datastore: fix MongoDB facet parser, #TASK-7151, #TASK-7134

opencb · Jan 17, 2025 · 1184be3 · 1184be3
1 parent 421d5ce
commit 1184be3
Show file tree

Hide file tree

Showing 2 changed files with 64 additions and 8 deletions.
diff --git a/...tastore-mongodb/src/main/java/org/opencb/commons/datastore/mongodb/MongoDBQueryUtils.java b/...tastore-mongodb/src/main/java/org/opencb/commons/datastore/mongodb/MongoDBQueryUtils.java
@@ -775,14 +775,18 @@ private static List<Bson> createFacet(Bson query, List<String> facetFields) {
                 facet = getMongoDBFacet(groupField, accumulator, accumulatorField, boundaries);
 
                 // Unwind in any case
-                String[] split = groupField.split("\\.");
-                String acc = "";
-                for (String s : split) {
-                    if (!StringUtils.isEmpty(acc)) {
-                        acc += ".";
-                    }
-                    acc += s;
-                    unwindList.add(Aggregates.unwind("$" + acc));
+                Set<String> unwindFields = new HashSet<>();
+                if (StringUtils.isNotEmpty(groupField)) {
+                    unwindFields.addAll(getUnwindFields(groupField));
+                }
+                if (StringUtils.isNotEmpty(accumulatorField)) {
+                    unwindFields.addAll(getUnwindFields(accumulatorField));
+                }
+                // We must order the "unwind" fields
+                List<String> unwindFieldList = new ArrayList<>(unwindFields);
+                unwindFieldList.sort(Comparator.comparingInt(s -> s.length() - s.replace(".", "").length()));
+                for (String unwindField : unwindFieldList) {
+                    unwindList.add(Aggregates.unwind("$" + unwindField));
                 }
             }
 
@@ -807,6 +811,20 @@ private static List<Bson> createFacet(Bson query, List<String> facetFields) {
         return result;
     }
 
+    private static Collection<String> getUnwindFields(String field) {
+        List<String> unwindFields = new ArrayList<>();
+        String[] split = field.split("\\.");
+        String acc = "";
+        for (String s : split) {
+            if (!StringUtils.isEmpty(acc)) {
+                acc += ".";
+            }
+            acc += s;
+            unwindFields.add(acc);
+        }
+        return unwindFields;
+    }
+
     private static Facet getMongoDBFacet(String groupField, Accumulator accumulator, String accumulatorField, List<Double> boundaries) {
         String groupFieldId = groupField;
         String accumulatorId = "$" + groupField;

diff --git a/...ore-mongodb/src/test/java/org/opencb/commons/datastore/mongodb/MongoDBCollectionTest.java b/...ore-mongodb/src/test/java/org/opencb/commons/datastore/mongodb/MongoDBCollectionTest.java
@@ -112,12 +112,14 @@ public String toString() {
 
         public static class Dog {
             public int age;
+            public List<Integer> years;
             public String color;
 
             @Override
             public String toString() {
                 final StringBuilder sb = new StringBuilder("Dog{");
                 sb.append("age=").append(age);
+                sb.append("years=").append(years);
                 sb.append("color=").append(color);
                 sb.append('}');
                 return sb.toString();
@@ -161,6 +163,14 @@ private static MongoDBCollection createTestCollection(String test, int size) {
             for (int j = 0 ; j < numDogs; j++) {
                 Document dog = new Document();
                 dog.put("age", random.nextInt(20));
+                int numYears = random.nextInt(3);
+                List<Integer> years = new ArrayList<>();
+                for (int k = 0 ; k < numYears; k++) {
+                    years.add(random.nextInt(100) + 1900);
+                }
+                if (years.size() > 1) {
+                    dog.put("years", years);
+                }
                 dog.put("color", COLORS.get(random.nextInt(COLORS.size())));
                 dogs.add(dog);
             }
@@ -701,6 +711,34 @@ public void testFacetAvgBucketsArray() {
         Assert.assertEquals(aggregate.getResults().get(0).get(0).getAggregationValues().get(0), 1.0d * acc / counter, 0.0001);
     }
 
+//    @Test
+//    public void testFacetAccumulatorMaxBucketsArray() {
+//        Document match = new Document("age", new BasicDBObject("$gt", 2));
+//        DataResult<Document> matchedResults = mongoDBCollection.find(match, null);
+//
+//        String fieldName = "dogs.color:max(dogs.years)";
+//        List<Bson> facets = MongoDBQueryUtils.createFacet(match, fieldName);
+//        System.out.println("facets = " + facets);
+//        MongoDBDocumentToFacetFieldsConverter converter = new MongoDBDocumentToFacetFieldsConverter();
+//        DataResult<List<FacetField>> aggregate = mongoDBCollection.aggregate(facets, converter, null);
+//        for (List<FacetField> facetFieldList : aggregate.getResults()) {
+//            System.out.println("facetFieldList = " + facetFieldList);
+//        }
+//
+//        int counter = 0;
+//        int acc = 0;
+//        for (Document doc : matchedResults.getResults()) {
+//            List<Document> dogs = (List<Document>) doc.get("dogs");
+//            for (Document dog : dogs) {
+//                counter++;
+//                acc += (int) dog.get("age");
+//            }
+//        }
+//        System.out.println("counter = " + counter);
+//        System.out.println("(acc/counter) = " + (1.0d * acc / counter));
+//        Assert.assertEquals(aggregate.getResults().get(0).get(0).getAggregationValues().get(0), 1.0d * acc / counter, 0.0001);
+//    }
+//
     @Test
     public void testFacetFilterAccumulatorBucketsArray() {
         Document match = new Document("age", new BasicDBObject("$gt", 2));