michael-s-molina commented on code in PR #36529: URL: https://github.com/apache/superset/pull/36529#discussion_r2619757223
########## superset/sql/execution/executor.py: ########## @@ -0,0 +1,1080 @@ +# 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. + +""" +SQL Executor implementation for Database.execute() and execute_async(). + +This module provides the SQLExecutor class that implements the query execution +methods defined in superset_core.api.models.Database. + +Implementation Features +----------------------- + +Query Preparation (applies to both sync and async): +- Jinja2 template rendering (via template_params in QueryOptions) +- SQL mutation via SQL_QUERY_MUTATOR config hook +- DML permission checking (requires database.allow_dml=True for DML) +- Disallowed functions checking via DISALLOWED_SQL_FUNCTIONS config +- Row-level security (RLS) via AST transformation (always applied) +- Result limit application via SQL_MAX_ROW config +- Catalog/schema resolution and validation + +Synchronous Execution (execute): +- Multi-statement SQL parsing and execution +- Progress tracking via Query model +- Result caching via cache_manager.data_cache +- Query logging via QUERY_LOGGER config hook +- Timeout protection via SQLLAB_TIMEOUT config +- Dry run mode (returns transformed SQL without execution) + +Asynchronous Execution (execute_async): +- Celery task submission for background execution +- Security validation before submission +- Query model creation with PENDING status +- Result caching check (returns cached if available) +- Background execution with timeout via SQLLAB_ASYNC_TIME_LIMIT_SEC +- Results stored in results backend for retrieval +- Handle-based progress tracking and cancellation + +See Database.execute() and Database.execute_async() docstrings in +superset_core.api.models for the public API contract. +""" + +from __future__ import annotations + +import logging +import time +from datetime import datetime +from typing import Any, TYPE_CHECKING + +from flask import current_app as app, g, has_app_context + +from superset import db +from superset.errors import ErrorLevel, SupersetError, SupersetErrorType +from superset.exceptions import ( + SupersetSecurityException, + SupersetTimeoutException, +) +from superset.extensions import cache_manager +from superset.sql.parse import SQLScript +from superset.utils import core as utils + +if TYPE_CHECKING: + from superset_core.api.types import ( + AsyncQueryHandle, + QueryOptions, + QueryResult, + ) + + from superset.models.core import Database + from superset.result_set import SupersetResultSet + +logger = logging.getLogger(__name__) + + +def execute_sql_with_cursor( + database: Database, + cursor: Any, + statements: list[str], + query: Any, + log_query_fn: Any | None = None, + check_stopped_fn: Any | None = None, + execute_fn: Any | None = None, +) -> list[tuple[str, SupersetResultSet | None, float, int]]: + """ + Execute SQL statements with a cursor and return all result sets. + + This is the shared execution logic used by both sync (SQLExecutor) and + async (celery_task) execution paths. It handles multi-statement execution + with progress tracking via the Query model. + + :param database: Database model to execute against + :param cursor: Database cursor to use for execution + :param statements: List of SQL statements to execute + :param query: Query model for progress tracking + :param log_query_fn: Optional function to log queries, called as fn(sql, schema) + :param check_stopped_fn: Optional function to check if query was stopped. + Should return True if stopped. Used by async execution for cancellation. + :param execute_fn: Optional custom execute function. If not provided, uses + database.db_engine_spec.execute(cursor, sql, database). Custom function + should accept (cursor, sql) and handle execution. + :returns: List of (statement_sql, result_set, execution_time_ms, rowcount) tuples + Returns empty list if stopped. Raises exception on error (fail-fast). + """ + from superset.result_set import SupersetResultSet + + total = len(statements) + if total == 0: + return [] + + results: list[tuple[str, SupersetResultSet | None, float, int]] = [] + + for i, statement in enumerate(statements): + # Check if query was stopped (async cancellation) + if check_stopped_fn and check_stopped_fn(): + return results + + stmt_start_time = time.time() + + # Apply SQL mutation + stmt_sql = database.mutate_sql_based_on_config( + statement, + is_split=True, + ) + + # Log query + if log_query_fn: + log_query_fn(stmt_sql, query.schema) + + # Execute - use custom function or default + if execute_fn: + execute_fn(cursor, stmt_sql) + else: + database.db_engine_spec.execute(cursor, stmt_sql, database) + + stmt_execution_time = (time.time() - stmt_start_time) * 1000 + + # Fetch results from ALL statements + description = cursor.description + if description: + rows = database.db_engine_spec.fetch_data(cursor) + result_set = SupersetResultSet( + rows, + description, + database.db_engine_spec, + ) + else: + # DML statement - no result set + result_set = None + + # Get row count for DML statements + rowcount = cursor.rowcount if hasattr(cursor, "rowcount") else 0 + + results.append((stmt_sql, result_set, stmt_execution_time, rowcount)) + + # Update progress on Query model + progress_pct = int(((i + 1) / total) * 100) + query.progress = progress_pct + query.set_extra_json_key( + "progress", + f"Running statement {i + 1} of {total}", + ) + db.session.commit() # pylint: disable=consider-using-transaction + + return results + + +class SQLExecutor: + """ + SQL query executor implementation. + + Implements Database.execute() and execute_async() methods. + See superset_core.api.models.Database for the full public API documentation. + """ + + def __init__(self, database: Database) -> None: + """ + Initialize the executor with a database. + + :param database: Database model instance to execute queries against + """ + self.database = database + + def execute( + self, + sql: str, + options: QueryOptions | None = None, + ) -> QueryResult: + """ + Execute SQL synchronously. + + If options.dry_run=True, returns the transformed SQL without execution. + All transformations (RLS, templates, limits) are still applied. + + See superset_core.api.models.Database.execute() for full documentation. + """ + from superset_core.api.types import ( + QueryOptions as QueryOptionsType, + QueryResult as QueryResultType, + QueryStatus, + StatementResult, + ) + + opts: QueryOptionsType = options or QueryOptionsType() + start_time = time.time() + + try: + # 1. Prepare SQL (assembly only, no security checks) + script, catalog, schema = self._prepare_sql(sql, opts) + + # 2. Security checks + self._check_security(script) + + # 3. Get mutation status and format SQL + has_mutation = script.has_mutation() + final_sql = script.format() + + # DRY RUN: Return transformed SQL without execution + if opts.dry_run: + total_execution_time_ms = (time.time() - start_time) * 1000 + # Create a StatementResult for each statement in dry-run mode + dry_run_statements = [ + StatementResult( + statement=stmt.format(), + data=None, + row_count=0, + execution_time_ms=0, + ) + for stmt in script.statements + ] + return QueryResultType( + status=QueryStatus.SUCCESS, + statements=dry_run_statements, + query_id=None, + total_execution_time_ms=total_execution_time_ms, + is_cached=False, + ) + + # 4. Check cache + cached_result = self._try_get_cached_result(has_mutation, final_sql, opts) + if cached_result: + return cached_result + + # 5. Create Query model for audit + query = self._create_query_record( + final_sql, opts, catalog, schema, status="running" + ) + + # 6. Execute with timeout + timeout = opts.timeout_seconds or app.config.get("SQLLAB_TIMEOUT", 30) + timeout_msg = f"Query exceeded the {timeout} seconds timeout." + + with utils.timeout(seconds=timeout, error_message=timeout_msg): + statement_results = self._execute_statements( + final_sql, + catalog, + schema, + query, + ) + + total_execution_time_ms = (time.time() - start_time) * 1000 + + # Calculate total row count for Query model + total_rows = sum(stmt.row_count for stmt in statement_results) + + # Update query record + query.status = "success" + query.rows = total_rows + query.progress = 100 + db.session.commit() # pylint: disable=consider-using-transaction + + result = QueryResultType( + status=QueryStatus.SUCCESS, + statements=statement_results, + query_id=query.id, + total_execution_time_ms=total_execution_time_ms, + ) + + # Store in cache (if SELECT and caching enabled) + if not has_mutation: + self._store_in_cache(result, final_sql, opts) + + return result + + except