codeant-ai-for-open-source[bot] commented on code in PR #40959: URL: https://github.com/apache/superset/pull/40959#discussion_r3477036493
########## superset/mcp_service/dashboard/tool/duplicate_dashboard.py: ########## @@ -0,0 +1,315 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. + +""" +MCP tool: duplicate_dashboard + +Duplicates an existing dashboard, optionally deep-copying its charts. +Canonical workflow: clone a template dashboard, then edit the copy +(e.g. to create a regional or staging variant). +""" + +import logging +from typing import Any + +from fastmcp import Context +from sqlalchemy.exc import SQLAlchemyError +from superset_core.mcp.decorators import tool, ToolAnnotations + +from superset.extensions import event_logger +from superset.mcp_service.dashboard.schemas import ( + _sanitize_dashboard_info_for_llm_context, + DashboardInfo, + DuplicateDashboardRequest, + DuplicateDashboardResponse, + serialize_chart_summary, +) +from superset.mcp_service.privacy import user_can_view_data_model_metadata +from superset.mcp_service.utils.url_utils import get_superset_base_url +from superset.utils import json + +logger = logging.getLogger(__name__) + + +def _positions_reference_charts(positions: dict[str, Any]) -> bool: + """Return whether a layout maps any chart into the dashboard. + + ``DashboardDAO.set_dash_metadata`` rebuilds the new dashboard's slice + list solely from the chart IDs found in ``positions``, so a layout + with no ``CHART`` entries yields an empty dashboard regardless of the + source's ``slices`` relationship. + """ + return any( + isinstance(value, dict) + and value.get("type") == "CHART" + and value.get("meta", {}).get("chartId") + for value in positions.values() + ) + + +def _build_copy_payload( + source: Any, dashboard_title: str, duplicate_slices: bool +) -> tuple[dict[str, Any], bool]: + """Build the data payload expected by ``CopyDashboardCommand``. + + Mirrors what the frontend "Save as" flow sends to the + ``/api/v1/dashboard/<id>/copy/`` endpoint: the source dashboard's + current ``json_metadata`` with a ``positions`` key holding the current + layout (``position_json``). ``DashboardCopySchema`` requires + ``json_metadata``, and ``DashboardDAO.copy_dashboard`` reads + ``positions`` from it to remap chart IDs when ``duplicate_slices`` + is enabled. + + Returns the payload and a flag indicating whether the layout maps any + chart, so the caller can refuse to produce a silently empty copy. + """ + try: + metadata = json.loads(source.json_metadata or "{}") + except (json.JSONDecodeError, TypeError): + metadata = {} + if not isinstance(metadata, dict): + metadata = {} + + try: + positions = json.loads(source.position_json or "{}") + except (json.JSONDecodeError, TypeError): + positions = {} + if not isinstance(positions, dict): + positions = {} + + metadata["positions"] = positions + + payload = { + "dashboard_title": dashboard_title, + "css": source.css, + "duplicate_slices": duplicate_slices, + "json_metadata": json.dumps(metadata), + } + return payload, _positions_reference_charts(positions) + + +def _serialize_new_dashboard(dashboard: Any) -> tuple[DashboardInfo, str]: + """Build the response ``DashboardInfo`` and URL for the new dashboard.""" + from superset.mcp_service.dashboard.schemas import serialize_tag_object + + dashboard_url = f"{get_superset_base_url()}/superset/dashboard/{dashboard.id}/" + include_data_model_metadata = user_can_view_data_model_metadata() + info = DashboardInfo( + id=dashboard.id, + dashboard_title=dashboard.dashboard_title, + slug=dashboard.slug, + description=dashboard.description, + published=dashboard.published, + created_on=dashboard.created_on, + changed_on=dashboard.changed_on, + uuid=str(dashboard.uuid) if dashboard.uuid else None, + url=dashboard_url, + chart_count=len(dashboard.slices), + tags=[ + obj + for tag in getattr(dashboard, "tags", []) + if (obj := serialize_tag_object(tag)) is not None + ], + charts=[ + obj + for chart in getattr(dashboard, "slices", []) + if ( + obj := serialize_chart_summary( + chart, + include_data_model_metadata=include_data_model_metadata, + ) + ) + is not None + ], + ) + return _sanitize_dashboard_info_for_llm_context(info), dashboard_url + + +@tool( + tags=["mutate"], + class_permission_name="Dashboard", + method_permission_name="write", + annotations=ToolAnnotations( + title="Duplicate