Delete Everything between $ including multiple lines in latex files

Question

I am working on laTex files, I need to delete everything between two $ including newlines and keep only english text.

I am using a command like this to process the files:

find "." -name "*.tex" | xargs perl -pi -e 's/\$[^\$].*?\$/ /g' *

Example:

Then use the naturality formula 

    $t_{G^{n-1}M} G^{i+1} (\epsilon_{G^{n-i}M}) 
    = G^{i+1} (\epsilon_{G^{n-i}M}) t_{G^n M}$ on the left-hand side.

OutPut:

Then use the naturality formula 
 on the left-hand side.

another example from file:

EXAMPLE:

\begin{itemize}
\item $M$ is atomic and finitely generated;
\item $M$ is cancellative;
\item $(M, \le_L)$ and $(M, \le_R)$ are lattices;
\item there exists an element $\Delta \in M$, called {\it Garside element}, such that the set 
$L(\Delta)= \{ x \in M; x\le_L \Delta\}$ generates $M$ and is equal to $R(\Delta)= \{ x\in M; 
x\le_R \Delta\}$.
\end{itemize}

OUTPUT:

\begin{itemize}
\item   is atomic and finitely generated;
\item   is cancellative;
\item   and   are lattices;
\item there exists an element  , called {\it Garside element}, such that the set 
  generates   and is equal to $R(\Delta)= \{ x\in M; 
x\le_R \Delta\}$.
\end{itemize} 

if you can notice ( $R(\Delta)= { x\in M; x\le_R \Delta}$.) can not be removed!!

Example 2 from different file, the input same as the output nothing has changed :

    Using the fact that   is atomic and that $L(\Delta)= 
\{x \in M; x \le_L \Delta\} M \pi_L(a) \neq 1 a \neq 
1 k \partial_L^k(a)=1 k$ be the

Answer 1

I'm guessing this is not matching when the text it should match spans multiple lines.

You have [^\$].*? which matches one character that is not $ using [^\$] and then matches .*? which matches any character that is not a newline character zero or more times lazily. This works on your single line cases because of the lazy modifier trying to match a $ before another ., but fails the multi-line cases because . will not match a newline.

Correct and more efficient would be [^\$]* which would match as many non-$ characters as possible, including newlines.

So your command would would be

s/\$[^\$]*\$/ /g

or cleaner looking in my opinion to use non-standard delimiters and avoid the 'fencepost' look /\

s~\$[^\$]*\$~ ~g

Demo

Perl is processing your file line by line, which is another cause for the failing matches across newlines. There are a number of documented answers to this problem already on SO and written by people who know perl better than I: How to match multiline data in perl