml(agent) { + const displayName = escapeHtml(agent.display_name || agent.displayName || agent.name || ""); + const name = escapeHtml(agent.name || ""); + const description = escapeHtml(agent.description || "No description available."); + const model = escapeHtml(agent.azure_openai_gpt_deployment || agent.model || "Default"); + const agentType = agent.agent_type === "aifoundry" ? "Azure AI Foundry" : "Local (Semantic Kernel)"; + const rawInstructions = agent.instructions || "No instructions defined."; + // Render instructions as Markdown (marked + DOMPurify are loaded globally in base.html) + const renderedInstructions = (typeof marked !== 'undefined' && typeof DOMPurify !== 'undefined') + ? DOMPurify.sanitize(marked.parse(rawInstructions)) + : escapeHtml(rawInstructions); + const isGlobal = agent.is_global; + const scopeBadge = isGlobal + ? 'Global' + : 'Personal'; + + return ` +
+
+ Basic Information +
+
+
+
+ + ${displayName} +
+
+ + ${name} +
+
+ + ${scopeBadge} +
+
+ + ${escapeHtml(agentType)} +
+
+ + ${description} +
+
+
+
+
+
+ Model Configuration +
+
+
+
+ + ${model} +
+
+
+
+
+
+ Instructions +
+
+
+${renderedInstructions} +
+
+
`; +} + +function buildActionViewHtml(action) { + const displayName = escapeHtml(action.display_name || action.displayName || action.name || ""); + const name = escapeHtml(action.name || ""); + const description = escapeHtml(action.description || "No description available."); + const type = escapeHtml(action.type || "unknown"); + const typeIcon = getTypeIcon(action.type); + const authType = escapeHtml(formatAuthType(action.auth?.type || action.auth_type || "")); + const endpoint = escapeHtml(action.endpoint || action.base_url || ""); + const isGlobal = action.is_global; + const scopeBadge = isGlobal + ? 'Global' + : 'Personal'; + + let configHtml = ""; + if (endpoint) { + configHtml = ` +
+
+ Configuration +
+
+
+
+ + ${endpoint} +
+
+ + ${authType || "None"} +
+
+
+
`; + } + + return ` +
+
+ Basic Information +
+
+
+
+ + ${displayName} +
+
+ + ${name} +
+
+ + ${humanizeName(type)} +
+
+ + ${scopeBadge} +
+
+ + ${description} +
+
+
+
+ ${configHtml}`; +} + +function formatAuthType(type) { + if (!type) return ""; + const map = { + "key": "API Key", + "identity": "Managed Identity", + "user": "User (Delegated)", + "servicePrincipal": "Service Principal", + "connection_string": "Connection String", + "basic": "Basic Auth", + "username_password": "Username / Password", + "NoAuth": "No Authentication" + }; + return map[type] || type; +} + +// ============================================================================ +// GRID CARD RENDERERS +// ============================================================================ + +/** + * Create a grid card element for an agent. + * @param {object} agent - Agent data object + * @param {object} options - { onChat, onView, onEdit, onDelete, canManage, isGroup } + * @returns {HTMLElement} + */ +export function createAgentCard(agent, options = {}) { + const { onChat, onView, onEdit, onDelete, canManage = false, isGroup = false } = options; + const col = document.createElement("div"); + col.className = "col-sm-6 col-md-4 col-lg-3"; + + const displayName = humanizeName(agent.display_name || agent.displayName || agent.name || ""); + const description = agent.description || "No description available."; + const isGlobal = agent.is_global; + + let badgeHtml = ""; + if (isGlobal) { + badgeHtml = 'Global'; + } + + let buttonsHtml = ` + + `; + + if (canManage && !isGlobal) { + buttonsHtml += ` + + `; + } + + col.innerHTML = ` +
+
+
+ +
+
${escapeHtml(displayName)}${badgeHtml}
+

${escapeHtml(truncateDescription(description, 120))}

+
+ ${buttonsHtml} +
+
+
`; + + // Bind button events + const chatBtn = col.querySelector(".item-card-chat-btn"); + const viewBtn = col.querySelector(".item-card-view-btn"); + const editBtn = col.querySelector(".item-card-edit-btn"); + const deleteBtn = col.querySelector(".item-card-delete-btn"); + + if (chatBtn && onChat) chatBtn.addEventListener("click", (e) => { e.stopPropagation(); onChat(agent); }); + if (viewBtn && onView) viewBtn.addEventListener("click", (e) => { e.stopPropagation(); onView(agent); }); + if (editBtn && onEdit) editBtn.addEventListener("click", (e) => { e.stopPropagation(); onEdit(agent); }); + if (deleteBtn && onDelete) deleteBtn.addEventListener("click", (e) => { e.stopPropagation(); onDelete(agent); }); + + // Clicking anywhere on the card opens the detail view + const cardEl = col.querySelector(".item-card"); + if (cardEl && onView) { + cardEl.style.cursor = "pointer"; + cardEl.addEventListener("click", () => onView(agent)); + } + + return col; +} + +/** + * Create a grid card element for an action/plugin. + * @param {object} plugin - Action/plugin data object + * @param {object} options - { onView, onEdit, onDelete, canManage, isAdmin } + * @returns {HTMLElement} + */ +export function createActionCard(plugin, options = {}) { + const { onView, onEdit, onDelete, canManage = true, isAdmin = false } = options; + const col = document.createElement("div"); + col.className = "col-sm-6 col-md-4 col-lg-3"; + + const displayName = humanizeName(plugin.display_name || plugin.displayName || plugin.name || ""); + const description = plugin.description || "No description available."; + const type = plugin.type || ""; + const typeIcon = getTypeIcon(type); + const isGlobal = plugin.is_global; + + let badgeHtml = ""; + if (isGlobal) { + badgeHtml = 'Global'; + } + + const typeBadge = type + ? `${escapeHtml(humanizeName(type))}` + : ""; + + let buttonsHtml = ` + `; + + if ((isAdmin || (canManage && !isGlobal))) { + buttonsHtml += ` + + `; + } + + col.innerHTML = ` +
+
+
+ +
+
${escapeHtml(displayName)}${badgeHtml}
+
${typeBadge}
+

