GNU bug report logs - #20626
Wishlist: M-x shell-command-on-rectangle-region

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 20626 in the body.
You can then email your comments to 20626 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: asparagus <at> comcast.net
To: bug-gnu-emacs <at> gnu.org
Subject: Wishlist: M-x shell-command-on-rectangle-region
Date: Thu, 21 May 2015 22:04:30 +0000 (UTC)

Wishlist:
M-x shell-command-on-rectangle-region

Message #8 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Juri Linkov <juri <at> linkov.net>
To: asparagus <at> comcast.net
Cc: 20626 <at> debbugs.gnu.org
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Fri, 22 May 2015 01:47:52 +0300

> Wishlist:
> M-x shell-command-on-rectangle-region

As you can see in bug#20070, the effort to make commands rectangleable
had stalled some time ago due to the need to decide how to handle
backward-compatibility of the existing region arguments, e.g. in

  (shell-command-on-region START END COMMAND &optional OUTPUT-BUFFER REPLACE
                           ERROR-BUFFER DISPLAY-ERROR-BUFFER)

how to send the boundaries of the rectangular region in START and END.

One idea is to handle it like recently we handled backward-compatibility
for saving dired positions in saveplace.el where we used a new format like

  ("~" (dired-filename . "~/.emacs.d/places"))

Using something like this means sending the rectangular bounds
either in START or END in the new format like

  (rect (1 . 2) (3 . 4))

Message #11 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Juri Linkov <juri <at> linkov.net>
To: asparagus <at> comcast.net
Cc: 20626 <at> debbugs.gnu.org
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Tue, 16 Jun 2015 00:45:27 +0300

>> Wishlist:
>> M-x shell-command-on-rectangle-region
>
> As you can see in bug#20070, the effort to make commands rectangleable
> had stalled some time ago due to the need to decide how to handle
> backward-compatibility of the existing region arguments, e.g. in
>
>   (shell-command-on-region START END COMMAND &optional OUTPUT-BUFFER REPLACE
>                            ERROR-BUFFER DISPLAY-ERROR-BUFFER)
>
> how to send the boundaries of the rectangular region in START and END.
>
> One idea is to handle it like recently we handled backward-compatibility
> for saving dired positions in saveplace.el where we used a new format like
>
>   ("~" (dired-filename . "~/.emacs.d/places"))
>
> Using something like this means sending the rectangular bounds
> either in START or END in the new format like
>
>   (rect (1 . 2) (3 . 4))

Sorry, I was wrong.  I realized now that query-replace has quite
a different requirement.  query-replace needs rectangular boundaries
to limit the search for replacements, whereas shell-command-on-region
should extract the rectangular region as strings and replace it with
the result of the command.  Here is a working prototype that demonstrates
its possible implementation:

(define-advice shell-command-on-region
    (:around (orig-fun start end command
                       &optional output-buffer replace
                       error-buffer display-error-buffer))
  (if (and (boundp 'rectangle-mark-mode) rectangle-mark-mode)
      (let ((input (mapconcat 'identity (delete-extract-rectangle start end) "\n"))
            output)
        (with-temp-buffer
          (insert input)
          (call-process-region (point-min) (point-max)
                               shell-file-name t t
                               nil shell-command-switch
                               command)
          (setq output (split-string (buffer-string) "\n")))
        (goto-char start)
        (insert-rectangle output))
    (funcall orig-fun start end command
             output-buffer replace
             error-buffer display-error-buffer)))

This is another case to take into account when designing the interface,
i.e. in this case the list of boundaries in the arg START is not necessary,
and I have no idea how to avoid `(if (and (boundp 'rectangle-mark-mode)
rectangle-mark-mode))'

Message #14 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Juri Linkov <juri <at> linkov.net>
To: asparagus <at> comcast.net
Cc: 20626 <at> debbugs.gnu.org
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Tue, 23 Jun 2015 01:37:53 +0300

> Sorry, I was wrong.  I realized now that query-replace has quite
> a different requirement.  query-replace needs rectangular boundaries
> to limit the search for replacements, whereas shell-command-on-region
> should extract the rectangular region as strings and replace it with
> the result of the command.  Here is a working prototype that demonstrates
> its possible implementation:
>
> (define-advice shell-command-on-region
>     (:around (orig-fun start end command
>                        &optional output-buffer replace
>                        error-buffer display-error-buffer))
>   (if (and (boundp 'rectangle-mark-mode) rectangle-mark-mode)
>       (let ((input (mapconcat 'identity (delete-extract-rectangle start end) "\n"))
>             output)
>         (with-temp-buffer
>           (insert input)
>           (call-process-region (point-min) (point-max)
>                                shell-file-name t t
>                                nil shell-command-switch
>                                command)
>           (setq output (split-string (buffer-string) "\n")))
>         (goto-char start)
>         (insert-rectangle output))
>     (funcall orig-fun start end command
>              output-buffer replace
>              error-buffer display-error-buffer)))
>
> This is another case to take into account when designing the interface,
> i.e. in this case the list of boundaries in the arg START is not necessary,
> and I have no idea how to avoid `(if (and (boundp 'rectangle-mark-mode)
> rectangle-mark-mode))'

For example, the command ‘kill-ring-save’ has the signature

  kill-ring-save (beg end &optional region)

with an additional boolean arg for the region.

