using HDF5
using CRC32c
using H5Zblosc
using Test

gatherf(dst_buf, dst_buf_bytes_used, op_data) = HDF5.API.herr_t(0)
gatherf_bad(dst_buf, dst_buf_bytes_used, op_data) = HDF5.API.herr_t(-1)
gatherf_data(dst_buf, dst_buf_bytes_used, op_data) = HDF5.API.herr_t((op_data == 9) - 1)

function scatterf(src_buf, src_buf_bytes_used, op_data)
    A = [1, 2, 3, 4]
    unsafe_store!(src_buf, pointer(A))
    unsafe_store!(src_buf_bytes_used, sizeof(A))
    return HDF5.API.herr_t(0)
end
scatterf_bad(src_buf, src_buf_bytes_used, op_data) = HDF5.API.herr_t(-1)
function scatterf_data(src_buf, src_buf_bytes_used, op_data)
    A = [1, 2, 3, 4]
    unsafe_store!(src_buf, pointer(A))
    unsafe_store!(src_buf_bytes_used, sizeof(A))
    return HDF5.API.herr_t((op_data == 9) - 1)
end

@testset "plain" begin

    # Create a new file
    fn = tempname()
    f = h5open(fn, "w")
    @test isopen(f)
    # Write scalars
    f["Float64"] = 3.2
    f["Int16"] = Int16(4)
    # compression of empty array (issue #246)
    f["compressedempty", shuffle=true, deflate=4] = Int64[]
    # compression of zero-dimensional array (pull request #445)
    f["compressed_zerodim", shuffle=true, deflate=4] = fill(Int32(42), ())
    f["bloscempty", blosc=4] = Int64[]
    # test creation of an anonymouse dataset
    f[nothing] = 5
    # Create arrays of different types
    A = randn(3, 5)
    write(f, "Afloat64", convert(Matrix{Float64}, A))
    write(f, "Afloat32", convert(Matrix{Float32}, A))
    Ai = rand(1:20, 2, 4)
    write(f, "Aint8", convert(Matrix{Int8}, Ai))
    f["Aint16"] = convert(Matrix{Int16}, Ai)
    write(f, "Aint32", convert(Matrix{Int32}, Ai))
    write(f, "Aint64", convert(Matrix{Int64}, Ai))
    write(f, "Auint8", convert(Matrix{UInt8}, Ai))
    write(f, "Auint16", convert(Matrix{UInt16}, Ai))

    # test writing multiple variable (issue #599)
    write(f, "Auint32", convert(Matrix{UInt32}, Ai), "Auint64", convert(Matrix{UInt64}, Ai))

    # Arrays of bools (pull request #540)
    Abool = [false, true, false]
    write(f, "Abool", Abool)

    salut = "Hi there"
    ucode = "uniçº∂e"
    write(f, "salut", salut)
    write(f, "ucode", ucode)
    # Manually write a variable-length string (issue #187)
    let
        dtype = HDF5.Datatype(HDF5.API.h5t_copy(HDF5.API.H5T_C_S1))
        HDF5.API.h5t_set_size(dtype, HDF5.API.H5T_VARIABLE)
        HDF5.API.h5t_set_cset(dtype, HDF5.cset(typeof(salut)))
        dspace = dataspace(salut)
        dset = create_dataset(f, "salut-vlen", dtype, dspace)
        GC.@preserve salut begin
            HDF5.API.h5d_write(
                dset,
                dtype,
                HDF5.API.H5S_ALL,
                HDF5.API.H5S_ALL,
                HDF5.API.H5P_DEFAULT,
                [pointer(salut)]
            )
        end
    end
    # Arrays of strings
    salut_split = ["Hi", "there"]
    write(f, "salut_split", salut_split)
    salut_2d = ["Hi" "there"; "Salut" "friend"]
    write(f, "salut_2d", salut_2d)
    # Arrays of strings as vlen
    vlen = HDF5.VLen(salut_split)
    write_dataset(f, "salut_vlen", vlen)
    # Arrays of scalars as vlen
    vlen_int = [[3], [1], [4]]
    vleni = HDF5.VLen(vlen_int)
    write_dataset(f, "int_vlen", vleni)
    write_attribute(f["int_vlen"], "vlen_attr", vleni)
    # Empty arrays
    empty = UInt32[]
    write(f, "empty", empty)
    write(f, nothing, empty)
    # Empty strings
    empty_string = ""
    write(f, "empty_string", empty_string)
    # Empty array of strings
    empty_string_array = String[]
    write(f, "empty_string_array", empty_string_array)
    # Array of empty string
    empty_array_of_strings = [""]
    write(f, "empty_array_of_strings", empty_array_of_strings)
    # attributes
    species = [["N", "C"]; ["A", "B"]]
    attributes(f)["species"] = species
    @test read(attributes(f)["species"]) == species
    @test attributes(f)["species"][] == species
    C∞ = 42
    attributes(f)["C∞"] = C∞
    dset = f["salut"]
    @test !isempty(dset)
    label = "This is a string"
    attributes(dset)["typeinfo"] = label
    @test read(attributes(dset)["typeinfo"]) == label
    @test attributes(dset)["typeinfo"][] == label
    @test dset["typeinfo"][] == label
    close(dset)
    # Scalar reference values in attributes
    attributes(f)["ref_test"] = HDF5.Reference(f, "empty_array_of_strings")
    @test read(attributes(f)["ref_test"]) === HDF5.Reference(f, "empty_array_of_strings")
    # Group
    g = create_group(f, "mygroup")
    # Test dataset with compression
    R = rand(1:20, 20, 40)
    g["CompressedA", chunk=(5, 6), shuffle=true, deflate=9] = R
    g["BloscA", chunk=(5, 6), shuffle=true, blosc=9] = R
    close(g)
    # Copy group containing dataset
    copy_object(f, "mygroup", f, "mygroup2")
    # Copy dataset
    g = create_group(f, "mygroup3")
    copy_object(f["mygroup/CompressedA"], g, "CompressedA")
    copy_object(f["mygroup/BloscA"], g, "BloscA")
    close(g)
    # Writing hyperslabs
    dset = create_dataset(
        f, "slab", datatype(Float64), dataspace(20, 20, 5); chunk=(5, 5, 1)
    )
    Xslab = randn(20, 20, 5)
    for i in 1:5
        dset[:, :, i] = Xslab[:, :, i]
    end
    dset = create_dataset(
        f, nothing, datatype(Float64), dataspace(20, 20, 5); chunk=(5, 5, 1)
    )
    dset[:, :, :] = 3.0
    # More complex hyperslab and assignment with "incorrect" types (issue #34)
    d = create_dataset(f, "slab2", datatype(Float64), ((10, 20), (100, 200)); chunk=(1, 1))
    d[:, :] = 5
    d[1, 1] = 4
    # 1d indexing
    d = create_dataset(f, "slab3", datatype(Int), ((10,), (-1,)); chunk=(5,))
    @test d[:] == zeros(Int, 10)
    d[3:5] = 3:5
    # Create a dataset designed to be deleted
    f["deleteme"] = 17.2
    close(f)
    @test !isopen(f)
    # Test the h5read/write interface, with attributes
    W = copy(reshape(1:120, 15, 8))
    Wa = Dict("a" => 1, "b" => 2)
    h5write(fn, "newgroup/W", W)
    h5writeattr(fn, "newgroup/W", Wa)

