Regular expression help

I found a message from Randal Schwartz, Message-ID: <8cvi6k3cs5.fsf@gadget.c
scaper.com>#1/1
which gave a regular expression for a valid Unix name,

/^(?=.*?\D)[a-z\d]+$/

That works but why does it work?

/
^		# Start of a string
(?=		# 0 or 1 instance of
.*?		# anything but a newline
\D	# Non digit
)		#

[a-z\d]+	# All match a-z and any digit at
$	# End of a string
/


I tried breaking it down like above but it still doesn't say "Must not be al
l numbers and letters must be all lowercase"

Any help in turning that re into plain words would be appreciated


TIA



Owen

Owen <mailto:rcook@pcug.org.au> wrote:

: I found a message from Randal Schwartz, Message-ID:
: <8cvi6k3cs5.fsf@gadget.cscaper.com>#1/1
: which gave a regular expression for a valid Unix name,
:
:   /^(?=.*?\D)[a-z\d]+$/
:
: That works but why does it work?
:
:     /
:   ^       # Start of a string
:    (?=        # 0 or 1 instance of

(?=        # Zero-width positive look ahead assertion.


:       .*?     # anything but a newline
:    )      #
:
:   [a-z\d]+    # All match a-z and any digit at

[a-z\d]+    # Match a-z (lowercase only) and any digit


:       $   # End of a string
:     /
:
: I tried breaking it down like above but it still doesn't say
: "Must not be all numbers and letters must be all lowercase"
:
: Any help in turning that re into plain words would be
: appreciated

AFAIK, a (?= ... ) construct affects the contents of $&, $`,
and $'. Can you show any code immediately following this regular
expression and the whole line with it?


HTH,

Charles K. Clarkson
--
Mobile Homes Specialist
254 968-8328

Owen wrote:
> I found a message from Randal Schwartz, Message-ID:
> <8cvi6k3cs5.fsf@gadget.cscaper.com>#1/1
> which gave a regular expression for a valid Unix name,
>
> 	/^(?=.*?\D)[a-z\d]+$/
>
> That works but why does it work?
>
>     /
> 	^		# Start of a string
> 	 (?=		# 0 or 1 instance of
> 	    .*?		# anything but a newline
> 	 	\D	# Non digit
> 	 )		#
>
> 	[a-z\d]+	# All match a-z and any digit at
> 		$	# End of a string
>     /
>
>
> I tried breaking it down like above but it still doesn't say "Must not be
> all numbers and letters must be all lowercase"
>
> Any help in turning that re into plain words would be appreciated

(?=.*?\D) is a zero-width assertion so it doesn't affect the match which is
/^[a-z\d]+$/ -- match a string whose only characters are a-z0-9.  When you a
dd
the zero-width assertion it ensures that there is at least one \D character 
in
the match.


John
--
use Perl;
program
fulfillment

On 4/25/05, John W. Krahn <krahnj@telus.net> wrote:
> Owen wrote: 
be 
>=20
> (?=3D.*?\D) is a zero-width assertion so it doesn't affect the match whic=
h is
> /^[a-z\d]+$/ -- match a string whose only characters are a-z0-9.  When yo=
u add
> the zero-width assertion it ensures that there is at least one \D charact=
er in
> the match.
>=20
> John

I think this may be part of the confusion, too:
 

/.*?/ is not that same as /.?/.  /.?/ matches zero or 1 of any
character but the newline.  /.*?/, however, matches zero or more, just
like /.*/.  It's non-greedy, though, so it will stop matching at the
first occurance of \D instead of the last.  In this case it boils down
roughly to /(?=3D[^\D]{0,}\D/.

HTH,

--jay