One-liner (or few-liner) for massaging regexps?

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

Let's say I have a line with an undefined number of numbers in it, seperated by random text but surrounded by spaces. Something like this will do for an example:

/dev/hda5               303344     99042    188641  35% /

Basically, I want to take anything that matches this: m/ (\d+) /

...and do something to it (ie divide by 1024), then stick it back in where it was. It would be nice to be able to do this with perl -pe 'something', but I can't even figure out how to do it at all - how do I match once on a line, change something, then start matching again after the spot I changed on the same line?

For example, let's assume I wanted to take every big number i see in my input and turn it into that number divided by 1000. How do I do that?

Comment on One-liner (or few-liner) for massaging regexps? Download Code

Replies are listed 'Best First'.
Re: One-liner (or few-liner) for massaging regexps? by japhy (Canon) on Sep 17, 2002 at 18:03 UTC
`perl -pe 's{ (\d+) }{sprintf " %g ", $1/1000}'` _____________________________________________________ Jeff`[japhy]`Pinyan: Perl, regex, and perl hacker, who'd like a job (NYC-area) `s++=END;++y(;-P)}y js++=;shajsj<++y(p-q)}?print:??;`	[reply] [d/l]
Re: One-liner (or few-liner) for massaging regexps? by RMGir (Prior) on Sep 17, 2002 at 17:50 UTC
use df -m to get megabytes? :) Oh, you mean in perl... `perl -pe's{(\d+)}{($1>1024)?(int($1/1024)):$1}eg;'` [download] The only problem with this is that what you replace isn't going to be the same width as what you removed, which could mess up columns. `perl -pe's{(\d+)}{($1>1024)?(sprintf("%".length($1)."d",$1/1024)):$1}e +g;'` [download] -- Mike	[reply] [d/l] [select]
Re^2: One-liner (or few-liner) for massaging regexps? by Aristotle (Chancellor) on Sep 17, 2002 at 21:28 UTC
That variable sprintf can be easier written as `perl -pe's{\b(\d+)\b}{ sprintf "%d", length $1, $1/1024 }ge'` Update:* added `\b`s - thanks to sauoq. Makeshifts last the longest.	[reply] [d/l]
Re: One-liner (or few-liner) for massaging regexps? by sauoq (Abbot) on Sep 17, 2002 at 17:44 UTC
`perl -pe 's! (\d+)!" ".$1/1024!eg'` Update: `perl -pe 's!\b(\d+)(\s)!($1/1024).$2!eg'` would be better. -sauoq "My two cents aren't worth a dime.";	[reply] [d/l] [select]
Re: One-liner (or few-liner) for massaging regexps? by Aristotle (Chancellor) on Sep 18, 2002 at 07:05 UTC
sauoq just /msg'd me to point out that all the regexes here are broken: they will happily replace the 5 in `/dev/hda5` along with all the other numbers. It has to be `\b(\d+)\b` to work correctly. Makeshifts last the longest.	[reply]
Re: Re: One-liner (or few-liner) for massaging regexps? by sauoq (Abbot) on Sep 18, 2002 at 14:12 UTC
I did but I was wrong. I think japhy's is actually fine as long as the those are real spaces in the original and not tabs. I think most df commands do use spaces and mine relied on the same thing. Unfortunately, two `\b`s won't work either. They break on the disk utilization percentage. My original reply breaks on it too because I don't match a space after the number. Matching either a `\b` or a `\s` before the number and a `\s` after it would work. -sauoq "My two cents aren't worth a dime.";	[reply] [d/l] [select]