    # Read the file back in
    fr = h5open(fn)
    x = read(fr, "Float64")
    @test x == 3.2 && isa(x, Float64)
    y = read(fr, "Int16")
    @test y == 4 && isa(y, Int16)
    zerodim = read(fr, "compressed_zerodim")
    @test zerodim == 42 && isa(zerodim, Int32)
    bloscempty = read(fr, "bloscempty")
    @test bloscempty == Int64[] && isa(bloscempty, Vector{Int64})
    Af32 = read(fr, "Afloat32")
    @test convert(Matrix{Float32}, A) == Af32
    @test eltype(Af32) == Float32
    Af64 = read(fr, "Afloat64")
    @test convert(Matrix{Float64}, A) == Af64
    @test eltype(Af64) == Float64
    @test eltype(fr["Afloat64"]) == Float64  # issue 167
    Ai8 = read(fr, "Aint8")
    @test Ai == Ai8
    @test eltype(Ai8) == Int8
    Ai16 = read(fr, "Aint16")
    @test Ai == Ai16
    @test eltype(Ai16) == Int16
    Ai32 = read(fr, "Aint32")
    @test Ai == Ai32
    @test eltype(Ai32) == Int32
    Ai64 = read(fr, "Aint64")
    @test Ai == Ai64
    @test eltype(Ai64) == Int64
    Ai8 = read(fr, "Auint8")
    @test Ai == Ai8
    @test eltype(Ai8) == UInt8
    Ai16 = read(fr, "Auint16")
    @test Ai == Ai16
    @test eltype(Ai16) == UInt16
    Ai32 = read(fr, "Auint32")
    @test Ai == Ai32
    @test eltype(Ai32) == UInt32
    Ai64 = read(fr, "Auint64")
    @test Ai == Ai64
    @test eltype(Ai64) == UInt64

    Abool_read = read(fr, "Abool")
    @test Abool_read == Abool
    @test eltype(Abool_read) == Bool

    salutr = read(fr, "salut")
    @test salut == salutr
    salutr = read(fr, "salut-vlen")
    @test salut == salutr
    ucoder = read(fr, "ucode")
    @test ucode == ucoder
    salut_splitr = read(fr, "salut_split")
    @test salut_splitr == salut_split
    salut_2dr = read(fr, "salut_2d")
    @test salut_2d == salut_2dr
    salut_vlenr = read(fr, "salut_vlen")
    @test HDF5.vlen_get_buf_size(fr["salut_vlen"]) == 7
    @test HDF5.API.h5d_get_access_plist(fr["salut-vlen"]) != 0
    #@test salut_vlenr == salut_split
    vlen_intr = read(fr, "int_vlen")
    @test vlen_intr == vlen_int
    vlen_attrr = read(fr["int_vlen"]["vlen_attr"])
    @test vlen_attrr == vlen_int
    Rr = read(fr, "mygroup/CompressedA")
    @test Rr == R
    Rr2 = read(fr, "mygroup2/CompressedA")
    @test Rr2 == R
    Rr3 = read(fr, "mygroup3/CompressedA")
    @test Rr3 == R
    Rr4 = read(fr, "mygroup/BloscA")
    @test Rr4 == R
    Rr5 = read(fr, "mygroup2/BloscA")
    @test Rr5 == R
    Rr6 = read(fr, "mygroup3/BloscA")
    @test Rr6 == R
    dset = fr["mygroup/CompressedA"]
    @test HDF5.get_chunk(dset) == (5, 6)
    @test HDF5.name(dset) == "/mygroup/CompressedA"
    dset2 = fr["mygroup/BloscA"]
    @test HDF5.get_chunk(dset2) == (5, 6)
    @test HDF5.name(dset2) == "/mygroup/BloscA"
    Xslabr = read(fr, "slab")
    @test Xslabr == Xslab
    Xslabr = h5read(fn, "slab", (:, :, :))  # issue #87
    @test Xslabr == Xslab
    Xslab2r = read(fr, "slab2")
    target = fill(5, 10, 20)
    target[1] = 4
    @test Xslab2r == target
    dset = fr["slab3"]
    @test dset[3:5] == [3:5;]
    emptyr = read(fr, "empty")
    @test isempty(emptyr)
    empty_stringr = read(fr, "empty_string")
    @test empty_stringr == empty_string
    empty_string_arrayr = read(fr, "empty_string_array")
    @test empty_string_arrayr == empty_string_array
    empty_array_of_stringsr = read(fr, "empty_array_of_strings")
    @test empty_array_of_stringsr == empty_array_of_strings
    @test read_attribute(fr, "species") == species
    @test read_attribute(fr, "C∞") == C∞
    dset = fr["salut"]
    @test read_attribute(dset, "typeinfo") == label
    close(dset)
    # Test ref-based reading
    Aref = fr["Afloat64"]
    sel = (2:3, 1:2:5)
    Asub = Aref[sel...]
    @test Asub == A[sel...]
    close(Aref)
    # Test iteration, name, and parent
    for obj in fr
        @test HDF5.filename(obj) == fn
        n = HDF5.name(obj)
        p = parent(obj)
    end
    # Test reading multiple vars at once
    z = read(fr, "Float64", "Int16")
    @test z == (3.2, 4)
    @test typeof(z) == Tuple{Float64,Int16}
    # Test reading entire file at once
    z = read(fr)
    @test z["Float64"] == 3.2
    close(fr)

    # Test object deletion
    fr = h5open(fn, "r+")
    @test haskey(fr, "deleteme")
    delete_object(fr, "deleteme")
    @test !haskey(fr, "deleteme")
    close(fr)

    # Test object move
    h5open(fn, "r+") do io
        io["moveme"] = [1, 2, 3]
        create_group(io, "moveto")
    end

    h5open(fn, "r+") do io
        @test haskey(io, "moveme")
        @test haskey(io, "moveto") && !haskey(io, "moveto/moveme")
        move_link(io, "moveme", io["moveto"])
        @test haskey(io, "moveto/moveme") && !haskey(io, "moveme")
    end

    # Test the h5read interface
    Wr = h5read(fn, "newgroup/W")
    @test Wr == W
    rng = (2:3:15, 3:5)
    Wr = h5read(fn, "newgroup/W", rng)
    @test Wr == W[rng...]
    War = h5readattr(fn, "newgroup/W")
    @test War == Wa

    # issue #618
    # Test that invalid writes treat implicit creation as a transaction, cleaning up the partial
    # operation
    hid = h5open(fn, "w")
    A = rand(3, 3)'
    @test !haskey(hid, "A")
    @test_throws ArgumentError write(hid, "A", A)
    @test !haskey(hid, "A")
    dset = create_dataset(hid, "attr", datatype(Int), dataspace(0))
    @test !haskey(attributes(dset), "attr")
    # broken test - writing attributes does not check that the stride is correct
    @test_skip @test_throws ArgumentError write(dset, "attr", A)
    @test !haskey(attributes(dset), "attr")
    close(hid)

    # more do syntax
    h5open(fn, "w") do fid
        g = create_group(fid, "mygroup")
        write(g, "x", 3.2)
    end

    fid = h5open(fn, "r")
    @test keys(fid) == ["mygroup"]
    g = fid["mygroup"]
    @test keys(g) == ["x"]
    close(g)
    close(fid)
    rm(fn)

    # more do syntax: atomic rename version
    tmpdir = mktempdir()
    outfile = joinpath(tmpdir, "test.h5")

    # create a new file
    h5rewrite(outfile) do fid
        g = create_group(fid, "mygroup")
        write(g, "x", 3.3)
    end
    @test length(readdir(tmpdir)) == 1
    h5open(outfile, "r") do fid
        @test keys(fid) == ["mygroup"]
        @test keys(fid["mygroup"]) == ["x"]
    end

    # fail to overwrite
    @test_throws ErrorException h5rewrite(outfile) do fid
        g = create_group(fid, "mygroup")
        write(g, "oops", 3.3)
        error("failed")
    end
    @test length(readdir(tmpdir)) == 1
    h5open(outfile, "r") do fid
        @test keys(fid) == ["mygroup"]
        @test keys(fid["mygroup"]) == ["x"]
    end

