Add Athena filter() function support to SQLAlchemy dialect

[laughingman7743]

Summary

Implements support for Amazon Athena's filter() function with lambda expressions in PyAthena's SQLAlchemy dialect.

Fixes #480

Features

• ✅ Basic filter() function compilation with lambda expressions
• ✅ Support for complex lambda conditions and nested field access
• ✅ Comprehensive error handling for invalid argument counts
• ✅ Type-safe implementation using isinstance checks
• ✅ Full test coverage with 7 test cases

Examples

# Basic filtering
select(func.filter(table.c.numbers, literal('x -> x > 0')))

# Complex conditions  
select(func.filter(table.c.data, literal('x -> x IS NOT NULL AND x > 5')))

# Nested field access
select(func.filter(table.c.events, literal("x -> x['timestamp'] > '2023-01-01'")))

# GitHub issue example
lambda_expr = literal("x -> x['timestamp'] <= '2023-10-10' AND x['timestamp'] >= '2023-10-01' AND x['action_count'] >= 2")
select(func.count(func.filter(table.c.actions, lambda_expr)))

Implementation Details

• File: pyathena/sqlalchemy/compiler.py
• Method: AthenaStatementCompiler.visit_filter_func()
• Tests: tests/pyathena/sqlalchemy/test_compiler.py::TestAthenaStatementCompiler

Test Plan

• All existing tests pass
• New comprehensive test suite (7 test cases)
• Lint and type checks pass
• Manual SQL generation verification

Generated SQL Examples

-- Basic filter
SELECT filter(table.numbers, x -> x > 0) FROM table

-- Complex lambda  
SELECT filter(table.data, x -> x IS NOT NULL AND x > 5) FROM table

-- GitHub issue example
SELECT count(filter(table.actions, x -> x['timestamp'] <= '2023-10-10' AND x['timestamp'] >= '2023-10-01' AND x['action_count'] >= 2)) FROM table

🤖 Generated with Claude Code