ignore non-key dialect in MODEL block

Author: treysp

sqlmesh/core/dialect.py

@@ -803,8 +803,15 @@ def text_diff(
     return "\n".join(unified_diff(a_sql, b_sql))
 
 
+WS_OR_COMMENT = r"(?:\s|--[^\n]*\n|/\*.*?\*/)"
+HEADER = r"\b(?:model|audit)\b(?=\s*\()"
+KEY_BOUNDARY = r"(?:\(|,)"  # key is preceded by either '(' or ','
+DIALECT_VALUE = r"(?:'(?P<d_quoted>[a-z_][a-z0-9_]*)?'|(?P<d_unquoted>[a-z_][a-z0-9_]*))"  # value is single-quoted (maybe empty) or unquoted
+VALUE_BOUNDARY = r"(?=,|\))"  # value is followed by comma or closing paren
+
 DIALECT_PATTERN = re.compile(
-    r"(model|audit).*?\(.*?dialect\s+'?([a-z]*)", re.IGNORECASE | re.DOTALL
+    rf"{HEADER}.*?{KEY_BOUNDARY}{WS_OR_COMMENT}*dialect{WS_OR_COMMENT}+{DIALECT_VALUE}{WS_OR_COMMENT}*{VALUE_BOUNDARY}",
+    re.IGNORECASE | re.DOTALL,
 )
 
 
@@ -895,7 +902,8 @@ def parse(
         A list of the parsed expressions: [Model, *Statements, Query, *Statements]
     """
     match = match_dialect and DIALECT_PATTERN.search(sql[:MAX_MODEL_DEFINITION_SIZE])
-    dialect = Dialect.get_or_raise(match.group(2) if match else default_dialect)
+    dialect_str = (match.group("d_quoted") or match.group("d_unquoted")) if match else None
+    dialect = Dialect.get_or_raise(dialect_str or default_dialect)
 
     tokens = dialect.tokenize(sql)
     chunks: t.List[t.Tuple[t.List[Token], ChunkType]] = [([], ChunkType.SQL)]

tests/core/test_model.py

@@ -2727,6 +2727,131 @@ def test_parse(assert_exp_eq):
     )
 
 
+def test_dialect_pattern():
+    def make_test_sql(text: str) -> str:
+        return f"""
+        MODEL (
+            name test_model,
+            kind INCREMENTAL_BY_TIME_RANGE(
+                time_column ds
+            ),
+            {text}
+        );
+    
+        SELECT 1;  
+        """
+
+    def assert_match(test_sql: str, expected_value: t.Optional[str] = "duckdb"):
+        match = d.DIALECT_PATTERN.search(test_sql)
+        assert match
+        dialect_str = match.group("d_quoted") or match.group("d_unquoted")
+        assert dialect_str == expected_value
+
+    # single-quoted dialect
+    assert_match(
+        make_test_sql(
+            """
+            dialect 'duckdb',
+            description 'there's a dialect foo in here too!'
+            """
+        )
+    )
+
+    # bare dialect
+    assert_match(
+        make_test_sql(
+            """
+            dialect duckdb,
+            description 'there's a dialect foo in here too!'
+            """
+        )
+    )
+
+    # no dialect specified, "dialect" in description
+    test_sql = make_test_sql(
+        """
+        description 'there's a dialect foo in here too!'
+        """
+    )
+
+    matches = list(d.DIALECT_PATTERN.finditer(test_sql))
+    assert not matches
+
+    # line comment between properties
+    assert_match(
+        make_test_sql(
+            """
+            tag my_tag, -- comment
+            dialect duckdb
+            """
+        )
+    )
+
+    # block comment between properties
+    assert_match(
+        make_test_sql(
+            """
+            tag my_tag, /* comment */
+            dialect duckdb
+            """
+        )
+    )
+
+    # quoted empty dialect
+    assert_match(
+        make_test_sql(
+            """
+            dialect '',
+            tag my_tag
+            """
+        ),
+        None,
+    )
+
+    # trailing comment after dialect value
+    assert_match(
+        make_test_sql(
+            """
+            dialect duckdb -- trailing comment
+            """
+        )
+    )
+
+    # dialect value isn't terminated by ',' or ')'
+    test_sql = make_test_sql(
+        """
+        dialect duckdb -- trailing comment
+        tag my_tag
+        """
+    )
+
+    matches = list(d.DIALECT_PATTERN.finditer(test_sql))
+    assert not matches
+
+    # dialect first
+    assert_match(
+        """
+        MODEL(
+          dialect duckdb,
+          name my_name
+        );
+        """
+    )
+
+    # full parse
+    sql = """
+    MODEL (
+        name test_model,
+        description 'this text mentions dialect foo but is not a property'
+    );
+
+    SELECT 1;
+    """
+    expressions = d.parse(sql, default_dialect="duckdb")
+    model = load_sql_based_model(expressions)
+    assert model.dialect == ""
+
+
 CONST = "bar"