Description

Fixes an issue in awswrangler._sql_formatter._process_sql_params where double-colon type cast operators (e.g. col::text, col::timestamp, col::varchar, col::integer) are mistakenly parsed as parameter placeholders when a parameter name matches the cast target type name.

Root Cause

The parameter substitution regex _PATTERN was defined as re.compile(r":([A-Za-z0-9_]+)(?![A-Za-z0-9_])"). When a SQL query contains double colons :: (standard ANSI / PostgreSQL / Athena / Presto / Trino / CleanRooms cast syntax), the second colon was matched as a parameter placeholder. If params contained a key matching the type name (such as text in col::text), _process_sql_params substituted the cast operator, corrupting the SQL query to col:'value' and causing runtime SQL syntax errors.

Solution

Updated _PATTERN in awswrangler/_sql_formatter.py with a negative lookbehind (?<!:) to ensure single-colon parameters (e.g., :id) are matched without matching double-colon cast operators (::type):

_PATTERN = re.compile(r"(?<!:):([A-Za-z0-9_]+)(?![A-Za-z0-9_])")

Validation & Testing

• Added unit test test_process_sql_params_double_colon_cast in tests/unit/test_sql_params_formatter.py asserting that double-colon type casts are preserved while parameter placeholders are correctly substituted.
• Confirmed test failure on upstream/main baseline before applying fix.
• Ran quality checks:
  • pytest tests/unit/test_sql_params_formatter.py (Passed 13/13)
  • ruff format --check . (Passed)
  • ruff check . (Passed)
  • mypy awswrangler/_sql_formatter.py tests/unit/test_sql_params_formatter.py (Success)
  • git diff --check (Clean)

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.