ZIP ReaderFactory IEntry properties are not assigned

adamhathcock / sharpcompress

SharpCompress is a fully managed C# library to deal with many compression types and formats.

MIT License

2.26k stars 479 forks source link

ZIP ReaderFactory IEntry properties are not assigned #753

Open jbrockerville opened 1 year ago

jbrockerville commented 1 year ago

I have ZIP files created with a unknown app. I am accessing the IEntry entries via the ReaderFactory (a la USAGE.md). The entries contain their key, but all the other properties (e.g., CRC, uncompressed size, compressed size, various date-times, etc.) are not assigned (set to zero or null).

A couple interesting observations:

Despite the unassigned IEntry properties, opening the entry stream and copying it works fine.
If I use ArchiveFactory to access these ZIP files, their IEntry properties are assigned.
If I repack the ZIP files (with 7zip), ReaderFactory then works fine.

Unfortunately, I am not at liberty to share the offending archives. Instead, I have attached the output from Info-Zip from an original and repacked file. I hope some information can be gleaned from them. If I had to guess, I would say it is a problem with the zip headers.

archive.zip.txt archive-repack.zip.txt

Erior commented 1 year ago

The archive is probably written in a Stream mode, that is when you don't know that size of the file being added. CRC and sizes would then be 0, date time etc might or might not be set. And it comforms with your statements above and also the dump you generated with Info-Zip

Check the test file Zip.none.datadescriptors.zip

You can use zipdetails command if it is available to you, it should show the "Local Header" information.

adamhathcock commented 1 year ago

To echo what @Erior it saying: if the zips were made with Streaming mode, the File Headers have missing data. There should be Flie Trailers that have sizes and CRC.

These can't be known ahead of time with Reader because the file sizes, etc are unknown with Reader mode and Streaming zips

jbrockerville commented 1 year ago

Huh... I was unaware that you could do that. I guess I just learned something. I can accept that explanation. Thanks for looking into it and replying. I've made a solution to handle when we encounter an archive like this. But it's nice to know it's a feature, not a bug. ;)

Maybe add some words somewhere describing this possibility?

adamhathcock commented 1 year ago

It's not really a feature, just a known short coming of Reader (Streaming mode) and Streaming made zip files.

How do you suggest describing it? People always want more docs but I haven't found a good way to for this project as well as in my day job :)