dbmigrate: sync with latest ver

rhodecode/lib/dbmigrate/migrate/changeset/databases/sqlite.py

@@ -1,209 +1,230 b''
 """
    `SQLite`_ database specific implementations of changeset classes.
 
    .. _`SQLite`: http://www.sqlite.org/
 """
 try:  # Python 3
     from collections.abc import MutableMapping as DictMixin
 except ImportError:  # Python 2
     from UserDict import DictMixin
 from copy import copy
 import re
 
 from sqlalchemy.databases import sqlite as sa_base
 from sqlalchemy.schema import ForeignKeyConstraint
 from sqlalchemy.schema import UniqueConstraint
 
 from rhodecode.lib.dbmigrate.migrate import exceptions
 from rhodecode.lib.dbmigrate.migrate.changeset import ansisql
 import sqlite3
 
 SQLiteSchemaGenerator = sa_base.SQLiteDDLCompiler
 
 
 class SQLiteCommon(object):
 
     def _not_supported(self, op):
         raise exceptions.NotSupportedError("SQLite does not support "
             "%s; see http://www.sqlite.org/lang_altertable.html" % op)
 
 
 class SQLiteHelper(SQLiteCommon):
 
-    def _get_unique_constraints(self, table):
-        """Retrieve information about existing unique constraints of the table
+    def _filter_columns(self, cols, table):
+        """Splits the string of columns and returns those only in the table.
+
+        :param cols: comma-delimited string of table columns
+        :param table: the table to check
+        :return: list of columns in the table
+        """
+        columns = []
+        for c in cols.split(","):
+            if c in table.columns:
+                # There was a bug in reflection of SQLite columns with
+                # reserved identifiers as names (SQLite can return them
+                # wrapped with double quotes), so strip double quotes.
+                columns.extend(c.strip(' "'))
+        return columns
+
+    def _get_constraints(self, table):
+        """Retrieve information about existing constraints of the table
 
         This feature is needed for recreate_table() to work properly.
         """
 
         data = table.metadata.bind.execute(
             """SELECT sql
                FROM sqlite_master
                WHERE
                    type='table' AND
                    name=:table_name""",
             table_name=table.name
         ).fetchone()[0]
 
         UNIQUE_PATTERN = "CONSTRAINT (\w+) UNIQUE \(([^\)]+)\)"
         constraints = []
         for name, cols in re.findall(UNIQUE_PATTERN, data):
             # Filter out any columns that were dropped from the table.
-            columns = []
-            for c in cols.split(","):
-                if c in table.columns:
-                    # There was a bug in reflection of SQLite columns with
-                    # reserved identifiers as names (SQLite can return them
-                    # wrapped with double quotes), so strip double quotes.
-                    columns.extend(c.strip(' "'))
+            columns = self._filter_columns(cols, table)
             if columns:
                 constraints.extend(UniqueConstraint(*columns, name=name))
+
+        FKEY_PATTERN = "CONSTRAINT (\w+) FOREIGN KEY \(([^\)]+)\)"
+        for name, cols in re.findall(FKEY_PATTERN, data):
+            # Filter out any columns that were dropped from the table.
+            columns = self._filter_columns(cols, table)
+            if columns:
+                constraints.extend(ForeignKeyConstraint(*columns, name=name))
+
         return constraints
 
     def recreate_table(self, table, column=None, delta=None,
-                       omit_uniques=None):
+                       omit_constraints=None):
         table_name = self.preparer.format_table(table)
 
         # we remove all indexes so as not to have
         # problems during copy and re-create
         for index in table.indexes:
             index.drop()
 
