Adjusting the transformers for schema support

[LittleDan9]

When using a database with a schema defined in the SQLAlchemy metadata the resulting value of foreign_key.target_fullname would be [schema].[table_name].[column_name] The previous implementation split foreign_key.target_fullname on the . and defined left_table as element [0] ([schema]) and left_column as element [1] ([table_name]). This splitting does not take into account the possibility of a schema.

The table_name values defined in self.metadata.tables are stored as [schema].[table_name]. Therefore, the logic to enforce the existence of the table on the left side was always False when a schema is present because the names was not the same.

Changing the assumption to be that the last element [-1] of the resulting array when splitting foreign_key.target_fullname on a . is the column name, ensures that regardless of the use of a schema that left-column will be the correct value for column_name. Additionally, since the value of the table name stored in self.metadata.tables is foreign_key.target_fullname less .[column_name], by rejoining the elements of the split less the last element [:-1] by a . results in the correct value for left_table regardless of schema usage.

[tedivm]

If you run make chores it should resolve the formatting issues.

Tests currently aren't passing, but it looks like a fairly trivial issue there. If you can get that resolved so the formatting and testing passes I should be able to merge this.

Thank you!

[LittleDan9]

Thanks @tedivm for the direction. I've updated the tests and ran make chores to correct formatting issues.. Let me know if there is anything else.