${escapeHtml(truncateDescription(description, 120))}

+
+ ${buttonsHtml} +
+
+
`; + + // Bind button events + const viewBtn = col.querySelector(".item-card-view-btn"); + const editBtn = col.querySelector(".item-card-edit-btn"); + const deleteBtn = col.querySelector(".item-card-delete-btn"); + + if (viewBtn && onView) viewBtn.addEventListener("click", (e) => { e.stopPropagation(); onView(plugin); }); + if (editBtn && onEdit) editBtn.addEventListener("click", (e) => { e.stopPropagation(); onEdit(plugin); }); + if (deleteBtn && onDelete) deleteBtn.addEventListener("click", (e) => { e.stopPropagation(); onDelete(plugin); }); + + // Clicking anywhere on the card opens the detail view + const cardEl = col.querySelector(".item-card"); + if (cardEl && onView) { + cardEl.style.cursor = "pointer"; + cardEl.addEventListener("click", () => onView(plugin)); + } + + return col; +} diff --git a/application/single_app/static/js/workspace/workspace_agents.js b/application/single_app/static/js/workspace/workspace_agents.js index a0839b25..623be234 100644 --- a/application/single_app/static/js/workspace/workspace_agents.js +++ b/application/single_app/static/js/workspace/workspace_agents.js @@ -4,14 +4,22 @@ import { showToast } from "../chat/chat-toast.js"; import * as agentsCommon from '../agents_common.js'; import { AgentModalStepper } from '../agent_modal_stepper.js'; +import { + humanizeName, truncateDescription, escapeHtml, + setupViewToggle, switchViewContainers, + openViewModal, createAgentCard +} from './view-utils.js'; // --- DOM Elements & Globals --- const agentsTbody = document.getElementById('agents-table-body'); const agentsErrorDiv = document.getElementById('workspace-agents-error'); const createAgentBtn = document.getElementById('create-agent-btn'); const agentsSearchInput = document.getElementById('agents-search'); +const agentsListView = document.getElementById('agents-list-view'); +const agentsGridView = document.getElementById('agents-grid-view'); let agents = []; let filteredAgents = []; +let currentViewMode = 'list'; // --- Function Definitions --- @@ -43,104 +51,87 @@ function filterAgents(searchTerm) { }); } renderAgentsTable(filteredAgents); + renderAgentsGrid(filteredAgents); } -// --- Helper Functions --- - -function truncateDisplayName(displayName, maxLength = 12) { - if (!displayName || displayName.length <= maxLength) { - return displayName; +// Open the view modal for an agent with Chat/Edit/Delete actions in the footer +function openAgentViewModal(agent) { + const callbacks = { + onChat: (a) => chatWithAgent(a.name), + onDelete: !agent.is_global ? (a) => { if (confirm(`Delete agent '${a.name}'?`)) deleteAgent(a.name); } : null + }; + if (!agent.is_global) { + callbacks.onEdit = (a) => openAgentModal(a); } - return displayName.substring(0, maxLength) + '...'; + openViewModal(agent, 'agent', callbacks); } +// --- Rendering Functions --- function renderAgentsTable(agentsList) { if (!agentsTbody) return; agentsTbody.innerHTML = ''; if (!agentsList.length) { const tr = document.createElement('tr'); - tr.innerHTML = 'No agents found.'; + tr.innerHTML = 'No agents found.'; agentsTbody.appendChild(tr); return; } - // Fetch selected_agent from user settings (async) - fetch('/api/user/settings').then(res => { - if (!res.ok) throw new Error('Failed to load user settings'); - return res.json(); - }).then(settings => { - let selectedAgentObj = settings.selected_agent; - if (!selectedAgentObj && settings.settings && settings.settings.selected_agent) { - selectedAgentObj = settings.settings.selected_agent; - } - let selectedAgentName = typeof selectedAgentObj === 'object' ? selectedAgentObj.name : selectedAgentObj; - agentsTbody.innerHTML = ''; - for (const agent of agentsList) { - const tr = document.createElement('tr'); - - // Create action buttons - let actionButtons = ``; - - if (!agent.is_global) { - actionButtons += ` - - - `; - } - - const truncatedDisplayName = truncateDisplayName(agent.display_name || agent.name || ''); - - tr.innerHTML = ` - - ${truncatedDisplayName} - ${agent.is_global ? ' Global' : ''} - - ${agent.description || 'No description available'} - ${actionButtons} - `; - agentsTbody.appendChild(tr); - } - }).catch(e => { - renderError('Could not load agent settings: ' + e.message); - // Fallback: render table without settings - agentsTbody.innerHTML = ''; - for (const agent of agentsList) { - const tr = document.createElement('tr'); - - // Create action buttons - let actionButtons = ` + `; - - if (!agent.is_global) { - actionButtons += ` - - - `; - } - - const truncatedDisplayName = truncateDisplayName(agent.display_name || agent.name || ''); - - tr.innerHTML = ` - - ${truncatedDisplayName} - ${agent.is_global ? ' Global' : ''} - - ${agent.description || 'No description available'} - ${actionButtons} - `; - agentsTbody.appendChild(tr); + + if (!isGlobal) { + actionButtons += ` + + `; } - }); + + tr.innerHTML = ` + + ${escapeHtml(displayName)} + ${isGlobal ? ' Global' : ''} + + ${escapeHtml(truncatedDesc)} + ${actionButtons} + `; + agentsTbody.appendChild(tr); + } +} + +function renderAgentsGrid(agentsList) { + if (!agentsGridView) return; + agentsGridView.innerHTML = ''; + if (!agentsList.length) { + agentsGridView.innerHTML = '
No agents found.
'; + return; + } + + for (const agent of agentsList) { + const card = createAgentCard(agent, { + onChat: (a) => chatWithAgent(a.name), + onView: (a) => openAgentViewModal(a), + onEdit: (a) => openAgentModal(a), + onDelete: (a) => { if (confirm(`Delete agent '${a.name}'?`)) deleteAgent(a.name); }, + canManage: !agent.is_global + }); + agentsGridView.appendChild(card); + } } async function fetchAgents() { @@ -151,6 +142,7 @@ async function fetchAgents() { agents = await res.json(); filteredAgents = agents; // Initialize filtered list renderAgentsTable(filteredAgents); + renderAgentsGrid(filteredAgents); } catch (e) { renderError(e.message); } @@ -177,17 +169,14 @@ function attachAgentTableEvents() { } agentsTbody.addEventListener('click', function (e) { - console.log('Agent table clicked, target:', e.target); - // Find the button element (could be the target or a parent) const editBtn = e.target.closest('.edit-agent-btn'); const deleteBtn = e.target.closest('.delete-agent-btn'); const chatBtn = e.target.closest('.chat-agent-btn'); + const viewBtn = e.target.closest('.view-agent-btn'); if (editBtn) { - console.log('Edit agent button clicked, dataset:', editBtn.dataset); const agent = agents.find(a => a.name === editBtn.dataset.name); - console.log('Found agent:', agent); openAgentModal(agent); } @@ -201,33 +190,27 @@ function attachAgentTableEvents() { const agentName = chatBtn.dataset.name; chatWithAgent(agentName); } + + if (viewBtn) { + const agent = agents.find(a => a.name === viewBtn.dataset.name); + if (agent) openAgentViewModal(agent); + } }); } async function chatWithAgent(agentName) { try { - console.log('DEBUG: chatWithAgent called with agentName:', agentName); - console.log('DEBUG: Available agents:', agents); - - // Find the agent to get its is_global status const agent = agents.find(a => a.name === agentName); - console.log('DEBUG: Found agent:', agent); - if (!agent) { throw new Error('Agent not found'); } - console.log('DEBUG: Agent is_global flag:', agent.is_global); - console.log('DEBUG: !!agent.is_global:', !!agent.is_global); - - // Set the selected agent with proper is_global flag const payloadData = { selected_agent: { name: agentName, is_global: !!agent.is_global } }; - console.log('DEBUG: Sending payload:', payloadData); const resp = await fetch('/api/user/settings/selected_agent', { method: 