-        # reflect existing unique constraints
-        for uc in self._get_unique_constraints(table):
-            table.append_constraint(uc)
-        # omit given unique constraints when creating a new table if required
+        # reflect existing constraints
+        for constraint in self._get_constraints(table):
+            table.append_constraint(constraint)
+        # omit given constraints when creating a new table if required
         table.constraints = set([
             cons for cons in table.constraints
-            if omit_uniques is None or cons.name not in omit_uniques
+            if omit_constraints is None or cons.name not in omit_constraints
         ])
-        tup = sqlite3.sqlite_version_info
-        if tup[0] > 3 or (tup[0] == 3 and tup[1] >= 26):
+
+        # Use "PRAGMA legacy_alter_table = ON" with sqlite >= 3.26 when
+        # using "ALTER TABLE RENAME TO migration_tmp" to maintain legacy
+        # behavior. See: https://www.sqlite.org/src/info/ae9638e9c0ad0c36
+        if self.connection.engine.dialect.server_version_info >= (3, 26):
             self.append('PRAGMA legacy_alter_table = ON')
             self.execute()
-
         self.append('ALTER TABLE %s RENAME TO migration_tmp' % table_name)
         self.execute()
-        if tup[0] > 3 or (tup[0] == 3 and tup[1] >= 26):
+        if self.connection.engine.dialect.server_version_info >= (3, 26):
             self.append('PRAGMA legacy_alter_table = OFF')
             self.execute()
+
         insertion_string = self._modify_table(table, column, delta)
 
         table.create(bind=self.connection)
         self.append(insertion_string % {'table_name': table_name})
         self.execute()
         self.append('DROP TABLE migration_tmp')
         self.execute()
 
     def visit_column(self, delta):
         if isinstance(delta, DictMixin):
             column = delta.result_column
             table = self._to_table(delta.table)
         else:
             column = delta
             table = self._to_table(column.table)
-
         self.recreate_table(table,column,delta)
 
 class SQLiteColumnGenerator(SQLiteSchemaGenerator,
                             ansisql.ANSIColumnGenerator,
                             # at the end so we get the normal
                             # visit_column by default
                             SQLiteHelper,
                             SQLiteCommon
                             ):
     """SQLite ColumnGenerator"""
 
     def _modify_table(self, table, column, delta):
         columns = ' ,'.join(map(
                 self.preparer.format_column,
                 [c for c in table.columns if c.name!=column.name]))
         return ('INSERT INTO %%(table_name)s (%(cols)s) '
                 'SELECT %(cols)s from migration_tmp')%{'cols':columns}
 
     def visit_column(self,column):
         if column.foreign_keys:
             SQLiteHelper.visit_column(self,column)
         else:
             super(SQLiteColumnGenerator,self).visit_column(column)
 
 class SQLiteColumnDropper(SQLiteHelper, ansisql.ANSIColumnDropper):
     """SQLite ColumnDropper"""
 
     def _modify_table(self, table, column, delta):
 
         columns = ' ,'.join(map(self.preparer.format_column, table.columns))
         return 'INSERT INTO %(table_name)s SELECT ' + columns + \
             ' from migration_tmp'
 
     def visit_column(self,column):
         # For SQLite, we *have* to remove the column here so the table
         # is re-created properly.
         column.remove_from_table(column.table,unset_table=False)
         super(SQLiteColumnDropper,self).visit_column(column)
 
 
 class SQLiteSchemaChanger(SQLiteHelper, ansisql.ANSISchemaChanger):
     """SQLite SchemaChanger"""
 
     def _modify_table(self, table, column, delta):
         return 'INSERT INTO %(table_name)s SELECT * from migration_tmp'
 
     def visit_index(self, index):
         """Does not support ALTER INDEX"""
         self._not_supported('ALTER INDEX')
 
 
 class SQLiteConstraintGenerator(ansisql.ANSIConstraintGenerator, SQLiteHelper, SQLiteCommon):
 
     def visit_migrate_primary_key_constraint(self, constraint):
         tmpl = "CREATE UNIQUE INDEX %s ON %s ( %s )"
         cols = ', '.join(map(self.preparer.format_column, constraint.columns))
         tname = self.preparer.format_table(constraint.table)
         name = self.get_constraint_name(constraint)
         msg = tmpl % (name, tname, cols)
         self.append(msg)
         self.execute()
 
     def _modify_table(self, table, column, delta):
         return 'INSERT INTO %(table_name)s SELECT * from migration_tmp'
 
     def visit_migrate_foreign_key_constraint(self, *p, **k):
         self.recreate_table(p[0].table)
 
