evm-log-father 0.1.0

Fast EVM log decoding library

evm-log-father

Fast EVM log decoding library with Python bindings.

Performance

400-500k logs/second with parallel decoding on large parquet files.

Features

• Decode Ethereum event logs using alloy's dynamic ABI
• Read logs from parquet files (multiple schema formats supported)
• Parallel decoding with rayon
• Python bindings via PyO3
• CLI for quick testing

Installation

Python (from PyPI)

pip install evm-log-father

Python (from source)

pip install maturin
maturin develop --features python

Rust

[dependencies]
evm-log-father = "0.1"

Usage

Python

from evm_log_father import EventSchema, decode_parquet

# Create schema from event signature
schema = EventSchema("Transfer(address indexed from, address indexed to, uint256 value)")

# Decode logs from parquet file
logs = decode_parquet("transfers.parquet", schema, parallel=True)

for log in logs:
    print(f"Block {log['block_number']}: {log['params']['from']} -> {log['params']['to']}")

Rust

use evm_log_father::{EventSchema, decode_parquet_parallel};

let schema = EventSchema::new("Transfer(address indexed from, address indexed to, uint256 value)")?;
let logs = decode_parquet_parallel("transfers.parquet", &schema)?;

for log in logs {
    println!("Block {}: {:?}", log.block_number, log.params);
}

CLI

# Decode logs and output JSON
evm-log-father decode \
  --parquet transfers.parquet \
  --event "Transfer(address indexed from, address indexed to, uint256 value)" \
  --output decoded.json \
  --parallel \
  --timing

# Show event info
evm-log-father info --event "Transfer(address indexed from, address indexed to, uint256 value)"

Parquet Schema Support

Flexible schema support for various parquet formats:

Column Names

Both snake_case and camelCase supported:

• block_number / blockNumber
• transaction_hash / transactionHash / tx_hash
• log_index / logIndex
• contract / address

Topics Format

• Individual columns: topic0, topic1, topic2, topic3
• List column: topics (Spark format)

Data Types

• block_number: u64 or i64
• log_index: u32, u64, or i64
• data: binary or hex string

Benchmarking

python examples/benchmark.py transfers.parquet

License

MIT