    # overwrite
    h5rewrite(outfile) do fid
        g = create_group(fid, "mygroup")
        write(g, "y", 3.3)
    end
    @test length(readdir(tmpdir)) == 1
    h5open(outfile, "r") do fid
        @test keys(fid) == ["mygroup"]
        @test keys(fid["mygroup"]) == ["y"]
    end
    rm(tmpdir; recursive=true)

    test_files = joinpath(@__DIR__, "test_files")

    d = h5read(joinpath(test_files, "compound.h5"), "/data")
    @test typeof(d[1]) == NamedTuple{
        (:wgt, :xyz, :uvw, :E),Tuple{Float64,Array{Float64,1},Array{Float64,1},Float64}
    }

    # get-datasets
    fn = tempname()
    fd = h5open(fn, "w")
    fd["level_0"] = [1, 2, 3]
    grp = create_group(fd, "mygroup")
    fd["mygroup/level_1"] = [4, 5]
    grp2 = create_group(grp, "deep_group")
    fd["mygroup/deep_group/level_2"] = [6.0, 7.0]
    datasets = HDF5.get_datasets(fd)
    @test sort(map(HDF5.name, datasets)) ==
        sort(["/level_0", "/mygroup/deep_group/level_2", "/mygroup/level_1"])
    close(fd)
    rm(fn)

    # File creation and access property lists
    fid = h5open(
        fn,
        "w";
        userblock=1024,
        libver_bounds=(HDF5.API.H5F_LIBVER_EARLIEST, HDF5.API.H5F_LIBVER_LATEST)
    )
    write(fid, "intarray", [1, 2, 3])
    close(fid)
    h5open(
        fn, "r"; libver_bounds=(HDF5.API.H5F_LIBVER_EARLIEST, HDF5.API.H5F_LIBVER_LATEST)
    ) do fid
        intarray = read(fid, "intarray")
        @test intarray == [1, 2, 3]
    end

    # Test null terminated ASCII string (e.g. exported by h5py) #332
    h5open(joinpath(test_files, "nullterm_ascii.h5"), "r") do fid
        str = read(fid["test"])
        @test str == "Hello World"
    end

    @test HDF5.unpad(UInt8[0x43, 0x43, 0x41], 1) == "CCA"

    # Test the h5read/write interface with a filename as a first argument, when
    # the file does not exist
    rm(fn)
    h5write(fn, "newgroup/W", W)
    Wr = h5read(fn, "newgroup/W")
    @test Wr == W
    close(f)
    rm(fn)

    # Test dataspace convenience versions of create_dataset
    try
        h5open(fn, "w") do f
            create_dataset(f, "test", Int, (128, 32))
            create_dataset(f, "test2", Float64, 128, 64)
            @test size(f["test"]) == (128, 32)
            @test size(f["test2"]) == (128, 64)
        end
    finally
        rm(fn)
    end

    @testset "h5d_fill" begin
        val = 5
        h5open(fn, "w") do f
            d = create_dataset(f, "dataset", datatype(Int), dataspace(6, 6); chunk=(2, 3))
            buf = Array{Int,2}(undef, (6, 6))
            dtype = datatype(Int)
            HDF5.API.h5d_fill(Ref(val), dtype, buf, datatype(Int), dataspace(d))
            @test all(buf .== 5)
            HDF5.API.h5d_write(
                d, dtype, HDF5.API.H5S_ALL, HDF5.API.H5S_ALL, HDF5.API.H5P_DEFAULT, buf
            )
        end
        h5open(fn, "r") do f
            @test all(f["dataset"][:, :] .== 5)
        end
        rm(fn)
    end # testset "Test h5d_fill

    @testset "h5d_gather" begin
        src_buf = rand(Int, (4, 4))
        dst_buf = Array{Int,2}(undef, (4, 4))
        h5open(fn, "w") do f
            d = create_dataset(f, "dataset", datatype(Int), dataspace(4, 4); chunk=(2, 2))
            @test isnothing(
                HDF5.API.h5d_gather(
                    dataspace(d),
                    src_buf,
                    datatype(Int),
                    sizeof(dst_buf),
                    dst_buf,
                    C_NULL,
                    C_NULL
                )
            )
            @test src_buf == dst_buf
            gatherf_ptr = @cfunction(
                gatherf, HDF5.API.herr_t, (Ptr{Nothing}, Csize_t, Ptr{Nothing})
            )
            @test isnothing(
                HDF5.API.h5d_gather(
                    dataspace(d),
                    src_buf,
                    datatype(Int),
                    sizeof(dst_buf) ÷ 2,
                    dst_buf,
                    gatherf_ptr,
                    C_NULL
                )
            )
            gatherf_bad_ptr = @cfunction(
                gatherf_bad, HDF5.API.herr_t, (Ptr{Nothing}, Csize_t, Ptr{Nothing})
            )
            @test_throws HDF5.API.H5Error HDF5.API.h5d_gather(
                dataspace(d),
                src_buf,
                datatype(Int),
                sizeof(dst_buf) ÷ 2,
                dst_buf,
                gatherf_bad_ptr,
                C_NULL
            )
            gatherf_data_ptr = @cfunction(
                gatherf_data, HDF5.API.herr_t, (Ptr{Nothing}, Csize_t, Ref{Int})
            )
            @test isnothing(
                HDF5.API.h5d_gather(
                    dataspace(d),
                    src_buf,
                    datatype(Int),
                    sizeof(dst_buf) ÷ 2,
                    dst_buf,
                    gatherf_data_ptr,
                    Ref(9)
                )
            )
            @test_throws HDF5.API.H5Error HDF5.API.h5d_gather(
                dataspace(d),
                src_buf,
                datatype(Int),
                sizeof(dst_buf) ÷ 2,
                dst_buf,
                gatherf_data_ptr,
                10
            )
        end
        rm(fn)
    end

    @testset "h5d_scatter" begin
        h5open(fn, "w") do f
            dst_buf = Array{Int,2}(undef, (4, 4))
            d = create_dataset(f, "dataset", datatype(Int), dataspace(4, 4); chunk=(2, 2))
            scatterf_ptr = @cfunction(
                scatterf, HDF5.API.herr_t, (Ptr{Ptr{Nothing}}, Ptr{Csize_t}, Ptr{Nothing})
            )
            @test isnothing(
                HDF5.API.h5d_scatter(
                    scatterf_ptr, C_NULL, datatype(Int), dataspace(d), dst_buf
                )
            )
            scatterf_bad_ptr = @cfunction(
                scatterf_bad,
                HDF5.API.herr_t,
                (Ptr{Ptr{Nothing}}, Ptr{Csize_t}, Ptr{Nothing})
            )
            @test_throws HDF5.API.H5Error HDF5.API.h5d_scatter(
                scatterf_bad_ptr, C_NULL, datatype(Int), dataspace(d), dst_buf
            )
            scatterf_data_ptr = @cfunction(
                scatterf_data, HDF5.API.herr_t, (Ptr{Ptr{Int}}, Ptr{Csize_t}, Ref{Int})
            )
            @test isnothing(
                HDF5.API.h5d_scatter(
                    scatterf_data_ptr, Ref(9), datatype(Int), dataspace(d), dst_buf
                )
            )
        end
        rm(fn)
    end

    # Test that switching time tracking off results in identical files
    fn1 = tempname()
    fn2 = tempname()
    h5open(fn1, "w") do f
        f["x", obj_track_times=false] = [1, 2, 3]
    end
    sleep(1)
    h5open(fn2, "w") do f
        f["x", obj_track_times=false] = [1, 2, 3]
    end
    @test open(crc32c, fn1) == open(crc32c, fn2)
    rm(fn1)
    rm(fn2)
end # testset plain