     def visit_migrate_unique_constraint(self, *p, **k):
         self.recreate_table(p[0].table)
 
 
 class SQLiteConstraintDropper(ansisql.ANSIColumnDropper,
                               SQLiteHelper,
                               ansisql.ANSIConstraintCommon):
 
     def _modify_table(self, table, column, delta):
         return 'INSERT INTO %(table_name)s SELECT * from migration_tmp'
 
     def visit_migrate_primary_key_constraint(self, constraint):
         tmpl = "DROP INDEX %s "
         name = self.get_constraint_name(constraint)
         msg = tmpl % (name)
         self.append(msg)
         self.execute()
 
     def visit_migrate_foreign_key_constraint(self, *p, **k):
-        self._not_supported('ALTER TABLE DROP CONSTRAINT')
+        #self._not_supported('ALTER TABLE DROP CONSTRAINT')
+        self.recreate_table(p[0].table, omit_constraints=[p[0].name])
 
     def visit_migrate_check_constraint(self, *p, **k):
         self._not_supported('ALTER TABLE DROP CONSTRAINT')
 
     def visit_migrate_unique_constraint(self, *p, **k):
-        self.recreate_table(p[0].table, omit_uniques=[p[0].name])
+        self.recreate_table(p[0].table, omit_constraints=[p[0].name])
 
 
 # TODO: technically primary key is a NOT NULL + UNIQUE constraint, should add NOT NULL to index
 
 class SQLiteDialect(ansisql.ANSIDialect):
     columngenerator = SQLiteColumnGenerator
     columndropper = SQLiteColumnDropper
     schemachanger = SQLiteSchemaChanger
     constraintgenerator = SQLiteConstraintGenerator
     constraintdropper = SQLiteConstraintDropper

rhodecode/lib/dbmigrate/migrate/changeset/schema.py

@@ -1,715 +1,721 b''
 """
    Schema module providing common schema operations.
 """
 import abc
 try:  # Python 3
     from collections.abc import MutableMapping as DictMixin
 except ImportError:  # Python 2
     from UserDict import DictMixin
 import warnings
 
 import sqlalchemy
 
 from sqlalchemy.schema import ForeignKeyConstraint
 from sqlalchemy.schema import UniqueConstraint
 
 from rhodecode.lib.dbmigrate.migrate.exceptions import *
 from rhodecode.lib.dbmigrate.migrate.changeset import SQLA_07, SQLA_08
 from rhodecode.lib.dbmigrate.migrate.changeset import util
 from rhodecode.lib.dbmigrate.migrate.changeset.databases.visitor import (
     get_engine_visitor, run_single_visitor)
 
 
 __all__ = [
     'create_column',
     'drop_column',
     'alter_column',
     'rename_table',
     'rename_index',
     'ChangesetTable',
     'ChangesetColumn',
     'ChangesetIndex',
     'ChangesetDefaultClause',
     'ColumnDelta',
 ]
 
 
 def create_column(column, table=None, *p, **kw):
     """Create a column, given the table.
 
     API to :meth:`ChangesetColumn.create`.
     """
     if table is not None:
         return table.create_column(column, *p, **kw)
     return column.create(*p, **kw)
 
 
 def drop_column(column, table=None, *p, **kw):
     """Drop a column, given the table.
 
     API to :meth:`ChangesetColumn.drop`.
     """
     if table is not None:
         return table.drop_column(column, *p, **kw)
     return column.drop(*p, **kw)
 
 
 def rename_table(table, name, engine=None, **kw):
     """Rename a table.
 
     If Table instance is given, engine is not used.
 
     API to :meth:`ChangesetTable.rename`.
 
     :param table: Table to be renamed.
     :param name: New name for Table.
     :param engine: Engine instance.
     :type table: string or Table instance
     :type name: string
     :type engine: obj
     """
     table = _to_table(table, engine)
     table.rename(name, **kw)
 
 
 def rename_index(index, name, table=None, engine=None, **kw):
     """Rename an index.
 
     If Index instance is given,
     table and engine are not used.
 
     API to :meth:`ChangesetIndex.rename`.
 
