Re: [AUCTeX-devel] Regex best practice

[Arash Esbati]

Mosè Giordano  writes:

> 2015-11-24 21:13 GMT+01:00 Arash Esbati :
>>
>> Mosè Giordano  writes:
>>
>>> Probably yes: so you want to match the whole \declaretheoremstyle
>>> macro, but saving only the mandatory argument, right?
>>
>> Yes, exactly.
>
> The problem of your regex is that will match *everything* until the
> last closing bracket.  This is good when you test your regex in a
> simple case, but will fail in a real document with other commands
> taking optional and mandatory arguments.

Thanks for your response.  Your are right, I missed that completely.

> Is the non paired "]" in
>
>   anothername  = {valu-es[]}]{}  ,
>
> intended? 

No, it was just part of the experiment.

> If so, I don't have a solution at hand, otherwise, if you're sure you
> have always paired brackets, you can use something like
>
> 
> "declaresomething\\[\\(?:[^][]*\\(?:\\[[^][]*\\(?:\\[[^][]*\\(?:\\[[^][]*\\][^][]*\\)*\\][^][]*\\)*\\][^][]*\\)*\\)\\][
> \t\n\r]*{[ \t\n\r]*\\(.*\\)[ \t\n\r]*}"

The existence of brackets is not guaranteed, it is the optional
argument.  I will play around with your suggestion; I have to decode it
first ;-)

Best, Arash


___
auctex-devel mailing list
auctex-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/auctex-devel

Re: [AUCTeX-devel] Regex best practice

[Arash Esbati]

Hi Mosè,

Mosè Giordano  writes:

> Probably yes: so you want to match the whole \declaretheoremstyle
> macro, but saving only the mandatory argument, right?

Yes, exactly.

Best, Arash


___
auctex-devel mailing list
auctex-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/auctex-devel

Re: [AUCTeX-devel] Regex best practice

[Mosè Giordano]

Hi Arash,

2015-11-24 21:13 GMT+01:00 Arash Esbati :
> Hi Mosè,
>
> Mosè Giordano  writes:
>
>> Probably yes: so you want to match the whole \declaretheoremstyle
>> macro, but saving only the mandatory argument, right?
>
> Yes, exactly.

The problem of your regex is that will match *everything* until the
last closing bracket.  This is good when you test your regex in a
simple case, but will fail in a real document with other commands
taking optional and mandatory arguments. Consider for example

--8<---cut here---start->8---
\declaresomething[%
  key  = value   ,
  name = {[Optional]Value}   ,
  anothername  = {valu-es[]}]{}  ,
  colframe = red!75!black,
  fonttitle= \bfseries   ,
  enhanced   ,
  attach boxed = {yshift=-2mm}   ,
  title= #2
  ]
  {
some1thing
  }

\othercommand[optional]{mandatory}
--8<---cut here---end--->8---

`re-builder' highlights "mandatory", with your regex.  In addition, it
doesn't seem to work at all with XEmacs, but I don't know why.

Is the non paired "]" in

  anothername  = {valu-es[]}]{}  ,

intended?  If so, I don't have a solution at hand, otherwise, if
you're sure you have always paired brackets, you can use something
like


"declaresomething\\[\\(?:[^][]*\\(?:\\[[^][]*\\(?:\\[[^][]*\\(?:\\[[^][]*\\][^][]*\\)*\\][^][]*\\)*\\][^][]*\\)*\\)\\][
\t\n\r]*{[ \t\n\r]*\\(.*\\)[ \t\n\r]*}"

(I'm sorry for the hard-wrapping, replace the newline with a space)
which is inspired by `LaTeX-auto-index-regexp-list'.

