use Perl

• and the regex problem/solution was? (Score:1)

  by n1vux (1492)

  So what was the problem and the solution and the explanation for why it works?

  --
  Bill
  # I had a sig when sigs were cool
  use Sig;
  • Re:and the regex problem/solution was? (Score:2)

    by pudge (1) on 2004.12.14 14:39 (#36792) Homepage Journal

    It had to do with alternation and backtracking. We were matching any consecutive nonwhitespace chars, and treating HTML entities as single chars, for the purpose of counting (we were breaking up strings of $n chars) and to make sure we don't split in the middle of the entities. So something like this (simplified):

    $foo =~ /\G (( &\w+; | \S ){$n}) /gx;

    Consider $foo = 'abc&' and $n = 3. I would want 'abc' to match, and '&' to not match, since it is not 3 chars long. '&am' matches, but I don't want it to to match. So how to make it not match?
    
    I added negative lookahead:

    $foo =~ /\G (( &\w+; | (?!&\w+;)\S ){$n}) /gx;

    This way, when it backtracks after the &\w+; fails, it will try to proceed with \S, but fail because the lookahead sees the '&'.
    
    It's not the most complicated thing, but sometimes we have mental blocks ... and bathrooms are where you relieve blockages ... (sorry :)

    Reply to This

    Parent