tablediff-arrow 0.1.0

Fast, file-based diffs for Parquet/CSV/Arrow (local or S3) with keyed comparisons, per-column tolerances, and HTML/CSV reports—built on Apache Arrow.

tablediff-arrow

Fast, file-based diffs for Parquet/CSV/Arrow (local or S3) with keyed comparisons, per-column tolerances, and HTML/CSV reports—built on Apache Arrow.

Features

• Fast: Built on Apache Arrow for high-performance data processing
• Multiple Formats: Support for Parquet, CSV, and Arrow IPC files
• S3 Support: Read files directly from S3 (optional)
• Keyed Comparisons: Compare tables using one or more key columns
• Numeric Tolerances: Configure absolute and relative tolerances for numeric columns
• Rich Reports: Generate HTML and CSV reports with detailed differences
• Python 3.10+: Modern Python with type hints and clean APIs
• Well Tested: Comprehensive test suite with high coverage

Installation

pip install tablediff-arrow

For S3 support:

pip install tablediff-arrow[s3]

For development:

pip install -e ".[dev]"

Quick Start

Command Line Interface

Compare two Parquet files using id as the key column:

tablediff left.parquet right.parquet -k id

Compare with numeric tolerance:

tablediff left.csv right.csv -k id -t amount:0.01

Generate an HTML report:

tablediff left.parquet right.parquet -k id -o report.html

Compare S3 files:

tablediff s3://bucket/left.parquet s3://bucket/right.parquet -k id --s3

Python API

from tablediff_arrow import TableDiff

# Create a differ with key columns and tolerances
differ = TableDiff(
    key_columns=['id'],
    tolerance={'amount': 0.01},  # Absolute tolerance
    relative_tolerance={'price': 0.001}  # Relative tolerance (0.1%)
)

# Compare files
result = differ.compare_files('left.parquet', 'right.parquet')

# Print summary
print(result.summary())

# Check if there are differences
if result.has_differences:
    print(f"Found {result.changed_rows} changed rows")
    print(f"Found {result.left_only_rows} rows only in left")
    print(f"Found {result.right_only_rows} rows only in right")

# Generate reports
from tablediff_arrow.reports import generate_html_report, generate_csv_report

generate_html_report(result, 'report.html')
generate_csv_report(result, 'output_dir/', prefix='diff')

Usage Examples

Multiple Key Columns

Compare tables using composite keys:

tablediff left.parquet right.parquet -k year -k month -k product

differ = TableDiff(key_columns=['year', 'month', 'product'])
result = differ.compare_files('left.parquet', 'right.parquet')

Numeric Tolerances

Use absolute tolerance for monetary values:

tablediff left.csv right.csv -k id -t amount:0.01 -t balance:0.001

Use relative tolerance for percentages:

tablediff left.csv right.csv -k id -r rate:0.001 -r score:0.01

differ = TableDiff(
    key_columns=['id'],
    tolerance={'amount': 0.01, 'balance': 0.001},
    relative_tolerance={'rate': 0.001, 'score': 0.01}
)

Working with PyArrow Tables

import pyarrow as pa
from tablediff_arrow import TableDiff

# Create tables directly
left = pa.table({'id': [1, 2, 3], 'value': [10, 20, 30]})
right = pa.table({'id': [1, 2, 3], 'value': [10, 21, 30]})

# Compare
differ = TableDiff(key_columns=['id'])
result = differ.compare_tables(left, right)

print(result.summary())

S3 Files

import s3fs
from tablediff_arrow import TableDiff

# Create S3 filesystem
fs = s3fs.S3FileSystem()

# Compare S3 files
differ = TableDiff(key_columns=['id'])
result = differ.compare_files(
    's3://my-bucket/left.parquet',
    's3://my-bucket/right.parquet',
    filesystem=fs
)

CLI Options

Usage: tablediff [OPTIONS] LEFT RIGHT

  Compare two tables and generate diff reports.

Arguments:
  LEFT   Path to the left/source table file (local or s3://)
  RIGHT  Path to the right/target table file (local or s3://)

Options:
  -k, --key TEXT              Key column(s) for comparison (required, can be
                              specified multiple times)
  -t, --tolerance TEXT        Absolute tolerance for numeric columns
                              (format: column:value)
  -r, --relative-tolerance    Relative tolerance for numeric columns
                              (format: column:value)
  --left-format [parquet|csv|arrow]
                              Format of the left file
  --right-format [parquet|csv|arrow]
                              Format of the right file
  -o, --output TEXT           Output file path for HTML report
  --csv-output PATH           Output directory for CSV reports
  --s3                        Enable S3 filesystem support
  --help                      Show this message and exit.

Output Reports

HTML Report

The HTML report provides an interactive view of differences:

• Summary statistics (matched, changed, added, removed rows)
• Color-coded differences table
• Separate sections for left-only and right-only rows
• Change counts per column

CSV Reports

CSV output generates multiple files:

• {prefix}_summary.csv: Summary statistics
• {prefix}_changes.csv: Detailed changes with old and new values
• {prefix}_left_only.csv: Rows only in the left table
• {prefix}_right_only.csv: Rows only in the right table

Development

Setup

# Clone the repository
git clone https://github.com/psmman/tablediff-arrow.git
cd tablediff-arrow

# Install with development dependencies
pip install -e ".[dev]"

# Install pre-commit hooks
pre-commit install

Running Tests

# Run all tests
pytest

# Run with coverage
pytest --cov=tablediff_arrow --cov-report=html

# Run specific test file
pytest tests/test_compare.py

Code Quality

# Format code
black src tests

# Lint
ruff check src tests

# Type check
mypy src

Pre-commit Hooks

The project uses pre-commit hooks to ensure code quality:

• trailing-whitespace: Remove trailing whitespace
• end-of-file-fixer: Ensure files end with a newline
• check-yaml/json/toml: Validate config files
• black: Format Python code
• ruff: Lint Python code
• mypy: Type checking

Requirements

• Python 3.10 or higher
• pyarrow >= 14.0.0
• pandas >= 2.0.0
• click >= 8.0.0
• jinja2 >= 3.0.0
• s3fs >= 2023.0.0 (optional, for S3 support)

License

MIT License - see LICENSE file for details.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.