fmd5sum

fmd5sum is a small, cross-platform command-line tool and Python library for calculating standard MD5 checksums. It processes multiple files concurrently, keeps output in input order, and returns a non-zero exit status if any file cannot be read.

Requirements

• Python 3.10 or newer
• Windows, macOS, or Linux
• No runtime dependencies outside the Python standard library

Installation

Install the command with pipx

This is the recommended option for end users because it keeps the command in an isolated environment:

pipx install fmd5sum
fmd5sum --help

Install with pip

python -m pip install fmd5sum

Install from source

Clone the repository, create a virtual environment, and install the project.

Windows PowerShell:

git clone https://github.com/jlchen5/fmd5sum.git
cd fmd5sum
py -m venv .venv
.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip
python -m pip install .

macOS or Linux:

git clone https://github.com/jlchen5/fmd5sum.git
cd fmd5sum
python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install .

Install with Conda

conda env create -f environment.yml
conda activate fmd5sum-env

Usage

Calculate one or more checksums:

fmd5sum file1.txt file2.img

Wildcard patterns are expanded by fmd5sum, including on Windows shells that do not expand them. Quote the pattern to get consistent behavior:

fmd5sum "images/*.jpg" "images/*.png"

Choose the worker count and read block size:

fmd5sum --workers 4 --blocksize 1048576 file1.iso file2.iso

The package can also be run without installing its console script:

python -m fmd5sum file1.txt file2.img

Run fmd5sum --help for all options. A successful run returns exit code 0. If one or more files cannot be processed, valid checksums are still printed and the command returns exit code 1. Invalid arguments return exit code 2.

Python API

from fmd5sum import md5sum, process_files

checksum = md5sum("path/to/file.ext")
status = process_files(["file1.txt", "file2.img"], max_workers=4)

md5sum() returns None and reports an error to stderr when the file cannot be read. It raises ValueError when blocksize is not positive. process_files() returns a command-style status code: 0 for success and 1 when at least one file failed.

Performance

Concurrency is applied across independent files. A single file is read by one worker because standard MD5 has a sequential chaining dependency. The default worker count is capped at eight and never exceeds the number of input files or available CPUs. Use --workers to tune throughput for the storage device.

Thread-based workers avoid multiprocessing startup and serialization overhead, especially on Windows and macOS. Actual throughput depends mainly on storage, file sizes, cache state, and the number of independent devices. Benchmark with representative data before increasing concurrency.

Development and packaging

Install development dependencies and run the tests:

python -m pip install -e ".[dev]"
python -m pytest

Build a platform-independent wheel and source archive:

python -m build

The artifacts are written to dist/. The generated py3-none-any wheel can be installed on Windows, macOS, and Linux with a supported Python version.

License

This project is licensed under the MIT License. See LICENSE.