Re: Comparing Regular Expressions

There are some good benchmarking modules, but there may be easier ways:

perl -le '$text = <STDIN>; $T=time; $TestText=$text; for (1..$ARGV[0])
+ { $text =~ s/\s+,\s+|\s+,|,\s+/,/g; } print time - $T; $T=time; $Tes
+tText=$text; for (1..$ARGV[0]) { $text =~ s/\s*,\s*/,/g; } print time
+ - $T;' 1000000 </tmp/text.file
[download]

This oneliner just runs your two regexps on the same text for a specific number of loop runs. Use a number which takes at least 10, better more, seconds for each loop.
This could also be improved by using Time::HiRes.

Comment on Re: Comparing Regular Expressions Download Code

Replies are listed 'Best First'.
Re^2: Comparing Regular Expressions by JavaFan (Canon) on Sep 01, 2009 at 00:31 UTC
Your benchmark is bogus. Realize that `s///` actually modifies the string it's operating on. After the first iteration, there will be no spaces around commas left in `$text`. Furthermore, you assign to `$TestText`, but you never actually use its value.	[reply] [d/l] [select]
Re^3: Comparing Regular Expressions by Sewi (Friar) on Sep 01, 2009 at 07:13 UTC
You're completly right. Thank you for the hints. `perl -le '$text = <STDIN>; $T=time; for (1..$ARGV[0]) { my $TestText = + $text; $TestText =~ s/\s+,\s+\|\s+,\|,\s+/,/g; } print time - $T; $T=t +ime; $TestText=$text; for (1..$ARGV[0]) { my $TestText = $text; $Test +Text =~ s/\s,\s/,/g; } print time - $T;' 1000000 </tmp/text.file` [download] This should correct all bugs.	[reply] [d/l]