'POST', @@ -239,9 +222,6 @@ async function chatWithAgent(agentName) { throw new Error('Failed to select agent'); } - console.log('DEBUG: Agent selection saved successfully'); - - // Navigate to chat page window.location.href = '/chats'; } catch (err) { console.error('Error selecting agent for chat:', err); @@ -353,6 +333,17 @@ async function deleteAgent(name) { function initializeWorkspaceAgentUI() { window.agentModalStepper = new AgentModalStepper(false); attachAgentTableEvents(); + + // Set up view toggle + setupViewToggle('agents', 'agentsViewPreference', (mode) => { + currentViewMode = mode; + switchViewContainers(mode, agentsListView, agentsGridView); + // Re-render grid if switching to grid and we have data + if (mode === 'grid' && filteredAgents.length) { + renderAgentsGrid(filteredAgents); + } + }); + fetchAgents(); } diff --git a/application/single_app/static/js/workspace/workspace_plugins.js b/application/single_app/static/js/workspace/workspace_plugins.js index 30fef0d5..84f1eb46 100644 --- a/application/single_app/static/js/workspace/workspace_plugins.js +++ b/application/single_app/static/js/workspace/workspace_plugins.js @@ -1,10 +1,14 @@ // workspace_plugins.js (refactored to use plugin_common.js and new multi-step modal) -import { renderPluginsTable, ensurePluginsTableInRoot, validatePluginManifest } from '../plugin_common.js'; +import { renderPluginsTable, renderPluginsGrid, ensurePluginsTableInRoot, validatePluginManifest } from '../plugin_common.js'; import { showToast } from "../chat/chat-toast.js" +import { + setupViewToggle, switchViewContainers, openViewModal +} from './view-utils.js'; const root = document.getElementById('workspace-plugins-root'); let plugins = []; let filteredPlugins = []; +let currentViewMode = 'list'; function renderLoading() { root.innerHTML = `
`; @@ -14,6 +18,22 @@ function renderError(msg) { root.innerHTML = `
${msg}
`; } +function getViewHandlers() { + return { + onEdit: name => openPluginModal(plugins.find(p => p.name === name)), + onDelete: name => deletePlugin(name), + onView: name => { + const plugin = plugins.find(p => p.name === name); + if (plugin) { + openViewModal(plugin, 'action', { + onEdit: (item) => openPluginModal(item), + onDelete: (item) => deletePlugin(item.name) + }); + } + } + }; +} + function filterPlugins(searchTerm) { if (!searchTerm || !searchTerm.trim()) { filteredPlugins = plugins; @@ -26,14 +46,18 @@ function filterPlugins(searchTerm) { }); } - // Ensure table template is in place ensurePluginsTableInRoot(); + const handlers = getViewHandlers(); renderPluginsTable({ plugins: filteredPlugins, tbodySelector: '#plugins-table-body', - onEdit: name => openPluginModal(plugins.find(p => p.name === name)), - onDelete: name => deletePlugin(name) + ...handlers + }); + renderPluginsGrid({ + plugins: filteredPlugins, + containerSelector: '#plugins-grid-view', + ...handlers }); } @@ -47,12 +71,26 @@ async function fetchPlugins() { // Ensure table template is in place ensurePluginsTableInRoot(); + const handlers = getViewHandlers(); renderPluginsTable({ plugins: filteredPlugins, tbodySelector: '#plugins-table-body', - onEdit: name => openPluginModal(plugins.find(p => p.name === name)), - onDelete: name => deletePlugin(name) + ...handlers + }); + renderPluginsGrid({ + plugins: filteredPlugins, + containerSelector: '#plugins-grid-view', + ...handlers + }); + + // Set up view toggle (only once after template is in DOM) + setupViewToggle('plugins', 'pluginsViewPreference', (mode) => { + currentViewMode = mode; + switchViewContainers(mode, + document.getElementById('plugins-list-view'), + document.getElementById('plugins-grid-view') + ); }); // Set up the create action button diff --git a/application/single_app/static/json/schemas/sql_query.definition.json b/application/single_app/static/json/schemas/sql_query.definition.json index d38a41a8..6903c22a 100644 --- a/application/single_app/static/json/schemas/sql_query.definition.json +++ b/application/single_app/static/json/schemas/sql_query.definition.json @@ -1,6 +1,9 @@ { "$schema": "./plugin.definition.schema.json", "allowedAuthTypes": [ + "user", + "identity", + "servicePrincipal", "connection_string" ] } diff --git a/application/single_app/static/json/schemas/sql_query_plugin.additional_settings.schema.json b/application/single_app/static/json/schemas/sql_query_plugin.additional_settings.schema.json index 9e4f6d34..f7f46ebd 100644 --- a/application/single_app/static/json/schemas/sql_query_plugin.additional_settings.schema.json +++ b/application/single_app/static/json/schemas/sql_query_plugin.additional_settings.schema.json @@ -3,13 +3,13 @@ "title": "SQL Query Plugin Additional Settings", "type": "object", "properties": { - "connection_string__Secret": { + "connection_string": { "type": "string", "description": "Database connection string. Required if server/database not provided." }, "database_type": { "type": "string", - "enum": ["sqlserver", "postgresql", "mysql", "sqlite", "azure_sql", "azuresql"], + "enum": ["sqlserver", "postgresql", "mysql", "sqlite", "azure_sql"], "description": "Type of database engine." }, "server": { @@ -24,7 +24,7 @@ "type": "string", "description": "Username for authentication." }, - "password__Secret": { + "password": { "type": "string", "description": "Password for authentication." }, @@ -50,6 +50,6 @@ "description": "Query timeout in seconds." } }, - "required": ["database_type", "database"], + "required": ["database_type"], "additionalProperties": false } diff --git a/application/single_app/static/json/schemas/sql_schema.definition.json b/application/single_app/static/json/schemas/sql_schema.definition.json index d38a41a8..6903c22a 100644 --- a/application/single_app/static/json/schemas/sql_schema.definition.json +++ b/application/single_app/static/json/schemas/sql_schema.definition.json @@ -1,6 +1,9 @@ { "$schema": "./plugin.definition.schema.json", "allowedAuthTypes": [ + "user", + "identity", + "servicePrincipal", "connection_string" ] } diff --git a/application/single_app/static/json/schemas/sql_schema_plugin.additional_settings.schema.json b/application/single_app/static/json/schemas/sql_schema_plugin.additional_settings.schema.json index e97c7b4b..29fb6b3f 100644 --- a/application/single_app/static/json/schemas/sql_schema_plugin.additional_settings.schema.json +++ b/application/single_app/static/json/schemas/sql_schema_plugin.additional_settings.schema.json @@ -3,13 +3,13 @@ "title": "SQL Schema Plugin Additional Settings", "type": "object", "properties": { - "connection_string__Secret": { + "connection_string": { "type": "string", "description": "Database connection string. Required if server/database not provided." }, "database_type": { "type": "string", - "enum": ["sqlserver", "postgresql", "mysql", "sqlite", "azure_sql", "azuresql"], + "enum": ["sqlserver", "postgresql", "mysql", "sqlite", "azure_sql"], "description": "Type of database engine." }, "server": { @@ -24,7 +24,7 @@ "type": "string", "description": "Username for authentication." }, - "password__Secret": { + "password": { "type": "string", "description": "Password for authentication." }, @@ -33,6 +33,6 @@ "description": "ODBC or DB driver name." } }, - "required": ["database_type", "database"], + "required": ["database_type"], "additionalProperties": false } diff --git a/application/single_app/templates/_agent_examples_modal.html b/application/single_app/templates/_agent_examples_modal.html index 52f95cdc..398e930c 100644 --- a/application/single_app/templates/_agent_examples_modal.html +++ b/application/single_app/templates/_agent_examples_modal.html @@ -92,7 +92,7 @@
- 

