LinuxSA Mailing list archives

  From: Andrew Hill <list@fornax.net>
  To  : David Lloyd <lloy0076@rebel.net.au>
  Date: Mon, 06 Nov 2000 13:05:38 +1030

Regexp (was Re: extract marked infix from filename with bash script?)

David Lloyd wrote:
> Because I don't have an example of an actual file name (it would be
> useful), a perl regular expression such as:
> 
> s/.*\.D(d)+.*/$1/
> 
> ...reads "match any character, but when you see one single dot, followed
> by a non-digit, followed by one or more digits, save that for later and
> call it $1". I think you need something that will match the pattern and
> then allow you to extract just the digit parts.

Hi All,

Aren't regular expressions tricky? You need to be very careful about
what you say. (And David, I'm not having a go here - I can't recall the
number of times I've gotten a regexp wrong. I just want to help out
anyone that may try to get the regexp you've written to work, and can't
do so :-)

Firstly, in the regexp above, "D" and "d" will match the letters "D" and
"d", respectively. What you really want is "\D" and "\d" to match a non
digit and a digit. Or, if you prefer, "\D" is the same as [^0-9] and
"\d" is the same as [0-9].

So, a regexp of:

s/.*\.\D(\d)+.*/$1/

is more what you are looking for.

Now, the tricky part - what David has said about the way the regexp
works is slightly misleading (though I am sure this wasn't the intent
:-)

In Perl, the way that this regexp will work is as follows: "Match zero
or more characters that are valid for '.' (Often, this will mean match
zero or more of anything until a new line, but you can put Perl into
various modes, or use command modifiers to the regexp that will allow
new lines to be matched by '.') Then, match a literal '.'. If you can't
find a literal '.', then back off one character to the left, and try to
match that literal '.' again. If you can't, back off again, and again,
and so on."

So what this means in that ".*\." will not match anything up to the
first literal '.', and then the dot, but will match everything up to and
including the last (rightmost) literal '.' in the string.

The, it will continue, matching any non digit (or, if that fails, will
then back off another character, try to match the '.', then the non
digit, etc.), then match one or more digits, and then match zero or more
of anything that '.' will match after that.

So, in the assumed simple case that David was writing this regexp for,
it probably would have worked as expected. But move it to the big wide
world, and the explanation would have kept you wondering what was going
on.


Thought - I don't claim to be an absolute master of regexps, or a good
public speaker, but would people be interested in a LinuxSA talk about
regular expression and how they work (with most of the examples in Perl,
seeing that's what I know?)

Cheers,

-- 
Andrew Hill

"Right now, I'd happily snort gunk from the sink if it would take
my brain somewhere away from here...." - JB

-- 
LinuxSA WWW: http://www.linuxsa.org.au/  IRC: #linuxsa on irc.linux.org.au
To unsubscribe from the LinuxSA list:
  mail linuxsa-request@linuxsa.org.au with "unsubscribe" as the subject