Try using ChunkCodec as an HDF5 filter

Project.toml

@@ -16,7 +16,7 @@ UUIDs = "cf7118a7-6976-5b1a-9a39-7adc72f591a4"
 
 [compat]
 Blosc = "0.7.3"
-CodecBzip2 = "0.7, 0.8"
+ChunkCodecLibBzip2 = "0.3"
 CodecLz4 = "0.4"
 CodecZstd = "0.7, 0.8"
 Compat = "3.1.0, 4"
@@ -30,7 +30,7 @@ julia = "1.9"
 
 [extensions]
 BloscExt = "Blosc"
-CodecBzip2Ext = "CodecBzip2"
+ChunkCodecLibBzip2Ext = "ChunkCodecLibBzip2"
 CodecLz4Ext = "CodecLz4"
 CodecZstdExt = "CodecZstd"
 FileIOExt = "FileIO"
@@ -41,7 +41,6 @@ bitshuffle_jll_ext = "bitshuffle_jll"
 [extras]
 Blosc = "a74b3585-a348-5f62-a45c-50e91977d574"
 CRC32c = "8bf52ea8-c179-5cab-976a-9e18b702a9bc"
-CodecBzip2 = "523fee87-0ab8-5b00-afb7-3ecf72e48cfd"
 CodecLz4 = "5ba52731-8f18-5e0d-9241-30f10d1ec561"
 CodecZstd = "6b39b394-51ab-5f42-8807-6242bab2b4c2"
 Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
@@ -62,10 +61,10 @@ test = ["Test", "MPI", "Distributed", "LinearAlgebra", "OrderedCollections", "Pk
 
 [weakdeps]
 Blosc = "a74b3585-a348-5f62-a45c-50e91977d574"
-CodecBzip2 = "523fee87-0ab8-5b00-afb7-3ecf72e48cfd"
 CodecLz4 = "5ba52731-8f18-5e0d-9241-30f10d1ec561"
 CodecZstd = "6b39b394-51ab-5f42-8807-6242bab2b4c2"
 FileIO = "5789e2e9-d7fb-5bc7-8068-2c6fae9b9549"
 MPI = "da04e1cc-30fd-572f-bb4f-1f8673147195"
 OrderedCollections = "bac558e1-5e72-5ebc-8fee-abe8a469f55d"
+ChunkCodecLibBzip2 = "2b723af9-f480-4e8d-a1e4-4a9f5a906122"
 bitshuffle_jll = "228fe19c-1b83-5282-a626-13744502a320"

docs/make.jl

@@ -11,7 +11,7 @@ using MPI  # needed to generate docs for parallel HDF5 API
 const BloscExt = Base.get_extension(HDF5, :BloscExt)
 const bitshuffle_jll_ext = Base.get_extension(HDF5, :bitshuffle_jll_ext)
 const BloscExt = Base.get_extension(HDF5, :BloscExt)
-const CodecBzip2Ext = Base.get_extension(HDF5, :CodecBzip2Ext)
+const ChunkCodecLibBzip2Ext = Base.get_extension(HDF5, :ChunkCodecLibBzip2Ext)
 const CodecLz4Ext = Base.get_extension(HDF5, :CodecLz4Ext)
 const CodecZstdExt = Base.get_extension(HDF5, :CodecZstdExt)
 

