Append/Insert files into an existing Zip file?

[cdancy]

Append/Insert files into an existing Zip file?

[mholt]

Ah wow, I didn't know that was possible since even the official zip tooling doesn't support it.

Looks complicated. I can see why it's not obvious...

Let me see if I can get that library working with this one.

[mholt]

I mean, that lib works as far as I can tell, I don't know of another pure Go solution. You can try the latest commit I just pushed and see if it works for you. But if you're doing custom concurrency stuff maybe just best to use that lib directly...

[cdancy]

@mholt I'll give it a go and see how things fare. We haven't rolled out to production yet and so are still coding things up.

[mholt]

I don't think it's possible with the zip format, even the zip command line tool temporarily extracts the archive and re-creates it, apparently: https://wpguru.co.uk/2018/12/how-to-add-files-to-an-existing-zip-archive-on-macos-and-linux/

Tar is appendable because the format is basically just like delimiter-separated files, so you can easily add to it.

Closing, but feel free to continue discussion if needed.

[cdancy]

@mholt I found this library which does exactly what I need. I found some github issues online where this was requested by the below developer to be included in OOTB golang but core devs pushed back saying there wasn't enough need for it but if the community really wanted it they could look at putting it in.

https://github.com/STARRY-S/zip

[mholt]

Hmmm... I cannot get it to create a zip file that isn't corrupted (according to unzip -vl test.zip). Were you able to get it to work?

[mholt]

D'oh -- just kidding. I forgot to close the Updater.

I have a commit that works with a single test I did (:sweat_smile:) so you can use it if you want!

[cdancy]

@mholt nice! Let me know how it goes and if you're able to optimize things one way or another or "make the impl better" whatever that means in this context :) Once you've got something I can give it a go here. We've basically got a highly parallel process which keeps open the "zip updater", and as files are finished, we're appending them to the zip and then shipping that off. I don't think that other library is maintained or active anymore so if you can whip something up I'd be more than happy to take yours ;)

[cdancy]

@mholt with this new addition, and the way you implemented it, is it possible here to use a different compressor to get a smaller zip size? We're not necessarily concerned about how fast it takes to build the zip but more if we can get the zip size as small as possible. The files inside are all textual.

[mholt]

@cdancy Set the Compression field of your Zip struct: https://pkg.go.dev/github.com/mholt/archiver/v4#Zip.Compression (e.g. zip.Deflate)

[cdancy]

@mholt I'm having no luck :( No matter what I use I can't get the zip of the file to not be significantly larger than what I get with the starry-zip library. Same 4 files I'm using there and here results in a 57K versus 764K respectively. Maybe I'm doing something wrong?

        // have to use zip.NewWriter otherwise library was complaining that zip was not valid
        
	zipWriter := zip.NewWriter(zipFile)
	zipWriter.SetComment("Hello, World!")
	zipWriter.Close()

	zipper := archiver.Zip{
		Compression: flate.BestCompression,
	}

	err = zipper.Insert(context.Background(), zipFile, files)
	require.NoError(t, err)

[mholt]

@cdancy I think you might need to use zip.Deflate instead of flate.BestCompression, which is probably a uint8 that isn't recognized, so maybe it treats it as "store" instead of "compress".

[cdancy]

@mholt yeah I tried that as well but still nothing. When I open the zip written by the starry golang lib I see Defl:N compression used by default but no matter what I do here I only ever see Stored. I'm on mac-os so not sure if that plays into things at all.

L105342MUS:kadiv14103107114 dancc$ unzip -vl example.zip
Archive:  example.zip
Hello, World!
 Length   Method    Size  Cmpr    Date    Time   CRC-32   Name
--------  ------  ------- ---- ---------- ----- --------  ----
  687922  Stored   687922   0% 12-31-1979 00:00 2f6acde1  file1.txt
  354646  Stored   354646   0% 12-31-1979 00:00 9150eeac  file2.txt
   77921  Stored    77921   0% 12-31-1979 00:00 be9df062  file3.txt
  390349  Stored   390349   0% 12-31-1979 00:00 c6890ae4  file4.txt
--------          -------  ---                            -------
 1510838          1510838   0%                            4 files
L105342MUS:kadiv14103107114 dancc$ unzip -vl example-1.zip
Archive:  example-1.zip
Hello, World!
 Length   Method    Size  Cmpr    Date    Time   CRC-32   Name
--------  ------  ------- ---- ---------- ----- --------  ----
   77921  Defl:N     5752  93% 00-00-1980 00:00 be9df062  file1.txt
  390349  Defl:N    28739  93% 00-00-1980 00:00 c6890ae4  file2.txt
  354646  Defl:N    18367  95% 00-00-1980 00:00 9150eeac  file3.txt
  687922  Defl:N    40838  94% 00-00-1980 00:00 2f6acde1  file4.txt
--------          -------  ---                            -------
 1510838            93696  94%                            4 files

I don't know ...

[mholt]

I'm not sure if reusing the zip file after the zip writer wrote to it is a good idea. What if you have a fresh open file for the insert?