+          

         

         

           
@@ -427,7 +427,12 @@ 
                 

               

             
+
+            
+            

+              

+                
+                

+                  
+                  
+                

+              

+              

+                
+              

+            

           
         
 
         
         

           
Advanced

+          
Advanced settings are typically not required. Expand below if you need to customize metadata or additional fields.

           

-            
-            
-            
Optional metadata for this action.

+            
           

-          

-            
-            
-            
Additional configuration fields specific to this action type.

+          

+            

+              
+              
+              
Optional metadata for this action.

+            

+            

+              
+              
+              
Additional configuration fields specific to this action type.

+            

           

         

 
@@ -777,6 +802,15 @@ 

   background-color: #f8f9fa;
 }
 
+/* Advanced toggle chevron animation */
+#plugin-advanced-toggle-icon {
+  transition: transform 0.3s ease;
+}
+#plugin-advanced-collapse.show ~ .mb-3 #plugin-advanced-toggle-icon,
+[aria-expanded="true"] #plugin-advanced-toggle-icon {
+  transform: rotate(180deg);
+}
+
 .sql-connection-config, 
 .sql-auth-config {
   background-color: white;
diff --git a/application/single_app/templates/_sidebar_nav.html b/application/single_app/templates/_sidebar_nav.html
index a0bceee8..33a89b04 100644
--- a/application/single_app/templates/_sidebar_nav.html
+++ b/application/single_app/templates/_sidebar_nav.html
@@ -287,6 +287,11 @@
                   GPT Configuration
                 
               
+              
  • 
+                
+                  Processing Thoughts
+                
+              
    • 
               
  • 
                 
                   Embeddings
diff --git a/application/single_app/templates/admin_settings.html b/application/single_app/templates/admin_settings.html
index 7d01f7da..37467dd7 100644
--- a/application/single_app/templates/admin_settings.html
+++ b/application/single_app/templates/admin_settings.html
@@ -953,6 +953,12 @@ 
    Core Action Toggles
    
                                     
                                     
                                 
    • 
+                                
    
+                                    
+                                    
+                                    
+                                    Requires Enhanced Citations
+                                
    
                             
                             
@@ -1580,6 +1586,27 @@ 
    
                 
    
                 
 
+                
+                
    
+                    
    
+                        Processing Thoughts
+                    
    
+                    
    When enabled, real-time processing steps are shown to users during chat responses and persisted for later review.
    
+                    
    
+                        
+                        
+                        
+                    
    
+                
    
+
                 
                 
    
                     
    
@@ -3217,9 +3244,10 @@ 
    
                         
                     
    
                 
+
             
 
-            
+
             
    
                 
                  
    
diff --git a/application/single_app/templates/chats.html b/application/single_app/templates/chats.html
index b6c212cc..e845cc53 100644
--- a/application/single_app/templates/chats.html
+++ b/application/single_app/templates/chats.html
@@ -1039,7 +1039,8 @@ 
    Group Workspace
    
         
    
           You do not have permission to manage group agents.
         
    
-        
    
+        
    
           
+          
    
+            
+            
+            
+            
+          
    
         
    
-        
    
                                 
    
-          
-            
-              
-              
-              
-            
-          
-          
-            
-              
-            
-          
-        
    Display NameDescriptionActions
    
-                
-                Select a group to load agents.
-              
    
+        
    
+          
+            
+              
+                
+                
+                
+              
+            
+            
+              
+                
+              
+            
+          
    Display NameDescriptionActions
    
+                  
+                  Select a group to load agents.
+                
    
+        
    
+        
    
       
       
    
     
@@ -813,33 +822,42 @@ 
    Group Workspace
    
           
    
             
           
    
-          
    
+          
    
             
+            
    
+              
+              
+              
+              
+            
    
           
    
-          
-            
-              
-                
-                
-                
-              
-            
-            
-              
-                
-              
-            
-          
    Display NameDescriptionActions
    
-                  
-                  Select a group to load actions.
-                
    
+          
    
+            
+              
+                
+                  
+                  
+                  
+                
+              
+              
+                
+                  
+                
+              
+            
    Display NameDescriptionActions
    
+                    
+                    Select a group to load actions.
+                  
    
+          
    
+          
    
         
    
       
     
@@ -851,6 +869,22 @@ 
    Group Workspace
    
 
 
 
+
+
+
 
 
    Personal Workspace
                   
                   
                 
    
+                
    
+                  
+                  
+                  
+                  
+                
    
               
               
    
                 
               
    
-              
-                
-                
-                  
-                    
-                  
-                
-              
    Display NameDescriptionActions
    
-                      
-                      Loading agents...
-                    
    
+              
+              
    
+                
+                  
+                  
+                    
+                      
+                    
+                  
+                
    Display NameDescriptionActions
    
+                        
+                        Loading agents...
+                      
    
+              
    
+              
+              
    
             
             
    
           
@@ -730,16 +741,27 @@ 
    Personal Workspace
    
             
    
               
    
