in reply to most efficient regex to delete duplicate words

Here's one way.
$_="alpha beta beta gamma gamma gamma";
while (s/((\w+)\s\2)/$2/) {};
print $_;

[download]
the second line does something that may not be obvious to everyone, and seems to duplicate /g 's functionality. However, since you've (seemingly) got 3 gamma in a row, writing 
$_="alpha beta beta gamma gamma gamma";
s/((\w+)\s\2)/$2/g;
print $_;

[download]
Will leave you with an extra gamma. Using the 'useless' while loop allows the regex to check for multiple duplicates.

As for the regex you tried $string =~ s/(\w+)(.*)\b\1/$1 $2/sig;, we have :

Replies are listed 'Best First'.
Re: Re: most efficient regex to delete duplicate words (boo)
by blakem (Monsignor) on Aug 14, 2001 at 02:37 UTC
    I always perfer: 

    1 while (EXPR)

    [download]

    instead of:

    while (EXPR) {}

    [download]

    Simply because the '1 while' sticks out at the front where as the empty {} tends to get lost. I find that '1 while' immediately flags this perl idiom and makes it easier for me to pick it out.

    -Blake

[reply]
[d/l]
[select]
      1 while (EXPR): will also avoid the slight over head of entering and exiting the lexical context created by the BLOCK in while (EXPR) {} Perl might be smart enough to optimise the out the empty block though.
[reply]
[d/l]
[select]

      What's wrong with:

      "chill" while (EXPR);

      ???

[reply]
[d/l]

In Section Seekers of Perl Wisdom