`align-regexp' fails in Lisp code

`align-regexp' fails in Lisp code

[Raffaele Ricciardi]

Hello everyone,

I'm trying to run `align-regexp' from Lisp code, but it fails with
the error "align-region: Marker does not point anywhere".

Here is the specification of `align-regexp':

---

     (align-regexp BEG END REGEXP &optional GROUP SPACING REPEAT)

Align the current region using an ad-hoc rule read from the
minibuffer.  BEG and END mark the limits of the region.

---

Here is my Lisp code:

     (align-regexp (point-min) (point-max) " hides ")

Marking the whole buffer and calling `align-regexp' interactively does
work, though.

Tried with "emacs -Q".  I'm using GNU Emacs 24.3.1.

Thank you for your attention.

Re: `align-regexp' fails in Lisp code

[Leo Liu]

On 2013-07-10 08:11 +0800, Raffaele Ricciardi wrote:
> Marking the whole buffer and calling `align-regexp' interactively does
> work

Because the function works on a region as stated in its doc-string.

Leo

Re: `align-regexp' fails in Lisp code

[Le Wang]

I can find at least 4 variations of this question on stackoverflow.

This answer describes what's happening well:
http://stackoverflow.com/a/12975000/903943

Essentially, the parameters received by the function are pre-processed
by interactive forms, so you're not calling it correctly.

See also difference between function and command, interactive forms
sections of the docs.


On Wed, Jul 10, 2013 at 8:11 AM, Raffaele Ricciardi <rfflrccrd@gmail.com> wrote:
> Hello everyone,
>
> I'm trying to run `align-regexp' from Lisp code, but it fails with
> the error "align-region: Marker does not point anywhere".
>
> Here is the specification of `align-regexp':
>
> ---
>
>     (align-regexp BEG END REGEXP &optional GROUP SPACING REPEAT)
>
> Align the current region using an ad-hoc rule read from the
> minibuffer.  BEG and END mark the limits of the region.
>
> ---
>
> Here is my Lisp code:
>
>     (align-regexp (point-min) (point-max) " hides ")
>
> Marking the whole buffer and calling `align-regexp' interactively does
> work, though.
>
> Tried with "emacs -Q".  I'm using GNU Emacs 24.3.1.
>
> Thank you for your attention.
>
>
>
>



-- 
Le

Re: `align-regexp' fails in Lisp code

[Michael Heerdegen]

Leo Liu <sdl.web@gmail.com> writes:

> On 2013-07-10 08:11 +0800, Raffaele Ricciardi wrote:
> > Marking the whole buffer and calling `align-regexp' interactively does
> > work
>
> Because the function works on a region as stated in its doc-string.

If that's right (it seems so), then the doc-string should clearly say
that this command is not for non-interactive use, or how it can be
called from Lisp.


Michael.

Re: `align-regexp' fails in Lisp code

[andrea crotti]

"align-regexp is an interactive autoloaded Lisp function in `align.el'."

it says it's interactive in the doc..


2013/7/10 Michael Heerdegen <michael_heerdegen@web.de>

> Leo Liu <sdl.web@gmail.com> writes:
>
> > On 2013-07-10 08:11 +0800, Raffaele Ricciardi wrote:
> > > Marking the whole buffer and calling `align-regexp' interactively does
> > > work
> >
> > Because the function works on a region as stated in its doc-string.
>
> If that's right (it seems so), then the doc-string should clearly say
> that this command is not for non-interactive use, or how it can be
> called from Lisp.
>
>
> Michael.
>
>
>

Re: `align-regexp' fails in Lisp code

[Nicolas Richard]

andrea crotti <andrea.crotti.0@gmail.com> writes:

> "align-regexp is an interactive autoloaded Lisp function in `align.el'."
>
> it says it's interactive in the doc..

And so is goto-char :
"goto-char is an interactive built-in function in `editfns.c'."

This doesn't mean that goto-char shouldn't be used from lisp.

-- 
Nico.

Re: `align-regexp' fails in Lisp code

[Raffaele Ricciardi]

On 10/07/13 05:14, Michael Heerdegen wrote:
> Leo Liu <sdl.web@gmail.com> writes:
>
>> On 2013-07-10 08:11 +0800, Raffaele Ricciardi wrote:
>>> Marking the whole buffer and calling `align-regexp' interactively does
>>> work
>>
>> Because the function works on a region as stated in its doc-string.
>
> If that's right (it seems so), then the doc-string should clearly say
> that this command is not for non-interactive use, or how it can be
> called from Lisp.