ext/ChunkCodecLibBzip2Ext/ChunkCodecLibBzip2Ext.jl

@@ -0,0 +1,102 @@
+#=
+The code below has been ported to Julia from the original C source:
+https://github.com/nexusformat/HDF5-External-Filter-Plugins/blob/master/BZIP2/src/H5Zbzip2.c
+The filter function  H5Z_filter_bzip2 was adopted from:
+PyTables http://www.pytables.org.
+The plugin can be used with the HDF5 library version 1.8.11+ to read HDF5 datasets compressed with bzip2 created by PyTables.
+License: licenses/H5Zbzip2_LICENSE.txt
+
+The following license applies to the Julia port.
+Copyright (c) 2021 Mark Kittisopikul and Howard Hughes Medical Institute. License MIT, see LICENSE.txt
+=#
+module ChunkCodecLibBzip2Ext
+
+using ChunkCodecLibBzip2:
+    BZ2Codec,
+    BZ2EncodeOptions
+using ChunkCodecLibBzip2.ChunkCodecCore:
+    try_resize_decode!,
+    encode_bound,
+    try_encode!
+using HDF5.API
+import HDF5.Filters:
+    Filter, filterid, register_filter, filtername, filter_func, filter_cfunc, UnsafeBuffer
+
+export H5Z_FILTER_BZIP2, H5Z_filter_bzip2, Bzip2Filter
+
+const H5Z_FILTER_BZIP2 = API.H5Z_filter_t(307)
+const bzip2_name = "HDF5 bzip2 filter; see http://www.hdfgroup.org/services/contributions.html"
+
+function H5Z_filter_bzip2(
+    flags::Cuint,
+    cd_nelmts::Csize_t,
+    cd_values::Ptr{Cuint},
+    nbytes::Csize_t,
+    buf_size::Ptr{Csize_t},
+    buf::Ptr{Ptr{Cvoid}}
+)::Csize_t
+    dst = UnsafeBuffer()
+    outdatalen::Csize_t = Csize_t(0)
+    try
+        src = UnsafeBuffer(unsafe_load(buf), nbytes)
+        if flags & API.H5Z_FLAG_REVERSE != 0
+            # Decompress
+            outbuflen = Int64(nbytes) * 3 + 1
+            resize!(dst, outbuflen)
+            outdatalen = Int64(try_resize_decode!(BZ2Codec(), dst, src, typemax(Int64)))
+        else
+            # Compress data
+            blockSize100k = 9
+            # Get compression blocksize if present
+            if cd_nelmts > 0
+                blockSize100k = unsafe_load(cd_values)
+            end
+            encoder = BZ2EncodeOptions(;blockSize100k)
+            # Prepare the output buffer
+            resize!(dst, encode_bound(encoder, Int64(nbytes)))
+            outdatalen = Int64(try_encode!(encoder, dst, src))
+        end # if flags & API.H5Z_FLAG_REVERSE != 0
+        resize!(src, Int64(0))
+        unsafe_store!(buf, dst.p)
+        unsafe_store!(buf_size, dst.size)
+    catch err
+        #  "In the case of failure, the return value is 0 (zero) and all pointer arguments are left unchanged."
+        outdatalen = Csize_t(0)
+        resize!(dst, Int64(0))
+        @error "H5Zbzip2.jl Non-Fatal ERROR: " err
+        display(stacktrace(catch_backtrace()))
+    end # try - catch
+
+    return outdatalen
+end # function H5Z_filter_bzip2
+
+# Filters Module
+
+"""
+    Bzip2Filter(blockSize100k)
+
+Apply Bzip2 compression. The filter id is $H5Z_FILTER_BZIP2.
+
+# External Links
+* [BZIP2 HDF5 Filter ID 307](https://portal.hdfgroup.org/display/support/Filters#Filters-307)
+* [PyTables Repository (C code)](https://github.com/PyTables/PyTables)
+"""
+struct Bzip2Filter <: Filter
+    blockSize100k::Cuint
+end
+Bzip2Filter() = Bzip2Filter(9)
+
+filterid(::Type{Bzip2Filter}) = H5Z_FILTER_BZIP2
+filtername(::Type{Bzip2Filter}) = bzip2_name
+filter_func(::Type{Bzip2Filter}) = H5Z_filter_bzip2
+filter_cfunc(::Type{Bzip2Filter}) = @cfunction(
+    H5Z_filter_bzip2,
+    Csize_t,
+    (Cuint, Csize_t, Ptr{Cuint}, Csize_t, Ptr{Csize_t}, Ptr{Ptr{Cvoid}})
+)
+
+function __init__()
+    register_filter(Bzip2Filter)
+end
+
+end # module CodecBzip2Ext