s///g and capturing all matches

good $localtime fellow monks,

I am tearing my hair out over a small oddity. Why will this work:

perl -nle "my (@c) = /\:(\w+)\:/g; print join '.', @c"
 input> out:caputure1:out:capture2:out
output> caputure1.capture2
[download]

But this behaves so differently

perl -nle "my (@c) = s/\:(\w+)\:/X/g; print join '.', @c"
 input> out:caputure1:out:capture2:out
output> 2
[download]

I am trying to extract comments from a file I have in a buffer read in with local $/;. The comments are marked by a ; and continue to the end of the line. For various reasons I want the file as a sinlge line and want to extract the comments. The comments will be added back in later so I substiture them for a marker. I realise I could break the file into lines and put it back together again later but I would love to be able to do it in one line

Cheers,
R.

Pereant, qui ante nos nostra dixerunt!

Comment on s///g and capturing all matches Select or Download Code
Re: s///g and capturing all matches by Anonymous Monk on Dec 15, 2011 at 14:26 UTC
:D Because m// and s/// are different operators that behave differently :D s/// always returns the number of matches, unless you use /r, in which case it returns a string copy see perlop#s/PATTERN/REPLACEMENT/msixpodualgcer `my @c; s/(...)/ push @c, $1; q[X]; /ge` [download]	[reply] [d/l]
Re^2: s///g and capturing all matches by Random_Walk (Parson) on Dec 16, 2011 at 09:08 UTC
thank you anonymonk, very ellegant. I had forgoton the use of code in the RHS was possible. Cheers, R. Pereant, qui ante nos nostra dixerunt!	[reply]
Re: s///g and capturing all matches by jethro (Monsignor) on Dec 15, 2011 at 14:27 UTC
To quote the documentation of `s///` it " returns the number of substitutions made". You made two substitutions, so the result is 2	[reply] [d/l]
Re^2: s///g and capturing all matches by Random_Walk (Parson) on Dec 16, 2011 at 09:59 UTC
Hi Jethro, and the answer to life, the universe and everything is 42 ;) I was wondering why the implementations differ, as well as seeking an ellegant way to get the behaviour I wanted. If s///; returned a list of the items sub'd out then it would be easy enough to use that in scalar contect to have the number. Felt rather un-Perlish to miss out on potential utility. Cheers, R. Pereant, qui ante nos nostra dixerunt!	[reply]
Re: s///g and capturing all matches by johngg (Abbot) on Dec 15, 2011 at 15:06 UTC
Rather than a global match you could use a `while` statement modifier. `knoppix@Microknoppix:~$ perl -E ' > $file = <<EOD; > some data; comment 1 > more data; comment 2 > even more data; comment 3 > EOD > push @comments, $1 while $file =~ s{(;[^\n])}{MARKER}; > say $file; > say for @comments;' some dataMARKER* more dataMARKER even more dataMARKER ; comment 1 ; comment 2 ; comment 3 knoppix@Microknoppix:~$` [download] I hope this is of interest. Cheers, JohnGG	[reply] [d/l] [select]
Re^2: s///g and capturing all matches by Random_Walk (Parson) on Dec 16, 2011 at 09:52 UTC
Hi JohnGG, I had tried the while but could not get the syntax right. Thanks for pointing me down the right path. Cheers, R. Pereant, qui ante nos nostra dixerunt!	[reply]


Perl: the Markov chain saw
	PerlMonks

s///g and capturing all matches

When choosing user names for websites, I prefer to use: