Tar.extract_file: read the content of individual file(s),

that match a predicate, from the tarball archive.

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 +98,24 @@ 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(predicate::Function, tarball, out) -> Vector{Header}
+extract_file(predicate::AbstractString, tarball, out) -> Header
+```
+* `predicate :: Union{Function, AbstractString}`
+* `tarball   :: Union{AbstractString, AbstractCmd, IO}`
+* `out       :: Union{AbstractString, AbstractCmd, IO}`
+
+Read file(s) matching the predicate from `tarball` and write to `out`.
+Return the [`Header`](@ref)s of the matchin files.
+
+If `predicate::Function` it should take a `Header` as the only input
+argument and return `true`/`false`.
+If `predicate::String` it is interpreted as a path relative the
+tarball root and must only match a single entry.
+
 ### Tar.list
 
 ```jl

src/extract.jl

@@ -590,3 +590,46 @@ function read_data(
     r < n && error("premature end of tar file")
     return view(buf, 1:size)
 end
+
+
+"""
+    extract_file(predicate::Funtion, tarball, out) -> Vector{Header}
+    extract_file(predicate::AbstractString, tarball, out) -> Header
+
+Accepted argument types:
+ * `predicate :: Union{AbstractString, Function}`
+ * `tarball   :: Union{AbstractString, IO, Cmd}`
+ * `out       :: Union{AbstractString, IO, Cmd}`
+
+Read file(s) matching the predicate from `tarball` and write to `out`.
+Return the [`Header`](@ref)s of the matchin files.
+
+If `predicate::Function` it should take a `Header` as the only input
+argument and return `true`/`false`.
+If `predicate::String` it is interpreted as a path relative the
+tarball root and must only match a single entry.
+"""
+function extract_file(predicate::Function, tarball::ArgRead, out::ArgWrite)::Vector{Header}
+    headers = Header[]
+    arg_read(tarball) do tar; arg_write(out) do io
+        read_tarball(predicate, tar) do hdr, _
+            if hdr.type == :file # TODO: read symlinks??
+                push!(headers, hdr)
+                read_data(tar, io, size=hdr.size)
+            end
+        end
+    end end
+    return headers
+end
+function extract_file(predicate::AbstractString, tarball::ArgRead, out::ArgWrite)::Header
+    parts = filter!(x -> x != ".", splitpath(predicate))
+    headers = extract_file(tarball, out) do hdr
+        hdr_parts = filter!(x -> x != ".", splitpath(hdr.path))
+        hdr.type == :file && parts == hdr_parts
+    end
+    if length(headers) != 1
+        s = length(headers) == 0 ? "no" : "multiple"
+        throw(ArgumentError("$s files in the tarball matches the filename $predicate"))
+    end
+    return headers[1]
+end

test/runtests.jl

@@ -601,6 +601,75 @@ 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))
+            ## predicate::String
+            io = IOBuffer()
+            for pred in ("file.txt", "./file.txt")
+                hdr = Tar.extract_file(pred, tar(), io)
+                @test hdr.path == "file.txt"
+                @test hdr.size == 16
+                @test String(take!(io)) == "file at the root"
+            end
+
+            for pred in ("directory/file2.txt", "./directory/file2.txt")
+                hdr = Tar.extract_file(pred, tar(), io)
+                @test hdr.path == "directory/file2.txt"
+                @test hdr.size == 17
+                @test String(take!(io)) == "file in directory"
+            end
+
+            @test_throws ArgumentError("no files in the tarball matches the filename nope") Tar.extract_file("nope", tar(), io)
+
+            # predicate::Function
+            hdrs = Tar.extract_file(tar(), io) do hdr
+                hdr.path == "file.txt"
+            end
+            @test length(hdrs) == 1
+            @test hdrs[1].path == "file.txt"
+            @test hdrs[1].size == 16
+            @test String(take!(io)) == "file at the root"
+
+            hdrs = Tar.extract_file(hdr -> true, tar(), io)
+            @test length(hdrs) == 2
+            str = String(take!(io))
+            @test occursin("file at the root", str)
+            @test occursin("file in directory", str)
+            @test sum(h.size for h in hdrs) == sizeof(str)
+
+            hdrs = Tar.extract_file(hdr -> true, tar(), io)
+            @test length(hdrs) == 2
+            str = String(take!(io))
+            @test occursin("file at the root", str)
+            @test occursin("file in directory", str)
+            @test sum(h.size for h in hdrs) == sizeof(str)
+
+            hdrs = Tar.extract_file(hdr -> false, tar(), io)
+            @test length(hdrs) == 0
+            @test sizeof(String(take!(io))) == 0
+
+            # Non-IO output
+            mktempdir() do tmpd
+                o = joinpath(tmpd, "out.data")
+                Tar.extract_file("file.txt", tar(), o)
+                @test read(o, String) == "file at the root"
+                Tar.extract_file(hdr -> hdr.path == "file.txt", tar(), o)
+                @test read(o, String) == "file at the root"
+            end
+        end
+    end
+end
+
 @testset "API: rewrite" begin
     # reference standard tarball
     reference, hash₁ = make_test_tarball()