Quick Regex Query

ruby-talk
1

Hi all,

Really quick one for the regex gurus...

Anyone know of a regex string which will extract out the stuff between
"from" and "not" without the spaces between?

User root from toronto-hs-216-138-233-211.s-ip.magma.ca not allowed
because not listed in AllowUsers

I've found something which *almost* does the job...

\bfrom\W+(?:\w+\W+){1,6}not\b

This is limited however, as I can't know for certain how many "words"
will make up the hostname.

I've got a non-regex solution already, but you don't want to know how
that one works...

.... something to do with using split and relying on the fact that the
host name field always end up at the same index.

Seems more than a little clunky.

Any advise appreciated.

Cheers

~Neowulf

2

Neowulf wrote:

Hi all,

Really quick one for the regex gurus...

Anyone know of a regex string which will extract out the stuff between
"from" and "not" without the spaces between?

User root from toronto-hs-216-138-233-211.s-ip.magma.ca not allowed
because not listed in AllowUsers

I've found something which *almost* does the job...

\bfrom\W+(?:\w+\W+){1,6}not\b

This is limited however, as I can't know for certain how many "words"
will make up the hostname.

why do the number of words matter? i thought you only needed the
'stuff' between 'from' and 'not'. now when you say 'without the spaces
between', do you mean the spaces immediately after the from, and before
the not? what's wrong with:

/from\s+(\w+)\s+not/ && print "$1" # i'm more familiar with Perl, not
sure if $1 works in
                                                    # ruby, but it
should contain what's in capturing parens

can you provide several examples of input and expected output that
would illustrate your requirements?

3

Neowulf wrote:

Hi all,

Really quick one for the regex gurus...

Anyone know of a regex string which will extract out the stuff between
"from" and "not" without the spaces between?

User root from toronto-hs-216-138-233-211.s-ip.magma.ca not allowed
because not listed in AllowUsers

I've found something which *almost* does the job...

\bfrom\W+(?:\w+\W+){1,6}not\b

This is limited however, as I can't know for certain how many "words"
will make up the hostname.

I've got a non-regex solution already, but you don't want to know how
that one works...

... something to do with using split and relying on the fact that the
host name field always end up at the same index.

Seems more than a little clunky.

Any advise appreciated.

Cheers

~Neowulf

str = "User root from toronto-hs-216-138-233-211.s-ip.magma.ca not allowed because not listed in AllowUsers"

m = /from (.*?) not/.match(str)
puts m[1]

4

Neowulf wrote:

Hi all,

Really quick one for the regex gurus...

Anyone know of a regex string which will extract out the stuff between
"from" and "not" without the spaces between?

User root from toronto-hs-216-138-233-211.s-ip.magma.ca not allowed
because not listed in AllowUsers

I've found something which *almost* does the job...

\bfrom\W+(?:\w+\W+){1,6}not\b

host_name = /from\s+(\S+)\s+not/ =~ s && $1

This is limited however, as I can't know for certain how many "words"
will make up the hostname.

But there won't be any whitespace in the hostname, right? Then the RX
above will do it.

Kind regards

    robert

5

Robert Klemme wrote:

host_name = /from\s+(\S+)\s+not/ =~ s && $1

Tsk, those Perl guys... :wink:

host_name[/from\s+(\S+)\s+not/, 1] # Will either return nil or a String

···

--
http://flgr.0x42.net/

6

Thanks Guys,

Top effort all round.

Regex gives me a serious head ache sometimes...

Thanks again everyone for the help.

7

Robert Klemme wrote:

host_name = /from\s+(\S+)\s+not/ =~ s && $1

Tsk, those Perl guys... :wink:

Ssshhh! I'm trying to get over my Perl heritage but sometimes it still haunts me... :slight_smile:

host_name[/from\s+(\S+)\s+not/, 1] # Will either return nil or a
String

Thanks for reminding me - I always keep forgetting this variant, especially with the group selector!

Kind regards

    robert

···

Florian Groß <florgro@gmail.com> wrote: