Merge pull request #3330 from ingef/feature/faster-matching-stats

unroll span in MatchingStats.Entry#addEvent
ingef · Mar 6, 2024 · fecb070 · fecb070
2 parents a2f286a + b731823
commit fecb070
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 6 deletions.
diff --git a/backend/src/main/java/com/bakdata/conquery/models/datasets/concepts/MatchingStats.java b/backend/src/main/java/com/bakdata/conquery/models/datasets/concepts/MatchingStats.java
@@ -84,8 +84,20 @@ public static class Entry {
         @JsonIgnore
         private final Set<String> foundEntities = new HashSet<>();
         private long numberOfEntities;
-        private CDateRange span;
+		private int minDate = Integer.MAX_VALUE;
+		private int maxDate = Integer.MIN_VALUE;
 
+		@JsonIgnore
+		public CDateRange getSpan() {
+			if(minDate == Integer.MAX_VALUE && maxDate == Integer.MIN_VALUE) {
+				return null;
+			}
+
+			return CDateRange.of(
+					minDate == Integer.MAX_VALUE ? Integer.MIN_VALUE : minDate,
+					maxDate == Integer.MIN_VALUE ? Integer.MAX_VALUE : maxDate
+			);
+		}
 
         public void addEvent(Table table, Bucket bucket, int event, String entityForEvent) {
             numberOfEvents++;
@@ -103,7 +115,14 @@ public void addEvent(Table table, Bucket bucket, int event, String entityForEven
                 }
 
                 final CDateRange time = bucket.getAsDateRange(event, c);
-                span = time.spanClosed(span);
+
+				if (time.hasUpperBound()){
+					maxDate = Math.max(time.getMaxValue(), maxDate);
+				}
+
+				if (time.hasLowerBound()){
+					minDate = Math.min(time.getMinValue(), minDate);
+				}
             }
         }
     }

diff --git a/.../src/test/java/com/bakdata/conquery/models/datasets/concepts/tree/MatchingStatsTests.java b/.../src/test/java/com/bakdata/conquery/models/datasets/concepts/tree/MatchingStatsTests.java
@@ -2,7 +2,6 @@
 
 import static org.assertj.core.api.Assertions.assertThat;
 
-import com.bakdata.conquery.models.common.daterange.CDateRange;
 import com.bakdata.conquery.models.datasets.Column;
 import com.bakdata.conquery.models.datasets.Table;
 import com.bakdata.conquery.models.datasets.concepts.MatchingStats;
@@ -22,13 +21,13 @@ public void entitiesCountTest() {
 
         assertThat(stats.countEntities()).isEqualTo(0);
 
-        stats.putEntry(workerId1, new MatchingStats.Entry(5, 5, CDateRange.of(10, 20)));
+        stats.putEntry(workerId1, new MatchingStats.Entry(5, 5, 10, 20));
         assertThat(stats.countEntities()).isEqualTo(5);
 
-        stats.putEntry(workerId1, new MatchingStats.Entry(5, 8, CDateRange.of(10, 20)));
+        stats.putEntry(workerId1, new MatchingStats.Entry(5, 8, 10, 20));
         assertThat(stats.countEntities()).isEqualTo(8);
 
-        stats.putEntry(workerId2, new MatchingStats.Entry(5, 2, CDateRange.of(10, 20)));
+        stats.putEntry(workerId2, new MatchingStats.Entry(5, 2, 10, 20));
         assertThat(stats.countEntities()).isEqualTo(10);