Would you accept a PR to search inside archives?

[mqudsi]

Hello,

Would you be open to accepting a PR that searched within archives (starting with .zip files)?

This could be easily implemented using the zip crate, which would introduce dependencies on crc32fast, crossbeam-utils (already a crossbeam dependency in fd's own), indexmap, and memchr. It can obviously be gated behind a feature flag, but you'd have to figure out whether you'd want it supported by default or not (with the primary difference being how it ends up packaged by distros).

[tmccombs]

What would be the benefit of that over something like

unzip -l  file.zip| grep ...

fd's parallelization wouldn't help that much, since the archive file has to be read in series. And the ignore functionality probably isn't as useful for inspecting archives.

Also, if we added support for searching zip files, we would probably also want support for tarballs (with various kinds of compression). And possibly other archive formats, such as 7zip as well.

[tavianator]

I think the benefit would be if you don't know exactly which .zip file contains the file you're looking for. But I'm not sure that potential benefit outweighs the added complexity to fd.

[mqudsi]

Yes, as @tavianator says, it would be to search for a file you know exists in a directory of uncompressed and compressed files, where the file could be in either. In my case, the Downloads directory.

[sergeevabc]

There is a folder full of *.tar.zst archives. Somewhere within these archives there is a FOO file I need. What's the easiest way to find it? If we were talking about finding a piece of text within the archives, I would use rg -z FOO or ug -z FOO, that is so simple.

$ fd -V
fd 10.2.0

$ fd -z FOO
error: unexpected argument '-z' found

[tavianator]

@sergeevabc Something like this should work:

$ fd -e tar.zst -x tar tf | grep FOO

[sergeevabc]

$ fd -e tar.zst -x tar tf | grep FOO
usr/bin/FOO

File is found, good to know it exists, but I still don't know which archive it is in. Ideas?

[tavianator]

Oh right. Uh, I'd probably do something like this:

$ fd -e tar.zst -x sh -c 'printf "\\n%s\\n" "$1" >&2 && tar tf "$1"' sh | grep FOO

[sergeevabc]

Meet zfind by @laktak. It searches for files, including inside tar, zip, 7z and rar archives, via SQL-syntax. Alas, it lacks zst and gz and xz support, some progress indicator, colors to distinguish the components of the path. It also fully utilizes 1 CPU core and is full of quotation marks. Nevertheless, it does the job. Better than nothing.

$ zfind "name like 'cygintl-8.dll'"
cygwin\x86_64\release\gettext\libintl8\libintl8-0.22.4-1.tar.xz//usr/bin/cygintl-8.dll

[sergeevabc]

Meet WinRAR. Its search is faster than zfind and supports more archive formats, including zst. However, this is a GUI app.

[sergeevabc]

Meet ugrep by @genivia-inc. It is bloody fast.

$ ug -z -l -g cygintl-8.dll ""
x86_64\
╰╴release\
│ ╰╴gettext\
│ │ ╰╴libintl8\
│ │ │ ╰╴libintl8-0.22.4-1.tar.xz{usr/bin/cygintl-8.dll}
▔ ▔ ▔