     :param index: Index to be renamed.
     :param name: New name for index.
     :param table: Table to which Index is reffered.
     :param engine: Engine instance.
     :type index: string or Index instance
     :type name: string
     :type table: string or Table instance
     :type engine: obj
     """
     index = _to_index(index, table, engine)
     index.rename(name, **kw)
 
 
 def alter_column(*p, **k):
     """Alter a column.
 
     This is a helper function that creates a :class:`ColumnDelta` and
     runs it.
 
     :argument column:
       The name of the column to be altered or a
       :class:`ChangesetColumn` column representing it.
 
     :param table:
       A :class:`~sqlalchemy.schema.Table` or table name to
       for the table where the column will be changed.
 
     :param engine:
       The :class:`~sqlalchemy.engine.base.Engine` to use for table
       reflection and schema alterations.
 
     :returns: A :class:`ColumnDelta` instance representing the change.
 
 
     """
 
     if 'table' not in k and isinstance(p[0], sqlalchemy.Column):
         k['table'] = p[0].table
     if 'engine' not in k:
         k['engine'] = k['table'].bind
 
     # deprecation
     if len(p) >= 2 and isinstance(p[1], sqlalchemy.Column):
         warnings.warn(
             "Passing a Column object to alter_column is deprecated."
             " Just pass in keyword parameters instead.",
             MigrateDeprecationWarning
             )
     engine = k['engine']
 
     # enough tests seem to break when metadata is always altered
     # that this crutch has to be left in until they can be sorted
     # out
     k['alter_metadata']=True
 
     delta = ColumnDelta(*p, **k)
 
     visitorcallable = get_engine_visitor(engine, 'schemachanger')
     _run_visitor(engine, visitorcallable, delta)
 
     return delta
 
 
 def _to_table(table, engine=None):
     """Return if instance of Table, else construct new with metadata"""
     if isinstance(table, sqlalchemy.Table):
         return table
 
     # Given: table name, maybe an engine
     meta = sqlalchemy.MetaData()
     if engine is not None:
         meta.bind = engine
     return sqlalchemy.Table(table, meta)
 
 
 def _to_index(index, table=None, engine=None):
     """Return if instance of Index, else construct new with metadata"""
     if isinstance(index, sqlalchemy.Index):
         return index
 
     # Given: index name; table name required
     table = _to_table(table, engine)
     ret = sqlalchemy.Index(index)
     ret.table = table
     return ret
 
 
 def _run_visitor(
     connectable, visitorcallable, element, connection=None, **kwargs
 ):
     if connection is not None:
         visitorcallable(
             connection.dialect, connection, **kwargs).traverse_single(element)
     else:
         conn = connectable.connect()
         try:
             visitorcallable(
                 conn.dialect, conn, **kwargs).traverse_single(element)
         finally:
             conn.close()
 
 
 # Python3: if we just use:
 #
 #     class ColumnDelta(DictMixin, sqlalchemy.schema.SchemaItem):
 #         ...
 #
 # We get the following error:
 # TypeError: metaclass conflict: the metaclass of a derived class must be a
 # (non-strict) subclass of the metaclasses of all its bases.
 #
 # The complete inheritance/metaclass relationship list of ColumnDelta can be
 # summarized by this following dot file:
 #
 # digraph test123 {
 #     ColumnDelta -> MutableMapping;
 #     MutableMapping -> Mapping;
 #     Mapping -> {Sized Iterable Container};
 #     {Sized Iterable Container} -> ABCMeta[style=dashed];
 #
 #     ColumnDelta -> SchemaItem;
 #     SchemaItem -> {SchemaEventTarget Visitable};
 #     SchemaEventTarget -> object;
 #     Visitable -> {VisitableType object} [style=dashed];
 #     VisitableType -> type;
 # }
 #
 # We need to use a metaclass that inherits from all the metaclasses of
 # DictMixin and sqlalchemy.schema.SchemaItem. Let's call it "MyMeta".
 class MyMeta(sqlalchemy.sql.visitors.VisitableType, abc.ABCMeta, object):
     pass
 
 
 class ColumnDelta(DictMixin, sqlalchemy.schema.SchemaItem, metaclass=MyMeta):
     """Extracts the differences between two columns/column-parameters
 
         May receive parameters arranged in several different ways:
 
         * **current_column, new_column, \*p, \*\*kw**
             Additional parameters can be specified to override column
             differences.
 