@testset "complex" begin
    HDF5.enable_complex_support()

    fn = tempname()
    f = h5open(fn, "w")

    f["ComplexF64"] = 1.0 + 2.0im
    attributes(f["ComplexF64"])["ComplexInt64"] = 1im

    Acmplx = rand(ComplexF64, 3, 5)
    write(f, "Acmplx64", convert(Matrix{ComplexF64}, Acmplx))
    write(f, "Acmplx32", convert(Matrix{ComplexF32}, Acmplx))

    dset = create_dataset(
        f, "Acmplx64_hyperslab", datatype(Complex{Float64}), dataspace(Acmplx)
    )
    for i in 1:size(Acmplx, 2)
        dset[:, i] = Acmplx[:, i]
    end

    HDF5.disable_complex_support()
    @test_throws ErrorException f["_ComplexF64"] = 1.0 + 2.0im
    @test_throws ErrorException write(f, "_Acmplx64", convert(Matrix{ComplexF64}, Acmplx))
    @test_throws ErrorException write(f, "_Acmplx32", convert(Matrix{ComplexF32}, Acmplx))
    HDF5.enable_complex_support()

    close(f)

    fr = h5open(fn)
    z = read(fr, "ComplexF64")
    @test z == 1.0 + 2.0im && isa(z, ComplexF64)
    z_attrs = attributes(fr["ComplexF64"])
    @test read(z_attrs["ComplexInt64"]) == 1im

    Acmplx32 = read(fr, "Acmplx32")
    @test convert(Matrix{ComplexF32}, Acmplx) == Acmplx32
    @test eltype(Acmplx32) == ComplexF32
    Acmplx64 = read(fr, "Acmplx64")
    @test convert(Matrix{ComplexF64}, Acmplx) == Acmplx64
    @test eltype(Acmplx64) == ComplexF64

    dset = fr["Acmplx64_hyperslab"]
    Acmplx64_hyperslab = zeros(eltype(dset), size(dset))
    for i in 1:size(dset, 2)
        Acmplx64_hyperslab[:, i] = dset[:, i]
    end
    @test convert(Matrix{ComplexF64}, Acmplx) == Acmplx64_hyperslab

    HDF5.disable_complex_support()
    z = read(fr, "ComplexF64")
    @test isa(z, NamedTuple{(:r, :i),Tuple{Float64,Float64}})

    Acmplx32 = read(fr, "Acmplx32")
    @test eltype(Acmplx32) == NamedTuple{(:r, :i),Tuple{Float32,Float32}}
    Acmplx64 = read(fr, "Acmplx64")
    @test eltype(Acmplx64) == NamedTuple{(:r, :i),Tuple{Float64,Float64}}

    close(fr)

    HDF5.enable_complex_support()
end

# test strings with null and undefined references
@testset "undefined and null" begin
    fn = tempname()
    f = h5open(fn, "w")

    # don't silently truncate data
    @test_throws ArgumentError write(f, "test", ["hello", "there", "\0"])
    @test_throws ArgumentError write(f, "trunc1", "\0")
    @test_throws ArgumentError write(f, "trunc2", "trunc\0ateme")

    # test writing uninitialized string arrays
    undefstrarr = similar(Vector(1:3), String) # strs = String[#undef, #undef, #undef]
    @test_throws UndefRefError write(f, "undef", undefstrarr)

    close(f)
    rm(fn)
end # testset null and undefined

# test writing abstract arrays
@testset "abstract arrays" begin

    # test writing reinterpreted data
    fn = tempname()
    try
        h5open(fn, "w") do f
            data = reinterpret(UInt8, [true, false, false])
            write(f, "reinterpret array", data)
        end

        @test h5open(fn, "r") do f
            read(f, "reinterpret array")
        end == UInt8[0x01, 0x00, 0x00]
    finally
        rm(fn)
    end

    # don't silently fail for arrays with a different stride
    fn = tempname()
    try
        data = rand(UInt16, 2, 3)
        pdv_data = PermutedDimsArray(data, (2, 1))

        @test_throws ArgumentError h5write(fn, "pdv_data", pdv_data)
    finally
        rm(fn)
    end

    # test alignment
    fn = tempname()
    h5open(fn, "w"; alignment=(0, 8)) do fid
        fid["x"] = zeros(10, 10)
    end
end # writing abstract arrays

# issue #705
@testset "empty and 0-size arrays" begin
    fn = tempname()
    hfile = h5open(fn, "w")

    # Write datasets with various 0-sizes
    write(hfile, "empty", HDF5.EmptyArray{Int64}()) # HDF5 empty
    write(hfile, "zerodim", fill(1.0π))   # 0-dimensional
    write(hfile, "zerovec", zeros(0))     # 1-dimensional, size 0
    write(hfile, "zeromat", zeros(0, 0))  # 2-dimensional, size 0
    write(hfile, "zeromat2", zeros(0, 1)) # 2-dimensional, size 0 with non-zero axis
    dempty = hfile["empty"]
    dzerodim = hfile["zerodim"]
    dzerovec = hfile["zerovec"]
    dzeromat = hfile["zeromat"]
    dzeromat2 = hfile["zeromat2"]

    # Test that eltype is preserved (especially for EmptyArray)
    @test eltype(dempty) == Int64
    @test eltype(dzerodim) == Float64
    @test eltype(dzerovec) == Float64
    @test eltype(dzeromat) == Float64
    @test eltype(dzeromat2) == Float64
    # Test sizes are as expected
    @test size(dempty) == ()
    @test size(dzerovec) == (0,)
    @test size(dzeromat) == (0, 0)
    @test size(dzeromat2) == (0, 1)
    @test HDF5.isnull(dempty)
    @test !HDF5.isnull(dzerovec)
    @test !HDF5.isnull(dzeromat)
    @test !HDF5.isnull(dzeromat2)
    # Reading back must preserve emptiness
    @test read(dempty) isa HDF5.EmptyArray
    # but 0-dimensional Array{T,0} are stored as HDF5 scalar
    @test size(dzerodim) == ()
    @test !HDF5.isnull(dzerodim)
    @test read(dzerodim) == 1.0π

    # Similar tests for writing to attributes
    write(dempty, "attr", HDF5.EmptyArray{Float64}())
    write(dzerodim, "attr", fill(1.0ℯ))
    write(dzerovec, "attr", zeros(Int64, 0))
    write(dzeromat, "attr", zeros(Int64, 0, 0))
    write(dzeromat2, "attr", zeros(Int64, 0, 1))
    aempty = dempty["attr"]
    azerodim = dzerodim["attr"]
    azerovec = dzerovec["attr"]
    azeromat = dzeromat["attr"]
    azeromat2 = dzeromat2["attr"]
    # Test that eltype is preserved (especially for EmptyArray)
    @test eltype(aempty) == Float64
    @test eltype(azerodim) == Float64
    @test eltype(azerovec) == Int64
    @test eltype(azeromat) == Int64
    @test eltype(azeromat2) == Int64
    # Test sizes are as expected
    @test size(aempty) == ()
    @test size(azerovec) == (0,)
    @test size(azeromat) == (0, 0)
    @test size(azeromat2) == (0, 1)
    @test HDF5.isnull(aempty)
    @test !HDF5.isnull(azerovec)
    @test !HDF5.isnull(azeromat)
    @test !HDF5.isnull(azeromat2)
    # Reading back must preserve emptiness
    @test read(aempty) isa HDF5.EmptyArray
    # but 0-dimensional Array{T,0} are stored as HDF5 scalar
    @test size(azerodim) == ()
    @test !HDF5.isnull(azerodim)
    @test read(azerodim) == 1.0ℯ

