[RFC PATCH 1/2] complete absolute paths for mpc add

[RFC PATCH 1/2] complete absolute paths for mpc add

[Karel Balej]

When connecting via Unix socket, mpc can queue files from outside the
music directory as well when given the absolute path.
---
mpc also supports explicitly prefixing the path with file:// which is
not covered here, however according to my testing this only works for
absolute paths too so there doesn't seem to be any benefit.

This implementation doesn't work with environment variables, such as
mpc add $HOME/somethi<tab>, I will appreciate tips on whether there is a
simple way to make this more robust to include situations such as this
instead of just checking for these two specific starting characters.

Thank you and kind regards,
K. B.

 Completion/Unix/Command/_mpc | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/Completion/Unix/Command/_mpc b/Completion/Unix/Command/_mpc
index c3f93878ca28..93100de09ca8 100644
--- a/Completion/Unix/Command/_mpc
+++ b/Completion/Unix/Command/_mpc
@@ -178,6 +178,14 @@ _mpc_helper_files() {
   fi
 }
 
+(( $+functions[_mpc_helper_all_files] )) ||
+_mpc_helper_all_files() {
+  if [[ $words[CURRENT] == [/~]* ]]; then
+	  _files
+  fi
+  _mpc_helper_files
+}
+
 (( $+functions[_mpc_helper_directories] )) ||
 _mpc_helper_directories() {
   if [[ -n $MPD_MUSIC_DIR ]]; then
@@ -204,7 +212,7 @@ _mpc_helper_outputs() {
 }
 
 _mpc_add() {
-  _mpc_helper_files
+  _mpc_helper_all_files
 }
 
 _mpc_albumart() {
-- 
2.46.0

[RFC PATCH 2/2] enable completion for mpc insert

[Karel Balej]

---
 Completion/Unix/Command/_mpc | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/Completion/Unix/Command/_mpc b/Completion/Unix/Command/_mpc
index 93100de09ca8..4566866ebfbd 100644
--- a/Completion/Unix/Command/_mpc
+++ b/Completion/Unix/Command/_mpc
@@ -215,6 +215,10 @@ _mpc_add() {
   _mpc_helper_all_files
 }
 
+_mpc_insert() {
+  _mpc_helper_all_files
+}
+
 _mpc_albumart() {
   _mpc_helper_files
 }
-- 
2.46.0

Re: [RFC PATCH 1/2] complete absolute paths for mpc add

[Karel Balej]

Hello,

Karel Balej, 2024-09-27T18:31:34+02:00:
> When connecting via Unix socket, mpc can queue files from outside the
> music directory as well when given the absolute path.
> ---
> mpc also supports explicitly prefixing the path with file:// which is
> not covered here, however according to my testing this only works for
> absolute paths too so there doesn't seem to be any benefit.
>
> This implementation doesn't work with environment variables, such as
> mpc add $HOME/somethi<tab>, I will appreciate tips on whether there is a
> simple way to make this more robust to include situations such as this
> instead of just checking for these two specific starting characters.

could I please get some feedback on these two patches?

Thank you,
K. B.

Re: [RFC PATCH 1/2] complete absolute paths for mpc add

[Bart Schaefer]

I didn't see this the first time around for some reason.

On Fri, Sep 27, 2024 at 9:34 AM Karel Balej <balejk@matfyz.cz> wrote:
>
> +(( $+functions[_mpc_helper_all_files] )) ||
> +_mpc_helper_all_files() {
> +  if [[ $words[CURRENT] == [/~]* ]]; then
> +         _files
> +  fi
> +  _mpc_helper_files
> +}

If _files succeeds ($? == 0) but _mpc_helper_files fails ($? != 0)
then _mpc_helper_all_files will appear to fail and completion will
proceed to the next possible alternative.

Does something like the following still give the result you expect?

_mpc_helper_all_files() {
  local ret=1
  if [[ $words[CURRENT] == [/~]* ]]; then
         _files
         ret=$?
  fi
  _mpc_helper_files || return ret
}

Re: [RFC PATCH 1/2] complete absolute paths for mpc add

[Karel Balej]

Bart Schaefer, 2024-11-09T11:44:08-08:00:
> I didn't see this the first time around for some reason.
>
> On Fri, Sep 27, 2024 at 9:34 AM Karel Balej <balejk@matfyz.cz> wrote:
> >
> > +(( $+functions[_mpc_helper_all_files] )) ||
> > +_mpc_helper_all_files() {
> > +  if [[ $words[CURRENT] == [/~]* ]]; then
> > +         _files
> > +  fi
> > +  _mpc_helper_files
> > +}
>
> If _files succeeds ($? == 0) but _mpc_helper_files fails ($? != 0)
> then _mpc_helper_all_files will appear to fail and completion will
> proceed to the next possible alternative.
>
> Does something like the following still give the result you expect?
>
> _mpc_helper_all_files() {
>   local ret=1
>   if [[ $words[CURRENT] == [/~]* ]]; then
>          _files
>          ret=$?
>   fi
>   _mpc_helper_files || return ret
> }

Thank you, this seems to give me the same results as the original,
however I don't immediately see any difference in behaviour even when I
force _mpc_helper_files to fail -- the _files results are still
completed even without the ret handling. Or do I just misunderstand how
this should work?

Again, I might misunderstand how this "next possible alternative"
mechanism works, however it occurred to me whether it might be possible
and possibly simpler to just have _mpc_helper_files fail on no matches
and have the completion fall back to _files only then?

Best regards,
K. B.

Re: [RFC PATCH 1/2] complete absolute paths for mpc add

[Bart Schaefer]

On Sun, Nov 10, 2024 at 10:56 AM Karel Balej <balejk@matfyz.cz> wrote:
>
> Thank you, this seems to give me the same results as the original,
> however I don't immediately see any difference in behaviour even when I
> force _mpc_helper_files to fail -- the _files results are still
> completed even without the ret handling. Or do I just misunderstand how
> this should work?

You won't get any fewer completions -- all those added by _files will
still be valid -- but if there is another (perhaps less specific)
possible completer after _mpc_helper_files fails, then additional
possible matches from that other completer will also be added.  In
ordinary circumstances, completion stops after a completer returns
success, unless specially programmed to continue.

>... possibly simpler to just have _mpc_helper_files fail on no matches
> and have the completion fall back to _files only then?

If, when _mpc_helper_files does find something, you don't mind
skipping files that _mpc_helper_files does not find, you can certainly
do that.

Re: [RFC PATCH 1/2] complete absolute paths for mpc add

[Karel Balej]

Bart Schaefer, 2024-11-10T15:10:55-08:00:
> On Sun, Nov 10, 2024 at 10:56 AM Karel Balej <balejk@matfyz.cz> wrote:
> >... possibly simpler to just have _mpc_helper_files fail on no matches
> > and have the completion fall back to _files only then?
>
> If, when _mpc_helper_files does find something, you don't mind
> skipping files that _mpc_helper_files does not find, you can certainly
> do that.

Actually that isn't really better because only absolute paths should be
completed via _files.

So if you have no further comments, I can submit the original patches
updated with the ret handling.

Thank you,
K. B.

Re: [RFC PATCH 1/2] complete absolute paths for mpc add

[Bart Schaefer]

I have no further comments.