Improve Unicode support for MSSQL (#6690)

* Implement unicode where cluases for mssql queries

* Add comment about unicode support on sqla 1.3+

docs/installation.rst

@@ -405,6 +405,11 @@ You can also use `PyAthena` library(no java required) like this ::
 
 See `PyAthena <https://github.com/laughingman7743/PyAthena#sqlalchemy>`_.
 
+MSSQL
+-----
+
+Full Unicode support requires SQLAlchemy 1.3 or later.
+
 Snowflake
 ---------
 

superset/connectors/sqla/models.py

@@ -115,7 +115,9 @@ class TableColumn(Model, BaseColumn):
         label = label if label else self.column_name
         label = self.table.get_label(label)
         if not self.expression:
-            col = column(self.column_name).label(label)
+            db_engine_spec = self.table.database.db_engine_spec
+            type_ = db_engine_spec.get_sqla_column_type(self.type)
+            col = column(self.column_name, type_=type_).label(label)
         else:
             col = literal_column(self.expression).label(label)
         return col

superset/db_engine_specs.py

@@ -46,6 +46,7 @@ from sqlalchemy.engine import create_engine
 from sqlalchemy.engine.url import make_url
 from sqlalchemy.sql import quoted_name, text
 from sqlalchemy.sql.expression import TextAsFrom
+from sqlalchemy.types import UnicodeText
 import sqlparse
 from werkzeug.utils import secure_filename
 
@@ -400,6 +401,15 @@ class BaseEngineSpec(object):
         label = cls.mutate_label(label)
         return quoted_name(label, True) if cls.force_column_alias_quotes else label
 
+    @classmethod
+    def get_sqla_column_type(cls, type_):
+        """
+        Return a sqlalchemy native column type that corresponds to the column type
+        defined in the data source (optional). Needs to be overridden if column requires
+        special handling (see MSSQL for example of NCHAR/NVARCHAR handling).
+        """
+        return None
+
     @staticmethod
     def mutate_label(label):
         """
@@ -1362,6 +1372,12 @@ class MssqlEngineSpec(BaseEngineSpec):
             data = [[elem for elem in r] for r in data]
         return data
 
+    @classmethod
+    def get_sqla_column_type(cls, type_):
+        if isinstance(type_, str) and re.match(r'^N(VAR){0-1}CHAR', type_):
+            return UnicodeText()
+        return None
+
 
 class AthenaEngineSpec(BaseEngineSpec):
     engine = 'awsathena'