Tar.extract: add set_permissions keyword argument (#113)

Co-authored-by: Stefan Karpinski <stefan@karpinski.org>

README.md

@@ -63,14 +63,19 @@ recreated. The `skeleton` and `predicate` arguments cannot be used together.
 ### Tar.extract
 
 ```jl
-extract([ predicate, ] tarball, [ dir ];
-        [ skeleton, ] [ copy_symlinks ]) -> dir
+extract(
+    [ predicate, ] tarball, [ dir ];
+    [ skeleton = <none>, ]
+    [ copy_symlinks = <auto>, ]
+    [ set_permissions = true, ]
+) -> dir
 ```
-* `predicate     :: Header --> Bool`
-* `tarball       :: Union{AbstractString, AbstractCmd, IO}`
-* `dir           :: AbstractString`
-* `skeleton      :: Union{AbstractString, AbstractCmd, IO}`
-* `copy_symlinks :: Bool`
+* `predicate       :: Header --> Bool`
+* `tarball         :: Union{AbstractString, AbstractCmd, IO}`
+* `dir             :: AbstractString`
+* `skeleton        :: Union{AbstractString, AbstractCmd, IO}`
+* `copy_symlinks   :: Bool`
+* `set_permissions :: Bool`
 
 Extract a tar archive ("tarball") located at the path `tarball` into the
 directory `dir`. If `tarball` is an IO object instead of a path, then the
@@ -105,6 +110,8 @@ 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.
 
+If `set_permissions` is `false`, no permissions are set on the extracted files.
+
 ### Tar.list
 
 ```jl

src/Tar.jl

@@ -166,14 +166,19 @@ function list(
 end
 
 """
-    extract([ predicate, ] tarball, [ dir ];
-            [ skeleton, ] [ copy_symlinks ]) -> dir
-
-        predicate     :: Header --> Bool
-        tarball       :: Union{AbstractString, AbstractCmd, IO}
-        dir           :: AbstractString
-        skeleton      :: Union{AbstractString, AbstractCmd, IO}
-        copy_symlinks :: Bool
+    extract(
+        [ predicate, ] tarball, [ dir ];
+        [ skeleton = <none>, ]
+        [ copy_symlinks = <auto>, ]
+        [ set_permissions = true, ]
+    ) -> dir
+
+        predicate       :: Header --> Bool
+        tarball         :: Union{AbstractString, AbstractCmd, IO}
+        dir             :: AbstractString
+        skeleton        :: Union{AbstractString, AbstractCmd, IO}
+        copy_symlinks   :: Bool
+        set_permissions :: Bool
 
 Extract a tar archive ("tarball") located at the path `tarball` into the
 directory `dir`. If `tarball` is an IO object instead of a path, then the
@@ -207,13 +212,16 @@ link to `/etc/passwd` will not be copied. Symlinks which are in any way cyclic
 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.
+
+If `set_permissions` is `false`, no permissions are set on the extracted files.
 """
 function extract(
     predicate::Function,
     tarball::ArgRead,
     dir::Union{AbstractString, Nothing} = nothing;
     skeleton::Union{ArgWrite, Nothing} = nothing,
     copy_symlinks::Union{Bool, Nothing} = nothing,
+    set_permissions::Bool = true,
 )
     predicate === true_predicate || skeleton === nothing ||
         error("extract: predicate and skeleton cannot be used together")
@@ -230,6 +238,7 @@ function extract(
                     predicate, tar, dir,
                     skeleton = skeleton,
                     copy_symlinks = copy_symlinks,
+                    set_permissions = set_permissions,
                 )
             end
         end
@@ -241,11 +250,13 @@ function extract(
     dir::Union{AbstractString, Nothing} = nothing;
     skeleton::Union{ArgWrite, Nothing} = nothing,
     copy_symlinks::Union{Bool, Nothing} = nothing,
+    set_permissions::Bool = true,
 )
     extract(
         true_predicate, tarball, dir,
         skeleton = skeleton,
         copy_symlinks = copy_symlinks,
+        set_permissions = set_permissions,
     )
 end
 

src/extract.jl

@@ -53,6 +53,7 @@ function extract_tarball(
     buf::Vector{UInt8} = Vector{UInt8}(undef, DEFAULT_BUFFER_SIZE),
     skeleton::IO = devnull,
     copy_symlinks::Bool = false,
+    set_permissions::Bool = true,
 )
     paths = read_tarball(predicate, tar; buf=buf, skeleton=skeleton) do hdr, parts
         # get the file system version of the path
@@ -82,7 +83,7 @@ function extract_tarball(
             error("unsupported tarball entry type: $(hdr.type)")
         end
         # apply tarball permissions
-        if hdr.type in (:file, :hardlink)
+        if set_permissions && hdr.type in (:file, :hardlink)
             exec = 0o100 & hdr.mode != 0
             tar_mode = exec ? 0o755 : 0o644
             sys_mode = filemode(sys_path)
@@ -139,7 +140,7 @@ function extract_tarball(
                 src = reduce(joinpath, init=root, split(what, '/'))
                 dst = reduce(joinpath, init=root, split(path, '/'))
                 cp(src, dst)
-                if Sys.iswindows()
+                if set_permissions && Sys.iswindows()
                     # our `cp` doesn't copy ACL properties, so manually set them via `chmod`
                     function copy_mode(src::String, dst::String)
                         chmod(dst, filemode(src))

test/runtests.jl

@@ -584,6 +584,7 @@ end
         @arg_test tar @test_throws ErrorException tar_count(tar, strict=true)
         @arg_test tar @test n + 1 == tar_count(tar, strict=false)
     end
+    rm(tarball)
 end
 
 @testset "API: extract" begin
@@ -708,6 +709,23 @@ end
             end
         end
     end
+    rm(tarball)
+
+    @testset "set_permissions" begin
+        tarball, _ = make_test_tarball()
+        dir = Tar.extract(tarball, set_permissions=false)
+        f_path = joinpath(dir, "0-ffffffff")
+        x_path = joinpath(dir, "0-xxxxxxxx")
+        @test isfile(f_path)
+        @test isfile(x_path)
+        if !Sys.iswindows()
+            @test !Sys.isexecutable(f_path)
+            @test !Sys.isexecutable(x_path)
+        end
+        @test Sys.isexecutable(f_path) == Sys.isexecutable(x_path)
+        rm(dir, recursive=true)
+        rm(tarball)
+    end
 end
 
 @testset "API: rewrite" begin
@@ -849,7 +867,7 @@ end
 
 if Sys.iswindows() && Sys.which("icacls") !== nothing && VERSION >= v"1.6"
     @testset "windows permissions" begin
-        tarball, hash = make_test_tarball()
+        tarball, _ = make_test_tarball()
         mktempdir() do dir
             Tar.extract(tarball, dir)
             f_path = joinpath(dir, "0-ffffffff")
@@ -865,6 +883,7 @@ if Sys.iswindows() && Sys.which("icacls") !== nothing && VERSION >= v"1.6"
             x_acl = readchomp(`icacls $(x_path)`)
             @test occursin("Everyone:(RX,WA)", x_acl)
         end
+        rm(tarball)
     end
 end