IterableData Development

Quick Setup

pip install -e ".[dev]"
pytest --verbose
ruff check iterable tests
ruff format iterable tests

Code Style

• Python 3.10+ with type hints where appropriate
• Max line length: 120 characters
• Use ruff for linting and formatting
• Double quotes for strings consistently
• Always use context managers for file operations
• Import order: standard library, third-party, local imports

Project Structure

• iterable/helpers/ - Utility functions (detect, schema, utils)
• iterable/datatypes/ - Format-specific implementations
• iterable/codecs/ - Compression codec implementations
• iterable/engines/ - Processing engines (DuckDB, internal)
• iterable/convert/ - Format conversion utilities
• iterable/pipeline/ - Data pipeline processing
• tests/ - Test suite (one test file per format/feature)

Import Patterns

• Main entry: from iterable.helpers.detect import open_iterable
• Format-specific: from iterable.datatypes.csv import CSVIterable
• Codecs: from iterable.codecs.gzipcodec import GZIPCodec
• Always use open_iterable() for user-facing code

File Handling

• Always use context managers: with open_iterable('file.csv') as source:
• Never call .close() when using with statements
• Reset iterators with .reset() method when needed
• Handle compression automatically via filename detection

Error Handling

• Format detection failures: provide helpful error messages
• Missing optional dependencies: raise clear ImportError with installation instructions
• Invalid file formats: raise appropriate exceptions (ValueError, TypeError)
• Always handle file I/O errors gracefully

Code Conventions

• Use open_iterable() for automatic format detection
• Prefer bulk operations (read_bulk, write_bulk) for performance
• Use DuckDB engine when appropriate (CSV, JSONL files)
• Handle encoding automatically via chardet or user specification

Pre-Commit Checks

Before committing:

1. Run tests: pytest --verbose
2. Run linter: ruff check iterable tests
3. Format code: ruff format iterable tests
4. Type check: mypy iterable (warnings allowed)

Quality Tools

• Security: bandit -r iterable -ll
• Dead code: vulture iterable --min-confidence 80
• Complexity: radon cc iterable --min B
• Coverage: pytest --cov=iterable --cov-report=html

Known Constraints

• DuckDB engine supports: CSV, JSONL, JSON formats and GZIP, ZStandard codecs
• Large files: use streaming (iterator interface) to avoid memory issues
• XML parsing: requires iterableargs={'tagname': 'item'}
• Some formats require optional dependencies (see pyproject.toml)