http://www.perlmonks.org?node_id=782109


in reply to Re: How do I quickly strip blank space from the beginning/end of a string?
in thread How do I quickly strip blank space from the beginning/end of a string?

You're right, in my haste to get this question out there I made the jump from my original problem to japhy's post to talking about his benchmarks. One of my coworkers pointed out the same thing to me as I was headed out the door. Obviously I should be benchmarking the exact problem I want to solve, not some generally similar example. After I get the kids to bed I'll write a better benchmark and try your \K suggestion below too. Thanks.

Also, you made the point that none of my input ends with spaces. I think that's generally true in real life usage too. It's frustrating that we have this pervasive idiom in our code of "strip whitespace just in case", but I think most of the time the input is already just fine. In fact, I think much of the time the input is short and has no spaces at all. I wonder if I should be checking it with index() first to quickly rule out that case.

update: added paragraph spacing

  • Comment on Re^2: How do I quickly strip blank space from the beginning/end of a string?

Replies are listed 'Best First'.
Re^3: How do I quickly strip blank space from the beginning/end of a string?
by ikegami (Patriarch) on Jul 21, 2009 at 23:06 UTC

    Also, you made the point that none of my input ends with spaces. I think that's generally true in real life usage too.

    Then shouldn't you be benchmarking space detection?

[reply]
[d/l]
      Good point. I've tried the 3 methods below: 
      if ($x =~ /\s/)

      [download]
      and 
      if (substr($x, -1) =~ /^\s/

      [download]
      and 
      if (rindex($x," ") == 0 || rindex($x,"\r") == 0 ||  rindex($x,"\n") ==
+ 0 ||  rindex($x,"\t") == 0)

      [download]
      The rindex method is much faster on the data that has spaces; otherwise the substr method is the fastest. Any other good ideas, keeping in mind that the data (I think) won't often contain any trailing spaces?
[reply]
[d/l]
[select]

        $x =~ /\s/
        doesn't work. It should be
        $x =~ /\s\z/
        Bonus: The speed of fixed version won't depend on the length of the string like the broken one did.

        if (rindex($x," ") == 0 || rindex($x,"\r") == 0 || rindex($x,"\n") == 0 || rindex($x,"\t") == 0)
        doesn't work. It should be
        if (rindex($x," ") == length($x)-1 || rindex($x,"\n") == length($x)-1 || rindex($x,"\t") == length($x)-1 || rindex($x,"\r") == length($x)-1)
        Or just
        my $ch = substr($x, -1); $ch eq " " || $ch eq "\n" || $ch eq "\t" || $ch eq "\r"

        And then there's
        length($x) && index(" \n\t\r", substr($x, -1)) >= 0

        You should be concentrating on writing code that actually works before worrying about operations that take 0.00001 second.

[reply]
[d/l]
[select]

In Section Seekers of Perl Wisdom