Exclude tests and testdata from published crate

[AlyoshaVasilieva]

Reduces size of crate from 8975KB to 17KB
crates.io limits crates to 10MB max

https://doc.rust-lang.org/cargo/reference/manifest.html#the-exclude-and-include-fields

[bojand]

Sigh... Hmm this was actually added in 0.7.0 to solve #47 and #54. I do think this is more appropriate (hence original implementation), and Linux distro packaging should use GitHub release archive perhaps...? But frankly I am not familiar with those processes.

@remilauzier @janbaudisch

[ghost]

On fedora we use a python script that take the source from crates.io and create a .spec file for fedora to ease management. If the files are no longer in the crates source, the policy of fedora is to deactivate the tests to not have to import all the file manually.

[AlyoshaVasilieva]

An alternate option might be to trim the testdata files to only about as long as their header/use smaller files, to avoid hitting the crates.io limit in the future

[janbaudisch]

I can see that the crate size right now is definitely not ideal. I didn't intent to force outside guidelines onto the project (I also intend to package this crate for Fedora). It just felt right for a crate to be the complete package with everything needed, be it running it or testing it.
Maybe reducing the size of the test files could be a solution? I don't really know much about the internals but that should be possible to do, or would it break anything?

[AlyoshaVasilieva]

For at least some formats it should be possible. For example, the MP4 matcher currently checks 12 bytes but the test file is about 2MB. I'd guess most test files can be cut to 4KB or less without issues (but I am by no means an expert on file signatures)

edit: Truncating all test files >4KB to 4KB, except DOC/PPT/XLS, keeps all tests passing and reduces size of .crate to 154KB

[ghost]

@janbaudisch
Infer is already package in fedora.
https://src.fedoraproject.org/rpms/rust-infer

[ghost]

@AlyoshaVasilieva
Thanks for maintaining the test passing in crates.io!

[janbaudisch]

@remilauzier Oh, I didn't realize that you already packaged it. Thanks for the heads-up.