Two positions make a region, hence the call:

     (align-regexp (point-min) (point-max) " hides ")

is correct.

The problem is that - I had to look into Lisp code to understand - the 
regexp that you enter in the mini-buffer is not the regexp that 
`align-regexp' receives, because of a conversion applied by the 
`interactive' form inside the command.  Hence I agree with you - but for 
a different reason - that the documentation should say how the command 
could be called from Lisp.

Here is a wrapper to call `align-regexp' from Lisp with no prefix arg :

(defun align-regexp-function (^start ^end ^regexp)
   "Like `align-regexp' when no prefix arg was specified,
but callable from Lisp."
   ;; Convert arguments like the `interactive' form
   ;; in `align-regexp' does when no prefix arg was specified.
   (align-regexp ^start
                 ^end
                 (concat "\\(\\s-*\\)" ^regexp)
                 1
                 align-default-spacing
                 nil))

Re: `align-regexp' fails in Lisp code

[Michael Heerdegen]

Raffaele Ricciardi <rfflrccrd@gmail.com> writes:

> Two positions make a region, hence the call:
>
>     (align-regexp (point-min) (point-max) " hides ")
>
> is correct.
>
> The problem is that - I had to look into Lisp code to understand - the
> regexp that you enter in the mini-buffer is not the regexp that
> align-regexp' receives, because of a conversion applied by the
> interactive' form inside the command.  Hence I agree with you - but
> for a different reason - that the documentation should say how the
> command could be called from Lisp.

Interesting.  Do you also understand why

  (align-regexp (point-min) (point-max) " hides ")

is illegitimate?  After having read the doc, I don't understand why this
call must lead to an error.


Regards,

Michael.

Re: `align-regexp' fails in Lisp code

[Raffaele Ricciardi]

On 12/07/13 02:26, Michael Heerdegen wrote:
> Interesting.  Do you also understand why
>
>    (align-regexp (point-min) (point-max) " hides ")
>
> is illegitimate?  After having read the doc, I don't understand why this
> call must lead to an error.

Indeed, the doc doesn't touch the issue.  If you call `align-regexp' 
with the regexp that you would have typed in the mini-buffer, then you 
won't be calling `align-regexp' with the regexp that it expects, because 
a regexp typed in the mini-buffer would have been converted into 
something else.

Re: `align-regexp' fails in Lisp code

[Jambunathan K]

Run the command as usual

    M-x align-regexp RET 

Let Emacs tell you what the lisp representation of it is.  Do

   M-x list-command-history RET

you will find your very recent command there.

Copy, paste that stuff, do some modifications for the region portion and
your are done.

----------------------------------------------------------------

You can also get the lisp representation with

    M-x repeat-complex-command RET

    M-p
    M-p
    
    etc

until you find what you want.



Raffaele Ricciardi <rfflrccrd@gmail.com> writes:

> Hello everyone,
>
> I'm trying to run `align-regexp' from Lisp code, but it fails with
> the error "align-region: Marker does not point anywhere".
>
> Here is the specification of `align-regexp':
>
> ---
>
>     (align-regexp BEG END REGEXP &optional GROUP SPACING REPEAT)
>
> Align the current region using an ad-hoc rule read from the
> minibuffer.  BEG and END mark the limits of the region.
>
> ---
>
> Here is my Lisp code:
>
>     (align-regexp (point-min) (point-max) " hides ")
>
> Marking the whole buffer and calling `align-regexp' interactively does
> work, though.
>
> Tried with "emacs -Q".  I'm using GNU Emacs 24.3.1.
>
> Thank you for your attention.

Re: `align-regexp' fails in Lisp code

[Michael Heerdegen]

Raffaele Ricciardi <rfflrccrd@gmail.com> writes:

> > Interesting.  Do you also understand why
> >
> >    (align-regexp (point-min) (point-max) " hides ")
> >
> > is illegitimate?  After having read the doc, I don't understand why this
> > call must lead to an error.
>
> Indeed, the doc doesn't touch the issue.  If you call `align-regexp'
> with the regexp that you would have typed in the mini-buffer, then you
> won't be calling `align-regexp' with the regexp that it expects,
> because a regexp typed in the mini-buffer would have been converted
> into something else.

The doc is just incomplete, so I filed a bug report.  It's bug 14857
now.


Regards,

Michael.