                 
+                
    
+                  
+                  
+                  
+                  
+                
    
               
    
               
    
                 
               
    
-              
-                
-                
-                  
-                
-              
    Display NameDescriptionActions
    
+              
+              
    
+                
+                  
+                  
+                    
+                  
+                
    Display NameDescriptionActions
    
+              
    
+              
+              
    
             
    
           
           
@@ -754,6 +776,24 @@ 
    Personal Workspace
    
   
   
 
+  
+  
+
   
   
    /messages//thoughts` | Fetch persisted thoughts for a specific assistant message (historical viewing) |
+| GET | `/api/conversations//thoughts/pending` | Fetch latest in-progress thoughts (polling while waiting for response) |
+
+Both endpoints return `{"thoughts": [...], "enabled": true/false}`. When `enable_thoughts` is off, they return `{"thoughts": [], "enabled": false}`.
+
+#### Instrumentation Points (`route_backend_chats.py`)
+
+| Step Type | Content Example | When |
+|-----------|----------------|------|
+| `content_safety` | "Checking content safety..." | Before content safety check |
+| `search` | "Searching personal documents for 'query'..." | Before hybrid search |
+| `search` | "Found 5 results from 3 documents" | After search results |
+| `tabular_analysis` | "Found tabular data — evaluating analysis..." | When tabular data detected |
+| `web_search` | "Searching the web for 'query'..." | Before web search |
+| `web_search` | "Got 8 web results" | After web search |
+| `agent_tool_call` | "Sending to agent 'Data Analyst'..." | Before agent invocation |
+| `generation` | "Generating response..." | Before GPT call |
+
+### Frontend
+
+#### Streaming Mode
+Thought events are embedded in the SSE stream as `{"type": "thought", ...}` JSON payloads. The streaming handler in `chat-streaming.js` passes these to `handleStreamingThought()` which updates the streaming placeholder badge.
+
+#### Non-Streaming Mode
+A polling mechanism in `chat-thoughts.js` fetches `/thoughts/pending` every 2 seconds while waiting for a response. The loading indicator text is updated with the latest thought step.
+
+#### Per-Message History
+Each assistant message footer includes a lightbulb toggle button (when thoughts exist). Clicking it opens a collapsible section that lazy-loads thoughts from the API. Each step shows an icon, content text, and optional duration.
+
+#### Icon Map
+| Step Type | Bootstrap Icon |
+|-----------|---------------|
+| `search` | `bi-search` |
+| `tabular_analysis` | `bi-table` |
+| `web_search` | `bi-globe` |
+| `agent_tool_call` | `bi-robot` |
+| `generation` | `bi-lightning` |
+| `content_safety` | `bi-shield-check` |
+
+## Configuration
+
+### Admin Settings
+- **Toggle**: `enable_thoughts` (default: `false`)
+- **Location**: Admin Settings > Optional Features tab > "Processing Thoughts" section
+- **Effect**: When disabled, no thoughts are recorded and no UI elements are shown
+
+### Cosmos DB Containers
+| Container | Partition Key | Purpose |
+|-----------|--------------|---------|
+| `thoughts` | `/user_id` | Active thought records |
+| `archive_thoughts` | `/user_id` | Archived thoughts from deleted conversations |
+
+## Archive and Cleanup
+
+When a conversation is deleted:
+- **Archiving enabled**: Thoughts are copied to `archive_thoughts` container, then deleted from `thoughts`
+- **Archiving disabled**: Thoughts are permanently deleted from `thoughts`
+
+This applies to both single conversation delete and bulk delete operations.
+
+## File Structure
+
+### Files Created
+| File | Purpose |
+|------|---------|
+| `functions_thoughts.py` | ThoughtTracker class, Cosmos CRUD helpers |
+| `route_backend_thoughts.py` | API endpoints for fetching thoughts |
+| `static/js/chat/chat-thoughts.js` | Frontend polling, rendering, toggle |
+
+### Files Modified
+| File | Change |
+|------|--------|
+| `config.py` | Added `thoughts` + `archive_thoughts` Cosmos containers, bumped VERSION |
+| `functions_settings.py` | Added `enable_thoughts` default setting |
+| `app.py` | Imported and registered thought routes |
+| `route_backend_chats.py` | Instrumented ~8 thought points per chat path |
+| `route_backend_conversations.py` | Added archive/delete thoughts on conversation delete |
+| `templates/admin_settings.html` | Added Processing Thoughts toggle card |
+| `static/js/admin/admin_settings.js` | Added `enable_thoughts` to settings collection |
+| `static/js/chat/chat-messages.js` | Integrated thoughts toggle in footer, polling start/stop |
+| `static/js/chat/chat-streaming.js` | Handle `type: "thought"` in SSE data |
+| `static/js/chat/chat-loading-indicator.js` | Added `updateLoadingIndicatorText()` for thought display |
+| `static/css/chats.css` | Added thought indicator, toggle, container, and dark mode styles |
+
+## Testing
+
+1. **Enable feature**: Set `enable_thoughts: True` in admin settings
+2. **Non-streaming**: Send a message with document search — verify loading indicator updates with thought steps, lightbulb icon appears after response
+3. **Streaming**: Send a message — verify streaming placeholder shows thought badges, lightbulb available after finalization
+4. **History**: Reload page, open old conversation — click lightbulb to verify lazy-loaded thoughts
+5. **Disabled**: Set `enable_thoughts: False` — verify no thoughts generated, no lightbulb icons
+6. **Archive**: Delete a conversation with archiving enabled — verify thoughts moved to `archive_thoughts`
diff --git a/docs/explanation/fixes/v0.239.008/CHAT_TABULAR_SK_TRIGGER_FIX.md b/docs/explanation/fixes/v0.239.008/CHAT_TABULAR_SK_TRIGGER_FIX.md
new file mode 100644
index 00000000..d525e2d1
--- /dev/null
+++ b/docs/explanation/fixes/v0.239.008/CHAT_TABULAR_SK_TRIGGER_FIX.md
@@ -0,0 +1,67 @@
+# Chat-Uploaded Tabular File SK Mini-Agent Trigger Fix
+
+## Issue Description
+
+When a user uploads a tabular file (CSV, XLSX, XLS, XLSM) directly to a chat conversation and asks a question in model-only mode (no agent selected), the SK mini-agent (`run_tabular_sk_analysis`) did not trigger. The model would see instructions to "use plugin functions" but could not call them without an agent, resulting in the model describing what it would do instead of providing actual analysis results.
+
+The full agent mode worked correctly because the agent has direct access to the `TabularProcessingPlugin` and can call its functions.
+
+## Root Cause
+
+Three gaps prevented the mini SK agent from activating for chat-uploaded tabular files:
+
+1. **Streaming path ignored `file` role messages**: The streaming conversation history loop (`/api/chat/stream`) only processed `user` and `assistant` roles, making chat-uploaded files completely invisible to the model.
+
+2. **Mini SK only triggered from search results**: Both streaming and non-streaming paths only invoked `run_tabular_sk_analysis()` when tabular files appeared in hybrid search results (`combined_documents`). Chat-uploaded files are stored in blob storage as `file` role messages and are not indexed in Azure AI Search, so they never appeared in search results.
+
+3. **Model-only mode can't call plugin functions**: The non-streaming path's file handler injected "Use the tabular_processing plugin functions" as a system message, but in model-only mode the model has no function-calling capability.
+
+## Technical Details
+
+### Files Modified
+
+- `application/single_app/route_backend_chats.py` — All code changes
+- `application/single_app/config.py` — Version bump to 0.239.008
+
+### Code Changes
+
+#### Non-streaming path (`/api/chat`)
+
+1. Added `chat_tabular_files = set()` tracker before the conversation history loop (~line 1896)
+2. Added `chat_tabular_files.add(filename)` inside the `if is_table and file_content_source == 'blob':` block (~line 1936)
+3. After the history loop, added a block that checks `chat_tabular_files` and calls `run_tabular_sk_analysis(source_hint="chat")`, injecting pre-computed results as a system message (~line 2027)
+
+#### Streaming path (`/api/chat/stream`)
+
+4. Replaced the simple 8-line history loop (which only handled `user`/`assistant`) with expanded logic that mirrors the non-streaming path's `file` role handling, including blob tabular file tracking (~line 3687)
+5. Added the same mini SK trigger block after the expanded loop (~line 3751)
+
+### How It Works After Fix
+
+1. User uploads `sales.xlsx` to chat, asks "analyze sales/profit"
+2. During conversation history building, the `file` role message with `is_table=True` and `file_content_source='blob'` is detected
+3. The filename is collected into `chat_tabular_files`
+4. After the history loop, `run_tabular_sk_analysis()` is called with `source_hint="chat"`, which resolves the file from the `personal-chat` blob container
+5. The mini SK agent pre-loads the file schema, calls plugin functions (aggregate, filter, etc.), and returns computed results
+6. Results are injected as a system message so the model can present accurate numbers
+7. Plugin invocation citations are collected for transparency
+
+## Testing
+
+1. Upload a tabular file (xlsx/csv) directly to chat
+2. With no agent selected, send a data analysis question
+3. Verify the response contains actual computed data (not just a description of steps)
+4. Check logs for `[Chat Tabular SK]` entries confirming the mini SK trigger
+5. Verify agent mode still works as before
+
+## Impact
+
+- Enables tabular data analysis in model-only chat mode for chat-uploaded files
+- No changes to existing search-result-based tabular detection
+- No changes to full agent mode behavior
+- Streaming and non-streaming paths both fixed
+
+## Version
+
+- **Version**: 0.239.008
+- **Implemented in**: 0.239.008
diff --git a/docs/explanation/release_notes.md b/docs/explanation/release_notes.md
index 8077ff3f..bba394db 100644
--- a/docs/explanation/release_notes.md
+++ b/docs/explanation/release_notes.md
@@ -2,6 +2,123 @@
 
