Skip to content

BioJulia/Libz.jl

Repository files navigation

Build Status Build status codecov.io

NOTE: If you are starting a new project on Julia 0.6 or later, it is recommended to use the CodecZlib.jl package instead. CodecZlib.jl and other packages offer more unified interfaces for a wide range of file formats.

This is yet another zlib interface for Julia. It's intended to replace the two prior zlib packages.

Both have shortcomings that this package aims to address, specifically:

  • Zlib.jl is very slow.
  • GZip.jl is not as slow as Zlib.jl, but still slower than it could to be.
  • GZip.jl only supports file I/O.
  • GZip.jl doesn't support reading/writing plain zlib data.

API

This library exports four stream types:

Type Description
ZlibInflateOutputStream write and decompress data
ZlibDeflateOutputStream write and compress data
ZlibInflateInputStream read and decompress data
ZlibDeflateInputStream read and compress data

These work like regular IO objects. Each takes as a parameter either in input or output source.

Examples

# read lines from a compressed file
for line in eachline(open("data.txt.gz") |> ZlibInflateInputStream)
    # do something...
end

# write compressed data to a file
io = open("data.txt.gz", "w")
stream = ZlibDeflateOutputStream(io)
for c in rand(UInt8, 10000)
    write(stream, c)
end
close(stream)  # this closes not only `stream` but also `io`

# pointlessly compress and decompress some data (use `read` on v0.5)
readbytes(rand(UInt8, 10000) |> ZlibDeflateInputStream |> ZlibInflateInputStream)

Other functions

There are convenience Libz.inflate(::Vector{UInt8}) and Libz.deflate(::Vector{UInt8}) functions that take a byte array and return another compressed or decompressed byte array.

Checksum functions are exposed as Libz.crc32(::Vector{UInt8}) and Libz.adler32(::Vector{UInt8}).

See BufferedStreams.jl for benchmarks of this library.

Low-level APIs are defined in src/lowlevel.jl. These constants and functions are not exported but available if necessary. At the moment, function wrappers are minimal but feel free to add and send functions you need as pull requests.