pycugrep 0.1.2

A GPU-accelerated grep tool built with cuDF.

PyCuGrep

PyCuGrep is a Python package that provides a GPU-accelerated grep utility, leveraging the power of NVIDIA GPUs and the cuDF library from RAPIDS. It's designed for rapidly searching large text files or datasets by offloading the pattern matching to the GPU, significantly speeding up operations compared to traditional CPU-based grep for suitable workloads.

Features

• GPU Accelerated: Utilizes cuDF for high-performance string matching on NVIDIA GPUs.
• Regex Support: Supports standard regular expressions for flexible pattern matching.
• Case Insensitive Search: Option to perform case-insensitive searches.
• Invert Match: Find lines that do not match the pattern.
• Line Numbers: Display line numbers for matched (or non-matched) lines.
• Count Only: Only output the count of matching lines.

Installation

To install pycugrep, you'll need a system with an NVIDIA GPU and CUDA installed. It's recommended to install via pip:

pip install pycugrep

Usage

PyCuGrep can be used directly from the command line after installation:

cugrep [OPTIONS] PATTERN FILE...

Example: Basic search

cugrep "error" /path/to/logfile.txt

Example: Case-insensitive search across multiple files

cugrep -i "warning" file1.txt file2.txt

Example: Count only, inverted match

cugrep -c -v "success" all_logs.txt

For more details on available options, run cugrep --help.

Development

For local development, after cloning the repository, navigate to the python/ directory and install in editable mode:

cd cugrep/python
pip install -e .

License

PyCuGrep is licensed under the Apache 2.0 License. See the LICENSE file for more details.