Re: [Savannah-hackers-public] Assaf's SAE regexp remote method

[Assaf Gordon]

Hello Bruno,

On 09/15/2015 10:49 PM, Bruno Félix Rezende Ribeiro wrote:
> For instance, when evaluating the blimpserver[1] submission I saw that
> SAE detects XML files as "not-text" and skips the license check on
> them.  That's not correct as XML may be handcrafted source.

Detection is done with "file", and patches are welcomed.

> Then I went to your online remote SAE service[2] so I could run it on
> blimpserver's source tree and easily point you to the results in my
> bug report, but found out two problems preventing me from using it:
> zip is yet not supported

Zip is indeed not yet supported, and patches are welcomed.

> and the address Savannah gives to the package
> is
>
>    https://savannah.gnu.org/task/download.php?file_id=34880
>
> which doesn't match the expected regular expression
>
>    ^(https?|ftp)://[A-Za-z0-9\_\.\/-]*\.tar\.(gz|bz2|xz)$
>
> Why not just fetch whatever is at the given URL and then run 'file -i'
> on it and only then discriminate for the method of access?

Since the code that runs (on my machine) will do a 'wget' on which ever URL you 
give it,
I wanted to limit it to a (somewhat) reasonable subset.
It's just a trade-off - not really secure or bullet-proof, and doesn't answer 
all cases.
But it's sufficient for most cases of simple project registration.

As before, patches and improvements are welcomed.

The code is here:
  http://git.housegordon.org/cgit/gsv-eval.git/


regards,
 - assaf