Work Item / Issue Reference

AB#34927

Summary

This pull request adds support for retrieving column metadata from SQL Server using the ODBC SQLColumns API, and exposes this functionality to Python through the columns() method on the cursor class. The implementation includes both backend C++ bindings and a high-level Python interface, as well as careful handling of cursor state and result set mapping.

The most important changes are:

Python interface:

• Added a new columns() method to the cursor class in mssql_python/cursor.py to retrieve detailed column metadata for tables using the ODBC SQLColumns function. This method includes extensive documentation, result mapping, and warnings about potentially expensive queries. It also wraps and overrides the fetch methods to provide specialized column mappings for the result set.

• Updated the execute() method in cursor.py to restore the original fetch methods if they were previously overridden, ensuring correct behavior after using columns().

C++ bindings and backend:

• Added the SQLColumnsFunc typedef, function pointer, and symbol loading for the SQLColumnsW ODBC API in ddbc_bindings.h and ddbc_bindings.cpp. This ensures the backend can call the ODBC function for column metadata.

• Implemented the SQLColumns_wrap C++ function in ddbc_bindings.cpp to call the ODBC SQLColumns API, handling both Windows and Unix platforms, and converting Python objects to wide strings as needed.

• Exposed a new DDBCSQLColumns method in the Python extension module, allowing Python code to call the new backend functionality.