@Test
public void whenInsertDuplicatesInMultiSet_thenInserted() {
final Multiset<String> names = HashMultiset.create();
names.add("John");
names.add("Adam", 3);
names.add("John");
assertEquals(2, names.count("John"));
names.remove("John");
assertEquals(1, names.count("John"));
assertEquals(3, names.count("Adam"));
names.remove("Adam", 2);
assertEquals(1, names.count("Adam"));
}
public void testCreateWithSize() {
Multiset<String> multiset = HashMultiset.create(50);
multiset.add("foo", 2);
multiset.add("bar");
assertEquals(3, multiset.size());
assertEquals(2, multiset.count("foo"));
}
public void testCreateWithComparator() {
Multiset<String> multiset = TreeMultiset.create(Collections.reverseOrder());
multiset.add("foo", 2);
multiset.add("bar");
assertEquals(3, multiset.size());
assertEquals(2, multiset.count("foo"));
assertEquals("[foo x 2, bar]", multiset.toString());
}
@Override
public int count(Object element) {
int count = unfiltered.count(element);
if (count > 0) {
@SuppressWarnings("unchecked") // element is equal to an E
E e = (E) element;
return predicate.apply(e) ? count : 0;
}
return 0;
}
/** An implementation of {@link Multiset#setCount(Object, int, int)}. */
static <E> boolean setCountImpl(Multiset<E> self, E element, int oldCount, int newCount) {
checkNonnegative(oldCount, "oldCount");
checkNonnegative(newCount, "newCount");
if (self.count(element) == oldCount) {
self.setCount(element, newCount);
return true;
} else {
return false;
}
}
/**
* Returns {@code true} if {@code subMultiset.count(o) <= superMultiset.count(o)} for all {@code
* o}.
*
* @since 10.0
*/
public static boolean containsOccurrences(Multiset<?> superMultiset, Multiset<?> subMultiset) {
checkNotNull(superMultiset);
checkNotNull(subMultiset);
for (Entry<?> entry : subMultiset.entrySet()) {
int superCount = superMultiset.count(entry.getElement());
if (superCount < entry.getCount()) {
return false;
}
}
return true;
}
/** An implementation of {@link Multiset#setCount(Object, int)}. */
static <E> int setCountImpl(Multiset<E> self, E element, int count) {
checkNonnegative(count, "count");
int oldCount = self.count(element);
int delta = count - oldCount;
if (delta > 0) {
self.add(element, delta);
} else if (delta < 0) {
self.remove(element, -delta);
}
return oldCount;
}
/**
* For each occurrence of an element {@code e} in {@code occurrencesToRemove}, removes one
* occurrence of {@code e} in {@code multisetToModify}.
*
* <p>Equivalently, this method modifies {@code multisetToModify} so that {@code
* multisetToModify.count(e)} is set to {@code Math.max(0, multisetToModify.count(e) -
* occurrencesToRemove.count(e))}.
*
* <p>This is <i>not</i> the same as {@code multisetToModify.} {@link Multiset#removeAll
* removeAll}{@code (occurrencesToRemove)}, which removes all occurrences of elements that appear
* in {@code occurrencesToRemove}. However, this operation <i>is</i> equivalent to, albeit
* sometimes more efficient than, the following:
*
* <pre>{@code
* for (E e : occurrencesToRemove) {
* multisetToModify.remove(e);
* }
* }</pre>
*
* @return {@code true} if {@code multisetToModify} was changed as a result of this operation
* @since 10.0 (missing in 18.0 when only the overload taking an {@code Iterable} was present)
*/
public static boolean removeOccurrences(
Multiset<?> multisetToModify, Multiset<?> occurrencesToRemove) {
checkNotNull(multisetToModify);
checkNotNull(occurrencesToRemove);
boolean changed = false;
Iterator<? extends Entry<?>> entryIterator = multisetToModify.entrySet().iterator();
while (entryIterator.hasNext()) {
Entry<?> entry = entryIterator.next();
int removeCount = occurrencesToRemove.count(entry.getElement());
if (removeCount >= entry.getCount()) {
entryIterator.remove();
changed = true;
} else if (removeCount > 0) {
multisetToModify.remove(entry.getElement(), removeCount);
changed = true;
}
}
return changed;
}
/** Delegate implementation which cares about the element type. */
private static <E> boolean retainOccurrencesImpl(
Multiset<E> multisetToModify, Multiset<?> occurrencesToRetain) {
checkNotNull(multisetToModify);
checkNotNull(occurrencesToRetain);
// Avoiding ConcurrentModificationExceptions is tricky.
Iterator<Entry<E>> entryIterator = multisetToModify.entrySet().iterator();
boolean changed = false;
while (entryIterator.hasNext()) {
Entry<E> entry = entryIterator.next();
int retainCount = occurrencesToRetain.count(entry.getElement());
if (retainCount == 0) {
entryIterator.remove();
changed = true;
} else if (retainCount < entry.getCount()) {
multisetToModify.setCount(entry.getElement(), retainCount);
changed = true;
}
}
return changed;
}
/** An implementation of {@link Multiset#equals}. */
static boolean equalsImpl(Multiset<?> multiset, Object object) {
if (object == multiset) {
return true;
}
if (object instanceof Multiset) {
Multiset<?> that = (Multiset<?>) object;
/*
* We can't simply check whether the entry sets are equal, since that
* approach fails when a TreeMultiset has a comparator that returns 0
* when passed unequal elements.
