StaticHDF5

A simple, lightweight Julia interface to HDF5 files focused on array operations. It's main goal is to support juliac

Features

• Simple API for reading and writing arrays to HDF5 files
• Support for various data types (Float64, Float32, Int64, Int32, Int16, Int8, UInt64, UInt32, UInt16, UInt8, Bool)
• Group operations for organizing data
• Type-stable operations for --trim

Installation

using Pkg
Pkg.add("StaticHDF5")

Basic Usage

using StaticHDF5

# Create a file and write some arrays
file = create_file("example.h5")
write_object(file, "integers", [1, 2, 3, 4, 5])
write_object(file, "matrix", reshape(1:12, 3, 4))
close_file(file)

# Read arrays back
file_id = open_file("example.h5")
integers = read_object(file_id, "integers", Vector{Int64})
matrix = read_object(file_id, "matrix", Matrix{Int64})
close_file(file_id)

Working with Groups

# Create a file with groups
file = create_file("grouped_example.h5")

# Create a group and write data to it
group = create_group(file, "measurements")
write_object(group, "temperatures", [22.1, 22.3, 22.0, 21.8])

# Create a subgroup
subgroup = create_group(group, "day1")
write_object(subgroup, "morning", [20.1, 20.5, 21.0])

# Close groups and file
close_group(subgroup)
close_group(group)
close_file(file_id)

Type-Stable Reading

For --trim use StaticHDF5 has a way to pass in the expected return type, this makes it type stable

1. Using parametric array types (recommended):

   # For 1D arrays (vectors)
   vector = read_object(file_id, "vector", Vector{Float64})
   
   # For 2D arrays (matrices)
   matrix = read_object(file_id, "matrix", Matrix{Int})
   
   # For N-dimensional arrays
   tensor = read_object(file_id, "tensor", Array{Float32, 3})

The standard version of read_array without the return type is not stable because it's not possible to know what kind of array is in the file

Differences from HDF5.jl

Firstly we would like to thank the HDF5.jl developers for making such a featureful and powerful package. The goal for StaticHDF5.jl is not to replace it, but to have an interface that is compatible with trimming. This means cutting down on some of the features and also forcing the user to be explicit about the type they expect to load out of the hdf5 file.