Fix: parse & generate Length properly in clickhouse

Author: georgesittas

sqlglot/dialects/clickhouse.py

@@ -11,6 +11,7 @@
     inline_array_sql,
     json_extract_segments,
     json_path_key_only_name,
+    length_or_char_length_sql,
     no_pivot_sql,
     build_json_extract_path,
     rename_func,
@@ -261,6 +262,7 @@ class Parser(parser.Parser):
             "JSONEXTRACTSTRING": build_json_extract_path(
                 exp.JSONExtractScalar, zero_based_indexing=False
             ),
+            "LENGTH": lambda args: exp.Length(this=seq_get(args, 0), binary=True),
             "MAP": parser.build_var_map,
             "MATCH": exp.RegexpLike.from_arg_list,
             "RANDCANONICAL": exp.Rand.from_arg_list,
@@ -979,6 +981,7 @@ class Generator(generator.Generator):
             exp.JSONExtractScalar: json_extract_segments("JSONExtractString", quoted_index=False),
             exp.JSONPathKey: json_path_key_only_name,
             exp.JSONPathRoot: lambda *_: "",
+            exp.Length: length_or_char_length_sql,
             exp.Map: lambda self, e: _lower_func(var_map_sql(self, e)),
             exp.Median: rename_func("median"),
             exp.Nullif: rename_func("nullIf"),

sqlglot/dialects/dialect.py

@@ -1754,3 +1754,8 @@ def no_make_interval_sql(self: Generator, expression: exp.MakeInterval, sep: str
         args.append(f"{value} {unit}")
 
     return f"INTERVAL '{self.format_args(*args, sep=sep)}'"
+
+
+def length_or_char_length_sql(self: Generator, expression: exp.Length) -> str:
+    length_func = "LENGTH" if expression.args.get("binary") else "CHAR_LENGTH"
+    return self.func(length_func, expression.this)

sqlglot/dialects/mysql.py

@@ -11,6 +11,7 @@
     datestrtodate_sql,
     build_formatted_time,
     isnull_to_is_null,
+    length_or_char_length_sql,
     locate_to_strposition,
     max_or_greatest,
     min_or_least,
@@ -730,6 +731,7 @@ class Generator(generator.Generator):
             e: f"""GROUP_CONCAT({self.sql(e, "this")} SEPARATOR {self.sql(e, "separator") or "','"})""",
             exp.ILike: no_ilike_sql,
             exp.JSONExtractScalar: arrow_json_extract_sql,
+            exp.Length: length_or_char_length_sql,
             exp.LogicalOr: rename_func("MAX"),
             exp.LogicalAnd: rename_func("MIN"),
             exp.Max: max_or_greatest,
@@ -1263,7 +1265,3 @@ def attimezone_sql(self, expression: exp.AtTimeZone) -> str:
 
         def isascii_sql(self, expression: exp.IsAscii) -> str:
             return f"REGEXP_LIKE({self.sql(expression.this)}, '^[[:ascii:]]*$')"
-
-        def length_sql(self, expression: exp.Length) -> str:
-            length_func = "LENGTH" if expression.args.get("binary") else "CHAR_LENGTH"
-            return self.func(length_func, expression.this)

sqlglot/expressions.py

@@ -6154,7 +6154,7 @@ class Right(Func):
 
 class Length(Func):
     arg_types = {"this": True, "binary": False, "encoding": False}
-    _sql_names = ["LENGTH", "LEN"]
+    _sql_names = ["LENGTH", "LEN", "CHAR_LENGTH", "CHARACTER_LENGTH"]
 
 
 class Levenshtein(Func):

sqlglot/parser.py

@@ -188,8 +188,6 @@ class Parser(metaclass=_Parser):
             this=seq_get(args, 0), nulls_excluded=dialect.ARRAY_AGG_INCLUDES_NULLS is None or None
         ),
         "CHAR": lambda args: exp.Chr(expressions=args),
-        "CHAR_LENGTH": exp.Length.from_arg_list,
-        "CHARACTER_LENGTH": exp.Length.from_arg_list,
         "CHR": lambda args: exp.Chr(expressions=args),
         "COUNT": lambda args: exp.Count(this=seq_get(args, 0), expressions=args[1:], big_int=True),
         "CONCAT": lambda args, dialect: exp.Concat(