[TxMt] LaTeX Syntax coloring problem

[Paul Hagstrom]

On Fri Apr 18, Kai von Fintel wrote:

> The minimal test case below shows that something in the parsing of the
> newenvironment command is throwing off the syntax coloring by the
> LaTeX bundle, at least on my system (TM 1.5.7 1464 on Leopard). The
> text in the document is incorrectly parsed as being in a math
> environment.
>
> \documentclass{article}
>
> \usepackage{amsmath}
>
> \newenvironment{choice}
>      {\left\lbrace\begin{gathered}}
>      {\end{gathered}\right\rbrace}
>
> \begin{document}
>
> Text incorrectly parsed as being in a math environment.
>
> \end{document}
>
> -----
>
> Any ideas of how to fix the parsing?


Hi Kai -- This isn't exactly the correct fix, but if you go to Bundles- 
 >Bundle Editor->Edit Languages... and pick the LaTeX->LaTeX syntax  
definition, down about 1/3rd of the way down, you'll find the place  
where it is looking for these mathy environments:

align|equation|eqnarray
| multline|aligned|alignat
| split|gather|gathered

soon below that, you'll find

end = '(?x)
  (?:\s*) # Optional whitespace

If you change the (?:\s*) to (?:.*?) then you can get a sort of  
approximation of the correct syntax coloring.

I really don't quite understand why it works, but the idea was  
essentially to allow the { that precedes the \end{gathered} to be  
ignored (it wouldn't have counted as whitespace). In the original  
formulation, TextMate couldn't find the \end{gathered}, and so just  
assumes that the environment ends when the document does.

This isn't actually the right solution, but it at least keeps you from  
having the entire body of your document math-colored. The symptom of  
why this isn't the right solution is that everything following  
\begin{gathered} (including the closing brace of the first argument to  
\newenvironment, as well as anything from then on until just before  
the line containing \end{gathered}) is math colored.

Ideally, the LaTeX bundle would be able to handle \(re)newenvironment  
specially, since each of the two arguments to these commands will  
almost always contain a lone \begin or \end, and scope of any  
mathification should presumably be restricted to the argument itself.   
But I haven't got a clue how to approach this the "right" way.

  -Paul