Re: regex for any word from a set of words

The original post was also looking the "positive" form of this search --
a regular expression which matches the entire content of a string which
contains at least one of a set of string literals.

That would look like this:

	^(.*(?:foo|bar|baz).*)$

Where:
^			# Anchor to beginning of string
 (			# Capture
  .*			# Any characters
  (?:foo|bar|baz)	# Followed by complex expression matching string
literal "foo" or "bar" or "baz"
  .*			# Any characters
 )			# End capture
$			# Anchor to end of string


-----Original Message-----
From: Mark Merrill [mailto:Mark.Merrill-r/[email protected]] 
Sent: Monday, April 21, 2008 2:01 PM
To: Uttormark, Tim
Cc: [email protected]; [email protected]; Tim Jowers
Subject: Re: [Juglist] regex for any word from a set of words

I like that Tim.  Very elegent.

I was trying to get a negative lookup to work, making sure that a
certain
string didn't exist before (and I guess you'd have to use positive
lookup
to make sure it didn't exist after), but I was obviously making it far
too
difficult on myself.

Mark Merrill
Development Engineer


|------------>
| From:      |
|------------>
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|
  |"Uttormark, Tim" <[email protected]>
|
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|
|------------>
| To:        |
|------------>
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|
  |"Tim Jowers" <[email protected]>
|
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|
|------------>
| Cc:        |
|------------>
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|
  |[email protected]
|
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|
|------------>
| Date:      |
|------------>
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|
  |04/21/2008 01:52 PM
|
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|
|------------>
| Subject:   |
|------------>
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|
  |Re: [Juglist] regex for any word from a set of words
|
 
>-----------------------------------------------------------------------
------------------------------------------------------------------------
---|





I presume the problem is to define a regular expression which returns
the
entire content of a string for all strings not containing any of a set
of
string literals.

How about this:

^((?:(?!foo)(?!bar)(?!baz).)+)$

Where:?!

^               # Anchor to beginning of string
  (             # Capture
   (?:          # Complex expression:
      (?!foo)  #   make sure we're not at the beginning of string
literal
"foo"
      (?!bar)  #     and we're not at the beginning of string literal
"bar"
      (?!baz)  #     and we're not at the beginning of string literal
"baz"
      .         #   accept any character
   )+           # at least once (to ensure string is not empty)
  )            # End capture
$              # anchor to end of string

-- Tim Uttormark
   Principal Java Architect, Peopleclick, Inc.



From: [email protected] [mailto:[email protected]] On
Behalf Of Tim Jowers
Sent: Monday, April 21, 2008 10:01 AM
To: Java Users Group,
Subject: [Juglist] regex for any word from a set of words


Hi,

I'm seeking to define a regular expression to match on one of a set of
words. Does anyone have a trick for this? Does anyone have a trick for
negative search (find all but strings having a word)?

Specifically, EMC Smarts network monitoring software allows subscription
to
events based on a regular expression. We do not want noise events like
"Unresponsive" or "Duplicate".

To match all, I can use:
.*
To match the ones I do NOT want I could use:
.*Unresponsive$  to find matches to with Unresponsive
.*[U|D][n|u][r|p][e|l][s|i][p|c][o|a][n|t]s?i?v?e"

But of course the last one also matches other letter combinations like
"Unplicate". :-) This could be OK for our scenario as we have defined a
small set of event types.

But what about NOT having a word. I've searched for this on the net for
about 1.5 hours and tried various things but does anyone know a good way
to
write a regular expression to match any strings *without* a word from a
set
of words?

I can contrive something to match anything except the word
"Unresponsive"
for instance:
".+?_(?!U(?:n(?!r(?:e(?:s(?!p(?:o(?!s)))))))).+?"

Unfortunately, the "groups" created with parenthesis really are not much
use in Pattern.matches but only when iterating for matches. E.g. I can
find
matches with:
    String regex = ".+?com_(?!Unresponsive)";
    Pattern pattern = Pattern.compile(regex);
    Matcher matcher = pattern.matcher(candidateString);
    while (matcher.find()) {
      System.out.println("MATCH:" + matcher.group() );
    }
The (?!  signifies the negative search and works fine for making groups
but
does not seem to work for testing if an expression matches or not. I can
try the ^ operator but like everything else it only seems to apply to
the
next character.

My reasoning is the regular expression logic creates a match graph based
on
characters so any "word" matching really has to be distilled to
character
matching which make the use of a Regular Expression for this task a
tedious
challenge at best and error prone at least. Has anyone found a class or
library with a Regular Expression style functionality but operating on
Strings? Or Strings as well as characters?


Thanks!
TimJowers

_______________________________________________
Juglist mailing list
[email protected]
http://trijug.org/mailman/listinfo/juglist_trijug.org