in reply to Perlre interpretation required

It means that the rules for parsing (?{}) and (??{}) are different than parsing normal perl code. The main difference is that all brackets must be either backslashed or balanced. For instance, this re is invalid:

/(??{ "\d+" if $1 eq "{" })/

It would need to be changed to:

/(??{ "\d+" if $1 eq "\{" })/

Replies are listed 'Best First'.
Re: Re: Perlre interpretation required
by BrowserUk (Patriarch) on May 02, 2003 at 21:54 UTC

    Thankyou. That explains not only the text, but some peculiarities I have been encountering and couldn't get to the bottom of.

    I sincerly think your explaination deserves adding to perlre.


    Examine what is said, not who speaks.
    1) When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong.
    2) The only way of discovering the limits of the possible is to venture a little way past them into the impossible
    3) Any sufficiently advanced technology is indistinguishable from magic.
    Arthur C. Clarke.
      I might have encountered Peculiarities in using (??{...}) too, and I intended to inquire about them, but didn't want to deal with it yet (guess I'm about to though!). I wanted to match either non-single or non-double quote characters in a regex, depending on which type of quotes surrounded it. For example, since
      $var =~ /(["'])[^\1]*?\1/;

      doesn't work due to [^\1] not working, I tried something like
      $var =~ /(["'])[^(??{"$1"})]*?\1/;
      which worked unless $1 contained a special character (forgot the character, but it wasn't a quote), when it gave me an error. I tried various fixes, and finally got
      $var =~ /(["'])([^"']|[^(??{"$1"})])*?\1/;
      to work. I figured the [^"'] alternative that I added would match the problem character, eliminating the need for parsing the (??{"$1"}) construct, which had given me the error. Well, it worked until it didn't, for whatever reason. I think I created another patch (placed \Q and \E somewhere), which didn't work consistantly either. Then I spent several hours deleting all of the (??{...}) constructs and stuck with a simple .*? between the quotes. I'm just a beginner, so it's possible I did something wrong, but I don't know what.

        $var =~ /(["'])[^(??{"$1"})]*?\1/; which worked unless $1 contained a special character (forgot the character, but it wasn't a quote),

        I don't quite understand this? How could $1 contain anything other than ["']?

        I'm just a beginner

        As am I.

        This is my attempt at capturing simple quoted elements - Ie. those without embedded quotes

        # Doesn't handle embedded like-quotes my $re_simpleQuoted = qr[ ( ["'] ) ( (??{ "[^\1]+?" }) ) \1 ]x;

        In order to handle embedded quotes you need to go a stage further. The code below seems to handle most cases I have tried of either or both embedded quotes like this "...""..." or '...''....''....', which IMO is the rational way to do it outside of perl source code, and also done this way, "...\"..." and '...\'...'.

        It doesn't handle basket cases where an escaped quote is immediately preceded by an escaped escape. Eg. "...\\\"...\"...". This could be addressed, but I think sufficiently rare that I wouldn't bother unless I had specific knowledge that it would arise. There are probably others that it would fail on too.

        Code

        Output

        The PITA of using capturing parens to establish which quote to look for is that it limits the use when combining regexes together as you never know which $n var you captured into.

        The 5.8 addition of the $^N variable has made this easier and the use of the experimental features more useful.

        The logical extension (IMO) to this would be to extend the non-capturing group syntax to allow capturing to a user defined variable (our vars make sense I think).

        Something like

        (?: ... :$captured)

        The other extension that I would like to see is a way of specifying a variable length match-anything-except-this construct. Like [^"] but longer

        (?>>stop when you encounter this)*

        Essentially, a non-zero length positive look ahead assertion! :)

        Makes sense to me, but then I have no idea of the costs/difficulties these would impose on the regex engine.

        T'would be nice though.


        Examine what is said, not who speaks.
        1) When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong.
        2) The only way of discovering the limits of the possible is to venture a little way past them into the impossible
        3) Any sufficiently advanced technology is indistinguishable from magic.
        Arthur C. Clarke.
Re: Re: Perlre interpretation required
by diotalevi (Canon) on May 02, 2003 at 21:51 UTC

    No - you only have to do that if the curly brackets are unbalanced. If you use only balanced brackets then you don't have to escape them.

      Isn't that what I just said? (-:
      The main difference is that all brackets must be either backslashed or balanced.

        And I didn't see that you said balanced. Oops!