[
https://issues.apache.org/jira/browse/COLLECTIONS-511?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Brent Worden updated COLLECTIONS-511:
-------------------------------------
Attachment: collections-511-proposal.diff
collections-511-proposal.diff contains a proposed API for partitioning a
collection using one or more predicates.
Thoughts?
> CollectionUtils.bisect(...), this would be a combination of Select and
> SelectRejected
> -------------------------------------------------------------------------------------
>
> Key: COLLECTIONS-511
> URL: https://issues.apache.org/jira/browse/COLLECTIONS-511
> Project: Commons Collections
> Issue Type: New Feature
> Components: Collection
> Reporter: Nathan Blomquist
> Priority: Trivial
> Attachments: collections-511-proposal.diff
>
>
> I recently needed a way to use a Predicate to select things from a list, but
> I also wanted to know which ones failed the predicate test.
> I wanted the following, but with one iteration instead of two:
> {code}
> List originalList = (...)
> Predicate p = (...)
> Collection selected = CollectionUtils.select(originalList, p);
> Collection rejected = CollectionUtils.selectRejected(originalList, p);
> // handle the selected cases (save them or whatnot)
> // now throw an error message or handle the rejected cases
> {code}
> This is what I came up with based on the CollectionUtils.select(...) method:
> {code:java}
> public static <O, R extends Collection<? super O>> void bisect(
> final Iterable<? extends O> inputCollection,
> final Predicate<? super O> predicate,
> final R selectedCollection,
> final R rejectedCollection) {
> if (inputCollection != null && predicate != null) {
> for (final O item : inputCollection) {
> if (predicate.evaluate(item)) {
> selectedCollection.add(item);
> }else{
> rejectedCollection.add(item);
> }
> }
> }
> }
>
> public static void main(String[] args){
> // this will test the bisection code
> List<String> original = Arrays.asList(
> "testString1",
> "testString2",
> "testString3",
> "String1",
> "String2",
> "String3",
> "testString4",
> "String4",
> "testString5"
> );
>
> List<String> selected = new ArrayList<String>();
> List<String> rejected = new ArrayList<String>();
>
> Predicate<String> beginsWithTestPredicate =
> new Predicate<String>() {
> public boolean evaluate(String object) {
> return object.startsWith("test");
> }
> };
>
> bisect(original, beginsWithTestPredicate, selected, rejected);
>
> System.out.println("Size of selected (should be 5):"
> + selected.size());
> System.out.println("Size of rejected (should be 4):"
> + rejected.size());
> }
> {code}
> This will of course throw a NullPointerException if either output collection
> is null. This seems appropriate since we need to return two outputs anyway.
> Not sure if *bisect* is the best name, but this method will split the
> original into two pieces. https://www.google.com/#q=define+bisect
--
This message was sent by Atlassian JIRA
(v6.2#6252)
