Skip NULL aggregate values from shards with no matching rows (#1331)

src/python/txtai/database/sql/aggregate.py

@@ -123,16 +123,14 @@ class Aggregate(SQL):
         # Compute column values
         rows = []
         for result in results:
-            # Row counts for this group, if a count(*) column was selected
-            counts = [r[countcolumn] for r in result] if countcolumn else None
-
             # Calculate/copy column values
             row = {}
             for column in columns:
                 if column in aggcolumns:
-                    # Calculate aggregate value
-                    values = [r[column] for r in result]
-                    row[column] = self.avg(values, counts) if column.lower().startswith("avg(") else aggcolumns[column](values)
+                    # Calculate aggregate value, skip NULLs from shards with no matching rows
+                    matches = [r for r in result if r[column] is not None]
+                    values, counts = [r[column] for r in matches], [r[countcolumn] for r in matches] if countcolumn else None
+                    row[column] = (self.avg(values, counts) if column.lower().startswith("avg(") else aggcolumns[column](values)) if values else None
                 else:
                     # Non aggregate column value repeat, use first value
                     row[column] = result[0][column]

test/python/testdatabase/testsql.py

@@ -62,6 +62,19 @@ class TestSQL(unittest.TestCase):
         with self.assertRaises(SQLError):
             aggregate("select avg(price) from txtai", [{"avg(price)": 100.0}, {"avg(price)": 10.0}])
 
+    def testAggregateNull(self):
+        """
+        Test Aggregate skips NULL values returned by shards with no matching rows
+        """
+
+        aggregate = Aggregate()
+
+        query = "select count(*), max(price), avg(price) from txtai where price > 50"
+        empty, result = {"count(*)": 0, "max(price)": None, "avg(price)": None}, {"count(*)": 2, "max(price)": 100.0, "avg(price)": 80.0}
+
+        self.assertEqual(aggregate(query, [empty, result]), [result])
+        self.assertEqual(aggregate(query, [empty, empty]), [empty])
+
     def testAlias(self):
         """
         Test alias clauses