 # Feature Release
 
+### **(v0.239.008)**
+
+#### Bug Fixes
+
+*   **Chat-Uploaded Tabular Files Now Trigger SK Mini-Agent in Model-Only Mode**
+    *   Fixed an issue where tabular files (CSV, XLSX, XLS, XLSM) uploaded directly to a chat conversation were not analyzed by the SK mini-agent when no agent was selected. The model would describe what analysis it would perform instead of returning actual computed results.
+    *   **Root Cause**: The mini SK agent only triggered from search results, but chat-uploaded files are stored in blob storage and not indexed in Azure AI Search. Additionally, the streaming path completely ignored `file` role messages in conversation history.
+    *   **Fix**: Both streaming and non-streaming chat paths now detect chat-uploaded tabular files during conversation history building and trigger `run_tabular_sk_analysis(source_hint="chat")` to pre-compute results. The streaming path also now properly handles `file` role messages (tabular and non-tabular) matching the non-streaming path's behavior.
+    *   (Ref: `route_backend_chats.py`, `run_tabular_sk_analysis()`, `collect_tabular_sk_citations()`)
+
+---
+### **(v0.239.013)**
+
+#### New Features
+
+*   **Agent & Action User Tracking (created_by / modified_by)**
+    *   All agent and action documents (personal, group, and global) now include `created_by`, `created_at`, `modified_by`, and `modified_at` fields that track which user created or last modified the entity.
+    *   On updates, the original `created_by` and `created_at` values are preserved while `modified_by` and `modified_at` are refreshed with the current user and timestamp.
+    *   New optional `user_id` parameter added to `save_group_agent`, `save_global_agent`, `save_group_action`, and `save_global_action` for caller-supplied user tracking (backward-compatible, defaults to `None`).
+    *   (Ref: `functions_personal_agents.py`, `functions_group_agents.py`, `functions_global_agents.py`, `functions_personal_actions.py`, `functions_group_actions.py`, `functions_global_actions.py`)
+
+*   **Activity Logging for Agent & Action CRUD Operations**
+    *   Every create, update, and delete operation on agents and actions now generates an activity log record in the `activity_logs` Cosmos DB container and Application Insights.
+    *   Six new logging functions: `log_agent_creation`, `log_agent_update`, `log_agent_deletion`, `log_action_creation`, `log_action_update`, `log_action_deletion`.
+    *   Activity records include: `user_id`, `activity_type`, `entity_type` (agent/action), `operation` (create/update/delete), `workspace_type` (personal/group/global), and `workspace_context` (group_id when applicable).
+    *   Logging is fire-and-forget — failures never break the CRUD operation.
+    *   All personal, group, and admin routes for both agents and actions are wired up.
+    *   (Ref: `functions_activity_logging.py`, `route_backend_agents.py`, `route_backend_plugins.py`)
+
+### **(v0.239.008)**
+
+#### User Interface Enhancements
+
+*   **List/Grid View Toggle for Agents and Actions**
+    *   Added a list/grid view toggle to all four workspace areas: personal agents, personal actions, group agents, and group actions.
+    *   **Grid View**: Large cards with type icon, humanized name, truncated description, and action buttons (Chat, View, Edit, Delete as applicable).
+    *   **List View**: Improved table layout with fixed column widths (28%/47%/25%), humanized display names, and truncated descriptions with hover tooltips for full text.
+    *   **View Button**: New eye-icon button on every agent and action that opens a read-only detail modal with gradient-header summary cards (Basic Information, Model Configuration, Instructions for agents; Basic Information, Configuration for actions).
+    *   **Name Humanization**: Display names are now automatically parsed — underscores and camelCase/PascalCase boundaries are converted to properly spaced, title-cased words (e.g., `myCustomAgent` → `My Custom Agent`).
+    *   **Persistent Preference**: View mode selection (list/grid) is saved per area in localStorage and restored on page load.
+    *   New shared utility module `view-utils.js` provides reusable functions for all four workspace areas.
+    *   (Ref: `view-utils.js`, `workspace_agents.js`, `workspace_plugins.js`, `plugin_common.js`, `group_agents.js`, `group_plugins.js`, `workspace.html`, `group_workspaces.html`, `styles.css`)
+
+### **(v0.239.007)**
+
+#### New Features
+
+*   **Tabular Data Analysis — SK Mini-Agent for Normal Chat**
+    *   Tabular files (CSV, XLSX, XLS, XLSM) detected in search results now trigger a lightweight Semantic Kernel mini-agent that pre-computes data analysis before the main LLM response. This brings the same analytical depth previously only available in full agent mode to every normal chat conversation.
+    *   **Automatic Detection**: When AI Search results include tabular files from any workspace (personal, group, or public) or chat-uploaded documents, the system automatically identifies them via the `TABULAR_EXTENSIONS` configuration and routes the query through the SK mini-agent pipeline.
+    *   **Unified Workspace and Chat Handling**: Tabular files are processed identically regardless of their storage location. The plugin resolves blob paths across all four container types (`user-documents`, `group-documents`, `public-documents`, `personal-chat`) with automatic fallback resolution if the primary source lookup fails. A user asking about an Excel file in their personal workspace gets the same analytical treatment as one asking about a CSV uploaded directly to a chat.
+    *   **Six Data Analysis Functions**: The `TabularProcessingPlugin` exposes `describe_tabular_file`, `aggregate_column` (sum, mean, count, min, max, median, std, nunique, value_counts), `filter_rows` (==, !=, >, <, >=, <=, contains, startswith, endswith), `query_tabular_data` (pandas query syntax), `group_by_aggregate`, and `list_tabular_files` — all registered as Semantic Kernel functions that the mini-agent orchestrates autonomously.
+    *   **Pre-Computed Results Injected as Context**: The mini-agent's computed analysis (exact numerical results, aggregations, filtered data) is injected into the main LLM's system context so it can present accurate, citation-backed answers without hallucinating numbers.
+    *   **Graceful Degradation**: If the mini-agent analysis fails for any reason, the system falls back to instructing the main LLM to use the tabular processing plugin functions directly, preserving full functionality.
+    *   **Non-Streaming and Streaming Support**: Both chat modes are supported. The mini-agent runs synchronously before the main LLM call in both paths.
+    *   **Requires Enhanced Citations**: The tabular processing plugin depends on the blob storage client initialized by the enhanced citations system. The `enable_enhanced_citations` admin setting must be enabled for tabular data analysis to activate.
+    *   **Files Modified**: `route_backend_chats.py`, `semantic_kernel_plugins/tabular_processing_plugin.py`, `config.py`.
+    *   (Ref: `run_tabular_sk_analysis()`, `TabularProcessingPlugin`, `collect_tabular_sk_citations()`, `TABULAR_EXTENSIONS`)
+
+*   **Tabular Tool Execution Citations**
+    *   Every tool call made by the SK mini-agent during tabular analysis is captured and surfaced as an agent citation, providing full transparency into the data analysis pipeline.
+    *   **Automatic Capture**: The existing `@plugin_function_logger` decorator on all `TabularProcessingPlugin` functions records each invocation including function name, input parameters, returned results, execution duration, and success/failure status.
+    *   **Citation Format**: Tool execution citations appear in the same "Agent Tool Execution" modal used by full agent mode, showing `tool_name` (e.g., `TabularProcessingPlugin.aggregate_column`), `function_arguments` (the exact parameters passed), and `function_result` (the computed data returned).
+    *   **End-to-End Auditability**: Users can verify exactly which aggregations, filters, or queries were run against their data, what parameters were used, and what raw results were returned — before the LLM summarized them into the final response.
+    *   **Files Modified**: `route_backend_chats.py`.
+    *   (Ref: `collect_tabular_sk_citations()`, `plugin_invocation_logger.py`)
+
+*   **SK Mini-Agent Performance Optimization**
+    *   Reduced typical tabular analysis time from ~74 seconds to an estimated ~30-33 seconds (55-60% reduction) through three complementary optimizations.
+    *   **DataFrame Caching**: Per-request in-memory cache eliminates redundant blob downloads. Previously, each of the ~8 tool calls in a typical analysis downloaded and parsed the same file independently. Now the file is downloaded once and subsequent calls read from cache. Cache is automatically scoped to the request (new plugin instance per analysis) and garbage-collected afterward.
+    *   **Pre-Dispatch Schema Injection**: File schemas (columns, data types, row counts, and a 3-row preview) are pre-loaded and injected into the SK mini-agent's system prompt before execution begins. This eliminates 2 LLM round-trips that were previously spent on file discovery (`list_tabular_files`) and schema inspection (`describe_tabular_file`), allowing the model to jump directly to analysis tool calls.
+    *   **Async Plugin Functions**: All six `@kernel_function` methods converted to `async def` using `asyncio.to_thread()`. This enables Semantic Kernel's built-in `asyncio.gather()` to truly parallelize batched tool calls (e.g., 3 simultaneous `aggregate_column` calls) instead of executing them serially on the event loop.
+    *   **Batching Instructions**: The system prompt now instructs the model to batch multiple independent function calls in a single response, reducing LLM round-trips further.
+    *   **Files Modified**: `tabular_processing_plugin.py`, `route_backend_chats.py`, `config.py`.
+    *   (Ref: `_df_cache`, `asyncio.to_thread`, pre-dispatch schema injection in `run_tabular_sk_analysis()`)
+
+---
+*   **Chat with Agent Button for Group Agents**
+    *   Added a "Chat" button to each group agent row, allowing users to quickly select a group agent and navigate to the chat page.
+    *   (Ref: `group_agents.js`, `group_workspaces.html`)
+
+### **(v0.239.006)**
+
+#### User Interface Enhancements
+
+*   **Hidden Deprecated Action Types**
+    *   Deprecated action types (`sql_schema`, `ui_test`, `queue_storage`, `blob_storage`, `embedding_model`) are now hidden from the action creation wizard type selector. Existing actions of these types remain functional.
+    *   (Ref: `plugin_modal_stepper.js`)
+
+### **(v0.239.005)**
+
+#### Bug Fixes
+
+*   **Group SQL Action/Plugin Save Failure**
+    *   Fixed group SQL actions (sql_query and sql_schema types) failing to save correctly due to missing endpoint placeholder. Group routes now apply the same `sql://sql_query` / `sql://sql_schema` endpoint logic as personal action routes.
+    *   Fixed Step 4 (Advanced) dynamic fields overwriting Step 3 (Configuration) SQL values with empty strings during form data collection. SQL types now skip the dynamic field merge entirely since Step 3 already provides all necessary configuration.
+    *   Fixed auth type definition schemas (`sql_query.definition.json`, `sql_schema.definition.json`) only allowing `connection_string` auth type, blocking `user`, `identity`, and `servicePrincipal` types that the UI and runtime support.
+    *   Fixed `__Secret` key suffix mismatch in additional settings schemas where `connection_string__Secret` and `password__Secret` didn't match the runtime's expected `connection_string` and `password` field names. Also removed duplicate `azuresql` enum value.
+    *   (Ref: `route_backend_plugins.py`, `plugin_modal_stepper.js`, `sql_query.definition.json`, `sql_schema.definition.json`, `sql_query_plugin.additional_settings.schema.json`, `sql_schema_plugin.additional_settings.schema.json`)
+
+#### New Features
+
+*   **SQL Test Connection Button**
+    *   Added a "Test Connection" button to the SQL Database Configuration section (Step 3) of the action wizard, allowing users to validate database connectivity before saving.
+    *   Supports all database types: SQL Server, Azure SQL (with managed identity), PostgreSQL, MySQL, and SQLite.
+    *   Shows inline success/failure alerts with a 15-second timeout cap and sanitized error messages.
+    *   New backend endpoint: `POST /api/plugins/test-sql-connection`.
+    *   (Ref: `route_backend_plugins.py`, `plugin_modal_stepper.js`, `_plugin_modal.html`)
+
+#### User Interface Enhancements
+
+*   **Advanced Settings Collapse Toggle**
+    *   Step 4 (Advanced) content is now hidden behind a collapsible toggle button ("Show Advanced Settings") instead of being displayed by default. Reduces visual noise for most users.
+    *   For SQL action types, the redundant additional fields UI in Step 4 is hidden entirely since all SQL configuration is already handled in Step 3.
+    *   Step 5 (Summary) no longer shows the raw additional fields JSON dump for SQL types, since that data is already shown in the SQL Database Configuration summary card.
+    *   (Ref: `_plugin_modal.html`, `plugin_modal_stepper.js`)
+
 ### **(v0.239.001)**
 
