fix: ensure automatically generated filename is a valid filename

[Monirzadeh]

some characters are not valid in some os like ? in windows.
we check filename and replace that kind of charachter with - so ``go-epub` always use valid filename in any os.
Fix #26
@fmartingr can you please test this in shiori before we merge that too?
if it is ok i will just add more unit test to cover all characters in test too

[codecov-commenter]

Codecov Report

Attention: 9 lines in your changes are missing coverage. Please review.

Comparison is base (2f66845) 83.37% compared to head (7080e4d) 82.87%.

Files	Patch %	Lines
epub.go	52.63%	4 Missing and 5 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #27      +/-   ##
==========================================
- Coverage   83.37%   82.87%   -0.51%     
==========================================
  Files          10       10              
  Lines        1209     1226      +17     
==========================================
+ Hits         1008     1016       +8     
- Misses        141      145       +4     
- Partials       60       65       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[fmartingr]

From a personal opinion, I wouldn't add a new dependency just for this. It's easier to just create a random string a-z0-9 of a fixed length and append the extension. I checked and the other library do not have any subdependencies, but in the end I think the filename inside the epub does not matter anyway.

[Monirzadeh]

From a personal opinion, I wouldn't add a new dependency just for this. It's easier to just create a random string a-z0-9 of a fixed length and append the extension. I checked and the other library do not have any subdependencies, but in the end I think the filename inside the epub does not matter anyway.

the why i try to avoid to just create a random string is that lots of other part of the code (legacy part) use that filename (extension part) so it breaks lots of things.
i need to review that more carefully.
the invalid character is limited in any os what do you think if i done that as function inside go-epub for now?

[Monirzadeh]

@fmartingr some other place in code go-epub use github.com/gabriel-vasile/mimetype do you think we use that here instead of mime is better?

[fmartingr]

@fmartingr some other place in code go-epub use github.com/gabriel-vasile/mimetype do you think we use that here instead of mime is better?

I didn't notice that dependency, was just doing some tests. If go-eupb already have it and it provides improvements over the standard library, yeah, absolutely.

[Monirzadeh]

@fmartingr some other place in code go-epub use github.com/gabriel-vasile/mimetype do you think we use that here instead of mime is better?

I didn't notice that dependency, was just doing some tests. If go-eupb already have it and it provides improvements over the standard library, yeah, absolutely.

I check that again, just because in the current solution we don't check the actual byte of the file (check only the header), it doesn't have any specific advantage.
If every things is ok. please approve this PR.

[fmartingr]

LGTM. I'd say this is a good compromise considering we are not downloading the files twice nor doing a big refactor in the code to guess the file type by its contents . 👍