6
$\begingroup$

I am not able to figure the StringPattern to use to remove markers in string.

This is the input.

lst = {{1, 2, "this is a test", 4}, {Pi, 20, xy, 10}};
buf = ToString@TeXForm@lst

which gives

\left(
\begin{array}{cccc}
     1 & 2 & \text{this is a test} & 4 \\
     \pi  & 20 & \text{xy} & 10 \\
    \end{array}
\right)

I need to remove all the places where this pattern shows up \text{.....} and replace it with just what is inside ..... i.e. strip out the \text{ and the closing } on the other side. For each such instance in the input.

So the above should become

\left(
\begin{array}{cccc}
     1 & 2 & this is a test & 4 \\
     \pi  & 20 & xy & 10 \\
    \end{array}
\right)

I tried many things. Tried also using RegularExpression.

One attempt:

StringReplace[buf, "\\text{" ~~ x___ ~~ "}" .. :> x]

But this has a problem. It does not stop at the first closing }, but goes all the way to the ending } in the string, ending up with

\left(
\begin{array}{cccc}
 1 & 2 & this is a test} & 4 \\
 \pi  & 20 & \text{xy} & 10 \\
\end{array
\right)

Notice, it went all the way to the end, and removed the } after {array.

I did not know how to tell it to stop at the first } it sees after it sees \text{. And that is what I am struggling with. I know I wrote x__ but I needed to do this, so I can pick out the x.

Any idea how to do this? Either using StringPattern or ReqgularExpression will work.

$\endgroup$
1

2 Answers 2

9
$\begingroup$

With both StringPattern and RegularExpression the problem is greediness: wildcards will try to match as much as possible. With StringPattern this can be fixed using Shortest:

StringReplace[buf, "\\text{" ~~ Shortest[x___] ~~ "}" :> x]

With a regular expression a quantifier can be made ungreed with ? (e.g. {(.*?)}), but when you're going that way, you can actually write a safer regular expression using a negated character class:

StringReplace[buf, RegularExpression["\\\\text{(.*?)}"] :> "$1"]

Which gives the same result.

Both of these have one issue though: they're not entirely safe. When your actual string contains }, then they will stop at that. Consider:

lst = {"abc", "x}y", "123"};
buf = ToString@TeXForm@lst

This gives:

\{\text{abc},\text{x$\}$y},123\}

And using either solution will turn it into:

\{abc,x$\$y},123\}

I think to fix this, only a regular expression approach is viable, which knows exactly what characters (or combinations) are allowed within the {...}:

StringReplace[buf, RegularExpression["\\\\text{((?:\\\\.|[^\\\\}])*)}"] :> "$1"]

Which gives

\{abc,x$\}$y,123\}

as expected.

$\endgroup$
3
  • $\begingroup$ Thanks. It is the Shortest which I did not know to use. $\endgroup$
    – Nasser
    Apr 6, 2016 at 10:25
  • 1
    $\begingroup$ +1, but all those escapes in the penultimate snippet are making me dizzy. :) $\endgroup$ Apr 6, 2016 at 10:29
  • 2
    $\begingroup$ @J.M. hm yeah, it's definitely annoying me that there is no sort of verbatim string syntax in Mathematica whenever working with regex. :/ $\endgroup$ Apr 6, 2016 at 10:53
0
$\begingroup$

A form without RegularExpression that I believe works on Martin's second example:

lst = {"abc", "x}y", "123"};
buf = ToString @ TeXForm @ lst;

StringReplace[buf, "\\text{" ~~ Shortest[a___] ~~ b : Except["\\"] ~~ "}" :> a <> b]
"\\{abc,x$\\}$y,123\\}"
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.