xpatch-rs 0.3.0

High-performance delta compression library with automatic algorithm selection

xpatch-rs - Python Bindings

High-performance delta compression library for Python with automatic algorithm selection.

Installation

pip install xpatch-rs

Note: The package is named xpatch-rs on PyPI, but you import it as xpatch in your code.

Or build from source:

cd crates/xpatch-python
pip install maturin
maturin develop

Quick Start

import xpatch

# Create a delta patch
base = b"Hello, World!"
new = b"Hello, Rust!"
delta = xpatch.encode(tag=0, base_data=base, new_data=new)

# Apply the patch
reconstructed = xpatch.decode(base_data=base, delta=delta)
assert reconstructed == new

# Extract metadata tag
tag = xpatch.get_tag(delta)
print(f"Tag: {tag}")

API Reference

encode(tag, base_data, new_data, enable_zstd=True) -> bytes

Creates a delta patch between base_data and new_data.

Parameters:

• tag (int): Metadata tag to embed (0-15 for no overhead, larger values supported)
• base_data (bytes): Original data
• new_data (bytes): New data
• enable_zstd (bool): Enable zstd compression (default: True)

Returns: bytes - The encoded delta patch

decode(base_data, delta) -> bytes

Reconstructs new_data from base_data and a delta patch.

Parameters:

• base_data (bytes): Original data
• delta (bytes): Delta patch created by encode()

Returns: bytes - The reconstructed new data

Raises: ValueError if delta is invalid

get_tag(delta) -> int

Extracts the metadata tag from a delta patch without decoding.

Parameters:

• delta (bytes): Delta patch

Returns: int - The embedded tag

Raises: ValueError if delta is invalid

Performance

xpatch achieves exceptional compression ratios on real-world data:

• 99.8% compression on typical code changes
• 2 byte median delta for sequential edits
• Instant decoding (<1µs for most patches)
• 40-55 GB/s throughput for encoding

Use Cases

Perfect for:

• Version control systems
• Document synchronization
• Incremental backups
• Network-efficient updates
• Real-time collaborative editing

License

This project is dual-licensed:

• AGPL-3.0-or-later for open-source use
• Commercial license available at [e-mail]

See LICENSE-AGPL.txt and LICENSE-COMMERCIAL.txt for details.

Links

• GitHub Repository
• Demo Editor
• PyPI Package
• Rust Documentation