util.regex.Pattern anomaly

[Roger Marquis]

Finally figured out an odd behavior in 1.4's util.regex.Pattern
class. Specifically, the matcher() method does not work as expected.
Unlike Perl, grep, sed, and most other regex engines Java's
matcher() must match the _entire_ line to return true.

The documentation does have a section titled "Comparison to Perl
5" but for some reason it doesn't list this important difference
(though it does provide a hint in the MULTILINE detail). The fix
is inelegant but straightforward: simply add a wildcard (.*) prefix
and/or suffix to each Pattern.compile. For example the perl/sed/grep:

/somestring/

is equivalent to Java's:

/.*somestring.*/

Here's an example in this hostname validation method:

public static boolean isValidFQDN(String FQDN) {
Pattern legal = Pattern.compile("[0-9a-zA-Z\\.\\-]+");
Matcher isLegal = legal.matcher(FQDN);
if ( ! isLegal.matches() ) {
return false;
}
Pattern illegal = Pattern.compile(".*\\.\\..*|.*--.*|^\\..*|^-.*|.*\\.$|.*-$");
// -- note java regex wildcards ---^^------^^-^^--^^-----^^---^^-^^-----^^
Matcher isIllegal = illegal.matcher(FQDN);
if ( isIllegal.matches() ) {
return false;
}
return true;
}

 

[Roger Marquis]

Pattern illegal = Pattern.compile(".*\\.\\..*|.*--.*|^\\..*|^-.*|.*\\.$|.*-$");
// -- note java regex wildcards ---^^------^^-^^--^^-----^^---^^-^^-----^^

That should read:

Pattern illegal = Pattern.compile(".*\\.\\..*|.*--.*|^\\..*|^-.*|.*\\.$|.*-$");
// -- note java regex wildcards ---^^------^^-^^--^^-----^^---^^-^^-----^^

 

[Alan Moore]

A much better fix is to use find() instead of matches(). It looks for
a match *within* the target string, like Perl, instead of trying to
match the whole string.

 

[Roedy Green]

A much better fix is to use find() instead of matches(). It looks for
a match *within* the target string, like Perl, instead of trying to
match the whole string.

for some coding examples see http://mindprod.com/jgloss/regex.html

 

[Roger Marquis]

It seems you read Pattern's docs but not Matcher's - there are more
methods in that class than the one used by the example in Pattern's
javadoc.

That was the problem. That and a lack of of online documentation.
Thanks!