Fix(snowflake, bigquery): parse COLLATE as a func instead of a binary…

… operator (#2343)

* Fix(snowflake): parse COLLATE as a func instead of a binary operator

* BigQuery also has the COLLATE function

sqlglot/dialects/bigquery.py

@@ -420,6 +420,7 @@ class Generator(generator.Generator):
         ESCAPE_LINE_BREAK = True
         NVL2_SUPPORTED = False
         UNNEST_WITH_ORDINALITY = False
+        COLLATE_IS_FUNC = True
 
         TRANSFORMS = {
             **generator.Generator.TRANSFORMS,

sqlglot/dialects/snowflake.py

@@ -419,6 +419,7 @@ class Generator(generator.Generator):
         QUERY_HINTS = False
         AGGREGATE_FILTER_SUPPORTED = False
         SUPPORTS_TABLE_COPY = False
+        COLLATE_IS_FUNC = True
 
         TRANSFORMS = {
             **generator.Generator.TRANSFORMS,

sqlglot/expressions.py

@@ -4279,7 +4279,7 @@ class CastToStrType(Func):
     arg_types = {"this": True, "to": True}
 
 
-class Collate(Binary):
+class Collate(Binary, Func):
     pass
 
 

sqlglot/generator.py

@@ -213,6 +213,9 @@ class Generator:
     # Whether or not parentheses are required around the table sample's expression
     TABLESAMPLE_REQUIRES_PARENS = True
 
+    # Whether or not COLLATE is a function instead of a binary operator
+    COLLATE_IS_FUNC = False
+
     TYPE_MAPPING = {
         exp.DataType.Type.NCHAR: "CHAR",
         exp.DataType.Type.NVARCHAR: "VARCHAR",
@@ -2321,6 +2324,8 @@ def currentdate_sql(self, expression: exp.CurrentDate) -> str:
         return f"CURRENT_DATE({zone})" if zone else "CURRENT_DATE"
 
     def collate_sql(self, expression: exp.Collate) -> str:
+        if self.COLLATE_IS_FUNC:
+            return self.function_fallback_sql(expression)
         return self.binary(expression, "COLLATE")
 
     def command_sql(self, expression: exp.Command) -> str:

sqlglot/parser.py

@@ -340,6 +340,7 @@ class Parser(metaclass=_Parser):
     TRIM_TYPES = {"LEADING", "TRAILING", "BOTH"}
 
     FUNC_TOKENS = {
+        TokenType.COLLATE,
         TokenType.COMMAND,
         TokenType.CURRENT_DATE,
         TokenType.CURRENT_DATETIME,

tests/dialects/test_snowflake.py

@@ -75,6 +75,13 @@ def test_snowflake(self):
         self.validate_all("CAST(x AS CHAR VARYING)", write={"snowflake": "CAST(x AS VARCHAR)"})
         self.validate_all("CAST(x AS CHARACTER VARYING)", write={"snowflake": "CAST(x AS VARCHAR)"})
         self.validate_all("CAST(x AS NCHAR VARYING)", write={"snowflake": "CAST(x AS VARCHAR)"})
+        self.validate_all(
+            "SELECT COLLATE('B', 'und:ci')",
+            write={
+                "bigquery": "SELECT COLLATE('B', 'und:ci')",
+                "snowflake": "SELECT COLLATE('B', 'und:ci')",
+            },
+        )
         self.validate_all(
             "SELECT * FROM x START WITH a = b CONNECT BY c = PRIOR d",
             read={