    # Concatenation of EmptyArrays is not supported
    x = HDF5.EmptyArray{Float64}()
    @test_throws ErrorException [x x]
    @test_throws ErrorException [x; x]
    @test_throws ErrorException [x x; x x]

    close(hfile)
    rm(fn)

    # check that printing EmptyArray doesn't error
    buf = IOBuffer()
    show(buf, HDF5.EmptyArray{Int64}())
    @test String(take!(buf)) == "HDF5.EmptyArray{Int64}()"
    show(buf, MIME"text/plain"(), HDF5.EmptyArray{Int64}())
    @test String(take!(buf)) == "HDF5.EmptyArray{Int64}()"
end # empty and 0-size arrays

@testset "generic read of native types" begin
    fn = tempname()
    hfile = h5open(fn, "w")

    dtype_varstring = HDF5.Datatype(HDF5.API.h5t_copy(HDF5.API.H5T_C_S1))
    HDF5.API.h5t_set_size(dtype_varstring, HDF5.API.H5T_VARIABLE)

    write(hfile, "uint8_array", UInt8[(1:8)...])
    write(hfile, "bool_scalar", true)

    fixstring = "fix"
    varstring = "var"
    write(hfile, "fixed_string", fixstring)
    vardset = create_dataset(
        hfile, "variable_string", dtype_varstring, dataspace(varstring)
    )
    GC.@preserve varstring begin
        HDF5.API.h5d_write(
            vardset,
            dtype_varstring,
            HDF5.API.H5S_ALL,
            HDF5.API.H5S_ALL,
            HDF5.API.H5P_DEFAULT,
            [pointer(varstring)]
        )
    end
    flush(hfile)
    close(dtype_varstring)

    # generic read() handles concrete types with definite sizes transparently
    d = read(hfile["uint8_array"], UInt8)
    @test d isa Vector{UInt8}
    @test d == 1:8
    d = read(hfile["bool_scalar"], Bool)
    @test d isa Bool
    @test d == true
    d = read(hfile["fixed_string"], HDF5.FixedString{length(fixstring),0})
    @test d isa String
    @test d == fixstring
    d = read(hfile["variable_string"], Cstring)
    @test d isa String
    @test d == varstring
    # will also accept memory-compatible reinterpretations
    d = read(hfile["uint8_array"], Int8)
    @test d isa Vector{Int8}
    @test d == 1:8
    d = read(hfile["bool_scalar"], UInt8)
    @test d isa UInt8
    @test d == 0x1
    # but should throw on non-compatible types
    @test_throws ErrorException("""
                                Type size mismatch
                                sizeof(UInt16) = 2
                                sizeof($(sprint(show, datatype(UInt8)))) = 1
                                """) read(hfile["uint8_array"], UInt16)

    # Strings are not fixed size, but generic read still handles them if given the correct
    # underlying FixedString or Cstring type; a method overload makes String work, too.
    d = read(hfile["fixed_string"], String)
    @test d isa String
    @test d == fixstring
    d = read(hfile["variable_string"], String)
    @test d isa String
    @test d == varstring

    close(hfile)
    rm(fn)
end # generic read of native types

@testset "show" begin
    fn = tempname()

    # First create data objects and sure they print useful outputs

    hfile = h5open(fn, "w"; swmr=true)
    @test sprint(show, hfile) == "HDF5.File: (read-write, swmr) $fn"

    group = create_group(hfile, "group")
    @test sprint(show, group) == "HDF5.Group: /group (file: $fn)"

    dset = create_dataset(group, "dset", datatype(Int), dataspace((1,)))
    @test sprint(show, dset) == "HDF5.Dataset: /group/dset (file: $fn xfer_mode: 0)"

    meta = create_attribute(dset, "meta", datatype(Bool), dataspace((1,)))
    @test sprint(show, meta) == "HDF5.Attribute: meta"

    dsetattrs = attributes(dset)
    @test sprint(show, dsetattrs) ==
        "Attributes of HDF5.Dataset: /group/dset (file: $fn xfer_mode: 0)"

    prop = HDF5.init!(HDF5.LinkCreateProperties())
    @test sprint(show, prop) == """
    HDF5.LinkCreateProperties(
      create_intermediate_group = false,
      char_encoding   = :ascii,
    )"""

    prop = HDF5.DatasetCreateProperties()
    @test sprint(show, prop) == "HDF5.DatasetCreateProperties()"

    dtype = HDF5.Datatype(HDF5.API.h5t_copy(HDF5.API.H5T_IEEE_F64LE))
    @test sprint(show, dtype) == "HDF5.Datatype: H5T_IEEE_F64LE"
    commit_datatype(hfile, "type", dtype)
    @test sprint(show, dtype) == "HDF5.Datatype: /type H5T_IEEE_F64LE"

    dtypemeta = create_attribute(dtype, "dtypemeta", datatype(Bool), dataspace((1,)))
    @test sprint(show, dtypemeta) == "HDF5.Attribute: dtypemeta"

    dtypeattrs = attributes(dtype)
    @test sprint(show, dtypeattrs) == "Attributes of HDF5.Datatype: /type H5T_IEEE_F64LE"

    dspace_null = HDF5.Dataspace(HDF5.API.h5s_create(HDF5.API.H5S_NULL))
    dspace_scal = HDF5.Dataspace(HDF5.API.h5s_create(HDF5.API.H5S_SCALAR))
    dspace_norm = dataspace((100, 4))
    dspace_maxd = dataspace((100, 4); max_dims=(256, 4))
    dspace_slab = HDF5.hyperslab(dataspace((100, 4)), 1:20:100, 1:4)
    if HDF5.libversion ≥ v"1.10.7"
        dspace_irrg = HDF5.Dataspace(
            HDF5.API.h5s_combine_select(
                HDF5.API.h5s_copy(dspace_slab),
                HDF5.API.H5S_SELECT_OR,
                HDF5.hyperslab(dataspace((100, 4)), 2, 2)
            )
        )
        @test sprint(show, dspace_irrg) == "HDF5.Dataspace: (100, 4) [irregular selection]"
    end
    @test sprint(show, dspace_null) == "HDF5.Dataspace: H5S_NULL"
    @test sprint(show, dspace_scal) == "HDF5.Dataspace: H5S_SCALAR"
    @test sprint(show, dspace_norm) == "HDF5.Dataspace: (100, 4)"
    @test sprint(show, dspace_maxd) == "HDF5.Dataspace: (100, 4) / (256, 4)"
    @test sprint(show, dspace_slab) == "HDF5.Dataspace: (1:20:81, 1:4) / (1:100, 1:4)"

    # Now test printing after closing each object

    close(dspace_null)
    @test sprint(show, dspace_null) == "HDF5.Dataspace: (invalid)"

    close(dtype)
    @test sprint(show, dtype) == "HDF5.Datatype: (invalid)"

    close(prop)
    @test sprint(show, prop) == "HDF5.DatasetCreateProperties: (invalid)"

    close(meta)
    @test sprint(show, meta) == "HDF5.Attribute: (invalid)"

    close(dtypemeta)
    @test sprint(show, dtypemeta) == "HDF5.Attribute: (invalid)"

    close(dset)
    @test sprint(show, dset) == "HDF5.Dataset: (invalid)"
    @test sprint(show, dsetattrs) == "Attributes of HDF5.Dataset: (invalid)"

    close(group)
    @test sprint(show, group) == "HDF5.Group: (invalid)"

    close(hfile)
    @test sprint(show, hfile) == "HDF5.File: (closed) $fn"