dashboard", + readOnlyHint=False, + destructiveHint=False, + ), +) +async def duplicate_dashboard( + request: DuplicateDashboardRequest, ctx: Context +) -> DuplicateDashboardResponse: + """ + Duplicate an existing dashboard under a new title. + + By default the copy references the same charts as the source. + Set duplicate_slices=true to also deep-copy every chart into new + chart objects owned by you, so edits to the copies never affect + the originals. + + The source dashboard can be identified by numeric ID, UUID, or slug. + Returns the new dashboard's ID, title, and URL. + """ + await ctx.info( + "Duplicating dashboard: dashboard_id=%s, duplicate_slices=%s" + % (request.dashboard_id, request.duplicate_slices) + ) + + from superset.commands.dashboard.copy import CopyDashboardCommand + from superset.commands.dashboard.exceptions import ( + DashboardAccessDeniedError, + DashboardCopyError, + DashboardForbiddenError, + DashboardInvalidError, + DashboardNotFoundError, + ) + from superset.daos.dashboard import DashboardDAO + + try: + with event_logger.log_context(action="mcp.duplicate_dashboard.lookup"): + try: + source = DashboardDAO.get_by_id_or_slug(str(request.dashboard_id)) + except DashboardNotFoundError: + await ctx.warning( + "Dashboard not found for duplication: dashboard_id=%s" + % (request.dashboard_id,) + ) + return DuplicateDashboardResponse( + error=( + f"Dashboard '{request.dashboard_id}' not found. " + "Use list_dashboards to get valid dashboard IDs." + ), + ) + except DashboardAccessDeniedError: + await ctx.warning( + "Dashboard access denied for duplication: dashboard_id=%s" + % (request.dashboard_id,) + ) + return DuplicateDashboardResponse( + error=( + f"You don't have access to dashboard " + f"'{request.dashboard_id}', so it cannot be duplicated." + ), + ) + + data, layout_has_charts = _build_copy_payload( + source, request.dashboard_title, request.duplicate_slices + ) + + if getattr(source, "slices", None) and not layout_has_charts: + await ctx.warning( + "Source layout maps no charts; refusing to duplicate to " + "avoid an empty copy: dashboard_id=%s" % (request.dashboard_id,) + ) + return DuplicateDashboardResponse( + error=( + f"Dashboard '{request.dashboard_id}' has charts but its " + "saved layout is missing or invalid, so duplicating it " + "would produce a dashboard with no charts. Open and " + "re-save the source dashboard to repair its layout, then " + "try again." + ), + ) + + with event_logger.log_context(action="mcp.duplicate_dashboard.copy"): + new_dashboard = CopyDashboardCommand(source, data).run() + + # Re-fetch with eager-loaded relationships to avoid lazy-loading on + # a session that the command's commit may have invalidated. + from sqlalchemy.orm import subqueryload + + from superset.models.dashboard import Dashboard + from superset.models.slice import Slice + + try: + new_dashboard = ( + DashboardDAO.find_by_id( + new_dashboard.id, + query_options=[ + subqueryload(Dashboard.slices).subqueryload(Slice.tags), + subqueryload(Dashboard.tags), + ], + ) + or new_dashboard + ) + info, dashboard_url = _serialize_new_dashboard(new_dashboard) + except SQLAlchemyError: + logger.warning( + "Re-fetch of dashboard %s failed; returning minimal response", + new_dashboard.id, + exc_info=True, + ) + dashboard_url = ( + f"{get_superset_base_url()}/superset/dashboard/{new_dashboard.id}/" + ) + info = _sanitize_dashboard_info_for_llm_context( + DashboardInfo( + id=new_dashboard.id, + dashboard_title=request.dashboard_title, + url=dashboard_url, + ) + ) Review Comment: **Suggestion:** The SQLAlchemy error fallback for the post-copy re-fetch does not roll back the failed transaction before continuing. If the failed query leaves the session in an invalid transaction state, later DB use in the same request lifecycle (including logging hooks or downstream ORM access) can fail with "transaction rolled back"/"can't reconnect" errors. Add a `db.session.rollback()` in this `except SQLAlchemyError` block before building the minimal response, matching the existing recovery pattern used in other MCP dashboard tools. [missing cleanup] <details> <summary><b>Severity Level:</b> Major ⚠️</summary> ```mdx - ⚠️ MCP duplicate_dashboard leaves db.session in failed state on error. - ⚠️ Subsequent ORM work in same lifecycle may hit rollback errors. - ⚠️ Inconsistent with other MCP dashboard tools' SQLAlchemyError handling. ``` </details> <details> <summary><b>Steps of Reproduction ✅ </b></summary> ```mdx 1. Start the MCP service (FastMCP app created in `superset/mcp_service/app.py:18-29`) and call the `duplicate_dashboard` tool via an MCP client as done in tests (see `tests/unit_tests/mcp_service/dashboard/tool/test_duplicate_dashboard.py:18-22`, which calls `client.call_tool(\"duplicate_dashboard\", {\"request\": {...