Bye,
Mosè

___
auctex-devel mailing list
auctex-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/auctex-devel

Re: [AUCTeX-devel] Regex best practice

[Mosè Giordano]

Hi Arash,

2015-11-21 23:20 GMT+01:00 Arash Esbati :
> Hi Mosè,
>
> Mosè Giordano  writes:
>
>> 2015-11-19 20:58 GMT+01:00 Arash Esbati :
>
>>> looking at some packages, the optional arguments for some commands are
>>> getting more and more complex, e.g.
>>>
>>> --8<---cut here---start->8---
>>> \declaresomething[%
>>>   key  = value   ,
>>>   name = {[Optional]Value}   ,
>>>   anothername  = {valu-es[]}]{}  ,
>>>   colframe = red!75!black,
>>>   fonttitle= \bfseries   ,
>>>   enhanced   ,
>>>   attach boxed = {yshift=-2mm}   ,
>>>   title= #2
>>>   ]
>>>   {
>>> some1thing
>>>   }
>>> --8<---cut here---end--->8---
>>>
>>> I was thinking about a general regex-solution how to add these kind of
>>> beasts to AUCTeX parser.  My solution is currently:
>>>
>>> --8<---cut here---start->8---
>>> declaresomething[ \t\n\r%]*\\(?:\\[\\(?:[\t\n\r[:print:]]*\\)\\]\\)?[ 
>>> \t\n\r%]*{[ \t\n\r%]*\\([[:print:]]+\\)[ \t\n\r%]*}
>>> --8<---cut here---end--->8---
>>>
>>> Any comments?  Admittedly, the last two `[ \t\n\r%]' are excessive, but
>>> you never know.
>>
>> Ehm, sorry but I didn't get what you want to match (re-builder
>> highlights "some1thing") and where you want to add this regex.
>
> My apologies for being brief and not clear.  Please consider the
> following MWE:
>
> --8<---cut here---start->8---
> \documentclass[a4paper,10pt]{article}
>
> \usepackage{amsthm}
> \usepackage{thmtools}
>
> \declaretheoremstyle[%
>   spaceabove= 6pt  ,
>   spacebelow= 6pt  ,
>   headfont  = \normalfont\bfseries ,
>   notefont  = \mdseries\scshape,
>   notebraces= {[}{]}   ,
>   bodyfont  = \normalfont\itshape  ,
>   postheadspace = 1em  ,
>   qed   = \qedsymbol
>   ]{mystyle}
>
> \declaretheorem[style=mystyle]{styledtheorem}
>
> \begin{document}
>
> \begin{styledtheorem}[Euclid]
>   For every prime $p$\dots
> \end{styledtheorem}
>
> \end{document}
> --8<---cut here---end--->8---
>
> Suppose I want to write a `thmtools.el'.  If I want to catch and process
> user defined styles (in this case "mystyle"), I would currently do:
>
> --8<---cut here---start->8---
> (defvar LaTeX-thmtools-declaretheoremstyle-regexp
>   `(,(concat "declaretheoremstyle[ \t\n\r%]*"
>  "\\(?:\\[\\(?:[\t\n\r[:print:]]*\\)\\]\\)?"
>  "[ \t\n\r%]*"
>  "{"
>  "[ \t\n\r%]*\\([[:print:]]+\\)[ \t\n\r%]*"
>  "}")))
>
> (TeX-auto-add-type "thmtools-declaretheoremstyle" "LaTeX")
>
> (TeX-add-style-hook
>  "thmtools"
>  (lambda ()
>(TeX-auto-add-regexp LaTeX-thmtools-declaretheoremstyle-regexp)
>...))
> --8<---cut here---end--->8---
>
> My question is: Is there any other, better solution for this regexp?  I
> am looking a general solution to handle this kind of cases.  I hope it
> is more clear now.

Probably yes: so you want to match the whole \declaretheoremstyle
macro, but saving only the mandatory argument, right?

Bye,
Mosè

___
auctex-devel mailing list
auctex-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/auctex-devel

Re: [AUCTeX-devel] Regex best practice

[Mosè Giordano]

Hi Arash,

2015-11-19 20:58 GMT+01:00 Arash Esbati :
> Hi all,
>
> looking at some packages, the optional arguments for some commands are
> getting more and more complex, e.g.
>
> --8<---cut here---start->8---
> \declaresomething[%
>   key  = value   ,
>   name = {[Optional]Value}   ,
>   anothername  = {valu-es[]}]{}  ,
>   colframe = red!75!black,
>   fonttitle= \bfseries   ,
>   enhanced   ,
>   attach boxed = {yshift=-2mm}   ,
>   title= #2
>   ]
>   {
> some1thing
>   }
> --8<---cut here---end--->8---
>
> I was thinking about a general regex-solution how to add these kind of
> beasts to AUCTeX parser.  My solution is currently:
>
> --8<---cut here---start->8---
> declaresomething[ \t\n\r%]*\\(?:\\[\\(?:[\t\n\r[:print:]]*\\)\\]\\)?[ 
> \t\n\r%]*{[ \t\n\r%]*\\([[:print:]]+\\)[ \t\n\r%]*}
> --8<---cut here---end--->8---
>
> Any comments?  Admittedly, the last two `[ \t\n\r%]' are excessive, but
> you never know.

