Hi Daniel Shahaf,

Thank you very much.

*Thanks and Best Regards,Ahmad Ismail*


On Tue, Nov 17, 2020 at 1:53 AM Daniel Shahaf <d.s@daniel.shahaf.name>
wrote:

> Ahmad Ismail wrote on Sat, 14 Nov 2020 03:45 +0600:
> > Hi All,
> >
> > I have the following functions that rename files from camel to sort of
> > kebab case.
> >
> > function ctokfiles() {
> >     # Transform [Capital][Capital][Small] to [Capital]-[Capital][Small]
> >     zmv -Q '(**/)(*[A-Z][A-Z][a-z]*)(.adoc|.txt)'
> '$1${2//(#b)([A-Z])([A-Z][a-z])/$match[1]-$match[2]}$3'
> >
> >     # Put - Between [Small][Capital]
> >     zmv -Q '(**/)(*[a-z][A-Z]*)(.adoc|.txt)'
> '$1${2//(#b)([a-z])([A-Z])/$match[1]-$match[2]}$3'
> >
> >     # Change [Capital][Small] to Lovercase
> >     zmv -Q '(**/)(*[A-Z][a-z]*)(.adoc|.txt)'
> '$1${2//(#m)[A-Z][a-z]/${(L)MATCH}}$3'
> > }
> >
> > So, IF the input name is "ThisIsMyOCDTalking", it becomes
> > "this-is-my-OCD-talking".
> >
> > Now I want to rename directories, adoc & txt files only. For
> directories, I
> > think I have to use
> >
> >     zmv -Q '(**/)(*[A-Z][A-Z][a-z]*)(/)'
> >
> > How to make a function that will work on adoc, txt and directories. I
> tried
> > (/|.adoc|.txt), apparently not working.
> >
>
> Consider how in «*foo*(bar)», the «foo» are matched against the
> filename but the «bar» are parsed as glob qualifiers.  You can't just
> juxtapose them within the same set of parentheses; they're apples and
> oranges.
>
> I don't see a good solution for "Everything in «**/*» that's
> a directory or named *.foo or *.bar".  I'd consider find(1) for this.
>
> > One more thing, is there any way I can reduce the steps in this
> functions.
>
> As a rule, I wouldn't recommend reducing steps in functions in the
> first place, unless you're coding for sport rather than robustness — in
> which case, I'd probably opt for Perl:
>
> % print -rNC1 -- *.(adoc|txt)(N) *(/N) | rename -0 -n '$_ = join "-", map
> { /[a-z]/ ? lc : $_ } split /(?<=[a-z])(?=[A-Z])|(?=[A-Z][a-z])/'
> rename(ThisIsMyOCDTalking.adoc, this-is-my-OCD-talking.adoc)
> %
>
> The main algorithmic difference is splitting the filename into an array
> on zero-width lookaround matches.  I don't think there's a native zsh
> equivalent of this.
>
> Note that changing «print -rNC1» to «ls» would introduce two separate bugs.
>
> I'm not sure how to extend this to nested hierarchies, given the
> possibility of foo/bar where both foo and bar have to be changed.
> Perhaps with «rename -d» and ensuring the input is sorted children
> before their parents.
>
> Cheers,
>
> Daniel
>