Match whitespace or start-of-string with lookbehind

BigLug has asked for the wisdom of the Perl Monks concerning the following question:

I'm trying to curl my quotes. To do that I need to match a quote after whitespace followed by a character. However what if the quote is the first thing in the string? In that case I need to match start-of-string or whitespace. Here's code that works in my head:

$text =~ s/(?<=(^|\s))"(?=\w)/“/gs;
[download]

However because ^ is zero-width and \s isn't, I get an error about variable width lookbehind not being supported.

How do I do this? Or will I need to use two regexes for this?

Cheers!
Rick
If this is a root node: Before responding, please ensure your clue bit is set.
If this is a reply: This is a discussion group, not a helpdesk ... If the discussion happens to answer a question you've asked, that's incidental.

Comment on Match whitespace or start-of-string with lookbehind Select or Download Code

Replies are listed 'Best First'.
Re: Match whitespace or start-of-string with lookbehind by Enlil (Parson) on Aug 24, 2004 at 02:54 UTC
I am not quite sure of what you are asking but here is something that I think is functionally equivalent to what you are showing doing: `$text =~ s/(^\|\s)"(?=\w)/$1“/gs;` [download] In short I am just changing the variable length lookbehind assertion and making it a regular captured pattern and putting it back after the match. HTH -enlil	[reply] [d/l]
Re^2: Match whitespace or start-of-string with lookbehind by BigLug (Chaplain) on Aug 24, 2004 at 02:58 UTC
Yeah, that works. I just hate capturing something just to stick it back in the same place later! If this is a root node: Before responding, please ensure your clue bit is set. If this is a reply: This is a discussion group, not a helpdesk ... If the discussion happens to answer a question you've asked, that's incidental.	[reply]
Re: Match whitespace or start-of-string with lookbehind by etcshadow (Priest) on Aug 24, 2004 at 03:50 UTC
It's not precisely the same thing as you're asking for... but would word-boundary (\b) be sufficient? If not, you can just re-nest the look-behind and the alternation into: `$text =~ s/(^\|(?<=\s))"(?=\w)/“/gs;` [download] That works. They key is that the ^ doesn't have to be look-behind, since it's zero-width, anyway. `------------ :Wq Not an editor command: Wq` [download]	[reply] [d/l] [select]
Re: Match whitespace or start-of-string with lookbehind by ysth (Canon) on Aug 24, 2004 at 05:07 UTC
In general you need one alternative lookbehind for each possible length: `(?:(?<=^)\|(?<=\s))`. In this case you can negate it to make the lookbehind always one character: `(?<!\S)`.	[reply] [d/l] [select]
Re: Match whitespace or start-of-string with lookbehind by Your Mother (Archbishop) on Aug 24, 2004 at 06:28 UTC
I think this does what you're after for this: `$text =~ s/(?: (?<=\s) \| (?<=\A) ) " (?=\w) /“/gx;` [download] (update, I think etcshadow's answer is better in this case, though the non-capture \|'d lookbehind is useful sometimes) (update 2: yikes, I don't know how I missed ysth's post, sorry for the duplication.)	[reply] [d/l]
Re: Match whitespace or start-of-string with lookbehind by diotalevi (Canon) on Aug 24, 2004 at 13:35 UTC
I like ysth's response best but here's another: size each alternative to be identical. `(?<= (?s: ^. ) \| \s )` [download]	[reply] [d/l]