Omnizip

Compression and archive toolkit for Ruby: LZMA/LZMA2, Zstandard, BZip2, PPMd7/8, Deflate/Deflate64 codecs; ZIP, 7z, RAR4/RAR5, TAR, CPIO, ISO, XAR, RPM, OLE, PAR2 formats; deterministic creation, AES encryption, PAR2 recovery records, format conversion and repair. Where a prebuilt binary exists the codecs and archives run on the omnizip-rs Rust engine — with zero compilation for the user; everywhere else the gem stays fully functional on its pure-Ruby core.

Installation

gem install omnizip          # picks a prebuilt Rust cdylib per platform

Supported prebuilt platforms: x86_64/aarch64/arm Linux (glibc + musl), arm64/x86_64 macOS, x64/arm64 Windows (ucrt). JRuby and TruffleRuby bind through the bundled ffi gem; MRI uses stdlib Fiddle. Escape hatches:

OMNIZIP_NO_RUST=1 — force the pure-Ruby core everywhere.
OMNIZIP_FFI_DYLIB=/path/to/libomnizip_ffi — load a specific build.
rake rust:build — build and vendor the cdylib from a checkout.

A Rust-side failure falls back to the Ruby core mid-operation, so acceleration is never worse than pure Ruby.

External code must require "omnizip". Files under lib/omnizip/** are internal — requiring them directly bypasses the autoload chain.

Command line

omnizip compress input.txt output.lzma --level 9
omnizip archive create backup.7z documents/ photos/
omnizip archive extract backup.7z output/
omnizip archive list backup.7z
omnizip archive verify backup.rar
omnizip archive repair damaged.rar
omnizip parity create backup.rar --percent 10
omnizip parity verify backup.par2
omnizip parity repair backup.par2
omnizip profile list
omnizip metadata archive.rar

Ruby API

One-liners

require "omnizip"

Omnizip.compress_file("input.txt", "output.zip")
Omnizip.extract_archive("archive.zip", "output/")
Omnizip.list_archive("archive.zip")

In-memory archives

Omnizip::Buffer creates and reads ZIP and 7z archives entirely in memory:

buffer = Omnizip::Buffer.create(:zip) do |archive|
  archive.add("readme.txt", "Hello World")
end
names = Omnizip::Backends.archive_entry_names(buffer.to_s)

Archive files

Omnizip::Zip::File.open("in.zip") { |z| z.extract("a.txt", "out.txt") }
Omnizip::Zip::File.open("out.zip", create: true) do |z|
  z.add_file("a.txt")
end

names = Omnizip::Backends.archive_entry_names("archive.rar")
data  = Omnizip::Backends.archive_read_entry("archive.rar", "doc.txt",
                                             password: "secret")

Codecs directly

compressed = Omnizip::Algorithms::Lzma.compress(data, level: 9)
plain      = Omnizip::Algorithms::Lzma.decompress(compressed)

Omnizip::Algorithms::Zstandard.compress(data, level: 19)
Omnizip::Algorithms::Bzip2.compress(data)
Omnizip::Algorithms::Ppmd7.compress(data, order: 8)

SDK-compatible LZMA

File.open("out.lzma", "wb") do |f|
  Omnizip::Algorithms::LZMA::Encoder.new(f, sdk_compatible: true)
    .encode_stream("Hello, World!")
end

Profiles, conversion, streaming, repair

Content-class profiles pick the codec automatically (Omnizip::Profile::DEFAULT and friends); Omnizip::Converter moves archives between formats; the chunked reader/writer processes inputs far larger than RAM; Omnizip::ArchiveRepairCommand routes structural repair. See the guides below.

Documentation

Installation · CLI usage · API usage
Compression algorithms · Profiles · Format converter · Progress, ETA, parallel · Performance profiler
Archive formats · RAR4/RAR5 guide · PAR2 parity archives · Encryption and checksums
Architecture · Rust backend · Compatibility · RAR write support

Development

bundle install
bundle exec rspec                       # suite
bundle exec rspec spec/omnizip/algorithms/lzma_spec.rb   # one file
bundle exec rubocop                     # lint
rake rust:build                         # vendor the Rust cdylib locally
rake docs:build                         # render docs/site via Coradoc
rake docs:links                         # lychee link check

Copyright and license

Copyright (c) Ribose Inc. Released under the MIT License; see the LICENSE file. The Rust engine (omnizip-rs) is MIT OR Apache-2.0.

omnizip

Runtime