*/
if (multiset.size() != that.size() || multiset.entrySet().size() != that.entrySet().size()) {
return false;
}
for (Entry<?> entry : that.entrySet()) {
if (multiset.count(entry.getElement()) != entry.getCount()) {
return false;
}
}
return true;
}
return false;
}
public void testCreateFromIterable() {
Multiset<String> multiset = HashMultiset.create(Arrays.asList("foo", "bar", "foo"));
assertEquals(3, multiset.size());
assertEquals(2, multiset.count("foo"));
}
public void testCreateFromIterable() {
Multiset<String> multiset = TreeMultiset.create(Arrays.asList("foo", "bar", "foo"));
assertEquals(3, multiset.size());
assertEquals(2, multiset.count("foo"));
assertEquals("[bar, foo x 2]", multiset.toString());
}
/**
* The core scoring method for statistics queries
*
* @param statisticsQuery query to be peformed on statisticsStorage
* @param statisticsStorage core data for Statistics qeries
* @param scoringExps an out parameter.
* <p>- If null, experiment counts result of statisticsQuery should be returned. if - If
* non-null, it serves as a flag that an optimised statisticsQuery should be performed to just
* collect Experiments for which non-zero counts exist for Statistics query. A typical call
* scenario in this case is just one efv per statisticsQuery, in which we can both: 1. check
* if the efv Attribute itself is a scoring one 2. map this Attribute and Experimeants in
* scoringExps to efo terms - via the reverse mapping efv-experiment-> efo term in EfoIndex
* (c.f. atlasStatisticsQueryService.getScoringAttributesForGenes())
* @return Multiset of aggregated experiment counts, where the set of scores genes is intersected
* across statisticsQuery.getConditions(), and union-ed across attributes within each
* condition in statisticsQuery.getConditions().
*/
public static Multiset<Integer> scoreQuery(
StatisticsQueryCondition statisticsQuery,
final StatisticsStorage statisticsStorage,
Set<ExperimentInfo> scoringExps) {
// gatherScoringExpsOnly -> experiment counts should be calculated for statisticsQuery
// !gatherScoringExpsOnly -> scoring experiments should be collected (into scoringExps) only
boolean gatherScoringExpsOnly = scoringExps != null;
Set<StatisticsQueryOrConditions<StatisticsQueryCondition>> andStatisticsQueryConditions =
statisticsQuery.getConditions();
Multiset<Integer> results = null;
if (andStatisticsQueryConditions.isEmpty()) { // End of recursion
Set<Integer> bioEntityIdRestrictionSet = statisticsQuery.getBioEntityIdRestrictionSet();
Set<EfAttribute> attributes = statisticsQuery.getAttributes();
if (attributes.isEmpty()) {
// No attributes were provided - we have to use pre-computed scores across all attributes
Multiset<Integer> scoresAcrossAllEfos =
statisticsStorage.getScoresAcrossAllEfos(statisticsQuery.getStatisticsType());
results = intersect(scoresAcrossAllEfos, bioEntityIdRestrictionSet);
} else {
results = HashMultiset.create();
setQueryExperiments(statisticsQuery, statisticsStorage);
// For each experiment in the query, traverse through all attributes and add all gene
// indexes into one ConciseSet. This way a gene can score
// only once for a single experiment - across all OR attributes in this query. Once all
// attributes have been traversed for a single experiment,
// add ConciseSet to Multiset results
for (ExperimentInfo exp : statisticsQuery.getExperiments()) {
FastSet statsForExperiment = new FastSet();
for (EfAttribute attr : attributes) {
Map<ExperimentInfo, ConciseSet> expsToStats =
getStatisticsForAttribute(
statisticsQuery.getStatisticsType(), attr, statisticsStorage);
if (expsToStats != null) {
if (expsToStats.isEmpty()) {
log.debug(
"Failed to retrieve stats for stat: "
+ statisticsQuery.getStatisticsType()
+ " and attr: "
+ attr);
} else {
if (expsToStats.get(exp) != null) {
if (!gatherScoringExpsOnly) {
statsForExperiment.addAll(
intersect(expsToStats.get(exp), bioEntityIdRestrictionSet));
} else if (containsAtLeastOne(expsToStats.get(exp), bioEntityIdRestrictionSet)) {
// exp contains at least one non-zero score for at least one gene index in
// bioEntityIdRestrictionSet -> add it to scoringExps
scoringExps.add(exp);
}
} else {
log.debug(
"Failed to retrieve stats for stat: "
+ statisticsQuery.getStatisticsType()
+ " exp: "
+ exp.getAccession()
+ " and attr: "
+ attr);
}
}
}
}
if (!gatherScoringExpsOnly) {
results.addAll(statsForExperiment);
}
}
}
} else {
// run over all AND conditions, do "OR" inside (cf. scoreOrStatisticsQueryConditions()) ,
// "AND"'ing over the whole thing
for (StatisticsQueryOrConditions<StatisticsQueryCondition> orConditions :
andStatisticsQueryConditions) {
// Pass gene restriction set down to orConditions
orConditions.setGeneRestrictionSet(statisticsQuery.getBioEntityIdRestrictionSet());
// process OR conditions
Multiset<Integer> condGenes =
getScoresForOrConditions(orConditions, statisticsStorage, scoringExps);
if (results == null) results = condGenes;
else {
Iterator<Multiset.Entry<Integer>> resultGenes = results.entrySet().iterator();
while (resultGenes.hasNext()) {
Multiset.Entry<Integer> entry = resultGenes.next();
if (!condGenes.contains(
entry.getElement())) // AND operation between different top query conditions
resultGenes.remove();
else
// for all gene ids belonging to intersection of all conditions seen so far, we
// accumulate experiment counts
results.setCount(
entry.getElement(), entry.getCount() + condGenes.count(entry.getElement()));
}
}
}
}
if (results == null) {
results = HashMultiset.create();
}
return results;
}
