Is the usage of streams problematic?

[adhanji8]

Sorry for creating an issue out of this, you can close it right after. I just wanted clarification regarding the usage of streams in this library.

Yauzl has a point regarding streams that said:
Follow the spec. Don't scan for local file headers. Read the central directory for file metadata. (see No Streaming Unzip API).

I noticed that this library uses Streams. Is this an implementation detail that differs from the original Yauzl, or are streams being used in a different way that doesn't break the spec? Apologies for the potentially stupid question, appreciate it!

[overlookmotel]

Hi. Interesting question.

This library does not accept the input ZIP file as a stream. It requires random access, for the reasons you've quoted.

It starts at the end of the ZIP file and works backwards to find the Central Directory.

But once the Central Directory is found, that directory identifies the start and end positions of each file in the ZIP. At that point, yauzl-promise switches to streams in order to read each individual file from the ZIP and stream it out to the user.

This is exactly how yauzl works too, and in no way violates the ZIP spec.

I hope that makes sense. If it does, please close the issue. But if it doesn't, feel free to ask more questions!

[adhanji8]

Thank you for the amazing explanation and also for maintaining is incredible project. You are sincerely appreciated 🙏. Thank you.