         * **current_column, \*p, \*\*kw**
             Additional parameters alter current_column. Table name is extracted
             from current_column object.
             Name is changed to current_column.name from current_name,
             if current_name is specified.
 
         * **current_col_name, \*p, \*\*kw**
             Table kw must specified.
 
         :param table: Table at which current Column should be bound to.\
         If table name is given, reflection will be used.
         :type table: string or Table instance
 
         :param metadata: A :class:`MetaData` instance to store
                          reflected table names
 
         :param engine: When reflecting tables, either engine or metadata must \
         be specified to acquire engine object.
         :type engine: :class:`Engine` instance
         :returns: :class:`ColumnDelta` instance provides interface for altered attributes to \
         `result_column` through :func:`dict` alike object.
 
         * :class:`ColumnDelta`.result_column is altered column with new attributes
 
         * :class:`ColumnDelta`.current_name is current name of column in db
 
 
     """
 
     # Column attributes that can be altered
     diff_keys = ('name', 'type', 'primary_key', 'nullable',
         'server_onupdate', 'server_default', 'autoincrement')
     diffs = dict()
     __visit_name__ = 'column'
 
     def __init__(self, *p, **kw):
         # 'alter_metadata' is not a public api. It exists purely
         # as a crutch until the tests that fail when 'alter_metadata'
         # behaviour always happens can be sorted out
         self.alter_metadata = kw.pop("alter_metadata", False)
 
         self.meta = kw.pop("metadata", None)
         self.engine = kw.pop("engine", None)
 
         # Things are initialized differently depending on how many column
         # parameters are given. Figure out how many and call the appropriate
         # method.
         if len(p) >= 1 and isinstance(p[0], sqlalchemy.Column):
             # At least one column specified
             if len(p) >= 2 and isinstance(p[1], sqlalchemy.Column):
                 # Two columns specified
                 diffs = self.compare_2_columns(*p, **kw)
             else:
                 # Exactly one column specified
                 diffs = self.compare_1_column(*p, **kw)
         else:
             # Zero columns specified
             if not len(p) or not isinstance(p[0], str):
                 raise ValueError("First argument must be column name")
             diffs = self.compare_parameters(*p, **kw)
 
         self.apply_diffs(diffs)
 
     def __repr__(self):
         return '<ColumnDelta altermetadata=%r, %s>' % (
             self.alter_metadata,
             super(ColumnDelta, self).__repr__()
             )
 
     def __getitem__(self, key):
         if key not in list(self.keys()):
             raise KeyError("No such diff key, available: %s" % self.diffs )
         return getattr(self.result_column, key)
 
     def __setitem__(self, key, value):
         if key not in list(self.keys()):
             raise KeyError("No such diff key, available: %s" % self.diffs )
         setattr(self.result_column, key, value)
 
     def __delitem__(self, key):
         raise NotImplementedError
 
     def __len__(self):
         raise NotImplementedError
 
     def __iter__(self):
         raise NotImplementedError
 
     def keys(self):
         return list(self.diffs.keys())
 
     def compare_parameters(self, current_name, *p, **k):
         """Compares Column objects with reflection"""
         self.table = k.pop('table')
         self.result_column = self._table.c.get(current_name)
         if len(p):
             k = self._extract_parameters(p, k, self.result_column)
         return k
 
     def compare_1_column(self, col, *p, **k):
         """Compares one Column object"""
         self.table = k.pop('table', None)
         if self.table is None:
             self.table = col.table
         self.result_column = col
         if len(p):
             k = self._extract_parameters(p, k, self.result_column)
         return k
 
     def compare_2_columns(self, old_col, new_col, *p, **k):
         """Compares two Column objects"""
         self.process_column(new_col)
         self.table = k.pop('table', None)
         # we cannot use bool() on table in SA06
         if self.table is None:
             self.table = old_col.table
         if self.table is None:
             new_col.table
         self.result_column = old_col
 
