archive/zip: Differentiate ReadDir result types

[cugu]

What version of Go are you using (go version)?

Current HEAD

$ go version
go version devel +cd99385ff4 Sat Jan 23 12:38:15 2021 +0000 darwin/amd64

What did you do?

reader, _ := zip.OpenReader("test.zip")
root, _ := reader.Open(".")
entries, _ := root.(fs.ReadDirFile).ReadDir(0)
for _, entry := range entries {
    fmt.Printf("%s: %T\n", entry.Name(), entry)
}

Results in:

folder: *zip.fileListEntry
folder: zip.headerFileInfo

I understand that you might want both results (s. #40354), so my question is:

Is there a good way to differentiate those internal types, if you just need one?

It escalates a little with fs.WalkDir:

reader, _ := zip.OpenReader("test.zip")
fs.WalkDir(reader, ".", func(path string, d fs.DirEntry, err error) error {
    fmt.Println(path)
    return nil
})

folder
folder/file.txt
folder/subfolder
folder/subfolder/subfile.txt
folder/subfolder/subsubfolder
folder/subfolder/subsubfolder/subsubfile.txt
folder/subfolder/subsubfolder
folder/subfolder/subsubfolder/subsubfile.txt
folder/subfolder
folder/subfolder/subfile.txt
folder/subfolder/subsubfolder
folder/subfolder/subsubfolder/subsubfile.txt
folder/subfolder/subsubfolder
folder/subfolder/subsubfolder/subsubfile.txt
folder
folder/file.txt
folder/subfolder
folder/subfolder/subfile.txt
folder/subfolder/subsubfolder
folder/subfolder/subsubfolder/subsubfile.txt
folder/subfolder/subsubfolder
folder/subfolder/subsubfolder/subsubfile.txt
folder/subfolder
folder/subfolder/subfile.txt
folder/subfolder/subsubfolder
folder/subfolder/subsubfolder/subsubfile.txt
folder/subfolder/subsubfolder
folder/subfolder/subsubfolder/subsubfile.txt

@rsc

[cugu]

Ok this seem like an issue, fs.Stat chooses one of both types at random.

for i := 0; i < 10; i++ {
	reader, _ := zip.OpenReader("go1.15.7.windows-386.zip")
	info, _ := fs.Stat(reader, "go")
	fmt.Printf("%T\n", info)
}

Results in:

zip.headerFileInfo
*zip.fileListEntry
zip.headerFileInfo
*zip.fileListEntry
zip.headerFileInfo
*zip.fileListEntry
*zip.fileListEntry
*zip.fileListEntry
*zip.fileListEntry
zip.headerFileInfo

[seankhliao]

I think there should be a continue in the inner for loop?

go/src/archive/zip/reader.go

Lines 676 to 685 in b634f5d

	for _, file := range r.File {
	name := toValidName(file.Name)
	for dir := path.Dir(name); dir != "."; dir = path.Dir(dir) {
	dirs[dir] = true
	}
	r.fileList = append(r.fileList, fileListEntry{name, file})
	}
	for dir := range dirs {
	r.fileList = append(r.fileList, fileListEntry{dir + "/", nil})
	}

[cugu]

zip.headerFileInfo objects do have a meaningful ModTime() while *zip.fileListEntry do not. So I would prefer to get the zip.headerFileInfo objects on default and maybe have another method for *zip.fileListEntry?

[gopherbot]

Change https://golang.org/cl/311530 mentions this issue: archive/zip: only return directory once via io/fs.FS

[gopherbot]

Change https://golang.org/cl/315249 mentions this issue: [release-branch.go1.16] archive/zip: only return directory once via io/fs.FS