[XERCESJ-1126] Regular Expressions starting with "." in non-XML Schema mode use wrong first character optimization

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.7.1
• Fix Version/s: 2.8.1
• Component/s: XML Schema 1.0 Datatypes
• Assignee: Michael Glavassevich
• Reporter: Martin Probst
• Created: 05/Jan/06 23:35
• Updated: 08/Sep/06 17:40
• Resolved: 21/Jun/06 02:59

Description

The following Java snippet prints "not matched", but should print "matched".

RegularExpression regex = new RegularExpression(".oo", "");
if (regex.matches("foo")) System.out.println("matched");
else System.out.println("not matched");

It uses the class org.apache.xerces.impl.xpath.regex.RegularExpression.java. I believe this happens because of the first character optimization kicking in and checks for a first character of "o", does not match the 'f' and then consequently returns false. This may be caused by this code snippet from Token.java:493

case DOT: // ****
if (isSet(options, RegularExpression.SINGLE_LINE))

{ return FC_CONTINUE; // **** We can not optimize. }

else

{ return FC_CONTINUE; /* * result.addRange(0, RegularExpression.LINE_FEED-1); * result.addRange(RegularExpression.LINE_FEED+1, * RegularExpression.CARRIAGE_RETURN-1); * result.addRange(RegularExpression.CARRIAGE_RETURN+1, * RegularExpression.LINE_SEPARATOR-1); * result.addRange(RegularExpression.PARAGRAPH_SEPARATOR+1, UTF16_MAX); * return 1; */ }

I think it should unconditionally return FC_ANY for DOT, at least in the case of a starting '.'