check eof before read bytes (#346)

Author: johnnychen94

Project.toml

@@ -13,11 +13,12 @@ julia = "0.7, 1"
 
 [extras]
 CSVFiles = "5d742f6a-9f54-50ce-8119-2520741973ca"
+CodecZlib = "944b1d66-785c-5afd-91f1-9de20f533193"
 ColorTypes = "3da002f7-5984-5a60-b8a6-cbb66c0b333f"
 FilePathsBase = "48062228-2e41-5def-b9a4-89aafe57970f"
 HTTP = "cd3eb016-35fb-5094-929b-558a96fad6f3"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["ColorTypes", "CSVFiles", "FilePathsBase", "HTTP", "Random", "Test"]
+test = ["ColorTypes", "CodecZlib", "CSVFiles", "FilePathsBase", "HTTP", "Random", "Test"]

src/registry.jl

@@ -59,23 +59,44 @@ detect_compressed(io, len=getlength(io); kwargs...) = detect_compressor(io, len;
 # test for RD?n magic sequence at the beginning of R data input stream
 function detect_rdata(io)
     seekstart(io)
-    getlength(io) <= 4 && return false
-    b = read(io, UInt8)
-    if b == UInt8('R')
-        return read(io, UInt8) == UInt8('D') &&
-               read(io, UInt8) in (UInt8('A'), UInt8('B'), UInt8('X')) &&
-               read(io, UInt8) in (UInt8('2'), UInt8('3')) &&
-               (c = read(io, UInt8); c == UInt8('\n') || (c == UInt8('\r') && read(io, UInt8) == UInt8('\n')))
+    function checked_match(io)
+        for m in (
+                (UInt8('R'), ),
+                (UInt8('D'), ),
+                (UInt8('A'), UInt8('B'), UInt8('X')),
+                (UInt8('2'), UInt8('3')))
+            eof(io) && return false
+            read(io, UInt8) in m || return false
+        end
+        c = read(io, UInt8)
+        if c == UInt8('\r')
+            eof(io) && return false
+            c = read(io, UInt8)
+        end
+        c == UInt8('\n') || return false
+        return true
     end
+    checked_match(io) && return true
     return detect_compressed(io; formats=["GZIP", "BZIP2", "XZ"])
 end
 
 add_format(format"RData", detect_rdata, [".rda", ".RData", ".rdata"], [idRData, LOAD])
 
 function detect_rdata_single(io)
     seekstart(io)
-    res = read(io, UInt8) in (UInt8('A'), UInt8('B'), UInt8('X')) &&
-        (c = read(io, UInt8); c == UInt8('\n') || (c == UInt8('\r') && read(io, UInt8) == UInt8('\n')))
+    function checked_match(io)
+        eof(io) && return false
+        read(io, UInt8) in (UInt8('A'), UInt8('B'), UInt8('X')) || return false
+        c = read(io, UInt8)
+        if c == UInt8('\r')
+            eof(io) && return false
+            c = read(io, UInt8)
+        end
+        c == UInt8('\n') || return false
+        return true
+    end
+    
+    res = checked_match(io)
     if !res
         res = detect_compressed(io; formats=["GZIP", "BZIP2", "XZ"])
     end

test/query.jl

@@ -518,4 +518,22 @@ let file_dir = joinpath(@__DIR__, "files"), file_path = Path(file_dir)
         @test FileIO.formatname(q) == :PNG
         rm("test.png")
     end
+
+    @testset "issue #345" begin
+        iris = joinpath("files", "iris.rda")
+
+        q = query(iris)
+        @test typeof(q) <: File{format"RData"}
+
+        io = open(iris)
+        q = query(io)
+        @test typeof(q) <: Stream{format"GZIP"} # FIXME: should be RData
+        @test FileIO.detect_rdata(io)
+        
+        # issue #345: it errors here
+        io = CodecZlib.GzipDecompressorStream(open(iris))
+        q = query(io)
+        @test FileIO.unknown(q) # FIXME: should be RData
+        @test FileIO.detect_rdata(io)
+    end
 end

test/runtests.jl

@@ -2,6 +2,7 @@ using FileIO
 using FilePathsBase
 using Test
 using UUIDs
+using CodecZlib
 
 Threads.nthreads() <= 1 && @info "Threads.nthreads() = $(Threads.nthreads()), multithread tests will be disabled"