 #### New Features
diff --git a/functional_tests/test_chat_tabular_sk_trigger.py b/functional_tests/test_chat_tabular_sk_trigger.py
new file mode 100644
index 00000000..9de91b39
--- /dev/null
+++ b/functional_tests/test_chat_tabular_sk_trigger.py
@@ -0,0 +1,319 @@
+#!/usr/bin/env python3
+# test_chat_tabular_sk_trigger.py
+"""
+Functional test for chat-uploaded tabular file SK mini-agent trigger fix.
+Version: 0.239.008
+Implemented in: 0.239.008
+
+This test ensures that when tabular data files (CSV, XLSX) are uploaded directly
+to a chat conversation, the SK mini-agent (`run_tabular_sk_analysis`) is properly
+triggered in model-only mode (no agent selected) to pre-compute analysis results.
+
+Previously, the mini SK agent only triggered from search results, missing
+chat-uploaded files entirely. The streaming path also ignored file role messages.
+
+This test validates:
+1. Chat-uploaded tabular files are detected in conversation history (file role)
+2. Filenames are collected into chat_tabular_files set for both blob and inline cases
+3. The streaming path properly handles file role messages (not just user/assistant)
+4. The mini SK trigger block activates when chat_tabular_files is non-empty
+"""
+
+import sys
+import os
+
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+sys.path.append(os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), 'application', 'single_app'))
+
+
+def test_chat_tabular_file_detection():
+    """Test that chat-uploaded tabular files are detected from file role messages."""
+    print("Testing chat-uploaded tabular file detection...")
+
+    try:
+        # Simulate conversation messages as they would appear from Cosmos DB
+        recent_messages = [
+            {
+                'role': 'file',
+                'filename': 'Sample_-_Superstore_1.xlsx',
+                'file_content': '',
+                'is_table': True,
+                'file_content_source': 'blob',
+                'metadata': {},
+            },
+            {
+                'role': 'user',
+                'content': 'analyze sales/profit',
+                'metadata': {},
+            },
+        ]
+
+        # Simulate the conversation history building logic (streaming path)
+        conversation_history_for_api = []
+        allowed_roles_in_history = ['user', 'assistant']
+        max_file_content_length_in_history = 50000
+        max_tabular_content_length_in_history = 50000
+        chat_tabular_files = set()
+
+        for message in recent_messages:
+            role = message.get('role')
+            content = message.get('content', '')
+
+            if role in allowed_roles_in_history:
+                conversation_history_for_api.append({
+                    'role': role,
+                    'content': content
+                })
+            elif role == 'file':
+                filename = message.get('filename', 'uploaded_file')
+                file_content = message.get('file_content', '')
+                is_table = message.get('is_table', False)
+                file_content_source = message.get('file_content_source', '')
+
+                if is_table and file_content_source == 'blob':
+                    chat_tabular_files.add(filename)
+                    conversation_history_for_api.append({
+                        'role': 'system',
+                        'content': (
+                            f"[User uploaded a tabular data file named '{filename}'. "
+                            f"The file is stored in blob storage and available for analysis. "
+                            f"The file source is 'chat'.]"
+                        )
+                    })
+
+        # Verify file was detected
+        assert 'Sample_-_Superstore_1.xlsx' in chat_tabular_files, \
+            f"Expected xlsx file in chat_tabular_files, got: {chat_tabular_files}"
+
+        # Verify system message was added for the file
+        system_msgs = [m for m in conversation_history_for_api if m['role'] == 'system']
+        assert len(system_msgs) == 1, f"Expected 1 system message, got {len(system_msgs)}"
+        assert 'Sample_-_Superstore_1.xlsx' in system_msgs[0]['content'], \
+            "System message should reference the uploaded file"
+
+        # Verify user message was included
+        user_msgs = [m for m in conversation_history_for_api if m['role'] == 'user']
+        assert len(user_msgs) == 1, f"Expected 1 user message, got {len(user_msgs)}"
+        assert user_msgs[0]['content'] == 'analyze sales/profit', \
+            "User message content should be preserved"
+
+        print("Test passed!")
+        return True
+
+    except Exception as e:
+        print(f"Test failed: {e}")
+        import traceback
+        traceback.print_exc()
+        return False
+
+
+def test_non_tabular_file_handling():
+    """Test that non-tabular files are handled with content preview."""
+    print("Testing non-tabular file handling in streaming path...")
+
+    try:
+        recent_messages = [
+            {
+                'role': 'file',
+                'filename': 'report.txt',
+                'file_content': 'This is the report content.',
+                'is_table': False,
+                'file_content_source': 'inline',
+                'metadata': {},
+            },
+            {
+                'role': 'user',
+                'content': 'summarize the report',
+                'metadata': {},
+            },
+        ]
+
+        conversation_history_for_api = []
+        allowed_roles_in_history = ['user', 'assistant']
+        max_file_content_length_in_history = 50000
+        max_tabular_content_length_in_history = 50000
+        chat_tabular_files = set()
+
+        for message in recent_messages:
+            role = message.get('role')
+            content = message.get('content', '')
+
+            if role in allowed_roles_in_history:
+                conversation_history_for_api.append({
+                    'role': role,
+                    'content': content
+                })
+            elif role == 'file':
+                filename = message.get('filename', 'uploaded_file')
+                file_content = message.get('file_content', '')
+                is_table = message.get('is_table', False)
+                file_content_source = message.get('file_content_source', '')
+
+                if is_table and file_content_source == 'blob':
+                    chat_tabular_files.add(filename)
+                    conversation_history_for_api.append({
+                        'role': 'system',
+                        'content': f"[User uploaded a tabular data file named '{filename}'.]"
+                    })
+                else:
+                    content_limit = (
+                        max_tabular_content_length_in_history if is_table
+                        else max_file_content_length_in_history
+                    )
+                    display_content = file_content[:content_limit]
+                    if len(file_content) > content_limit:
+                        display_content += "..."
+
+                    conversation_history_for_api.append({
+                        'role': 'system',
+                        'content': (
+                            f"[User uploaded a file named '{filename}'. "
+                            f"Content preview:\n{display_content}]\n"
+                            f"Use this file context if relevant."
+                        )
+                    })
+
+        # Non-tabular files should NOT be in chat_tabular_files
+        assert len(chat_tabular_files) == 0, \
+            f"Non-tabular files should not be tracked, got: {chat_tabular_files}"
+
+        # System message should contain file content preview
+        system_msgs = [m for m in conversation_history_for_api if m['role'] == 'system']
+        assert len(system_msgs) == 1, f"Expected 1 system message, got {len(system_msgs)}"
+        assert 'report.txt' in system_msgs[0]['content'], \
+            "System message should reference the file"
+        assert 'This is the report content.' in system_msgs[0]['content'], \
+            "System message should include file content preview"
+
+        print("Test passed!")
+        return True
+
+    except Exception as e:
+        print(f"Test failed: {e}")
+        import traceback
+        traceback.print_exc()
+        return False
+
+
+def test_multiple_tabular_files():
+    """Test detection of multiple tabular files uploaded to chat."""
+    print("Testing multiple tabular file detection...")
+
+    try:
+        recent_messages = [
+            {
+                'role': 'file',
+                'filename': 'sales_2024.xlsx',
+                'file_content': '',
+                'is_table': True,
+                'file_content_source': 'blob',
+                'metadata': {},
+            },
+            {
+                'role': 'file',
+                'filename': 'inventory.csv',
+                'file_content': 'item,count\nwidget,100',
+                'is_table': True,
+                'file_content_source': 'blob',
+                'metadata': {},
+            },
+            {
+                'role': 'user',
+                'content': 'compare sales to inventory',
+                'metadata': {},
+            },
+        ]
+
+        chat_tabular_files = set()
+        conversation_history_for_api = []
+        allowed_roles_in_history = ['user', 'assistant']
+
+        for message in recent_messages:
+            role = message.get('role')
+            content = message.get('content', '')
+
+            if role in allowed_roles_in_history:
+                conversation_history_for_api.append({'role': role, 'content': content})
+            elif role == 'file':
+                filename = message.get('filename', 'uploaded_file')
+                is_table = message.get('is_table', False)
+                file_content_source = message.get('file_content_source', '')
+
+                if is_table and file_content_source == 'blob':
+                    chat_tabular_files.add(filename)
+                    conversation_history_for_api.append({
+                        'role': 'system',
+                        'content': f"[Tabular file '{filename}' available for analysis.]"
+                    })
+
+        assert len(chat_tabular_files) == 2, \
+            f"Expected 2 tabular files, got {len(chat_tabular_files)}"
+        assert 'sales_2024.xlsx' in chat_tabular_files, "Missing sales_2024.xlsx"
+        assert 'inventory.csv' in chat_tabular_files, "Missing inventory.csv"
+
+        print("Test passed!")
+        return True
+
+    except Exception as e:
+        print(f"Test failed: {e}")
+        import traceback
+        traceback.print_exc()
+        return False
+
+
+def test_inline_tabular_not_tracked_for_sk():
+    """Test that inline tabular files (not blob) are NOT tracked for SK analysis."""
+    print("Testing inline tabular files are not tracked for SK analysis...")
+
+    try:
+        recent_messages = [
+            {
+                'role': 'file',
+                'filename': 'data.csv',
+                'file_content': 'col1,col2\n1,2\n3,4',
+                'is_table': True,
+                'file_content_source': 'inline',  # Not blob
+                'metadata': {},
+            },
+        ]
+
+        chat_tabular_files = set()
+
+        for message in recent_messages:
+            role = message.get('role')
+            if role == 'file':
+                is_table = message.get('is_table', False)
+                file_content_source = message.get('file_content_source', '')
+                filename = message.get('filename', '')
+
+                if is_table and file_content_source == 'blob':
+                    chat_tabular_files.add(filename)
+
+        assert len(chat_tabular_files) == 0, \
+            f"Inline tabular files should not be tracked for SK, got: {chat_tabular_files}"
+
+        print("Test passed!")
+        return True
+
+    except Exception as e:
+        print(f"Test failed: {e}")
+        import traceback
+        traceback.print_exc()
+        return False
+
+
+if __name__ == "__main__":
+    results = []
+    results.append(test_chat_tabular_file_detection())
+    results.append(test_non_tabular_file_handling())
+    results.append(test_multiple_tabular_files())
+    results.append(test_inline_tabular_not_tracked_for_sk())
+
+    print(f"\n{'='*50}")
+    print(f"Results: {sum(results)}/{len(results)} tests passed")
+    if all(results):
+        print("All tests passed!")
+        sys.exit(0)
+    else:
+        print("Some tests failed!")
+        sys.exit(1)
diff --git a/functional_tests/test_thoughts_feature.py b/functional_tests/test_thoughts_feature.py
new file mode 100644
index 00000000..c07df2da
--- /dev/null
+++ b/functional_tests/test_thoughts_feature.py
@@ -0,0 +1,520 @@
+#!/usr/bin/env python3
+"""
+Functional test for Processing Thoughts feature.
+Version: 0.239.003
+Implemented in: 0.239.003
+
+This test ensures that the Processing Thoughts feature is properly implemented
+across backend (ThoughtTracker, API endpoints, chat instrumentation) and
+frontend (polling, streaming, toggle, rendering) components.
+"""
+
+import sys
+import os
+
+sys.path.insert(0, os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), 'application', 'single_app'))
+
+
+def test_thoughts_backend_module():
+    """Test that functions_thoughts.py has proper ThoughtTracker and CRUD functions."""
