Merge pull request #77 from ClickHouse/fix_regex_in_column

Fix brackets regex in column.py

dbt/adapters/clickhouse/__version__.py

@@ -1 +1 @@
-version = '1.1.6'
+version = '1.1.7'

dbt/adapters/clickhouse/column.py

@@ -17,28 +17,25 @@ class ClickhouseColumn(Column):
         'INTEGER': 'Int64',
     }
     is_nullable: bool = False
-    _brackets_regex = re.compile(r'(Nullable|LowCardinality)\((.*?\))\)')
+    _brackets_regex = re.compile(r'^(Nullable|LowCardinality)\((.*)\)$')
     _fix_size_regex = re.compile(r'FixedString\((.*?)\)')
     _decimal_regex = re.compile(r'Decimal\((\d+), (\d+)\)')
 
-    def __init__(
-        self,
-        column: str,
-        dtype: str,
-        is_nullable: bool = False,
-    ) -> None:
+    def __init__(self, column: str, dtype: str) -> None:
         char_size = None
         numeric_precision = None
         numeric_scale = None
 
-        match_brackets = self._brackets_regex.search(dtype)
-        if match_brackets:
-            self.is_nullable = True
-            dtype = match_brackets.group(2)
+        inner_dtype = self.match_brackets(dtype)
+        if inner_dtype:
+            dtype = inner_dtype
+            if not self.is_nullable:
+                # Support LowCardinality(Nullable(dtype))
+                inner_dtype = self.match_brackets(dtype)
+                dtype = inner_dtype if inner_dtype else dtype
 
         if dtype.lower().startswith('fixedstring'):
             match_sized = self._fix_size_regex.search(dtype)
-            char_size = 0
             if match_sized:
                 char_size = int(match_sized.group(1))
 
@@ -86,7 +83,7 @@ def is_string(self) -> bool:
             'tinyblob',
             'char',
             'mediumtext',
-        ]
+        ] or self.dtype.lower().startswith('fixedstring')
 
     def is_integer(self) -> bool:
         return self.dtype.lower().startswith('int') or self.dtype.lower().startswith('uint')
@@ -101,7 +98,7 @@ def string_size(self) -> int:
         if not self.is_string():
             raise RuntimeException('Called string_size() on non-string field!')
 
-        if self.dtype.lower() != 'fixedstring' or self.char_size is None:
+        if not self.dtype.lower().startswith('fixedstring') or self.char_size is None:
             return 256
         else:
             return int(self.char_size)
@@ -122,3 +119,9 @@ def can_expand_to(self, other_column: 'Column') -> bool:
             return False
 
         return other_column.string_size() > self.string_size()
+
+    def match_brackets(self, dtype):
+        match = self._brackets_regex.search(dtype.strip())
+        if match:
+            self.is_nullable = match.group(1) == 'Nullable'
+            return match.group(2)

