perf: refactor SIP-68 db migrations with INSERT SELECT FROM

superset/columns/models.py

@@ -23,7 +23,6 @@
 
 These models are not fully implemented, and shouldn't be used yet.
 """
-
 import sqlalchemy as sa
 from flask_appbuilder import Model
 
@@ -33,6 +32,8 @@
     ImportExportMixin,
 )
 
+UNKOWN_TYPE = "UNKNOWN"
+
 
 class Column(
     Model,
@@ -52,51 +53,58 @@ class Column(
 
     id = sa.Column(sa.Integer, primary_key=True)
 
+    # Assuming the column is an aggregation, is it additive? Useful for determining which
+    # aggregations can be done on the metric. Eg, ``COUNT(DISTINCT user_id)`` is not
+    # additive, so it shouldn't be used in a ``SUM``.
+    is_additive = sa.Column(sa.Boolean, default=False)
+
+    # Is this column an aggregation (metric)?
+    is_aggregation = sa.Column(sa.Boolean, default=False)
+
+    is_filterable = sa.Column(sa.Boolean, nullable=False, default=True)
+    is_dimensional = sa.Column(sa.Boolean, nullable=False, default=False)
+
+    # Is an increase desired? Useful for displaying the results of A/B tests, or setting
+    # up alerts. Eg, this is true for "revenue", but false for "latency".
+    is_increase_desired = sa.Column(sa.Boolean, default=True)
+
+    # Column is managed externally and should be read-only inside Superset
+    is_managed_externally = sa.Column(sa.Boolean, nullable=False, default=False)
+
+    # Is this column a partition? Useful for scheduling queries and previewing the latest
+    # data.
+    is_partition = sa.Column(sa.Boolean, default=False)
+
+    # Does the expression point directly to a physical column?
+    is_physical = sa.Column(sa.Boolean, default=True)
+
+    # Is this a spatial column? This could be leveraged in the future for spatial
+    # visualizations.
+    is_spatial = sa.Column(sa.Boolean, default=False)
+
+    # Is this a time column? Useful for plotting time series.
+    is_temporal = sa.Column(sa.Boolean, default=False)
+
     # We use ``sa.Text`` for these attributes because (1) in modern databases the
     # performance is the same as ``VARCHAR``[1] and (2) because some table names can be
     # **really** long (eg, Google Sheets URLs).
     #
     # [1] https://www.postgresql.org/docs/9.1/datatype-character.html
     name = sa.Column(sa.Text)
-    type = sa.Column(sa.Text)
+    # Raw type as returned and used by db engine.
+    type = sa.Column(sa.Text, default=UNKOWN_TYPE)
 
     # Columns are defined by expressions. For tables, these are the actual columns names,
     # and should match the ``name`` attribute. For datasets, these can be any valid SQL
     # expression. If the SQL expression is an aggregation the column is a metric,
     # otherwise it's a computed column.
     expression = sa.Column(sa.Text)
-
-    # Does the expression point directly to a physical column?
-    is_physical = sa.Column(sa.Boolean, default=True)
+    unit = sa.Column(sa.Text)
 
     # Additional metadata describing the column.
     description = sa.Column(sa.Text)
     warning_text = sa.Column(sa.Text)
-    unit = sa.Column(sa.Text)
-
-    # Is this a time column? Useful for plotting time series.
-    is_temporal = sa.Column(sa.Boolean, default=False)
-
-    # Is this a spatial column? This could be leveraged in the future for spatial
-    # visualizations.
-    is_spatial = sa.Column(sa.Boolean, default=False)
-
-    # Is this column a partition? Useful for scheduling queries and previewing the latest
-    # data.
-    is_partition = sa.Column(sa.Boolean, default=False)
-
-    # Is this column an aggregation (metric)?
-    is_aggregation = sa.Column(sa.Boolean, default=False)
-
-    # Assuming the column is an aggregation, is it additive? Useful for determining which
-    # aggregations can be done on the metric. Eg, ``COUNT(DISTINCT user_id)`` is not
-    # additive, so it shouldn't be used in a ``SUM``.
-    is_additive = sa.Column(sa.Boolean, default=False)
-
-    # Is an increase desired? Useful for displaying the results of A/B tests, or setting
-    # up alerts. Eg, this is true for "revenue", but false for "latency".
-    is_increase_desired = sa.Column(sa.Boolean, default=True)
-
-    # Column is managed externally and should be read-only inside Superset
-    is_managed_externally = sa.Column(sa.Boolean, nullable=False, default=False)
     external_url = sa.Column(sa.Text, nullable=True)
+
+    def __repr__(self) -> str:
+        return f"<Column id={self.id}>"

superset/connectors/base/models.py

@@ -31,7 +31,7 @@
 from superset.models.slice import Slice
 from superset.superset_typing import FilterValue, FilterValues, QueryObjectDict
 from superset.utils import core as utils
-from superset.utils.core import GenericDataType
+from superset.utils.core import GenericDataType, MediumText
 
 METRIC_FORM_DATA_PARAMS = [
     "metric",
@@ -586,7 +586,7 @@ class BaseColumn(AuditMixinNullable, ImportExportMixin):
     type = Column(Text)
     groupby = Column(Boolean, default=True)
     filterable = Column(Boolean, default=True)
-    description = Column(Text)
+    description = Column(MediumText())
     is_dttm = None
 
     # [optional] Set this to support import/export functionality
@@ -672,7 +672,7 @@ class BaseMetric(AuditMixinNullable, ImportExportMixin):
     metric_name = Column(String(255), nullable=False)
     verbose_name = Column(String(1024))
     metric_type = Column(String(32))
-    description = Column(Text)
+    description = Column(MediumText())
     d3format = Column(String(128))
     warning_text = Column(Text)