UNBOUND_SQL_PARAMETER with Runtime 15.4

オープン

#466 2024年11月8日に作成コメント 0 件リアクション 0 件担当者 0 名 GitHub で見る

まだ誰も着手していません。

評価

難易度: 4/5
見積もり時間: 3〜5日
初心者へのやさしさ: 35/100

issue の種類: バグ
明瞭さ: おおむね明確
活発さ: 停滞
技術スタック: python, sql, sqlalchemy
領域: backend-api-design, databases

調査の方向性

まず databricks/sql/client.py の Cursor.execute でクエリを再現し、次に準備された操作とパラメータを databricks/sql/thrift_backend.py の ThriftBackend.execute_command まで追跡します。この経路を、動作している notebook の挙動および issue 288 と比較します。:myparam マッピングが値をインライン化せずに正常にバインドされれば完了です。

索引モデルが issue の本文から書いたものです。

説明

Hello, I have trouble using parameters with databricks-sql-python, using a Databricks cluster with Unity Catalog with Runtime 15.4.

This will fail:

from databricks import sql as databricks_sql

with databricks_sql.connect(
    server_hostname="...",
    http_path="...",
    auth_type="databricks-oauth",
    catalog="...",
    schema="...",
) as connection:
    with connection.cursor() as cursor:
        cursor.execute(
            "SELECT * FROM experimentation.time_series WHERE experimentation.time_series.name = :myparam",
            {"myparam": "myvalue"},
        )
        result = cursor.fetchall()
result

with the following exception: ServerOperationError: [UNBOUND_SQL_PARAMETER] Found the unbound parameter: myparam. Please, fix args and provide a mapping of the parameter to a SQL literal.; line 1 pos 83

Stacktrace

---------------------------------------------------------------------------
ServerOperationError                      Traceback (most recent call last)
Cell In[1], line 14
      6 with databricks_sql.connect(
      7     server_hostname="...",
      8     http_path="...",
   (...)
     11     schema="...",
     12 ) as connection:
     13     with connection.cursor() as cursor:
---> 14         cursor.execute(
     15             "SELECT * FROM experimentation.time_series WHERE experimentation.time_series.name = :myparam",
     16             {"myparam": "myvalue"},
     17         )
     18         result = cursor.fetchall()
     19 result

File ...\Lib\site-packages\databricks\sql\client.py:796, in Cursor.execute(self, operation, parameters)
    794 self._check_not_closed()
    795 self._close_and_clear_active_result_set()
--> 796 execute_response = self.thrift_backend.execute_command(
    797     operation=prepared_operation,
    798     session_handle=self.connection._session_handle,
    799     max_rows=self.arraysize,
    800     max_bytes=self.buffer_size_bytes,
    801     lz4_compression=self.connection.lz4_compression,
    802     cursor=self,
    803     use_cloud_fetch=self.connection.use_cloud_fetch,
    804     parameters=prepared_params,
    805 )
    806 self.active_result_set = ResultSet(
    807     self.connection,
    808     execute_response,
   (...)
    811     self.arraysize,
    812 )
    814 if execute_response.is_staging_operation:

File ...\Lib\site-packages\databricks\sql\thrift_backend.py:849, in ThriftBackend.execute_command(self, operation, session_handle, max_rows, max_bytes, lz4_compression, cursor, use_cloud_fetch, parameters)
    831 req = ttypes.TExecuteStatementReq(
    832     sessionHandle=session_handle,
    833     statement=operation,
   (...)
    846     parameters=parameters,
    847 )
    848 resp = self.make_request(self._client.ExecuteStatement, req)
--> 849 return self._handle_execute_response(resp, cursor)

File ...\Lib\site-packages\databricks\sql\thrift_backend.py:941, in ThriftBackend._handle_execute_response(self, resp, cursor)
    938 cursor.active_op_handle = resp.operationHandle
    939 self._check_direct_results_for_error(resp.directResults)
--> 941 final_operation_state = self._wait_until_command_done(
    942     resp.operationHandle,
    943     resp.directResults and resp.directResults.operationStatus,
    944 )
    946 return self._results_message_to_execute_response(resp, final_operation_state)

File ...\Lib\site-packages\databricks\sql\thrift_backend.py:787, in ThriftBackend._wait_until_command_done(self, op_handle, initial_operation_status_resp)
    785     poll_resp = self._poll_for_status(op_handle)
    786     operation_state = poll_resp.operationState
--> 787     self._check_command_not_in_error_or_closed_state(op_handle, poll_resp)
    788 return operation_state

File ...\Lib\site-packages\databricks\sql\thrift_backend.py:572, in ThriftBackend._check_command_not_in_error_or_closed_state(self, op_handle, get_operations_resp)
    570 if get_operations_resp.operationState == ttypes.TOperationState.ERROR_STATE:
    571     if get_operations_resp.displayMessage:
--> 572         raise ServerOperationError(
    573             get_operations_resp.displayMessage,
    574             {
    575                 "operation-id": op_handle
    576                 and self.guid_to_hex_id(op_handle.operationId.guid),
    577                 "diagnostic-info": get_operations_resp.diagnosticInfo,
    578             },
    579         )
    580     else:
    581         raise ServerOperationError(
    582             get_operations_resp.errorMessage,
    583             {
   (...)
    587             },
    588         )

Dependencies

sqlalchemy = "^2.0.36"
jupyter = "^1.1.1"
databricks-connect = "15.4.3"
databricks-sql-connector = {version = "^3.6.0", extras = ["sqlalchemy", "alembic"]}

However, this works using a Databricks notebook:

%python
myparam = 'myvalue'

%sql
SELECT * FROM experimentation.time_series 
WHERE experimentation.time_series.name = :myparam

Please note that this issue seems related to https://github.com/databricks/databricks-sql-python/issues/288.

As a temporary workaround, I can inline the parameters inside the query as documented here for SqlAlchemy: https://docs.sqlalchemy.org/en/20/faq/sqlexpressions.html#rendering-bound-parameters-inline However, this is not recommended for security reasons.

主要言語: Python
スター: 233
フォーク: 152
平均マージ: 21時間 5分
マージ済み PR（30日）: 10