$1 $2 Weirdness?

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

Replies are listed 'Best First'.
Re: $1 $2 Weirdness? by graff (Chancellor) on Jun 27, 2005 at 00:11 UTC
This substitution: `$name=~s/(\w)(\s)(\w)/$1/eig;` [download] is only removing the space and the first character of the second word. What you want instead is something like this (note that capturing parens are not needed): `$name =~ s/\s.*//; # replace space and all following characters wit +h empty string` [download] And you better read up on your regex qualifiers -- the "eig" at the end of that substitution is completely unnecessary: "e" means treat the replacement string as executable code -- I don't think that's what you really want "i" means use case-insensitive matching -- but you have no upper or lower case letters in the right-hand side (the match) part of the s/// operator "g" means apply the match/replacement globally -- but your sample input suggests that a single application will suffice. update: forgot to mention, the first substitution could be a lot simpler as well -- again, no capturing parens needed: `$temp =~ s/\s/+ +/;` [download]	[reply] [d/l] [select]
Re^2: $1 $2 Weirdness? by apt_get (Acolyte) on Jun 27, 2005 at 00:16 UTC
Thanks graff for pointing out the error of my ways. BTW, I know what the "e", "i" and "g" qualifiers are, but thanks anyways.	[reply]
Re: $1 $2 Weirdness? by sk (Curate) on Jun 27, 2005 at 00:11 UTC
Modifying only the second portion `$name=~s/(\w+)(\s+)(\w+)/$1/ig; print "$name\n";` [download] \w looks for just one word-character. You need look for one or more with the fix output - `Sam+ +Adams Sam Bud+ +weiser Bud Ice+ +House Ice` [download] cheers SK	[reply] [d/l] [select]
Re: $1 $2 Weirdness? by GrandFather (Saint) on Jun 27, 2005 at 00:11 UTC
Because in `~s/(\w)(\s)(\w)/$1/eig` (\w) matches one character, (\s) matches one character and (\w) matches one character. So in "Sam Adams" m A, gets matched and replaced with m. What you more likely want is: `$name=~s/(\w?)(\s)./$1/eig;` [download] Perl is Huffman encoded by design.	[reply] [d/l] [select]
Re^2: $1 $2 Weirdness? by bart (Canon) on Jun 27, 2005 at 08:40 UTC
What you more likely want is: `$name=~s/(\w?)(\s)./$1/eig;` [download] I don't think so. yet more cargo cult regexing, huh? sigh Let's strike all the unnecessary cruft. He wants `+`, not ``. What if his string started with a space? The `?` is unnecessary I see no reason to replace `\w+` with `.` The `/e` modifier is totally useless, though by accident, harmless The `/i` modifier is useless If you use `.*`, then the `/g` modifier is useless Granted, the modifiers you inherited from the OP, but for the rest I see very little excuse. IMO this is what the OP is after: `$name=~s/(\w+)(\s)(\w+)/$1/g;` [download] or even `$name=~s/(\w+)\s\w+/$1/g;` [download] though it would likely also do what he wants without the `/g`.	[reply] [d/l] [select]
Re^3: $1 $2 Weirdness? by GrandFather (Saint) on Jun 27, 2005 at 20:33 UTC
Yes to all of that. I was cruising through the afternoon and spotted his immediate problem without actually engaging my brain and looking at the big picture. sigh. By the time I noticed it was too late to retract and others had pointed out the problems anyway. Perl is Huffman encoded by design.	[reply]
Re^2: $1 $2 Weirdness? by apt_get (Acolyte) on Jun 27, 2005 at 00:18 UTC
Thanks all, for your replies. For some brain dead reason, I was assuming the "\w" to match a "more than one character" word, instead of just a single character. Now I know.	[reply]