deterministic-zip-go 6.0.3

deterministic-zip wrapped for usage with pip and/or python.

deterministic-zip

Simple (almost drop-in) replacement for zip that produces deterministic files.

Features

• Drop-in tool for zip
• Removes all metadata from files added for all platforms
• Creates immutable zip files
• Supports SOURCE_DATE_EPOCH for setting the modified date

Installation

Automatic install

bash <(curl -sS https://raw.githubusercontent.com/timo-reymann/deterministic-zip/main/installer)

Set the environment variable DETERMINISTIC_ZIP_VERSION to install a specific version

Manual

Linux (64-bit)

curl -LO https://github.com/timo-reymann/deterministic-zip/releases/download/$(curl -Lso /dev/null -w %{url_effective} https://github.com/timo-reymann/deterministic-zip/releases/latest | grep -o '[^/]*$')/deterministic-zip_linux-amd64 && \
chmod +x deterministic-zip_linux-amd64 && \
sudo mv deterministic-zip_linux-amd64 /usr/local/bin/deterministic-zip

Darwin (Intel)

brew

brew tap timo-reymann/deterministic-zip
brew install deterministic-zip

manual

curl -LO https://github.com/timo-reymann/deterministic-zip/releases/download/$(curl -Lso /dev/null -w %{url_effective} https://github.com/timo-reymann/deterministic-zip/releases/latest | grep -o '[^/]*$')/deterministic-zip_darwin-amd64 && \
chmod +x deterministic-zip_darwin-amd64 && \
sudo mv deterministic-zip_darwin-amd64 /usr/local/bin/deterministic-zip

Install with go

go install github.com/timo-reymann/deterministic-zip@latest

Install with pip(x)

Using pipx you can just use the following command use deterministic-zip as it is:

pipx install deterministic-zip-go

If you want to use it directly using the subprocess module you can install it with pip:

pip install deterministic-zip-go

And use the package like this:

import subprocess

from deterministic_zip_go import exec

# Run process and prefix stdout and stderr
exec.exec_with_templated_output(["--help"])

# Create a subprocess, specifying how to handle stdout, stderr
exec.create_subprocess(["--help"], stdout=subprocess.PIPE, stderr=subprocess.PIPE)

# Perform command with suppressed output and return finished proces instance,
# on that one can also check if the call was successfully
exec.exec_silently(["--version"])

Docker

Please check the Containerized section in Usage for more details.

Supported platforms

The following platforms are supported (and have prebuilt binaries / ready to use integration):

• Linux
  • 32-bit
  • 64-bit
  • ARM 64-bit
  • ARM 32-bit
• Darwin
  • 64-bit
  • ARM (M1/M2)
• Windows
  • ARM
  • 32-bit
  • 64-bit
• FreeBSD
  • 32-bit
  • 64-bit
  • ARM 64-bit
  • ARM 32-bit
• OpenBSD
  • 32-bit
  • 64-bit
• OCI compatible container engines (Docker, podman etc)
  • ARM
  • 64-bit
• CircleCI
• GitHub Actions

Where to find the latest release for your platform

Binaries

Binaries for all of these can be found on the latest release page.

Docker

For the docker image check the docker hub.

CI Provider

• CircleCI orb
• GitHub Action

Usage

Command Line

If you installed the binary via Releases, Install-Script or using go you can just run deterministic-zip as a command.

deterministic-zip -h

Containerized

Please be aware that the image contains just the binary, no OS, libs or anything else. It also runs as root to be able to zip files no matter the ownership, feel free to build your own images based on that as well.

Using the container directly

If you want to use the tool on a platform not supported yet or dont want to install the tool locally you can also mount your folder in /workspace which is the default working directory. Than you can just execute commands as you want to.

docker run -v $PWD:/workspace timoreymann/deterministic-zip:latest

Integrating into your CI image

If you want to integrate the tool directly into your build image, you can also utilize the auto updates from tools like renovatebot or dependabot. Using docker built in features you can just get the binary directly from the image.

FROM base-image:tag
# do your customizations
COPY --from=timoreymann/deterministic-zip:latest /deterministic-zip /usr/bin/deterministic-zip

Motivation

Why another zip-tool? What is this deterministic stuff?!

When we are talking about deterministic it means that the hash of the zip file won't change unless the contents of the zip file changes.

This means only the content, no metadata. You can achieve this with zip, yes.

The problem that still remains is that the order is almost unpredictable and zip is very platform specific, so you will end up with a bunch of crazy shell pipelines. And I am not even talking about windows at this point.

So this is where this tool comes in, it is intended to be a drop-in replacement for zip in your build process.

The use cases for this are primary:

• Zipping serverless code
• Backups or other files that get rsynced

Want to know more about the topic of deterministic/reproducible builds?

I can recommend the following resources:

• reproducible-builds.org
• Debian Wiki

Documentation

How reliable is it?

Of course, it is not as reliable as the battle-proven and billions of times executed zip.

Even though I am heavily relying on the go stdlib this software can of course have bugs. And you are welcome to report them and help make this even more stable. Of course there will be tests to cover most use cases but at the end this is still starting from scratch, so if you need advanced features or just dont feel comfortable about using this tool don't do it!

Differences between zip and deterministic-zip

Please see docs/differences

Contributing

I love your input! I want to make contributing to this project as easy and transparent as possible, whether it's:

• Reporting a bug
• Discussing the current state of the configuration
• Submitting a fix
• Proposing new features
• Becoming a maintainer

To get started please read the Contribution Guidelines.

Development

Requirements

• Go
• GNU make

Test

make test-coverage-report

Build

make build

Alternatives

As far as I know the following (GitHub) projects exist:

• bboe/deterministic_zip (Python)
  • You must list files explicitly
  • Changed order -> changed zip
  • You will need to install Python (no problem on Linux/Mac) and the package
• bitgenics/deterministic-zip (NodeJS/JavaScript)
  • Support for globs and ignores order
  • You need to install node.js, the package, and it has no cli interface
• orf/deterministic-zip (Rust)
  • has prebuilt binaries for all relevant platforms (and other can be built easily)
  • very basic, but you can customize compression (nice feature)

All in all they are just simply not what I needed. My favourite is Rust, because its just simply dropping in a binary. Something that's very convenient especially when it comes to Docker builds.

The main problem that all these solutions share is that it in my opinion cool things like excluding patterns, that I regularly use are simply not implemented, and i REALLY love glob patterns.

Credits

This whole project wouldnt be possible with the great work of the following libraries:

• glob by gobwas
• pflag by spf13
• go stdlib