matched by target/source fixes #1246

Author: tobymao

sqlglot/dialects/snowflake.py

@@ -254,6 +254,7 @@ class Tokenizer(tokens.Tokenizer):
     class Generator(generator.Generator):
         PARAMETER_TOKEN = "$"
         INTEGER_DIVISION = False
+        MATCHED_BY_SOURCE = False
 
         TRANSFORMS = {
             **generator.Generator.TRANSFORMS,  # type: ignore

sqlglot/expressions.py

@@ -3764,7 +3764,7 @@ class Merge(Expression):
 
 
 class When(Func):
-    arg_types = {"this": True, "then": True}
+    arg_types = {"matched": True, "source": False, "condition": False, "then": True}
 
 
 def _norm_args(expression):

sqlglot/generator.py

@@ -112,6 +112,9 @@ class Generator:
     # Whether or not to treat the division operator "/" as integer division
     INTEGER_DIVISION = True
 
+    # Whether or not MERGE ... WHEN MATCHED BY SOURCE is allowed
+    MATCHED_BY_SOURCE = True
+
     TYPE_MAPPING = {
         exp.DataType.Type.NCHAR: "CHAR",
         exp.DataType.Type.NVARCHAR: "VARCHAR",
@@ -1956,7 +1959,11 @@ def kwarg_sql(self, expression: exp.Kwarg) -> str:
         return self.binary(expression, "=>")
 
     def when_sql(self, expression: exp.When) -> str:
-        this = self.sql(expression, "this")
+        matched = "MATCHED" if expression.args["matched"] else "NOT MATCHED"
+        source = " BY SOURCE" if self.MATCHED_BY_SOURCE and expression.args.get("source") else ""
+        condition = self.sql(expression, "condition")
+        condition = f" AND {condition}" if condition else ""
+
         then_expression = expression.args.get("then")
         if isinstance(then_expression, exp.Insert):
             then = f"INSERT {self.sql(then_expression, 'this')}"
@@ -1969,7 +1976,7 @@ def when_sql(self, expression: exp.When) -> str:
                 then = f"UPDATE SET {self.expressions(then_expression, flat=True)}"
         else:
             then = self.sql(then_expression)
-        return f"WHEN {this} THEN {then}"
+        return f"WHEN {matched}{source}{condition} THEN {then}"
 
     def merge_sql(self, expression: exp.Merge) -> str:
         this = self.sql(expression, "this")

sqlglot/parser.py

@@ -3775,7 +3775,15 @@ def _parse_merge(self) -> exp.Expression:
 
         whens = []
         while self._match(TokenType.WHEN):
-            this = self._parse_conjunction()
+            matched = not self._match(TokenType.NOT)
+            self._match_text_seq("MATCHED")
+            source = (
+                False
+                if self._match_text_seq("BY", "TARGET")
+                else self._match_text_seq("BY", "SOURCE")
+            )
+            condition = self._parse_conjunction() if self._match(TokenType.AND) else None
+
             self._match(TokenType.THEN)
 
             if self._match(TokenType.INSERT):
@@ -3800,8 +3808,18 @@ def _parse_merge(self) -> exp.Expression:
                     )
             elif self._match(TokenType.DELETE):
                 then = self.expression(exp.Var, this=self._prev.text)
+            else:
+                then = None
 
-            whens.append(self.expression(exp.When, this=this, then=then))
+            whens.append(
+                self.expression(
+                    exp.When,
+                    matched=matched,
+                    source=source,
+                    condition=condition,
+                    then=then,
+                )
+            )
 
         return self.expression(
             exp.Merge,

tests/dialects/test_bigquery.py

@@ -390,3 +390,18 @@ def test_remove_precision_parameterized_types(self):
                 "bigquery": "INSERT INTO test (cola, colb) VALUES (CAST(7 AS STRING), CAST(14 AS STRING))",
             },
         )
+
+    def test_merge(self):
+        self.validate_all(
+            """
+            MERGE dataset.Inventory T
+            USING dataset.NewArrivals S ON FALSE
+            WHEN NOT MATCHED BY TARGET AND product LIKE '%a%'
+            THEN DELETE
+            WHEN NOT MATCHED BY SOURCE AND product LIKE '%b%'
+            THEN DELETE""",
+            write={
+                "bigquery": "MERGE INTO dataset.Inventory AS T USING dataset.NewArrivals AS S ON FALSE WHEN NOT MATCHED AND product LIKE '%a%' THEN DELETE WHEN NOT MATCHED BY SOURCE AND product LIKE '%b%' THEN DELETE",
+                "snowflake": "MERGE INTO dataset.Inventory AS T USING dataset.NewArrivals AS S ON FALSE WHEN NOT MATCHED AND product LIKE '%a%' THEN DELETE WHEN NOT MATCHED AND product LIKE '%b%' THEN DELETE",
+            },
+        )