regexp problems the evil \[

regexp problems the evil \[

[Uwe Brauer]

Hi

The following drives me crazy

  (query-replace "\\[" "\\begin{displaymath}\n" nil)

Replaces
\[

by
\begin{displaymath}

But 

  (query-replace-regexp "\\[\\([^}]+\\)\\]" "@\\1" nil))


Does not replace 

\[eq:testpandoc:1\] 

By @eq:testpandoc:1

What do I miss?

Thanks

Uwe Brauer

Re: regexp problems the evil \[

[Stefan Monnier]

>   (query-replace "\\[" "\\begin{displaymath}\n" nil)
     ^^^^^^^^^^^^^^^^^^^
     matches \[

>   (query-replace-regexp "\\[\\([^}]+\\)\\]" "@\\1" nil))
     ^^^^^^^^^^^^^^^^^^^^^^^^^
        matches [

Since [ in regexps has otherwise a special meaning.
To match \[ and \] you'll want

   (query-replace-regexp "\\\\\\[\\([^}]+\\)\\\\\\]" "@\\1" nil))


-- Stefan

Re: regexp problems the evil \[

[Uwe Brauer]

>>> "Stefan" == Stefan Monnier <monnier@iro.umontreal.ca> writes:

   >> (query-replace "\\[" "\\begin{displaymath}\n" nil)
   >      ^^^^^^^^^^^^^^^^^^^
   >      matches \[

   >> (query-replace-regexp "\\[\\([^}]+\\)\\]" "@\\1" nil))
   >      ^^^^^^^^^^^^^^^^^^^^^^^^^
   >         matches [

   > Since [ in regexps has otherwise a special meaning.
   > To match \[ and \] you'll want

   >    (query-replace-regexp "\\\\\\[\\([^}]+\\)\\\\\\]" "@\\1" nil))

Wow six \ I gave up after five.

Thanks

Uwe 

Re: regexp problems the evil \[

[Aurélien Aptel]

On Thu, Mar 23, 2017 at 11:11 PM, Uwe Brauer <oub@mat.ucm.es> wrote:
> Wow six \ I gave up after five.

Yes, it's ridiculous, I personally think emacs lisp regexes are
unusable once escapes are involved, especially interactively.
You can use rx instead of strings to make regexes from s-expressions.
Strings elements are always assumed to match literally (rx will always
escape whatever is needed to match the string literally).

    (query-replace-regexp (rx "\\[" (group (+ (not (any "}")))) "\\]") ...

Re: regexp problems the evil \[

[hector]

On Fri, Mar 24, 2017 at 12:38:18AM +0100, Aurélien Aptel wrote:
> On Thu, Mar 23, 2017 at 11:11 PM, Uwe Brauer <oub@mat.ucm.es> wrote:
> > Wow six \ I gave up after five.
> 
> Yes, it's ridiculous, I personally think emacs lisp regexes are
> unusable once escapes are involved, especially interactively.

It's not so serious. Easy:

    escape        *2
text ---> regexp ---> lisp

\[         \\\[       \\\\\\[

Re: regexp problems the evil \[

[Uwe Brauer]

>>> "hector" == hector  <hectorlahoz@gmail.com> writes:

   > On Fri, Mar 24, 2017 at 12:38:18AM +0100, Aurélien Aptel wrote:
   >> On Thu, Mar 23, 2017 at 11:11 PM, Uwe Brauer <oub@mat.ucm.es> wrote:
   >> > Wow six \ I gave up after five.
   >> 
   >> Yes, it's ridiculous, I personally think emacs lisp regexes are
   >> unusable once escapes are involved, especially interactively.

   > It's not so serious. Easy:

   >     escape        *2
   > text ---> regexp ---> lisp

   > \[         \\\[       \\\\\\[


But you must confess that it is a bit counter intuitive, no?

Re: regexp problems the evil \[

[Uwe Brauer]

>>> "Aurélien" == Aurélien Aptel <aurelien.aptel+emacs@gmail.com> writes:

   > On Thu, Mar 23, 2017 at 11:11 PM, Uwe Brauer <oub@mat.ucm.es> wrote:
   >> Wow six \ I gave up after five.

   > Yes, it's ridiculous, I personally think emacs lisp regexes are
   > unusable once escapes are involved, especially interactively.
   > You can use rx instead of strings to make regexes from s-expressions.
   > Strings elements are always assumed to match literally (rx will always
   > escape whatever is needed to match the string literally).

   >     (query-replace-regexp (rx "\\[" (group (+ (not (any "}")))) "\\]") ...


Ah thanks did not know about it.

Uwe 

Re: regexp problems the evil \[

[Stefan Monnier]

> Wow six \ I gave up after five.

Next exercise: write a regexp that matches this regexp.


        Stefan