Your example doesn't match your specification. If you were to match all
occurences of href=" that are not followed by the http:// sequence, with the
input:

<input>
href="http://www.mysite.com/test1.html" ... href="/test2.html" ...
</input>

you'd get one match:

<output>
 <match>href="</match>
</output>

you also mention a "replaceAll" but you don't say what you're replacing, and
with what.

Perhaps it'd help if you specified the goal, and not the method.

Are you trying to change all relative URLs in an HTML document to absolute
URLs?

   - Oliver

A pattern like "href=\"(?!http://).*" would exclude the string "http://"
after the "href=\"" part.  Depending on how everything's configured,
you've got to be sure to actually match the stuff you've just excluded
(as I do with the ".*").

The java.util.regex.Pattern doc on writing a pattern can be tough to
read.  Maybe unavoidable, regular expressions can be tough.  The (?!X)
construct is mentioned as a "zero-width negative lookahead" under
Special constructs.  By zero-width, they mean it doesn't actually
consume any characters.  It just asserts that at the current point in
the match, we must not be looking at X.

Cheers!
John