    # Go back and check different access modes for file printing
    hfile = h5open(fn, "r+"; swmr=true)
    @test sprint(show, hfile) == "HDF5.File: (read-write, swmr) $fn"
    close(hfile)
    hfile = h5open(fn, "r"; swmr=true)
    @test sprint(show, hfile) == "HDF5.File: (read-only, swmr) $fn"
    close(hfile)
    hfile = h5open(fn, "r")
    @test sprint(show, hfile) == "HDF5.File: (read-only) $fn"
    close(hfile)
    hfile = h5open(fn, "cw")
    @test sprint(show, hfile) == "HDF5.File: (read-write) $fn"
    close(hfile)

    rm(fn)

    # Make an interesting file tree
    hfile = h5open(fn, "w")
    # file level
    hfile["version"] = 1.0
    attributes(hfile)["creator"] = "HDF5.jl"
    # group level
    create_group(hfile, "inner")
    attributes(hfile["inner"])["dirty"] = true
    # dataset level
    hfile["inner/data"] = collect(-5:5)
    attributes(hfile["inner/data"])["mode"] = 1
    # non-trivial committed datatype
    # TODO: print more datatype information
    tmeta = HDF5.Datatype(
        HDF5.API.h5t_create(HDF5.API.H5T_COMPOUND, sizeof(Int) + sizeof(Float64))
    )
    HDF5.API.h5t_insert(tmeta, "scale", 0, HDF5.hdf5_type_id(Int))
    HDF5.API.h5t_insert(tmeta, "bias", sizeof(Int), HDF5.hdf5_type_id(Float64))
    tstr = datatype("fixed")
    t = HDF5.Datatype(
        HDF5.API.h5t_create(HDF5.API.H5T_COMPOUND, sizeof(tmeta) + sizeof(tstr))
    )
    HDF5.API.h5t_insert(t, "meta", 0, tmeta)
    HDF5.API.h5t_insert(t, "type", sizeof(tmeta), tstr)
    commit_datatype(hfile, "dtype", t)

    buf = IOBuffer()
    iobuf = IOContext(buf, :limit => true, :module => Main)
    show3(io::IO, x) = show(IOContext(io, iobuf), MIME"text/plain"(), x)

    HDF5.show_tree(iobuf, hfile)
    msg = String(take!(buf))
    @test occursin(
        r"""
🗂️ HDF5.File: .*$
├─ 🏷️ creator
├─ 📄 dtype
├─ 📂 inner
│  ├─ 🏷️ dirty
│  └─ 🔢 data
│     └─ 🏷️ mode
└─ 🔢 version"""m,
        msg
    )
    @test sprint(show3, hfile) == msg

    HDF5.show_tree(iobuf, hfile; attributes=false)
    @test occursin(
        r"""
🗂️ HDF5.File: .*$
├─ 📄 dtype
├─ 📂 inner
│  └─ 🔢 data
└─ 🔢 version"""m,
        String(take!(buf))
    )

    HDF5.show_tree(iobuf, attributes(hfile))
    msg = String(take!(buf))
    @test occursin(
        r"""
🗂️ Attributes of HDF5.File: .*$
└─ 🏷️ creator"""m,
        msg
    )
    @test sprint(show3, attributes(hfile)) == msg

    HDF5.show_tree(iobuf, hfile["inner"])
    msg = String(take!(buf))
    @test occursin(
        r"""
📂 HDF5.Group: /inner .*$
├─ 🏷️ dirty
└─ 🔢 data
   └─ 🏷️ mode"""m,
        msg
    )
    @test sprint(show3, hfile["inner"]) == msg

    HDF5.show_tree(iobuf, hfile["inner"]; attributes=false)
    @test occursin(
        r"""
📂 HDF5.Group: /inner .*$
└─ 🔢 data"""m,
        String(take!(buf))
    )

    HDF5.show_tree(iobuf, hfile["inner/data"])
    msg = String(take!(buf))
    @test occursin(
        r"""
🔢 HDF5.Dataset: /inner/data .*$
└─ 🏷️ mode"""m,
        msg
    )
    # xfer_mode changes between printings, so need regex again
    @test occursin(
        r"""
🔢 HDF5.Dataset: /inner/data .*$
└─ 🏷️ mode"""m,
        sprint(show3, hfile["inner/data"])
    )

    HDF5.show_tree(iobuf, hfile["inner/data"]; attributes=false)
    @test occursin(
        r"""
🔢 HDF5.Dataset: /inner/data .*$"""m,
        String(take!(buf))
    )

    HDF5.show_tree(iobuf, hfile["dtype"])
    @test occursin(
        r"""
📄 HDF5.Datatype: /dtype""",
        String(take!(buf))
    )

    HDF5.show_tree(iobuf, hfile["inner/data"]["mode"]; attributes=true)
    @test occursin(
        r"""
🏷️ HDF5.Attribute: mode""",
        String(take!(buf))
    )

    # configurable options

    # no emoji icons
    HDF5.SHOW_TREE_ICONS[] = false
    @test occursin(
        r"""
\[F\] HDF5.File: .*$
├─ \[A\] creator
├─ \[T\] dtype
├─ \[G\] inner
│  ├─ \[A\] dirty
│  └─ \[D\] data
│     └─ \[A\] mode
└─ \[D\] version"""m,
        sprint(show3, hfile)
    )
    HDF5.SHOW_TREE_ICONS[] = true

    # no tree printing
    show(IOContext(iobuf, :compact => true), MIME"text/plain"(), hfile)
    msg = String(take!(buf))
    @test msg == sprint(show, hfile)

    close(hfile)

    # Now test the print-limiting heuristics for large/complex datasets

    # group with a large number of children; tests child entry truncation heuristic
    h5open(fn, "w") do hfile
        dt, ds = datatype(Int), dataspace(())
        opts = Iterators.product('A':'Z', 1:9)
        for ii in opts
            create_dataset(hfile, string(ii...), dt, ds)
        end

        def = HDF5.SHOW_TREE_MAX_CHILDREN[]
        HDF5.SHOW_TREE_MAX_CHILDREN[] = 5

        HDF5.show_tree(iobuf, hfile)
        msg = String(take!(buf))
        @test occursin(
            r"""
🗂️ HDF5.File: .*$
├─ 🔢 A1
├─ 🔢 A2
├─ 🔢 A3
├─ 🔢 A4
├─ 🔢 A5
└─ \(229 more children\)"""m,
            msg
        )
        @test sprint(show3, hfile) == msg

        HDF5.SHOW_TREE_MAX_CHILDREN[] = def

        # IOContext can halt limiting
        HDF5.show_tree(IOContext(iobuf, :limit => false), hfile)
        @test countlines(seekstart(buf)) == length(opts) + 1
        truncate(buf, 0)
    end

    # deeply nested set of elements; test that the tree is truncated
    h5open(fn, "w") do hfile
        p = HDF5.root(hfile)::HDF5.Group
        opts = 'A':'Z'
        for ii in opts
            p = create_group(p, string(ii))
        end

        def = HDF5.SHOW_TREE_MAX_DEPTH[]
        HDF5.SHOW_TREE_MAX_DEPTH[] = 5

        HDF5.show_tree(iobuf, hfile)
        msg = String(take!(buf))
        @test occursin(
            r"""
🗂️ HDF5.File: .*$
└─ 📂 A
   └─ 📂 B
      └─ 📂 C
         └─ 📂 D
            └─ 📂 E
               └─ \(1 child\)"""m,
            msg
        )
        @test sprint(show3, hfile) == msg

