Question

I am using my texteditor (e-texteditor)’s regular expression to try to reverse some elements in a line of code. So here’s what I have:

  [[Date.today, "Today (#{Date.today})"], [Date.tomorrow, "Tomorrow (#{Date.tomorrow})"], [Date.today+3, "In 3 Days (#{Date.today+3})"], [this_weekend, "This Weekend (#{this_weekend})"]]

And I actually wanted it the reverse of it:

  [[this_weekend, "This Weekend (#{this_weekend})"], [Date.today+3, "In 3 Days (#{Date.today+3})"], [Date.tomorrow, "Tomorrow (#{Date.tomorrow})"], [Date.today, "Today (#{Date.today})"]]

By the way, the code is in Ruby in case you’re wondering. To visualize better:

Have: [A, B, C, D]
Want: [D, C, B, A]

So far, I am doing the dumb thing (or at least I think so) by searching for

 (\[[^[].+?\]),\s(\[[^[].+?\]),\s(\[[^[].+?\]),\s(\[[^[].+?\]) 

so it’s like search for (A),\s(B),\s(C),\s(D)

And then replace it with

  $4, $3, $2, $1

While that works, but it’s not very nice isn’t it, what if I have more than 4 elements. That’s a lot of typing up still. How will you approach it?

Here is the regex ‘engine/syntax’ that e-texteditor supports

Thank you!

Answer 1

This (\[[^[].+?\]), could be a problamatic subexpression.
It works in this case because the sample data you provide is
well formed. In ill-formed context, it could overflow to the next ],
if not preceded by a [, even though .+? is non greedy.

It might be better as this (\[[^\[\]]+\]).

To your question, I think that only .NET has variable amount of
capture buffers (its Capture Collection) and even then I’m not sure
how its variability could be used on the replacement side.
Example: \[(?:(\[[^\[\]]+\])(?:,\s*|(?=\])))+\] should make
a variable collection that you may be able to programatically
reverse on the replacement side. I’ve never done this so not
too sure.

Otherwise, it involves a couple of seperate intermediate steps
outside of a regex.

In Perl, it’d be like use global search and replace, where the
replacement side is a callback function that gets passed each individual record.

s/(\[(?:\[[^\[\]]+\](?:,\s*|(?=\])))+\])/ func($1) /eg;`

Then in the callback, parse the individual elements from the record passed, then
rebuild the string in reverse order, which then gets passed back to the original
regex to be used as the replacement for that record.

Perl code:

use strict;
use warnings;

my $samp = '[[Date.today, "Today (#{Date.today})"], [Date.tomorrow, "Tomorrow (#{Date.tomorrow})"], [Date.today+3, "In 3 Days (#{Date.today+3})"], [this_weekend, "This Weekend (#{this_weekend})"]]';

$samp =~ s/(\[(?:\[[^\[\]]+\](?:,\s*|(?=\])))+\])/ func($1) /eg;

print "\n$samp\n";


sub func
{
   my $rec = shift;
   my @elements = $rec =~ /(\[[^\[\]]+\])/g;
   return '['. join( ', ', reverse @elements ) . ']';
}

Output:

[[this_weekend, "This Weekend (#{this_weekend})"], [Date.today+3, "In 3 Days (#{Date.today+3})"], [Date.tomorrow, "Tomorrow (#{Date.tomorrow})"], [Date.today, "Today (#{Date.today})"]]