feat: Add support for CONTAINS(...) (#4399)

Author: VaggelisD

sqlglot/dialects/bigquery.py

@@ -312,6 +312,18 @@ def _builder(args: t.List) -> exp.TimeToStr:
     return _builder
 
 
+def _build_contains_substring(args: t.List) -> exp.Contains | exp.Anonymous:
+    if len(args) == 3:
+        return exp.Anonymous(this="CONTAINS_SUBSTRING", expressions=args)
+
+    # Lowercase the operands in case of transpilation, as exp.Contains
+    # is case-sensitive on other dialects
+    this = exp.Lower(this=seq_get(args, 0))
+    expr = exp.Lower(this=seq_get(args, 1))
+
+    return exp.Contains(this=this, expression=expr)
+
+
 class BigQuery(Dialect):
     WEEK_OFFSET = -1
     UNNEST_COLUMN_ONLY = True
@@ -457,6 +469,7 @@ class Parser(parser.Parser):
 
         FUNCTIONS = {
             **parser.Parser.FUNCTIONS,
+            "CONTAINS_SUBSTRING": _build_contains_substring,
             "DATE": _build_date,
             "DATE_ADD": build_date_delta_with_interval(exp.DateAdd),
             "DATE_SUB": build_date_delta_with_interval(exp.DateSub),
@@ -1153,3 +1166,13 @@ def version_sql(self, expression: exp.Version) -> str:
             if expression.name == "TIMESTAMP":
                 expression.set("this", "SYSTEM_TIME")
             return super().version_sql(expression)
+
+        def contains_sql(self, expression: exp.Contains) -> str:
+            this = expression.this
+            expr = expression.expression
+
+            if isinstance(this, exp.Lower) and isinstance(expr, exp.Lower):
+                this = this.this
+                expr = expr.this
+
+            return self.func("CONTAINS_SUBSTRING", this, expr)

sqlglot/expressions.py

@@ -5460,6 +5460,10 @@ class ConcatWs(Concat):
     _sql_names = ["CONCAT_WS"]
 
 
+class Contains(Func):
+    arg_types = {"this": True, "expression": True}
+
+
 # https://docs.oracle.com/cd/B13789_01/server.101/b10759/operators004.htm#i1035022
 class ConnectByRoot(Func):
     pass

tests/dialects/test_bigquery.py

@@ -1615,6 +1615,31 @@ def test_bigquery(self):
             },
         )
 
+        self.validate_identity(
+            "CONTAINS_SUBSTRING(a, b, json_scope => 'JSON_KEYS_AND_VALUES')"
+        ).assert_is(exp.Anonymous)
+
+        self.validate_all(
+            """CONTAINS_SUBSTRING(a, b)""",
+            read={
+                "": "CONTAINS(a, b)",
+                "spark": "CONTAINS(a, b)",
+                "databricks": "CONTAINS(a, b)",
+                "snowflake": "CONTAINS(a, b)",
+                "duckdb": "CONTAINS(a, b)",
+                "oracle": "CONTAINS(a, b)",
+            },
+            write={
+                "": "CONTAINS(LOWER(a), LOWER(b))",
+                "spark": "CONTAINS(LOWER(a), LOWER(b))",
+                "databricks": "CONTAINS(LOWER(a), LOWER(b))",
+                "snowflake": "CONTAINS(LOWER(a), LOWER(b))",
+                "duckdb": "CONTAINS(LOWER(a), LOWER(b))",
+                "oracle": "CONTAINS(LOWER(a), LOWER(b))",
+                "bigquery": "CONTAINS_SUBSTRING(a, b)",
+            },
+        )
+
     def test_errors(self):
         with self.assertRaises(TokenError):
             transpile("'\\'", read="bigquery")