         # set differences
         # leave out some stuff for later comp
         for key in (set(self.diff_keys) - set(('type',))):
             val = getattr(new_col, key, None)
             if getattr(self.result_column, key, None) != val:
                 k.setdefault(key, val)
 
         # inspect types
         if not self.are_column_types_eq(self.result_column.type, new_col.type):
             k.setdefault('type', new_col.type)
 
         if len(p):
             k = self._extract_parameters(p, k, self.result_column)
         return k
 
     def apply_diffs(self, diffs):
         """Populate dict and column object with new values"""
         self.diffs = diffs
         for key in self.diff_keys:
             if key in diffs:
                 setattr(self.result_column, key, diffs[key])
 
         self.process_column(self.result_column)
 
         # create an instance of class type if not yet
-        if 'type' in diffs and callable(self.result_column.type):
-            self.result_column.type = self.result_column.type()
+        if 'type' in diffs:
+            if callable(self.result_column.type):
+                self.result_column.type = self.result_column.type()
+            if self.result_column.autoincrement and \
+                    not issubclass(
+                        self.result_column.type._type_affinity,
+                        sqlalchemy.Integer):
+                self.result_column.autoincrement = False
 
         # add column to the table
         if self.table is not None and self.alter_metadata:
             self.result_column.add_to_table(self.table)
 
     def are_column_types_eq(self, old_type, new_type):
         """Compares two types to be equal"""
         ret = old_type.__class__ == new_type.__class__
 
         # String length is a special case
         if ret and isinstance(new_type, sqlalchemy.types.String):
             ret = (getattr(old_type, 'length', None) == \
                        getattr(new_type, 'length', None))
         return ret
 
     def _extract_parameters(self, p, k, column):
         """Extracts data from p and modifies diffs"""
         p = list(p)
         while len(p):
             if isinstance(p[0], str):
                 k.setdefault('name', p.pop(0))
             elif isinstance(p[0], sqlalchemy.types.TypeEngine):
                 k.setdefault('type', p.pop(0))
             elif callable(p[0]):
                 p[0] = p[0]()
             else:
                 break
 
         if len(p):
             new_col = column.copy_fixed()
             new_col._init_items(*p)
             k = self.compare_2_columns(column, new_col, **k)
         return k
 
     def process_column(self, column):
         """Processes default values for column"""
         # XXX: this is a snippet from SA processing of positional parameters
         toinit = list()
 
         if column.server_default is not None:
             if isinstance(column.server_default, sqlalchemy.FetchedValue):
                 toinit.append(column.server_default)
             else:
                 toinit.append(sqlalchemy.DefaultClause(column.server_default))
         if column.server_onupdate is not None:
             if isinstance(column.server_onupdate, FetchedValue):
                 toinit.append(column.server_default)
             else:
                 toinit.append(sqlalchemy.DefaultClause(column.server_onupdate,
                                             for_update=True))
         if toinit:
             column._init_items(*toinit)
 
     def _get_table(self):
         return getattr(self, '_table', None)
 
     def _set_table(self, table):
         if isinstance(table, str):
             if self.alter_metadata:
                 if not self.meta:
                     raise ValueError("metadata must be specified for table"
                         " reflection when using alter_metadata")
                 meta = self.meta
                 if self.engine:
                     meta.bind = self.engine
             else:
                 if not self.engine and not self.meta:
                     raise ValueError("engine or metadata must be specified"
                         " to reflect tables")
                 if not self.engine:
                     self.engine = self.meta.bind
                 meta = sqlalchemy.MetaData(bind=self.engine)
             self._table = sqlalchemy.Table(table, meta, autoload=True)
         elif isinstance(table, sqlalchemy.Table):
             self._table = table
             if not self.alter_metadata:
                 self._table.meta = sqlalchemy.MetaData(bind=self._table.bind)
     def _get_result_column(self):
         return getattr(self, '_result_column', None)
 
     def _set_result_column(self, column):
         """Set Column to Table based on alter_metadata evaluation."""
         self.process_column(column)
         if not hasattr(self, 'current_name'):
             self.current_name = column.name
         if self.alter_metadata:
             self._result_column = column
         else:
             self._result_column = column.copy_fixed()
 
     table = property(_get_table, _set_table)
     result_column = property(_get_result_column, _set_result_column)
 
 
 class ChangesetTable(object):
     """Changeset extensions to SQLAlchemy tables."""
 
     def create_column(self, column, *p, **kw):
         """Creates a column.
 
