Re: [Linuxsampler-devel] wav2gig naming-convention

[Kolja K. <ko...@fr...>]

> Could you please elaborate, i.e. examples of where you would see
> > issues by 
> > just supplying regex patterns from the command line?

Sure!

I don't know in which order the attributes might be.
The regex you implemented uses a fixed order:

'(Name1)(Name2)(Velocity)(Note_Nr)(Note_Name)'

Take the following filename:

'Name - 64 - Bb3 - articulation.wav'

A regex-pattern could look like this:

'(.*) - (\d*) - (.*) - (.*).wav'

But how would the code know, that '64' is meant to be the velocity?
Or that 'Bb3' is a NoteName and 'articulation' is Name2?

One would also somehow have to enter the correct order, like

'(Name1)(Velocity)(Note_Name)(Name2)',

or am I missing something?


By using format specifiers, like in printf, you would simply enter:

'%n - %v - %a - %m.wav'
(where %n = Name1, %v = velocity, %a = Note_Name, %m = Name2)



I implemented some more checks today, each attribute is now checked
against a given regex-string, e.g. NoteName:

(^[c|C|d|D|e|E|f|F|g|G|a|A|b|B].*-?\\d+$)

where a pattern for the signs ('.*' in the string) will have to be
supplied if no Note_Nr is available in the filename (defaults to '#'
and 'b').

Also, it is checked that all required values for further processing are
available (for now: Name1, Velocity, Note_Nr).

It 'should' be quite robust now, I believe.
Both updated files are attached (patch created against wav2gig.cpp from
today).


Of course, all the information and examples at the beginning could go
into the manpage, but I would recommend to at least print the available
specifiers (if you want to use the filenamescanner at all...;).


> The patch also adds
> > > "
> > >     } else {
> > >         s->MIDIUnityNote = wav->note;
> > > "
> > > as you proposed.
> > 
> > I committed that now, because no matter how the naming scheme
> > evolves, this 
> > change makes sense anyway.
Cool, thanks!


Cheers,
Kolja