Ehm, sorry but I didn't get what you want to match (re-builder
highlights "some1thing") and where you want to add this regex.

Bye,
Mosè

___
auctex-devel mailing list
auctex-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/auctex-devel

Re: [AUCTeX-devel] Regex best practice

[Arash Esbati]

Hi Mosè,

Mosè Giordano  writes:

> 2015-11-19 20:58 GMT+01:00 Arash Esbati :

>> looking at some packages, the optional arguments for some commands are
>> getting more and more complex, e.g.
>>
>> --8<---cut here---start->8---
>> \declaresomething[%
>>   key  = value   ,
>>   name = {[Optional]Value}   ,
>>   anothername  = {valu-es[]}]{}  ,
>>   colframe = red!75!black,
>>   fonttitle= \bfseries   ,
>>   enhanced   ,
>>   attach boxed = {yshift=-2mm}   ,
>>   title= #2
>>   ]
>>   {
>> some1thing
>>   }
>> --8<---cut here---end--->8---
>>
>> I was thinking about a general regex-solution how to add these kind of
>> beasts to AUCTeX parser.  My solution is currently:
>>
>> --8<---cut here---start->8---
>> declaresomething[ \t\n\r%]*\\(?:\\[\\(?:[\t\n\r[:print:]]*\\)\\]\\)?[ 
>> \t\n\r%]*{[ \t\n\r%]*\\([[:print:]]+\\)[ \t\n\r%]*}
>> --8<---cut here---end--->8---
>>
>> Any comments?  Admittedly, the last two `[ \t\n\r%]' are excessive, but
>> you never know.
>
> Ehm, sorry but I didn't get what you want to match (re-builder
> highlights "some1thing") and where you want to add this regex.

My apologies for being brief and not clear.  Please consider the
following MWE:

--8<---cut here---start->8---
\documentclass[a4paper,10pt]{article}

\usepackage{amsthm}
\usepackage{thmtools}

\declaretheoremstyle[%
  spaceabove= 6pt  , 
  spacebelow= 6pt  ,
  headfont  = \normalfont\bfseries ,
  notefont  = \mdseries\scshape, 
  notebraces= {[}{]}   ,
  bodyfont  = \normalfont\itshape  ,
  postheadspace = 1em  ,
  qed   = \qedsymbol
  ]{mystyle}

\declaretheorem[style=mystyle]{styledtheorem}

\begin{document}

\begin{styledtheorem}[Euclid]
  For every prime $p$\dots
\end{styledtheorem}

\end{document}
--8<---cut here---end--->8---

Suppose I want to write a `thmtools.el'.  If I want to catch and process
user defined styles (in this case "mystyle"), I would currently do:

--8<---cut here---start->8---
(defvar LaTeX-thmtools-declaretheoremstyle-regexp
  `(,(concat "declaretheoremstyle[ \t\n\r%]*"
 "\\(?:\\[\\(?:[\t\n\r[:print:]]*\\)\\]\\)?"
 "[ \t\n\r%]*"
 "{"
 "[ \t\n\r%]*\\([[:print:]]+\\)[ \t\n\r%]*"
 "}")))

(TeX-auto-add-type "thmtools-declaretheoremstyle" "LaTeX")

(TeX-add-style-hook
 "thmtools"
 (lambda ()
   (TeX-auto-add-regexp LaTeX-thmtools-declaretheoremstyle-regexp)
   ...))
--8<---cut here---end--->8---

My question is: Is there any other, better solution for this regexp?  I
am looking a general solution to handle this kind of cases.  I hope it
is more clear now.

Cheers, Arash


___
auctex-devel mailing list
auctex-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/auctex-devel