         The column parameter may be a column definition or the name of
         a column in this table.
 
         API to :meth:`ChangesetColumn.create`
 
         :param column: Column to be created
         :type column: Column instance or string
         """
         if not isinstance(column, sqlalchemy.Column):
             # It's a column name
             column = getattr(self.c, str(column))
         column.create(table=self, *p, **kw)
 
     def drop_column(self, column, *p, **kw):
         """Drop a column, given its name or definition.
 
         API to :meth:`ChangesetColumn.drop`
 
         :param column: Column to be droped
         :type column: Column instance or string
         """
         if not isinstance(column, sqlalchemy.Column):
             # It's a column name
             try:
                 column = getattr(self.c, str(column))
             except AttributeError:
                 # That column isn't part of the table. We don't need
                 # its entire definition to drop the column, just its
                 # name, so create a dummy column with the same name.
                 column = sqlalchemy.Column(str(column), sqlalchemy.Integer())
         column.drop(table=self, *p, **kw)
 
     def rename(self, name, connection=None, **kwargs):
         """Rename this table.
 
         :param name: New name of the table.
         :type name: string
         :param connection: reuse connection istead of creating new one.
         :type connection: :class:`sqlalchemy.engine.base.Connection` instance
         """
         engine = self.bind
         self.new_name = name
         visitorcallable = get_engine_visitor(engine, 'schemachanger')
         run_single_visitor(engine, visitorcallable, self, connection, **kwargs)
 
         # Fix metadata registration
         self.name = name
         self.deregister()
         self._set_parent(self.metadata)
 
     def _meta_key(self):
         """Get the meta key for this table."""
         return sqlalchemy.schema._get_table_key(self.name, self.schema)
 
     def deregister(self):
         """Remove this table from its metadata"""
         if SQLA_07:
             self.metadata._remove_table(self.name, self.schema)
         else:
             key = self._meta_key()
             meta = self.metadata
             if key in meta.tables:
                 del meta.tables[key]
 
 
 class ChangesetColumn(object):
     """Changeset extensions to SQLAlchemy columns."""
 
     def alter(self, *p, **k):
         """Makes a call to :func:`alter_column` for the column this
         method is called on.
         """
         if 'table' not in k:
             k['table'] = self.table
         if 'engine' not in k:
             k['engine'] = k['table'].bind
         return alter_column(self, *p, **k)
 
     def create(self, table=None, index_name=None, unique_name=None,
                primary_key_name=None, populate_default=True, connection=None, **kwargs):
         """Create this column in the database.
 
         Assumes the given table exists. ``ALTER TABLE ADD COLUMN``,
         for most databases.
 
         :param table: Table instance to create on.
         :param index_name: Creates :class:`ChangesetIndex` on this column.
         :param unique_name: Creates :class:\
 `~migrate.changeset.constraint.UniqueConstraint` on this column.
         :param primary_key_name: Creates :class:\
 `~migrate.changeset.constraint.PrimaryKeyConstraint` on this column.
         :param populate_default: If True, created column will be \
 populated with defaults
         :param connection: reuse connection istead of creating new one.
         :type table: Table instance
         :type index_name: string
         :type unique_name: string
         :type primary_key_name: string
         :type populate_default: bool
         :type connection: :class:`sqlalchemy.engine.base.Connection` instance
 
         :returns: self
         """
         self.populate_default = populate_default
         self.index_name = index_name
         self.unique_name = unique_name
         self.primary_key_name = primary_key_name
         for cons in ('index_name', 'unique_name', 'primary_key_name'):
             self._check_sanity_constraints(cons)
 
         self.add_to_table(table)
         engine = self.table.bind
         visitorcallable = get_engine_visitor(engine, 'columngenerator')
         _run_visitor(engine, visitorcallable, self, connection, **kwargs)
 
         # TODO: reuse existing connection
         if self.populate_default and self.default is not None:
             stmt = table.update().values({self: engine._execute_default(self.default)})
             engine.execute(stmt)
 
         return self
 
     def drop(self, table=None, connection=None, **kwargs):
         """Drop this column from the database, leaving its table intact.
 
