problems with regular expressions

[jestew]

Hi,

If this is a misunderstanding and not a bug, I appologize.

I am attempting to match a Global Property, SplopLocations, using the regular expression of a piece’s $BasicName$ property.

So the property may look like “EyeGlutForebostinator”, and the BasicName is “Glut”. In a trigger action I attempted this:

SplopLocations =~ $BasicName$

but this failed. However if I add ‘.*’ to the front and the back of the match it suddenly works:

SplopLocations =~ .*$BasicName$.*

As another example, if SplopLocations was “Eye”, then the following works:

SplopLocations =~ ^Eye

but if SplopLocations is “EyeGlut”, then the example does not work!! This is completely wrong.

Please help me understand if I am confused. Cheers, jas…

[jestew]

I seem to answer my own replies too often…

Brent has supplied part of the answer here:

vassalengine.org/forums/view … =7771#7771

I removed the spaces after the =~ and before the $BasicName$ but it still doesn’t solve the problem.

Also, the problem exists for regexp negation using !~

Any suggestions what I’m doing wrong? Thanks, jas…

[jestew]

OK, at least I understand that the problem is not a VASSAL problem. It is a bigger issue with how Java appears to implement regexes.

Here is the output of a Perl and Java program running the same regexes against the same strings:

jasons@jackal:~/Documents/VASSAL/code$ java Test
regex Glut doesn't match: EyeGlutForebostinator
regex Glut.* doesn't match: EyeGlutForebostinator
regex .*Glut.* matches: EyeGlutForebostinator

jasons@jackal:~/Documents/VASSAL/code$ perl ./test.pl
Regex Glut matches: EyeGlutForebostinator
Regex Glut.* matches: EyeGlutForebostinator
Regex .*Glut.* matches: EyeGlutForebostinator

Notice they are completely different. Java and Perl do not agree about regular expressions. I found the explanation here

The punch line is:

Note: Java is tricky here as matches() tries to match the WHOLE string, so matches(“ian”) actually means matches(“^ian$”), which wouldn’t find anything in this case! Remember: matches(“”) means matches(“^$”), so if you want to look for a substring, expand the “^$” regexp. Ex.: does the string contain “ian”? The regexp is: “^.ian.$”, that is: anything, “ian”, anything, where anything can also be an empty string.

Looking at src/VASSAL/counters/PropertiesPieceFilter.java

accept() is coded using Pattern.matches() - so this explains what is happening. If instead of using Pattern.matches() if it were to be recoded using Matcher.find() you would get a much more universal behavior that matches the expectations of users familiar with unix (grep) regexes and Perl regexes.

Using Matcher.find() the results are the same as for Perl:

jasons@jackal:~/Documents/VASSAL/code$ java Test2
regex Glut matches: EyeGlutForebostinator
regex Glut.* matches: EyeGlutForebostinator
regex .*Glut.* matches: EyeGlutForebostinator

Any thoughts about this?

Thanks for your time. Cheers, jas…

[Brent_Easton]

OK, at least I understand that the problem is not a VASSAL problem. It is
a bigger issue with how Java appears to implement regexes.

Here is the output of a Perl and Java program running the same regexes
against the same strings:

Code:

jasons@jackal:~/Documents/VASSAL/code$ java Test
regex Glut doesn’t match: EyeGlutForebostinator
regex Glut.* doesn’t match: EyeGlutForebostinator
regex .Glut. matches: EyeGlutForebostinator

jasons@jackal:~/Documents/VASSAL/code$ perl ./test.pl
Regex Glut matches: EyeGlutForebostinator
Regex Glut.* matches: EyeGlutForebostinator
Regex .Glut. matches: EyeGlutForebostinator

Notice they are completely different. Java and Perl do not agree about
regular expressions. I’m a bit bewildered with this result.

Exactly what Java code are you using to evaluate this?

To my naive eyes, the Java version is correct and the Perl expression is incorrect.

I believe Java is checking if the string matches the entire regular expression, whereas Perl is returning true if any subset of the string matches the regexp.

B.

---

Messages mailing list
Messages@forums.vassalengine.org
forums.vassalengine.org/mailman/ … engine.org

Post generated using Mail2Forum (mail2forum.com)

[jestew]

@Brent Easton

I believe Java is checking if the string matches the entire regular expression, whereas Perl is returning true if any subset of the string matches the regexp.

Hey Brent,

Sorry, I went back and edited the post with an explanation. Hope that clears things up.

Now that I understand what is happening, I can code around it. But I believe that using Matcher.find() would be backwards compatible, and it would be much more powerful.

Cheers, jas…

[tar]

On May 10, 2009, at 2:21 AM, jestew wrote:

OK, at least I understand that the problem is not a VASSAL problem.
It is a bigger issue with how Java appears to implement regexes.

Here is the output of a Perl and Java program running the same
regexes against the same strings:

Code:

jasons@jackal:~/Documents/VASSAL/code$ java Test
regex Glut doesn’t match: EyeGlutForebostinator
regex Glut.* doesn’t match: EyeGlutForebostinator
regex .Glut. matches: EyeGlutForebostinator

jasons@jackal:~/Documents/VASSAL/code$ perl ./test.pl
Regex Glut matches: EyeGlutForebostinator
Regex Glut.* matches: EyeGlutForebostinator
Regex .Glut. matches: EyeGlutForebostinator

This difference depends on whether the regular expression is only
looking for a matching subsequence, or if it needs to match the entire
sequence.

If you use Java, you can control this by choosing find/lookingAt
versus matches as the Matcher method.

I’m not sure how you would accomplish full string matching in Perl.
Perhaps only through the use of ^ and $ elements?

---

Messages mailing list
Messages@forums.vassalengine.org
forums.vassalengine.org/mailman/ … engine.org

Post generated using Mail2Forum (mail2forum.com)

[jestew]

@tar

I’m not sure how you would accomplish full string matching in Perl.
Perhaps only through the use of ^ and $ elements?

Yup! That’s what they are there for.

For example if you want to match an empty line:

$string =~ /^$/

Cheers, jas…