        HDF5.SHOW_TREE_MAX_DEPTH[] = def

        # IOContext can halt limiting
        HDF5.show_tree(IOContext(iobuf, :limit => false), hfile)
        @test countlines(seekstart(buf)) == length(opts) + 1
        truncate(buf, 0)
    end

    rm(fn)
end # show tests

@testset "split1" begin
    @test HDF5.split1("/") == ("/", "")
    @test HDF5.split1("a") == ("a", "")
    @test HDF5.split1("/a/b/c") == ("/", "a/b/c")
    @test HDF5.split1("a/b/c") == ("a", "b/c")
    @test HDF5.split1(GenericString("a")) == ("a", "")
    @test HDF5.split1(GenericString("/a/b/c")) == ("/", "a/b/c")
    @test HDF5.split1(GenericString("a/b/c")) == ("a", "b/c")

    # The following two paths have the same graphemes but different code unit structures:
    # the first one is
    #     <latin small letter a with circumflex> "/" <greek small leter alpha>
    # while the second one is
    #     "a" <combining circumflex accent> "/" <greek small letter alpha>
    circa = "â" # <latin small leter a with circumflex>
    acomb = "â" # "a" + <combining circumflex accent>
    path1 = circa * "/α"
    path2 = acomb * "/α"
    # Sanity checks that the two strings are different but equivalent under normalization
    @test path1 != path2
    @test Base.Unicode.normalize(path1, :NFC) == Base.Unicode.normalize(path2, :NFC)
    # Check split1 operates correctly
    @test HDF5.split1(path1) == (circa, "α")
    @test HDF5.split1(path2) == (acomb, "α")
    @test HDF5.split1("/" * path1) == ("/", path1)
    @test HDF5.split1("/" * path2) == ("/", path2)
end # split1 tests

# Also tests AbstractString interface
@testset "haskey" begin
    fn = tempname()
    hfile = h5open(fn, "w")

    group1 = create_group(hfile, "group1")
    group2 = create_group(group1, "group2")

    @test haskey(hfile, "/")
    @test haskey(hfile, GenericString("group1"))
    @test !haskey(hfile, GenericString("groupna"))
    @test haskey(hfile, "group1/group2")
    @test !haskey(hfile, "group1/groupna")
    @test_throws KeyError hfile["nothing"]

    dset1 = create_dataset(hfile, "dset1", datatype(Int), dataspace((1,)))
    dset2 = create_dataset(group1, "dset2", datatype(Int), dataspace((1,)))

    @test haskey(hfile, "dset1")
    @test !haskey(hfile, "dsetna")
    @test haskey(hfile, "group1/dset2")
    @test !haskey(hfile, "group1/dsetna")

    meta1 = create_attribute(dset1, "meta1", datatype(Bool), dataspace((1,)))
    @test haskey(dset1, "meta1")
    @test !haskey(dset1, "metana")
    @test_throws KeyError dset1["nothing"]

    attribs = attributes(hfile)
    attribs["test1"] = true
    attribs["test2"] = "foo"

    @test haskey(attribs, "test1")
    @test haskey(attribs, "test2")
    @test !haskey(attribs, "testna")
    @test_throws KeyError attribs["nothing"]

    attribs = attributes(dset2)
    attribs["attr"] = "foo"
    @test haskey(attribs, GenericString("attr"))

    close(hfile)
    rm(fn)
end # haskey tests

@testset "AbstractString" begin
    fn = GenericString(tempname())
    hfile = h5open(fn, "w")
    close(hfile)
    hfile = h5open(fn)
    close(hfile)
    hfile = h5open(fn, "w")

    @test_nowarn create_group(hfile, GenericString("group1"))
    @test_nowarn create_dataset(
        hfile, GenericString("dset1"), datatype(Int), dataspace((1,))
    )
    @test_nowarn create_dataset(hfile, GenericString("dset2"), 1)

    @test_nowarn hfile[GenericString("group1")]
    @test_nowarn hfile[GenericString("dset1")]

    dset1 = hfile["dset1"]
    @test_nowarn create_attribute(
        dset1, GenericString("meta1"), datatype(Bool), dataspace((1,))
    )
    @test_nowarn create_attribute(dset1, GenericString("meta2"), 1)
    @test_nowarn dset1[GenericString("meta1")]
    @test_nowarn dset1[GenericString("x")] = 2

    array_of_strings = ["test",]
    write(hfile, "array_of_strings", array_of_strings)
    @test_nowarn attributes(hfile)[GenericString("ref_test")] = HDF5.Reference(
        hfile, GenericString("array_of_strings")
    )
    @test read(attributes(hfile)[GenericString("ref_test")]) ===
        HDF5.Reference(hfile, "array_of_strings")

    hfile[GenericString("test")] = 17.2
    @test_nowarn delete_object(hfile, GenericString("test"))
    @test_nowarn delete_attribute(dset1, GenericString("meta1"))

    # transient types
    memtype_id = HDF5.API.h5t_copy(HDF5.API.H5T_NATIVE_DOUBLE)
    dt = HDF5.Datatype(memtype_id)
    @test !HDF5.API.h5t_committed(dt)
    commit_datatype(hfile, GenericString("dt"), dt)
    @test HDF5.API.h5t_committed(dt)

    dt = datatype(Int)
    ds = dataspace(0)
    d = create_dataset(hfile, GenericString("d"), dt, ds)
    g = create_group(hfile, GenericString("g"))
    a = create_attribute(hfile, GenericString("a"), dt, ds)

    for obj in (d, g)
        @test_nowarn write_attribute(obj, GenericString("a"), 1)
        @test_nowarn read_attribute(obj, GenericString("a"))
        @test_nowarn write(obj, GenericString("aa"), 1)
        @test_nowarn attributes(obj)["attr1"] = GenericString("b")
    end
    @test_nowarn write(d, "attr2", GenericString("c"))
    @test_nowarn write_dataset(g, GenericString("ag"), GenericString("gg"))
    @test_nowarn write_dataset(
        g, GenericString("ag_array"), [GenericString("a1"), GenericString("a2")]
    )

    genstrs = GenericString["fee", "fi", "foo"]
    @test_nowarn write_attribute(d, GenericString("myattr"), genstrs)
    @test genstrs == read(d["myattr"])

    for obj in (hfile,)
        @test_nowarn open_dataset(obj, GenericString("d"))
        @test_nowarn write_dataset(obj, GenericString("dd"), 1)
        @test_nowarn read_dataset(obj, GenericString("dd"))
        @test_nowarn read(obj, GenericString("dd"))
        @test_nowarn read(obj, GenericString("dd") => Int)
    end
    read(attributes(hfile), GenericString("a"))

    write(hfile, GenericString("ASD"), GenericString("Aa"))
    write(g, GenericString("ASD"), GenericString("Aa"))
    write(g, GenericString("ASD1"), [GenericString("Aa")])

    # test writing multiple variable
    @test_nowarn write(
        hfile, GenericString("a1"), rand(2, 2), GenericString("a2"), rand(2, 2)
    )

    # copy methods
    d1 = create_dataset(hfile, GenericString("d1"), dt, ds)
    d1["x"] = 32
    @test_nowarn copy_object(hfile, GenericString("d1"), hfile, GenericString("d1copy1"))
    @test_nowarn copy_object(d1, hfile, GenericString("d1copy2"))

    fn = GenericString(tempname())
    A = Matrix(reshape(1:120, 15, 8))
    @test_nowarn h5write(fn, GenericString("A"), A)
    @test_nowarn h5read(fn, GenericString("A"))
    @test_nowarn h5read(fn, GenericString("A"), (2:3:15, 3:5))

    @test_nowarn h5write(fn, GenericString("x"), 1)
    @test_nowarn h5read(fn, GenericString("x") => Int)

    @test_nowarn h5rewrite(fn) do fid
        g = create_group(fid, "mygroup")
        write(g, "x", 3.3)
    end
    @test_nowarn h5rewrite(fn) do fid
        g = create_group(fid, "mygroup")
        write(g, "y", 3.3)
    end

