DirectoryScanner & Excluding Directories

[David Medinets <me...@mtolive.com>]

I hope the following is understandable.

In the scandir method of DirectoryScanner, the code looks like this for
directories:

    if (isIncluded(name)) {
        if (!isExcluded(name)) {
            dirsIncluded.addElement(name);
            if (fast) {
              scandir(file, name+File.separator, fast);
            }
        } else {
            dirsExcluded.addElement(name);
        }
    } else {
        dirsNotIncluded.addElement(name);
        if (fast && couldHoldIncluded(name)) {
            scandir(file, name+File.separator, fast);
        }
    }

However, I'd like to use the following xml in my build.xml file:

      <AnalyzeRequest debug="yes" srcdir="${src}">
         <include name="*.cfm"/>
         <exclude name="images"/>
      </AnalyzeRequest>

I feel the above exclude should work, but it doesn't because if IsExcluded
test only occurs inside the IsIncluded test.

My solution was to wrap the whole above if statement with:

    if (isExcluded(name)) {
        dirsExcluded.addElement(name);
    }
    else { 
        ... the original if statement

        ... move the scandir call here so that
        ... we don't recurse through excluded directories.
    }

Comments? Is this a change that should find its way into the main code
base?

Re: DirectoryScanner & Excluding Directories

[Stefan Bodewig <bo...@bost.de>]

>>>>> "DM" == David Medinets <me...@mtolive.com> writes:

 DM> I feel the above exclude should work, but it doesn't because if
 DM> IsExcluded test only occurs inside the IsIncluded test.

It depends on what you mean by "exclude should work". The
file/directory named images should not be listed in one of the
getIncluded... methods.

DirectoryScanner assigns three states to files or directories:

(1) included - what you'd expect. This file/dir has matched an include
pattern.

(2) excluded - This file/dir matches both an include and an exclude
pattern, exclude wins.

(3) not included - This file/dir doesn't match an include pattern.

So the opposite of getIncludedFiles() is the union of
getExcludedFiles() and getNotIncludedFiles().

I think we could savely change the semantics if they seem
counterintuitive as no tasks I'm aware of call anything but the
getIncluded... methods - or maybe we just need better documentation?

Stefan