And ‘rectangle--extract-region’ contains a condition
that checks for ‘rectangle-mark-mode’.

Combining these two prerequisites we could do a similar thing in
‘shell-command-on-region’ by adding a new arg and using it in the
function body:

diff --git a/lisp/simple.el b/lisp/simple.el
index 1868077..d022504 100644
--- a/lisp/simple.el
+++ b/lisp/simple.el
@@ -3274,7 +3278,8 @@
 
 (defun shell-command-on-region (start end command
 				      &optional output-buffer replace
-				      error-buffer display-error-buffer)
+				      error-buffer display-error-buffer
+				      region)
   "Execute string COMMAND in inferior shell with region as input.
 Normally display output (if any) in temp buffer `*Shell Command Output*';
 Prefix arg means replace the region with it.  Return the exit code of
@@ -3337,7 +3342,8 @@ (defun shell-command-on-region (start end command
 		       current-prefix-arg
 		       current-prefix-arg
 		       shell-command-default-error-buffer
-		       t)))
+		       t
+		       rectangle-mark-mode)))
   (let ((error-file
 	 (if error-buffer
 	     (make-temp-file
@@ -3346,6 +3352,18 @@ (defun shell-command-on-region (start end command
 				    temporary-file-directory)))
 	   nil))
 	exit-status)
+    (if region
+        (let ((input (mapconcat 'identity (delete-extract-rectangle start end) "\n"))
+              output)
+          (with-temp-buffer
+            (insert input)
+            (call-process-region (point-min) (point-max)
+                                 shell-file-name t t
+                                 nil shell-command-switch
+                                 command)
+            (setq output (split-string (buffer-string) "\n")))
+          (goto-char start)
+          (insert-rectangle output))
     (if (or replace
 	    (and output-buffer
 		 (not (or (bufferp output-buffer) (stringp output-buffer)))))
@@ -3435,7 +3453,7 @@ (defun shell-command-on-region (start end command
 			      exit-status output))))
 	    ;; Don't kill: there might be useful info in the undo-log.
 	    ;; (kill-buffer buffer)
-	    ))))
+	    )))))
 
     (when (and error-file (file-exists-p error-file))
       (if (< 0 (nth 7 (file-attributes error-file)))

Message #17 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Juri Linkov <juri <at> linkov.net>
Cc: 20626 <at> debbugs.gnu.org, asparagus <at> comcast.net
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Mon, 22 Jun 2015 22:02:33 -0400

>  (defun shell-command-on-region (start end command
>  				      &optional output-buffer replace
> -				      error-buffer display-error-buffer)
> +				      error-buffer display-error-buffer
> +				      region)
>    "Execute string COMMAND in inferior shell with region as input.
>  Normally display output (if any) in temp buffer `*Shell Command Output*';
>  Prefix arg means replace the region with it.  Return the exit code of
> @@ -3337,7 +3342,8 @@ (defun shell-command-on-region (start end command
>  		       current-prefix-arg
>  		       current-prefix-arg
>  		       shell-command-default-error-buffer
> -		       t)))
> +		       t
> +		       rectangle-mark-mode)))

Doesn't make sense: if the value determine the use of rectangles, the
arg shouldn't be called `region' but something like `rectangle'.

Notice how kill-ring-save takes a `region' argument and doesn't have any
rectangle-specific code.

I still believe that shell-command-on-region should not have
rectangle-specific code.  The current `region-extract-function' does let
you extract the region (rectangular or not) in order to pass it to
a shell command.  So you don't need any rectangle-specific code for that
part of shell-command-on-region.
OTOH There is indeed some functionality missing there to let you insert
the output in a rectangular way (whatever that means).


        Stefan

Message #20 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Juri Linkov <juri <at> linkov.net>
To: Stefan Monnier <monnier <at> iro.umontreal.ca>
Cc: 20626 <at> debbugs.gnu.org, asparagus <at> comcast.net
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Wed, 24 Jun 2015 01:59:10 +0300

> I still believe that shell-command-on-region should not have
> rectangle-specific code.  The current `region-extract-function' does let
> you extract the region (rectangular or not) in order to pass it to
> a shell command.  So you don't need any rectangle-specific code for that
> part of shell-command-on-region.
> OTOH There is indeed some functionality missing there to let you insert
> the output in a rectangular way (whatever that means).

‘shell-command-on-region’ currently relies on the call to

   (call-process-region start end shell-file-name replace ...

but I'm not sure if ‘call-process-region’ is not too low level
to handle rectangular regions.  Otherwise, there should be a condition
in ‘shell-command-on-region’ to check for a rectangular region and
insert the output accordingly, so I see no way to avoid checking for
‘rectangle-mark-mode’ in ‘shell-command-on-region’.