tests/integration/adapter/test_column.py

@@ -0,0 +1,92 @@
+from dbt.adapters.clickhouse import ClickhouseColumn
+
+
+class TestColumn:
+    def test_base_types(self):
+        verify_column('name', 'UInt8', False, False, False, True)
+        verify_column('name', 'UInt16', False, False, False, True)
+        verify_column('name', 'UInt32', False, False, False, True)
+        verify_column('name', 'UInt64', False, False, False, True)
+        verify_column('name', 'UInt128', False, False, False, True)
+        verify_column('name', 'UInt256', False, False, False, True)
+        verify_column('name', 'Int8', False, False, False, True)
+        verify_column('name', 'Int16', False, False, False, True)
+        verify_column('name', 'Int32', False, False, False, True)
+        verify_column('name', 'Int64', False, False, False, True)
+        verify_column('name', 'Int128', False, False, False, True)
+        verify_column('name', 'Int256', False, False, False, True)
+        str_col = verify_column('name', 'String', True, False, False, False)
+        assert str_col.string_size() == 256
+        fixed_str_col = verify_column('name', 'FixedString', True, False, False, False)
+        assert fixed_str_col.string_size() == 256
+        fixed_str_col = verify_column('name', 'FixedString(16)', True, False, False, False)
+        assert fixed_str_col.string_size() == 16
+        verify_column('name', 'Decimal(6, 6)', False, True, False, False)
+        verify_column('name', 'Float32', False, False, True, False)
+        verify_column('name', 'Float64', False, False, True, False)
+        verify_column('name', 'Float64', False, False, True, False)
+        verify_column('name', 'Date', False, False, False, False)
+        verify_column('name', 'Date32', False, False, False, False)
+        verify_column('name', "DateTime('Asia/Istanbul')", False, False, False, False)
+        verify_column('name', "UUID", False, False, False, False)
+
+    def test_array_type(self):
+        # Test Array of Strings type
+        col = ClickhouseColumn(column='name', dtype='Array(String)')
+        verify_column_types(col, False, False, False, False)
+        assert repr(col) == '<ClickhouseColumn name (Array(String), is nullable: False)>'
+
+        # Test Array of Nullable Strings type
+        col = ClickhouseColumn(column='name', dtype='Array(Nullable(String))')
+        verify_column_types(col, False, False, False, False)
+        assert repr(col) == '<ClickhouseColumn name (Array(Nullable(String)), is nullable: False)>'
+
+        # Test Array of Nullable FixedStrings type
+        col = ClickhouseColumn(column='name', dtype='Array(Nullable(FixedString(16)))')
+        verify_column_types(col, False, False, False, False)
+        assert (
+            repr(col)
+            == '<ClickhouseColumn name (Array(Nullable(FixedString(16))), is nullable: False)>'
+        )
+
+    def test_low_cardinality_nullable_type(self):
+        col = ClickhouseColumn(column='name', dtype='LowCardinality(Nullable(String))')
+        verify_column_types(col, True, False, False, False)
+        assert repr(col) == '<ClickhouseColumn name (Nullable(String), is nullable: True)>'
+        col = ClickhouseColumn(column='name', dtype='LowCardinality(Nullable(FixedString(16)))')
+        verify_column_types(col, True, False, False, False)
+        assert repr(col) == '<ClickhouseColumn name (Nullable(String), is nullable: True)>'
+
+
+def verify_column(
+    name: str, dtype: str, is_string: bool, is_numeric: bool, is_float: bool, is_int: bool
+) -> ClickhouseColumn:
+    data_type = 'String' if is_string else dtype
+    col = ClickhouseColumn(column=name, dtype=dtype)
+    verify_column_types(col, is_string, is_numeric, is_float, is_int)
+    assert repr(col) == f'<ClickhouseColumn {name} ({data_type}, is nullable: False)>'
+
+    # Test Nullable dtype.
+    nullable_col = ClickhouseColumn(column=name, dtype=f'Nullable({dtype})')
+    verify_column_types(nullable_col, is_string, is_numeric, is_float, is_int)
+    assert (
+        repr(nullable_col)
+        == f'<ClickhouseColumn {name} (Nullable({data_type}), is nullable: True)>'
+    )
+
+    # Test low cardinality dtype
+    low_cardinality_col = ClickhouseColumn(column=name, dtype=f'LowCardinality({dtype})')
+    verify_column_types(low_cardinality_col, is_string, is_numeric, is_float, is_int)
+    assert (
+        repr(low_cardinality_col) == f'<ClickhouseColumn {name} ({data_type}, is nullable: False)>'
+    )
+    return col
+
+
+def verify_column_types(
+    col: ClickhouseColumn, is_string: bool, is_numeric: bool, is_float: bool, is_int: bool
+):
+    assert col.is_string() == is_string
+    assert col.is_numeric() == is_numeric
+    assert col.is_float() == is_float
+    assert col.is_integer() == is_int