use FileIO 1.6.5 detect_compressor()

- remove RData's own compressor format detection
- fix compression unit tests
- fix JuliaData#83
- bump versio to 0.8.0

Project.toml

@@ -1,6 +1,6 @@
 name = "RData"
 uuid = "df47a6cb-8c03-5eed-afd8-b6050d6c41da"
-version = "0.7.3"
+version = "0.8.0"
 
 [deps]
 CategoricalArrays = "324d7699-5711-5eae-9e2f-1d82baa6b597"
@@ -16,7 +16,7 @@ Unicode = "4ec0a83e-493e-50e2-b9ac-8f72acf5a8f5"
 CategoricalArrays = "0.8, 0.9"
 CodecZlib = "0.4, 0.5, 0.6, 0.7"
 DataFrames = "0.21, 0.22"
-FileIO = "1.0.5"
+FileIO = "1.6.5"
 Requires = "1.0.0"
 TimeZones = "0.7, 0.8, 0.9, 0.10, 1.0"
 julia = "1"

src/decompression.jl

@@ -2,7 +2,7 @@ using CodecZlib, Requires
 
 abstract type ArchiveFormat{S} end
 
-decompressor_stream(::Type{ArchiveFormat{:Gzip}}, io::IOStream) = GzipDecompressorStream(io)
+decompressor_stream(::Type{ArchiveFormat{:GZIP}}, io::IOStream) = GzipDecompressorStream(io)
 
 # Handles the case when we are loading a supported archive format but the codec is not loaded
 function decompressor_stream(::Type{ArchiveFormat{S}}, io::IOStream) where S
@@ -12,34 +12,19 @@ end
 # called by __init__() to enable support for optional codecs
 function define_optional_decompressor_streams()
     @require CodecBzip2="523fee87-0ab8-5b00-afb7-3ecf72e48cfd" begin
-        decompressor_stream(::Type{ArchiveFormat{:Bzip2}}, io::IOStream) = CodecBzip2.Bzip2DecompressorStream(io)
+        decompressor_stream(::Type{ArchiveFormat{:BZIP2}}, io::IOStream) = CodecBzip2.Bzip2DecompressorStream(io)
     end
     @require CodecXz="ba30903b-d9e8-5048-a5ec-d1f5b0d4b47b" begin
-        decompressor_stream(::Type{ArchiveFormat{:Xz}}, io::IOStream) = CodecXz.XzDecompressorStream(io)
+        decompressor_stream(::Type{ArchiveFormat{:XZ}}, io::IOStream) = CodecXz.XzDecompressorStream(io)
     end
 end
 
-# Magic numbers that identify the supported compression formats
-const MAGIC_GZIP = b"\x1F\x8B"
-const MAGIC_BZIP2 = b"BZh"
-const MAGIC_XZ = b"\xFD7zXZ\x00"
-
+# decompress the stream if it's compressed (by a supported codec)
 function decompress(io)
-    # Read the first 6 bytes to obtain the magic number if there is one.
-    buffer = zeros(UInt8, 6)
-    readbytes!(io, buffer)
-    seekstart(io)
-
-    # Check for any of the gzip, bzip2 or xz magic numbers
-    if buffer[1:2] == MAGIC_GZIP
-        return decompressor_stream(ArchiveFormat{:Gzip}, io)
-    elseif buffer[1:3] == MAGIC_BZIP2
-        return decompressor_stream(ArchiveFormat{:Bzip2}, io)
-    elseif buffer == MAGIC_XZ
-        return decompressor_stream(ArchiveFormat{:Xz}, io)
+    format = FileIO.detect_compressor(io, formats=["GZIP", "BZIP2", "XZ"])
+    if format !== nothing
+        return decompressor_stream(ArchiveFormat{Symbol(format)}, io)
+    else
+        return io # not compressed
     end
-
-    # If none of the magic numbers match, we assume the file is not compressed
-
-    return io
 end

test/decompression.jl

@@ -4,13 +4,27 @@ using DataFrames, RData, Test
 rdata_path = joinpath(dirname(@__FILE__), "data_v3")
 df = DataFrame(num = [1.1, 2.2])
 
+function test_missing_compressor(filename::AbstractString, codec::Symbol)
+    @test_throws CapturedException load(joinpath(rdata_path, filename))
+    caught = nothing
+    try
+        load(joinpath(rdata_path, filename))
+    catch ex
+        caught = ex
+    end
+
+    @test caught isa CapturedException
+    @test caught.ex isa CodecMissingError
+    @test caught.ex.formatName == codec
+end
+
 @testset "Loading compressed RData files" begin
     @testset "When the optional compression codecs are not loaded" begin
-        @test_warn "Error encountered while loading" begin
-            @test_throws CodecMissingError load(joinpath(rdata_path, "compressed_bzip2.rda"))
+        @test_warn "Error encountered while load" begin
+            test_missing_compressor("compressed_bzip2.rda", :BZIP2)
         end
-        @test_warn "Error encountered while loading" begin
-            @test_throws CodecMissingError load(joinpath(rdata_path, "compressed_xz.rda"))
+        @test_warn "Error encountered while load" begin
+            test_missing_compressor("compressed_xz.rda", :XZ)
         end
     end