Message #23 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> IRO.UMontreal.CA>
To: Juri Linkov <juri <at> linkov.net>
Cc: 20626 <at> debbugs.gnu.org, asparagus <at> comcast.net
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Tue, 23 Jun 2015 19:50:22 -0400

> in ‘shell-command-on-region’ to check for a rectangular region and
> insert the output accordingly, so I see no way to avoid checking for
> ‘rectangle-mark-mode’ in ‘shell-command-on-region’.

In any case checking specifically for a rectangular region is wrong.
Having a "fast-path" for the case of a simple contiguous region is fine,
but the "slow&complex" path is not just for rectangular regions:
the code should also work with a region made up of various
arbitrary chunks (there's currently no package that provides this
feature, but that's no excuse).


        Stefan

Message #26 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Juri Linkov <juri <at> linkov.net>
To: Stefan Monnier <monnier <at> IRO.UMontreal.CA>
Cc: 20626 <at> debbugs.gnu.org, asparagus <at> comcast.net
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Thu, 25 Jun 2015 01:27:18 +0300

>> in ‘shell-command-on-region’ to check for a rectangular region and
>> insert the output accordingly, so I see no way to avoid checking for
>> ‘rectangle-mark-mode’ in ‘shell-command-on-region’.
>
> In any case checking specifically for a rectangular region is wrong.
> Having a "fast-path" for the case of a simple contiguous region is fine,
> but the "slow&complex" path is not just for rectangular regions:
> the code should also work with a region made up of various
> arbitrary chunks (there's currently no package that provides this
> feature, but that's no excuse).

Then what about adding a new arg REGION to ‘shell-command-on-region’
and other region-sensitive commands that will provide the region
configuration including information about region type and region shape,
e.g. a list in the form ‘(rectangle …)’.  Then the command's body
will check the car of the arg REGION and act accordingly depending
on the region shape.

Message #29 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Juri Linkov <juri <at> linkov.net>
Cc: 20626 <at> debbugs.gnu.org, asparagus <at> comcast.net
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Wed, 24 Jun 2015 23:45:16 -0400

> e.g. a list in the form ‘(rectangle …)’.  Then the command's body
> will check the car of the arg REGION and act accordingly depending
> on the region shape.

Same difference.  It'd still have rectangle-specific code.

Instead it should call something like region-extract-function to do what
it needs to do and *this* thing will have rectangle-specific code (via
something like an add-function in rect.el).

IIUC the current region-extract-function doesn't satisfy all the needs
of shell-command-on-region, so either it will have to be extended again,
or we have to add more region-<foo>-function or equivalent methods.


        Stefan

Message #32 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Juri Linkov <juri <at> linkov.net>
To: Stefan Monnier <monnier <at> IRO.UMontreal.CA>
Cc: 20626 <at> debbugs.gnu.org, asparagus <at> comcast.net
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Fri, 26 Jun 2015 01:30:45 +0300

> In any case checking specifically for a rectangular region is wrong.
> Having a "fast-path" for the case of a simple contiguous region is fine,
> but the "slow&complex" path is not just for rectangular regions:
> the code should also work with a region made up of various
> arbitrary chunks (there's currently no package that provides this
> feature, but that's no excuse).

It's more-less clear how to write code for the "slow&complex" path
using the existing ‘region-extract-function’ and a new function like
‘region-insert-function’ like I demonstrated with define-advice
for shell-command-on-region in the beginning of this thread (where
specific ‘insert-rectangle’ should be replaced with ‘region-insert-function’).

What is not clear is how to distinguish "slow&complex" from the "fast-path"?
Maybe with a new arg REGION?

Message #35 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> IRO.UMontreal.CA>
To: Juri Linkov <juri <at> linkov.net>
Cc: 20626 <at> debbugs.gnu.org, asparagus <at> comcast.net
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Thu, 25 Jun 2015 21:53:42 -0400

> What is not clear is how to distinguish "slow&complex" from the "fast-path"?
> Maybe with a new arg REGION?

I thought it would be pretty easy: if region-extract-function returns
a single contiguous chunk or multiple chunks.


        Stefan

Message #38 received at 20626 <at> debbugs.gnu.org (full text, mbox):

From: Juri Linkov <juri <at> linkov.net>
To: Stefan Monnier <monnier <at> IRO.UMontreal.CA>
Cc: 20626 <at> debbugs.gnu.org, asparagus <at> comcast.net
Subject: Re: bug#20626: Wishlist: M-x shell-command-on-rectangle-region
Date: Tue, 30 Jun 2015 23:44:27 +0300

forcemerge 19829 20626
thanks

>> What is not clear is how to distinguish "slow&complex" from the "fast-path"?
>> Maybe with a new arg REGION?
>
> I thought it would be pretty easy: if region-extract-function returns
> a single contiguous chunk or multiple chunks.

‘query-replace’ will use the same design, so merged it with bug#19829
where I posted a composite patch.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.