SupersetTimeoutException: + return self._create_error_result( + QueryStatus.TIMED_OUT, + "Query exceeded the timeout limit", + sql, + start_time, + ) + except SupersetSecurityException as ex: + return self._create_error_result( + QueryStatus.FAILED, str(ex), sql, start_time + ) + except Exception as ex: + error_msg = self.database.db_engine_spec.extract_error_message(ex) + return self._create_error_result( + QueryStatus.FAILED, error_msg, sql, start_time + ) + + def execute_async( + self, + sql: str, + options: QueryOptions | None = None, + ) -> AsyncQueryHandle: + """ + Execute SQL asynchronously via Celery. + + If options.dry_run=True, returns the transformed SQL as a completed + AsyncQueryHandle without submitting to Celery. + + See superset_core.api.models.Database.execute_async() for full documentation. + """ + from superset_core.api.types import ( + QueryOptions as QueryOptionsType, + QueryResult as QueryResultType, + QueryStatus, + ) + + opts: QueryOptionsType = options or QueryOptionsType() + + # 1. Prepare SQL (assembly only, no security checks) + script, catalog, schema = self._prepare_sql(sql, opts) + + # 2. Security checks + self._check_security(script) + + # 3. Get mutation status and format SQL + has_mutation = script.has_mutation() + final_sql = script.format() + + # DRY RUN: Return transformed SQL as completed async handle + if opts.dry_run: + from superset_core.api.types import StatementResult + + dry_run_statements = [ + StatementResult( + statement=stmt.format(), + data=None, + row_count=0, + execution_time_ms=0, + ) + for stmt in script.statements + ] + dry_run_result = QueryResultType( + status=QueryStatus.SUCCESS, + statements=dry_run_statements, + query_id=None, + total_execution_time_ms=0, + is_cached=False, + ) + return self._create_cached_handle(dry_run_result) + + # 4. Check cache + if cached_result := self._try_get_cached_result(has_mutation, final_sql, opts): + return self._create_cached_handle(cached_result) + + # 5. Create Query model for audit + query = self._create_query_record( + final_sql, opts, catalog, schema, status="pending" + ) + + # 6. Submit to Celery + self._submit_query_to_celery(query, final_sql, opts) + + # 7. Create and return handle with bound methods + return self._create_async_handle(query.id) + + def _prepare_sql( + self, + sql: str, + opts: QueryOptions, + ) -> tuple[SQLScript, str | None, str | None]: + """ + Prepare SQL for execution (no side effects, no security checks). + + This method performs SQL preprocessing: + 1. Template rendering + 2. SQL parsing + 3. Catalog/schema resolution + 4. RLS application + 5. Limit application (if not mutation) + + Security checks (disallowed functions, DML permission) are performed + by the caller after receiving the prepared script. + + :param sql: Original SQL query + :param opts: Query options + :returns: Tuple of (prepared SQLScript, catalog, schema) + """ + # 1. Render Jinja2 templates + rendered_sql = self._render_sql_template(sql, opts.template_params) + + # 2. Parse SQL with SQLScript + script = SQLScript(rendered_sql, self.database.db_engine_spec.engine) + + # 3. Get catalog and schema + catalog = opts.catalog or self.database.get_default_catalog() + schema = opts.schema or self.database.get_default_schema(catalog) + + # 4. Apply RLS directly to script statements + self._apply_rls_to_script(script, catalog, schema) + + # 5. Apply limit only if not a mutation + if not script.has_mutation(): + self._apply_limit_to_script(script, opts) + + return script, catalog, schema + + def _check_security(self, script: SQLScript) -> None: + """ + Perform security checks on prepared SQL script. + + :param script: Prepared SQLScript + :raises SupersetSecurityException: If security checks fail + """ + # Check disallowed functions + if disallowed := self._check_disallowed_functions(script): + raise SupersetSecurityException( + SupersetError( + message=f"Disallowed SQL functions: {', '.join(disallowed)}", + error_type=SupersetErrorType.INVALID_SQL_ERROR, + level=ErrorLevel.ERROR, + ) + ) + + # Check DML permission + if script.has_mutation() and not self.database.allow_dml: + raise SupersetSecurityException( + SupersetError( + message="DML queries are not allowed on this database", Review Comment: Not an entity name in this context. -- 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]