}})`). 2. Within `duplicate_dashboard()` in `superset/mcp_service/dashboard/tool/duplicate_dashboard.py:152-165`, execution reaches the post-copy re-fetch block at lines 238-248 where `DashboardDAO.find_by_id(new_dashboard.id, query_options=[...])` is called to eagerly load `slices` and `tags`. 3. Simulate a database/connection failure so that this re-fetch raises `SQLAlchemyError` (e.g., by configuring the DB to drop connections or, in a unit test, patching `DashboardDAO.find_by_id` to raise `SQLAlchemyError`). The exception is caught by the `except SQLAlchemyError:` block at lines 249-255, which only logs a warning and builds a minimal `DashboardInfo` response (lines 255-264) without calling `db.session.rollback()`. 4. Because the MCP auth wrapper `mcp_auth_hook` in `superset/mcp_service/auth.py:112-127` only calls `_cleanup_session_on_error()` (which does `db.session.rollback()` and `db.session.remove()` at `auth.py:27-37`) when an exception propagates out of the tool, this caught `SQLAlchemyError` leaves the shared `superset.extensions.db.session` in a failed-transaction state for the remainder of the tool call. In contrast, other MCP dashboard tools explicitly roll back on the same "re-fetch failed" path (see `generate_dashboard` at `superset/mcp_service/dashboard/tool/generate_dashboard.py:452-461` and `add_chart_to_existing_dashboard` at `superset/mcp_service/dashboard/tool/add_chart_to_existing_dashboard.py:567-579`), demonstrating that the intended design is to clear the session after such errors; the missing rollback here is an inconsistent error-handling bug that would cause any additional DB usage in this tool (or shared-session consumers during the same lifecycle) to hit "transaction has been rolled back" errors until a later cleanup. ``` </details> [](https://app.codeant.ai/fix-in-ide?tool=cursor&prompt_id=0317cbedc9cb4e6a8ca97367f4005941&service=github&base_url=https%3A%2F%2Fgithub.com&org=apache&repo=apache%2Fsuperset) [](https://app.codeant.ai/fix-in-ide?tool=vscode-claude&prompt_id=0317cbedc9cb4e6a8ca97367f4005941&service=github&base_url=https%3A%2F%2Fgithub.com&org=apache&repo=apache%2Fsuperset) *(Use Cmd/Ctrl + Click for best experience)* <details> <summary><b>Prompt for AI Agent 🤖 </b></summary> ```mdx This is a comment left during a code review. **Path:** superset/mcp_service/dashboard/tool/duplicate_dashboard.py **Line:** 249:264 **Comment:** *Missing Cleanup: The SQLAlchemy error fallback for the post-copy re-fetch does not roll back the failed transaction before continuing. If the failed query leaves the session in an invalid transaction state, later DB use in the same request lifecycle (including logging hooks or downstream ORM access) can fail with "transaction rolled back"/"can't reconnect" errors. Add a `db.session.rollback()` in this `except SQLAlchemyError` block before building the minimal response, matching the existing recovery pattern used in other MCP dashboard tools. Validate the correctness of the flagged issue. If correct, How can I resolve this? If you propose a fix, implement it and please make it concise. Once fix is implemented, also check other comments on the same PR, and ask user if the user wants to fix the rest of the comments as well. if said yes, then fetch all the comments validate the correctness and implement a minimal fix ``` </details> <a href='https://app.codeant.ai/feedback?pr_url=https%3A%2F%2Fgithub.com%2Fapache%2Fsuperset%2Fpull%2F40959&comment_hash=cb993ed6a53f050f9175f8ee66b230ed814e0394bf0a256f85f1014ed837770d&reaction=like'>👍</a> | <a href='https://app.codeant.ai/feedback?pr_url=https%3A%2F%2Fgithub.com%2Fapache%2Fsuperset%2Fpull%2F40959&comment_hash=cb993ed6a53f050f9175f8ee66b230ed814e0394bf0a256f85f1014ed837770d&reaction=dislike'>👎</a> -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
