Add a function to read the data of a single file in the tarball.

Author: fredrikekre

Project.toml

@@ -1,7 +1,7 @@
 name = "Tar"
 uuid = "a4e569a6-e804-4fa4-b0f3-eef7a1d5b13e"
 authors = ["Stefan Karpinski <[email protected]>"]
-version = "1.9.0"
+version = "1.10.0"
 
 [deps]
 ArgTools = "0dad84c5-d112-42e6-8d28-ef12dabb789f"

README.md

@@ -98,6 +98,19 @@ will also not be copied and will instead be skipped. By default, `extract` will
 detect whether symlinks can be created in `dir` or not and will automatically
 copy symlinks if they cannot be created.
 
+### `Tar.extract_file`
+
+```jl
+extract_file(tarball, filepath) -> Vector{UInt8}
+```
+* `tarball  :: Union{AbstractString, AbstractCmd, IO}`
+* `filepath :: AbstractString`
+
+Read the content of a single file inside the tarball archive.
+Return a `Vector{UInt8}` with the data, or `nothing` if no
+matching file was found. `filepath` should be specified as a path
+relative the tarball root.
+
 ### Tar.list
 
 ```jl

src/extract.jl

@@ -590,3 +590,36 @@ function read_data(
     r < n && error("premature end of tar file")
     return view(buf, 1:size)
 end
+
+
+"""
+    extract_file(tarball::Union{AbstractString, IO, Cmd}, file::AbstractString) -> Vector{UInt8}
+
+Extract the content of a single file from the tarball.
+Return a `Vector{UInt8}` with the data, or `nothing` if no
+matching file was found. `file` should be specified as a path
+relative the tarball root.
+"""
+function extract_file(tarball::ArgRead, file::AbstractString)::Union{Nothing,Vector{UInt8}}
+    arg_read(tarball) do tar
+        # TODO: Correct to filter out "."?
+        parts = filter!(x -> x != ".", split(file, '/'; keepempty=false))
+        predicate = hdr -> begin
+            hdr_parts = filter!(x -> x != ".", split(hdr.path, '/'; keepempty=false))
+            hdr.type == :file && parts == hdr_parts
+        end
+        buf = IOBuffer()
+        found = false
+        Tar.read_tarball(predicate, tar) do hdr, _
+            found && throw(ArgumentError("multiple files in the tarball matches the filename $file"))
+            found = true
+            Tar.read_data(tar, buf, size=hdr.size)
+        end
+        if found
+            return take!(buf)
+        else
+            # TODO: Better to throw instead?
+            return nothing
+        end
+    end
+end

test/runtests.jl

@@ -601,6 +601,31 @@ end
     end
 end
 
+@testset "API: extract_file" begin
+    mktempdir() do dir
+        open(joinpath(dir, "file.txt"), "w") do io
+            write(io, "file at the root")
+        end
+        dir2 = mkdir(joinpath(dir, "directory"))
+        open(joinpath(dir2, "file2.txt"), "w") do io
+            write(io, "file in directory")
+        end
+        tarball = Tar.create(dir)
+        for tar in (()->tarball, ()->open(tarball))
+            bytes = Tar.extract_file(tar(), "file.txt")
+            @test String(bytes) == "file at the root"
+            bytes = Tar.extract_file(tar(), "./file.txt")
+            @test String(bytes) == "file at the root"
+            bytes = Tar.extract_file(tar(), "directory/file2.txt")
+            @test String(bytes) == "file in directory"
+            bytes = Tar.extract_file(tar(), "./directory/file2.txt")
+            @test String(bytes) == "file in directory"
+            bytes = Tar.extract_file(tar(), "non-existent")
+            @test bytes === nothing
+        end
+    end
+end
+
 @testset "API: rewrite" begin
     # reference standard tarball
     reference, hash₁ = make_test_tarball()