+    print("Testing functions_thoughts.py module structure...")
+    try:
+        backend_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'functions_thoughts.py'
+        )
+
+        with open(backend_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'ThoughtTracker class': 'class ThoughtTracker' in content,
+            'add_thought method': 'def add_thought' in content,
+            'complete_thought method': 'def complete_thought' in content,
+            'enabled property': 'def enabled' in content or 'enabled' in content,
+            'get_thoughts_for_message': 'def get_thoughts_for_message' in content,
+            'get_pending_thoughts': 'def get_pending_thoughts' in content,
+            'archive_thoughts_for_conversation': 'def archive_thoughts_for_conversation' in content,
+            'delete_thoughts_for_conversation': 'def delete_thoughts_for_conversation' in content,
+            'delete_thoughts_for_message': 'def delete_thoughts_for_message' in content,
+            'cosmos_thoughts_container import': 'cosmos_thoughts_container' in content,
+            'step_index tracking': 'step_index' in content,
+            'step_type field': 'step_type' in content,
+            'error handling (try/except)': 'except Exception' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_route_module():
+    """Test that route_backend_thoughts.py has proper API endpoints."""
+    print("\nTesting route_backend_thoughts.py API endpoints...")
+    try:
+        route_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'route_backend_thoughts.py'
+        )
+
+        with open(route_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'register function': 'def register_route_backend_thoughts' in content,
+            'message thoughts endpoint': '/messages/' in content and '/thoughts' in content,
+            'pending thoughts endpoint': '/thoughts/pending' in content,
+            'swagger_route decorator': 'swagger_route' in content,
+            'login_required decorator': 'login_required' in content,
+            'user_required decorator': 'user_required' in content,
+            'get_auth_security import': 'get_auth_security' in content,
+            'enable_thoughts check': 'enable_thoughts' in content,
+            'returns enabled flag': "'enabled'" in content or '"enabled"' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_chat_instrumentation():
+    """Test that route_backend_chats.py has thought instrumentation points."""
+    print("\nTesting route_backend_chats.py thought instrumentation...")
+    try:
+        chats_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'route_backend_chats.py'
+        )
+
+        with open(chats_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'ThoughtTracker import': 'ThoughtTracker' in content,
+            'thought_tracker instantiation': 'thought_tracker' in content and 'ThoughtTracker(' in content,
+            'add_thought calls': 'thought_tracker.add_thought' in content,
+            'content_safety thought': "'content_safety'" in content or '"content_safety"' in content,
+            'search thought': "add_thought('search'" in content or 'add_thought("search"' in content,
+            'web_search thought': "'web_search'" in content or '"web_search"' in content,
+            'generation thought': "'generation'" in content or '"generation"' in content,
+            'thoughts_enabled in response': 'thoughts_enabled' in content,
+            'SSE thought event (streaming)': '"type": "thought"' in content or "'type': 'thought'" in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_frontend_module():
+    """Test that chat-thoughts.js has polling, streaming, and toggle logic."""
+    print("\nTesting chat-thoughts.js frontend module...")
+    try:
+        js_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'static', 'js', 'chat', 'chat-thoughts.js'
+        )
+
+        with open(js_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'startThoughtPolling export': 'export function startThoughtPolling' in content,
+            'stopThoughtPolling export': 'export function stopThoughtPolling' in content,
+            'handleStreamingThought export': 'export function handleStreamingThought' in content,
+            'createThoughtsToggleHtml export': 'export function createThoughtsToggleHtml' in content,
+            'attachThoughtsToggleListener export': 'export function attachThoughtsToggleListener' in content,
+            'polling interval setup': 'setInterval' in content,
+            'polling endpoint fetch': '/thoughts/pending' in content,
+            'message thoughts fetch': '/thoughts' in content,
+            'icon map (search)': 'bi-search' in content,
+            'icon map (globe)': 'bi-globe' in content,
+            'icon map (robot)': 'bi-robot' in content,
+            'icon map (lightning)': 'bi-lightning' in content,
+            'icon map (shield)': 'bi-shield-check' in content,
+            'enable_thoughts check': 'enable_thoughts' in content,
+            'd-none toggle': 'd-none' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_chat_messages_integration():
+    """Test that chat-messages.js integrates thought polling and toggle."""
+    print("\nTesting chat-messages.js thoughts integration...")
+    try:
+        messages_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'static', 'js', 'chat', 'chat-messages.js'
+        )
+
+        with open(messages_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'chat-thoughts import': 'chat-thoughts' in content,
+            'startThoughtPolling import': 'startThoughtPolling' in content,
+            'stopThoughtPolling import': 'stopThoughtPolling' in content,
+            'createThoughtsToggleHtml import': 'createThoughtsToggleHtml' in content,
+            'attachThoughtsToggleListener import': 'attachThoughtsToggleListener' in content,
+            'startThoughtPolling call': 'startThoughtPolling(' in content,
+            'stopThoughtPolling call': 'stopThoughtPolling(' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_streaming_integration():
+    """Test that chat-streaming.js handles thought SSE events."""
+    print("\nTesting chat-streaming.js thought event handling...")
+    try:
+        streaming_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'static', 'js', 'chat', 'chat-streaming.js'
+        )
+
+        with open(streaming_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'handleStreamingThought import': 'handleStreamingThought' in content,
+            'thought type check': "'thought'" in content or '"thought"' in content,
+            'handleStreamingThought call': 'handleStreamingThought(' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_loading_indicator():
+    """Test that chat-loading-indicator.js supports thought text updates."""
+    print("\nTesting chat-loading-indicator.js thought text support...")
+    try:
+        indicator_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'static', 'js', 'chat', 'chat-loading-indicator.js'
+        )
+
+        with open(indicator_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'updateLoadingIndicatorText export': 'export function updateLoadingIndicatorText' in content
+                                                  or 'updateLoadingIndicatorText' in content,
+            'text element update': 'thought-live-text' in content or 'loading-text' in content
+                                   or 'textContent' in content or 'innerHTML' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_admin_settings():
+    """Test that admin settings UI includes the thoughts toggle."""
+    print("\nTesting admin settings UI for thoughts toggle...")
+    try:
+        html_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'templates', 'admin_settings.html'
+        )
+
+        with open(html_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'enable_thoughts checkbox': 'id="enable_thoughts"' in content,
+            'enable_thoughts name attr': 'name="enable_thoughts"' in content,
+            'Processing Thoughts label': 'Processing Thoughts' in content,
+            'lightbulb icon': 'bi-lightbulb' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_settings_default():
+    """Test that functions_settings.py includes enable_thoughts default."""
+    print("\nTesting functions_settings.py default setting...")
+    try:
+        settings_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'functions_settings.py'
+        )
+
+        with open(settings_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'enable_thoughts in defaults': 'enable_thoughts' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_cosmos_containers():
+    """Test that config.py defines thoughts Cosmos DB containers."""
+    print("\nTesting config.py Cosmos DB container definitions...")
+    try:
+        config_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'config.py'
+        )
+
+        with open(config_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'thoughts container name': 'cosmos_thoughts_container_name' in content,
+            'thoughts container object': 'cosmos_thoughts_container' in content,
+            'archive_thoughts container name': 'cosmos_archived_thoughts_container_name' in content,
+            'archive_thoughts container object': 'cosmos_archived_thoughts_container' in content,
+            'partition key /user_id': '/user_id' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_conversation_archive():
+    """Test that route_backend_conversations.py handles thought archive/delete."""
+    print("\nTesting route_backend_conversations.py thought archive support...")
+    try:
+        conv_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'route_backend_conversations.py'
+        )
+
+        with open(conv_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'archive_thoughts_for_conversation import': 'archive_thoughts_for_conversation' in content,
+            'delete_thoughts_for_conversation import': 'delete_thoughts_for_conversation' in content,
+            'archive_thoughts call': 'archive_thoughts_for_conversation(' in content,
+            'delete_thoughts call': 'delete_thoughts_for_conversation(' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_css_styles():
+    """Test that chats.css includes thought-related styles."""
+    print("\nTesting chats.css thought styles...")
+    try:
+        css_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'static', 'css', 'chats.css'
+        )
+
+        with open(css_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'thoughts-toggle-btn style': '.thoughts-toggle-btn' in content,
+            'thoughts-container style': '.thoughts-container' in content,
+            'thought-step style': '.thought-step' in content,
+            'animate-pulse animation': '.animate-pulse' in content,
+            'thought-pulse keyframes': '@keyframes thought-pulse' in content,
+            'dark mode thoughts toggle': '[data-bs-theme="dark"] .thoughts-toggle-btn' in content,
+            'dark mode thought step': '[data-bs-theme="dark"] .thought-step' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+def test_thoughts_app_registration():
+    """Test that app.py registers the thoughts route blueprint."""
+    print("\nTesting app.py route registration...")
+    try:
+        app_file = os.path.join(
+            os.path.dirname(os.path.dirname(__file__)),
+            'application', 'single_app', 'app.py'
+        )
+
+        with open(app_file, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+        checks = {
+            'import register function': 'register_route_backend_thoughts' in content,
+            'register call': 'register_route_backend_thoughts(app)' in content,
+        }
+
+        all_passed = True
+        for name, passed in checks.items():
+            status = 'PASS' if passed else 'FAIL'
+            print(f"  [{status}] {name}")
+            if not passed:
+                all_passed = False
+
+        return all_passed
+
+    except Exception as e:
+        print(f"  [FAIL] Exception: {e}")
+        return False
+
+
+if __name__ == "__main__":
+    print("=" * 60)
+    print("Processing Thoughts Feature - Functional Tests")
+    print("Version: 0.239.003")
+    print("=" * 60)
+
+    tests = [
+        test_thoughts_backend_module,
+        test_thoughts_route_module,
+        test_thoughts_chat_instrumentation,
+        test_thoughts_frontend_module,
+        test_thoughts_chat_messages_integration,
+        test_thoughts_streaming_integration,
+        test_thoughts_loading_indicator,
+        test_thoughts_admin_settings,
+        test_thoughts_settings_default,
+        test_thoughts_cosmos_containers,
+        test_thoughts_conversation_archive,
+        test_thoughts_css_styles,
+        test_thoughts_app_registration,
+    ]
+
+    results = []
+    for test_fn in tests:
+        try:
+            result = test_fn()
+            results.append(result)
+        except Exception as e:
+            print(f"  [FAIL] Unhandled exception: {e}")
+            results.append(False)
+
+    print("\n" + "=" * 60)
+    passed = sum(1 for r in results if r)
+    total = len(results)
+    print(f"Results: {passed}/{total} test groups passed")
+
+    if passed == total:
+        print("All tests passed!")
+    else:
+        print("Some tests failed. Review output above.")
+
+    print("=" * 60)
+    sys.exit(0 if passed == total else 1)