         ``ALTER TABLE DROP COLUMN``, for most databases.
 
         :param connection: reuse connection istead of creating new one.
         :type connection: :class:`sqlalchemy.engine.base.Connection` instance
         """
         if table is not None:
             self.table = table
         engine = self.table.bind
         visitorcallable = get_engine_visitor(engine, 'columndropper')
         _run_visitor(engine, visitorcallable, self, connection, **kwargs)
         self.remove_from_table(self.table, unset_table=False)
         self.table = None
         return self
 
     def add_to_table(self, table):
         if table is not None  and self.table is None:
             if SQLA_07:
                 table.append_column(self)
             else:
                 self._set_parent(table)
 
     def _col_name_in_constraint(self,cons,name):
         return False
 
     def remove_from_table(self, table, unset_table=True):
         # TODO: remove primary keys, constraints, etc
         if unset_table:
             self.table = None
 
         to_drop = set()
         for index in table.indexes:
             columns = []
             for col in index.columns:
                 if col.name!=self.name:
                     columns.append(col)
             if columns:
                 index.columns = columns
                 if SQLA_08:
                     index.expressions = columns
             else:
                 to_drop.add(index)
         table.indexes = table.indexes - to_drop
 
         to_drop = set()
         for cons in table.constraints:
             # TODO: deal with other types of constraint
             if isinstance(cons,(ForeignKeyConstraint,
                                 UniqueConstraint)):
                 for col_name in cons.columns:
                     if not isinstance(col_name, str):
                         col_name = col_name.name
                     if self.name==col_name:
                         to_drop.add(cons)
         table.constraints = table.constraints - to_drop
 
         if table.c.contains_column(self):
             if SQLA_07:
                 table._columns.remove(self)
             else:
                 table.c.remove(self)
 
     # TODO: this is fixed in 0.6
     def copy_fixed(self, **kw):
         """Create a copy of this ``Column``, with all attributes."""
         q = util.safe_quote(self)
         return sqlalchemy.Column(self.name, self.type, self.default,
             key=self.key,
             primary_key=self.primary_key,
             nullable=self.nullable,
             quote=q,
             index=self.index,
             unique=self.unique,
             onupdate=self.onupdate,
             autoincrement=self.autoincrement,
             server_default=self.server_default,
             server_onupdate=self.server_onupdate,
             *[c.copy(**kw) for c in self.constraints])
 
     def _check_sanity_constraints(self, name):
         """Check if constraints names are correct"""
         obj = getattr(self, name)
         if (getattr(self, name[:-5]) and not obj):
             raise InvalidConstraintError("Column.create() accepts index_name,"
             " primary_key_name and unique_name to generate constraints")
         if not isinstance(obj, str) and obj is not None:
             raise InvalidConstraintError(
             "%s argument for column must be constraint name" % name)
 
 
 class ChangesetIndex(object):
     """Changeset extensions to SQLAlchemy Indexes."""
 
     __visit_name__ = 'index'
 
     def rename(self, name, connection=None, **kwargs):
         """Change the name of an index.
 
         :param name: New name of the Index.
         :type name: string
         :param connection: reuse connection istead of creating new one.
         :type connection: :class:`sqlalchemy.engine.base.Connection` instance
         """
         engine = self.table.bind
         self.new_name = name
         visitorcallable = get_engine_visitor(engine, 'schemachanger')
         engine._run_visitor(visitorcallable, self, connection, **kwargs)
         self.name = name
 
 
 class ChangesetDefaultClause(object):
     """Implements comparison between :class:`DefaultClause` instances"""
 
     def __eq__(self, other):
         if isinstance(other, self.__class__):
             if self.arg == other.arg:
                 return True
 
     def __ne__(self, other):
         return not self.__eq__(other)