BinaryRecordReader

This package allow you to conveniently describe binary records for reading files. It require some extra dimensionally information for Julia Array fields inside a given type and creates a suitable read method.

Example usage

using BinaryRecordReader


struct TestS
    a::Int16
    b::Float64
    m::Vector{Int32}
    k::UInt8 
end

@construct_reader TestS (m=(500,),)

The first argument to the @construct_reader macro is the type we wish to generate a read method for, the second argument is a NameTuple to specify the dimensionality of the field names that require explicit definition.

In the above example, @construct_reader macro will construct a Base.read method for the TestS type with the size of the m having 500 elements. Concretely, the generated code will be:

function Base.read(io::IOStream, ::Type{TestS})
      a = read(io, Int16)
      b = read(io, Float64)
      m = read!(io, (Array{Int32,1})(undef, (500, 1)))
      k = read(io, UInt8)
      TestS(a, b, m, k)
  end

The generated read function can be explicitly called, as in

file_name="some_binary_file.dat"
open(file_name) do io
    read(io,TestS)
end

or implicitly, via read! while filling up an array, as in

myrecs=Vector{TestS}(undef,15)
read!(file_name,myrecs)

Other convenience methods

In addition to @construct_reader, which is intended for use at the top level, there are two convenience methods that can be called inside functions.

construct_reader_shallow is identical in form to @construct_reader, with added benefit of being executable inside any function/context.

construct_reader_deep has the added functionality where we can specify the dimensionality of internal types arrays. For example with following types

struct SimpleRecMat
    lat::Float32
    time::Matrix{Int32}
end

struct AnotherSimpleRecMat
    fun::UInt8
    funMat::Matrix{UInt16}
end

struct CompoundRecComplex
    long::Float32
    COG::UInt16
    funky::AnotherSimpleRecMat
    data::Matrix{SimpleRecMat}
end

We can fully a specify all three readers for each one on then by invoking

construct_reader_deep(CompoundRecComplex, (data=(5,5), AnotherSimpleRecMat__funMat=(2,2), SimpleRecMat__time =(3,3)))

Note that for internal type definition we have to be a explicit in the specifying the the internal field names via a Type__fieldName syntax.