assert that input and output buffers passed to transcode! are indepen…

…dent (#140)

* assert that input and output buffers passed to transcode! are independent

* Use Base.dataids

Co-authored-by: Mark Kittisopikul <mkitti@users.noreply.github.com>

* use Base.mightalias and add test

* fix noop codec implementation

* add unsafe_transcode

* add test

---------

Co-authored-by: Mark Kittisopikul <mkitti@users.noreply.github.com>

src/noop.jl

@@ -116,16 +116,15 @@ function Base.unsafe_write(stream::NoopStream, input::Ptr{UInt8}, nbytes::UInt)
     end
 end
 
-function Base.transcode(::Type{Noop}, data::ByteData)
-    return transcode(Noop(), data)
-end
+initial_output_size(codec::Noop, input::Memory) = length(input)
 
-function Base.transcode(codec::Noop, input::Buffer, output::Buffer = Buffer())
-    copydata!(output, input)
-    return output.data
+function process(codec::Noop, input::Memory, output::Memory, error::Error)
+    iszero(length(input)) && return (0, 0, :end)
+    n = min(length(input), length(output))
+    unsafe_copyto!(output.ptr, input.ptr, n)
+    (n, n, :ok)
 end
 
-
 # Stats
 # -----
 

src/transcode.jl

@@ -112,16 +112,34 @@ end
 """
     transcode!(output::Buffer, codec::Codec, input::Buffer)
 
-Transcode `input` by applying `codec` and storing the results in `output`.
-Note that this method does not initialize or finalize `codec`. This is
-efficient when you transcode a number of pieces of data, but you need to call
-[`TranscodingStreams.initialize`](@ref) and
+Transcode `input` by applying `codec` and storing the results in `output`
+with validation of input and output.  Note that this method does not initialize
+or finalize `codec`. This is efficient when you transcode a number of
+pieces of data, but you need to call [`TranscodingStreams.initialize`](@ref) and
 [`TranscodingStreams.finalize`](@ref) explicitly.
 """
 function transcode!(
     output::Buffer,
     codec::Codec,
     input::Buffer,
+)
+    @assert !Base.mightalias(input.data, output.data) "input and outbut buffers must be independent"
+    unsafe_transcode!(output, codec, input)
+end
+
+"""
+    unsafe_transcode!(output::Buffer, codec::Codec, input::Buffer)
+
+Transcode `input` by applying `codec` and storing the results in `output`
+without validation of input or output.  Note that this method does not initialize
+or finalize `codec`. This is efficient when you transcode a number of
+pieces of data, but you need to call [`TranscodingStreams.initialize`](@ref) and
+[`TranscodingStreams.finalize`](@ref) explicitly.
+"""
+function unsafe_transcode!(
+    output::Buffer,
+    codec::Codec,
+    input::Buffer,
 )
     error = Error()
     code = startproc(codec, :write, error)
@@ -171,6 +189,12 @@ Base.transcode(codec::Codec, data::Buffer, output::ByteData) =
 Base.transcode(codec::Codec, data::ByteData, args...) =
     transcode(codec, Buffer(data), args...)
 
+unsafe_transcode!(codec::Codec, data::Buffer, output::ByteData) =
+    unsafe_transcode!(Buffer(output), codec, data)
+
+unsafe_transcode!(codec::Codec, data::ByteData, args...) =
+    unsafe_transcode!(codec, Buffer(data), args...)
+
 # Return the initial output buffer size.
 function initial_output_size(codec::Codec, input::Memory)
     return max(

test/codecnoop.jl

@@ -189,6 +189,9 @@
     @test transcode(Noop, data)  == data
     @test transcode(Noop, data) !== data
 
+    data = Vector{UInt8}()
+    @test TranscodingStreams.unsafe_transcode!(Noop(), data, data) == data
+    @test_throws AssertionError transcode(Noop(), data, data)
     data = b""
     @test transcode(Noop(), data)  == data
     @test transcode(Noop(), data) !== data