regular expressions in perl explanation - regex

Question

I need help in understanding the below regular expressions.
Can somebody please tell what this means m/^(\d+\/\d+\/\d+\s\d+\:\d+\:\d+\.\d+)/msx
foreach my $line ( #{ $self->{'stdout'} } ) {
if ( $line =~ m/^(\d+\/\d+\/\d+\s\d+\:\d+\:\d+\.\d+)/msx ) {
$timestamp = $1;
}

Answer 1

This
if ( $line =~ m/^(\d+\/\d+\/\d+\s\d+\:\d+\:\d+\.\d+)/msx ) {
is quite unreadable, though the author made a start to make it readable as it has the /x flag (allowing whitespace, but not made use of), but it still suffers from backslashitis and doesn't limit the matches to what is really meant.
Rewriting it with different delimiters allows to get rid of some of the backslashes:
if ( $line =~ m{^(\d+/\d+/\d+\s\d+:\d+:\d+\.\d+)}msx ) {
Adding whitespace and using [.] instead to match a single dot and adding comments can provide a better idea of what would be matched:
if ( $line =~ m{^ # (start of line)
( # (capture group $1)
\d+ / \d+ / \d+ # digit(s) slash digit(s) slash digit(s)
\s # ANY whitespace character (space, tab, etc)
\d+ : \d+ : \d+ # digit(s) colon digit(s) colon digit(s)
[.] \d+ # dot digit(s)
) # (end capture group $1)
}msx ) {
Where digit(s) means one ore more digits 0-9 (or any utf8 digit). So this would happily match something like "00000/0000/0000000000 0000:0000000000000000000000:000000.0000", but it seems they meant to match e.g. "0000/00/00 00:00:00.000" (a time stamp including milliseconds).
A better regex (with a lower chance of matching something it shouldn't, though it is anchored to the start of the line so no real practical difference here but as a general rule it's highly advisable to be as specific as you can) would be something like this:
if ( $line =~ m{^
(
[0-9]{4} / [0-9]{2} / [0-9]{2}
[ ] # space character
[0-9]{2} : [0-9]{2} : [0-9]{2}
[.] [0-9]{3}
)
}msx ) {
With that in hand, the regex manpage others linked already should make more sense.

Answer 2

This original regex
m/^(\d+\/\d+\/\d+\s\d+\:\d+\:\d+\.\d+)/msx
is a very poorly-written pattern that matches a date/time string that looks like 2014/07/31 22:53:42.123
Since it contains no dots . the /s modifier is redundant.
The /x modifier allows whitespace layout to be added so we may as well do that, using a different delimiter so that slashes don't need escaping
m{ ^ ( \d+ / \d+ / \d+ \s \d+ : \d+ : \d+ \. \d+ ) }mx
So that matches
From the beginning of any line (i.e. at the start of the string or right after a newline because the /m modifier is in effect)
Capture the following
Some digits, a slash, some more digits, another slash, some more digits
A white space character (here I think a space was assumed)
Some digits, a colon, some more digits, another colon, some more digits, a dot, some more digits
Stop capturing
So, as I said, it would match (and capture)
2014/07/31 22:53:42.123
It would also match
0/1/2 3:4:5.6
I hope that helps