    @test_nowarn h5write(fn, "W", [1 2; 3 4])
    @test_nowarn h5writeattr(fn, GenericString("W"), Dict("a" => 1, "b" => 2))
    @test_nowarn h5readattr(fn, GenericString("W"))

    fn_external = GenericString(tempname())
    dset = HDF5.create_external_dataset(hfile, "ext", fn_external, Int, (10, 20))
    dcpl = HDF5.get_create_properties(dset)
    @test HDF5.API.h5p_get_external_count(dcpl) == 1
    ext_prop = HDF5.API.h5p_get_external(dcpl)
    @test ext_prop.name == fn_external
    @test ext_prop.offset == 0
    @test ext_prop.size == 10 * 20 * sizeof(Int)
    dapl = HDF5.get_access_properties(dset)
    dapl.efile_prefix = "efile_test"
    @test HDF5.API.h5p_get_efile_prefix(dapl) == "efile_test"
    close(hfile)
end

@testset "opaque data" begin
    mktemp() do path, io
        close(io)
        fid = h5open(path, "w")

        num   = 1
        olen  = 4
        otype = HDF5.Datatype(HDF5.API.h5t_create(HDF5.API.H5T_OPAQUE, olen))
        HDF5.API.h5t_set_tag(otype, "opaque test")

        # scalar
        dat0 = rand(UInt8, olen)
        create_dataset(fid, "scalar", otype, dataspace(()))
        write_dataset(fid["scalar"], otype, dat0)
        # vector
        dat1 = [rand(UInt8, olen) for _ in 1:4]
        buf1 = reduce(vcat, dat1)
        create_dataset(fid, "vector", otype, dataspace(dat1))
        write_dataset(fid["vector"], otype, buf1)
        # matrix
        dat2 = [rand(UInt8, olen) for _ in 1:4, _ in 1:2]
        buf2 = reduce(vcat, dat2)
        create_dataset(fid, "matrix", otype, dataspace(dat2))
        write_dataset(fid["matrix"], otype, buf2)

        # opaque data within a compound data type
        ctype = HDF5.Datatype(
            HDF5.API.h5t_create(HDF5.API.H5T_COMPOUND, sizeof(num) + sizeof(otype))
        )
        HDF5.API.h5t_insert(ctype, "v", 0, datatype(num))
        HDF5.API.h5t_insert(ctype, "d", sizeof(num), otype)
        cdat = vcat(reinterpret(UInt8, [num]), dat0)
        create_dataset(fid, "compound", ctype, dataspace(()))
        write_dataset(fid["compound"], ctype, cdat)

        opaque0 = read(fid["scalar"])
        @test opaque0.tag == "opaque test"
        @test opaque0.data == dat0
        opaque1 = read(fid["vector"])
        @test opaque1.tag == "opaque test"
        @test opaque1.data == dat1
        opaque2 = read(fid["matrix"])
        @test opaque2.tag == "opaque test"
        @test opaque2.data == dat2

        # Note: opaque tag is lost
        compound = read(fid["compound"])
        @test compound == (v=num, d=dat0)

        close(fid)
    end
end

@testset "FixedStrings and FixedArrays" begin
    # properties for FixedString
    fix = HDF5.FixedString{4,0}((b"test"...,))
    @test length(typeof(fix)) == 4
    @test length(fix) == 4
    @test HDF5.pad(typeof(fix)) == 0
    @test HDF5.pad(fix) == 0
    # issue #742, large fixed strings are readable
    mktemp() do path, io
        close(io)
        num = Int64(9)
        ref = join('a':'z')^1000
        fid = h5open(path, "w")
        # long string serialized as FixedString
        fid["longstring"] = ref

        # compound datatype containing a FixedString
        compound_dtype = HDF5.Datatype(
            HDF5.API.h5t_create(HDF5.API.H5T_COMPOUND, sizeof(num) + sizeof(ref))
        )
        HDF5.API.h5t_insert(compound_dtype, "n", 0, datatype(num))
        HDF5.API.h5t_insert(compound_dtype, "a", sizeof(num), datatype(ref))
        c = create_dataset(fid, "compoundlongstring", compound_dtype, dataspace(()))
        # normally this is done with a `struct name{N}; n::Int64; a::NTuple{N,Char}; end`,
        # but we need to not actually instantiate the `NTuple`.
        buf = IOBuffer()
        write(buf, num, ref)
        @assert position(buf) == sizeof(compound_dtype)
        write_dataset(c, compound_dtype, take!(buf))

        # Test reading without stalling
        d = fid["longstring"]
        T = HDF5.get_jl_type(d)
        @test T <: HDF5.FixedString
        @test length(T) == length(ref)
        @test read(d) == ref

        T = HDF5.get_jl_type(c)
        @test T <: NamedTuple
        @test fieldnames(T) == (:n, :a)
        @test read(c) == (n=num, a=ref)

        close(fid)
    end

    fix = HDF5.FixedArray{Float64,(2, 2),4}((1, 2, 3, 4))
    @test size(typeof(fix)) == (2, 2)
    @test size(fix) == (2, 2)
    @test eltype(typeof(fix)) == Float64
    @test eltype(fix) == Float64
    # large fixed arrays are readable
    mktemp() do path, io
        close(io)
        ref = rand(Float64, 3000)
        t = HDF5.Datatype(
            HDF5.API.h5t_array_create(datatype(Float64), ndims(ref), collect(size(ref)))
        )
        scalarspace = dataspace(())

        fid = h5open(path, "w")
        d = create_dataset(fid, "longnums", t, scalarspace)
        write_dataset(d, t, ref)

        T = HDF5.get_jl_type(d)
        @test T <: HDF5.FixedArray
        @test size(T) == size(ref)
        @test eltype(T) == eltype(ref)
        @test read(d) == ref

        close(fid)
    end
end

@testset "Object Exists" begin
    hfile = h5open(tempname(), "w")
    g1 = create_group(hfile, "group1")
    @test_throws HDF5.API.H5Error create_group(hfile, "group1")
    create_group(g1, "group1a")
    @test_throws HDF5.API.H5Error create_group(hfile, "/group1/group1a")
    @test_throws HDF5.API.H5Error create_group(g1, "group1a")

    # issue #1055
    create_group(hfile, SubString("abcdef", 1, 3))
    @test haskey(hfile, "abc")
    @test !haskey(hfile, "abcdef")

    create_dataset(hfile, "dset1", 1)
    create_dataset(hfile, "/group1/dset1", 1)

    @test_throws ErrorException create_dataset(hfile, "dset1", 1)
    @test_throws ErrorException create_dataset(hfile, "group1", 1)
    @test_throws ErrorException create_dataset(g1, "dset1", 1)

    close(hfile)
end

@testset "HDF5 existance" begin
    fn1 = tempname()
    fn2 = tempname()

    open(fn1, "w") do f
        write(f, "Hello text file")
    end

    @test !HDF5.ishdf5(fn1) # check that a non-hdf5 file retuns false
    @test !HDF5.ishdf5(fn2) # checks that a file that does not exist returns false

    @test_throws ErrorException h5write(fn1, "x", 1) # non hdf5 file throws
    h5write(fn2, "x", 1)

    @test HDF5.ishdf5(fn2)

    rm(fn1)
    rm(fn2)
end

@testset "bounds" begin

    # issue #954
    h5open(tempname(), "w") do f
        a, _ = create_attribute(f, "a", zeros(4))
        @test_throws ArgumentError write(a, ones(2))
        d, _ = create_dataset(f, "dd", zeros(4))
        @test_throws ArgumentError write(d, ones(2))
    end
end