x/sqlalchemy-migrate
Code Issues Proposed changes

Merge "Add DB2 10.5 Support"

Browse Source
This commit is contained in:
Jenkins 2014-02-23 19:16:52 +00:00 committed by Gerrit Code Review
commit 21fcdad0f4
13 changed files with 397 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
migrate/changeset/databases/__init__.py
View File

@ -7,4 +7,5 @@ __all__ = [
'sqlite',
'mysql',
'oracle',
'ibmdb2',
]

313
migrate/changeset/databases/ibmdb2.py Normal file
View File

@ -0,0 +1,313 @@
"""
DB2 database specific implementations of changeset classes.
"""
import logging
from ibm_db_sa import base
from sqlalchemy.schema import (AddConstraint,
CreateIndex,
DropConstraint)
from sqlalchemy.schema import (Index,
PrimaryKeyConstraint,
UniqueConstraint)
from migrate.changeset import ansisql
from migrate.changeset import constraint
from migrate import exceptions
LOG = logging.getLogger(__name__)
IBMDBSchemaGenerator = base.IBM_DBDDLCompiler
def get_server_version_info(dialect):
"""Returns the DB2 server major and minor version as a list of ints."""
return [int(ver_token) for ver_token in dialect.dbms_ver.split('.')[0:2]]
def is_unique_constraint_with_null_columns_supported(dialect):
"""Checks to see if the DB2 version is at least 10.5.
This is needed for checking if unique constraints with null columns
are supported.
"""
return get_server_version_info(dialect) >= [10, 5]
class IBMDBColumnGenerator(IBMDBSchemaGenerator,
ansisql.ANSIColumnGenerator):
def visit_column(self, column):
nullable = True
if not column.nullable:
nullable = False
column.nullable = True
table = self.start_alter_table(column)
self.append("ADD COLUMN ")
self.append(self.get_column_specification(column))
for cons in column.constraints:
self.traverse_single(cons)
if column.default is not None:
self.traverse_single(column.default)
self.execute()
#ALTER TABLE STATEMENTS
if not nullable:
self.start_alter_table(column)
self.append("ALTER COLUMN %s SET NOT NULL" %
self.preparer.format_column(column))
self.execute()
self.append("CALL SYSPROC.ADMIN_CMD('REORG TABLE %s')" %
self.preparer.format_table(table))
self.execute()
# add indexes and unique constraints
if column.index_name:
Index(column.index_name, column).create()
elif column.unique_name:
constraint.UniqueConstraint(column,
name=column.unique_name).create()
# SA bounds FK constraints to table, add manually
for fk in column.foreign_keys:
self.add_foreignkey(fk.constraint)
# add primary key constraint if needed
if column.primary_key_name:
pk = constraint.PrimaryKeyConstraint(
column, name=column.primary_key_name)
pk.create()
self.append("COMMIT")
self.execute()
self.append("CALL SYSPROC.ADMIN_CMD('REORG TABLE %s')" %
self.preparer.format_table(table))
self.execute()
class IBMDBColumnDropper(ansisql.ANSIColumnDropper):
def visit_column(self, column):
"""Drop a column from its table.
:param column: the column object
:type column: :class:`sqlalchemy.Column`
"""
#table = self.start_alter_table(column)
super(IBMDBColumnDropper, self).visit_column(column)
self.append("CALL SYSPROC.ADMIN_CMD('REORG TABLE %s')" %
self.preparer.format_table(column.table))
self.execute()
class IBMDBSchemaChanger(IBMDBSchemaGenerator, ansisql.ANSISchemaChanger):
def visit_table(self, table):
"""Rename a table; #38. Other ops aren't supported."""
self._rename_table(table)
self.append("TO %s" % self.preparer.quote(table.new_name, table.quote))
self.execute()
self.append("COMMIT")
self.execute()
def _rename_table(self, table):
self.append("RENAME TABLE %s " % self.preparer.format_table(table))
def visit_index(self, index):
old_name = self.preparer.quote(self._index_identifier(index.name),
index.quote)
new_name = self.preparer.quote(self._index_identifier(index.new_name),
index.quote)
self.append("RENAME INDEX %s TO %s" % (old_name, new_name))
self.execute()
self.append("COMMIT")
self.execute()
def _run_subvisit(self, delta, func, start_alter=True):
"""Runs visit method based on what needs to be changed on column"""
table = delta.table
if start_alter:
self.start_alter_table(table)
ret = func(table,
self.preparer.quote(delta.current_name, delta.quote),
delta)
self.execute()
self._reorg_table(self.preparer.format_table(delta.table))
def _reorg_table(self, delta):
self.append("CALL SYSPROC.ADMIN_CMD('REORG TABLE %s')" % delta)
self.execute()
def visit_column(self, delta):
keys = delta.keys()
tr = self.connection.begin()
column = delta.result_column.copy()
if 'type' in keys:
try:
self._run_subvisit(delta, self._visit_column_change, False)
except Exception as e:
LOG.warn("Unable to change the column type. Error: %s" % e)
if column.primary_key and 'primary_key' not in keys:
try:
self._run_subvisit(delta, self._visit_primary_key)
except Exception as e:
LOG.warn("Unable to add primary key. Error: %s" % e)
if 'nullable' in keys:
self._run_subvisit(delta, self._visit_column_nullable)
if 'server_default' in keys:
self._run_subvisit(delta, self._visit_column_default)
if 'primary_key' in keys:
self._run_subvisit(delta, self._visit_primary_key)
self._run_subvisit(delta, self._visit_unique_constraint)
if 'name' in keys:
try:
self._run_subvisit(delta, self._visit_column_name, False)
except Exception as e:
LOG.warn("Unable to change column %(name)s. Error: %(error)s" %
{'name': delta.current_name, 'error': e})
self._reorg_table(self.preparer.format_table(delta.table))
self.append("COMMIT")
self.execute()
tr.commit()
def _visit_unique_constraint(self, table, col_name, delta):
# Add primary key to the current column
self.append("ADD CONSTRAINT %s " % col_name)
self.append("UNIQUE (%s)" % col_name)
def _visit_primary_key(self, table, col_name, delta):
# Add primary key to the current column
self.append("ADD PRIMARY KEY (%s)" % col_name)
def _visit_column_name(self, table, col_name, delta):
column = delta.result_column.copy()
# Delete the primary key before renaming the column
if column.primary_key:
try:
self.start_alter_table(table)
self.append("DROP PRIMARY KEY")
self.execute()
except Exception:
LOG.debug("Continue since Primary key does not exist.")
self.start_alter_table(table)
new_name = self.preparer.format_column(delta.result_column)
self.append("RENAME COLUMN %s TO %s" % (col_name, new_name))
if column.primary_key:
# execute the rename before adding primary key back
self.execute()
self.start_alter_table(table)
self.append("ADD PRIMARY KEY (%s)" % new_name)
def _visit_column_nullable(self, table, col_name, delta):
self.append("ALTER COLUMN %s " % col_name)
nullable = delta['nullable']
if nullable:
self.append("DROP NOT NULL")
else:
self.append("SET NOT NULL")
def _visit_column_default(self, table, col_name, delta):
default_text = self.get_column_default_string(delta.result_column)
self.append("ALTER COLUMN %s " % col_name)
if default_text is None:
self.append("DROP DEFAULT")
else:
self.append("SET WITH DEFAULT %s" % default_text)
def _visit_column_change(self, table, col_name, delta):
column = delta.result_column.copy()
# Delete the primary key before
if column.primary_key:
try:
self.start_alter_table(table)
self.append("DROP PRIMARY KEY")
self.execute()
except Exception:
LOG.debug("Continue since Primary key does not exist.")
# Delete the identity before
try:
self.start_alter_table(table)
self.append("ALTER COLUMN %s DROP IDENTITY" % col_name)
self.execute()
except Exception:
LOG.debug("Continue since identity does not exist.")
column.default = None
if not column.table:
column.table = delta.table
self.start_alter_table(table)
self.append("ALTER COLUMN %s " % col_name)
self.append("SET DATA TYPE ")
type_text = self.dialect.type_compiler.process(
delta.result_column.type)
self.append(type_text)
class IBMDBConstraintGenerator(ansisql.ANSIConstraintGenerator):
def _visit_constraint(self, constraint):
constraint.name = self.get_constraint_name(constraint)
if (isinstance(constraint, UniqueConstraint) and
is_unique_constraint_with_null_columns_supported(
self.dialect)):
for column in constraint.columns._all_cols:
if column.nullable:
constraint.exclude_nulls = True
break
if getattr(constraint, 'exclude_nulls', None):
index = Index(constraint.name,
*(column for column in constraint.columns._all_cols),
unique=True)
sql = self.process(CreateIndex(index))
sql += ' EXCLUDE NULL KEYS'
else:
sql = self.process(AddConstraint(constraint))
self.append(sql)
self.execute()
class IBMDBConstraintDropper(ansisql.ANSIConstraintDropper,
ansisql.ANSIConstraintCommon):
def _visit_constraint(self, constraint):
constraint.name = self.get_constraint_name(constraint)
if (isinstance(constraint, UniqueConstraint) and
is_unique_constraint_with_null_columns_supported(
self.dialect)):
for column in constraint.columns._all_cols:
if column.nullable:
constraint.exclude_nulls = True
break
if getattr(constraint, 'exclude_nulls', None):
index_name = self.preparer.quote(
self._index_identifier(constraint.name), constraint.quote)
sql = 'DROP INDEX %s ' % index_name
else:
sql = self.process(DropConstraint(constraint,
cascade=constraint.cascade))
self.append(sql)
self.execute()
def visit_migrate_primary_key_constraint(self, constraint):
self.start_alter_table(constraint.table)
self.append("DROP PRIMARY KEY")
self.execute()
class IBMDBDialect(ansisql.ANSIDialect):
columngenerator = IBMDBColumnGenerator
columndropper = IBMDBColumnDropper
schemachanger = IBMDBSchemaChanger
constraintgenerator = IBMDBConstraintGenerator
constraintdropper = IBMDBConstraintDropper

4
migrate/changeset/databases/visitor.py
View File

@ -8,7 +8,8 @@ from migrate.changeset.databases import (sqlite,
postgres,
mysql,
oracle,
firebird)
firebird,
ibmdb2)
# Map SA dialects to the corresponding Migrate extensions
@ -20,6 +21,7 @@ DIALECTS = {
"mysql": mysql.MySQLDialect,
"oracle": oracle.OracleDialect,
"firebird": firebird.FBDialect,
"ibm_db_sa": ibmdb2.IBMDBDialect
}

0
migrate/tests/changeset/databases/__init__.py Normal file
View File

32
migrate/tests/changeset/databases/test_ibmdb2.py Normal file
View File

@ -0,0 +1,32 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import mock
import six
from migrate.changeset.databases import ibmdb2
from migrate.tests import fixture
class TestIBMDBDialect(fixture.Base):
"""
Test class for ibmdb2 dialect unit tests which do not require
a live backend database connection.
"""
def test_is_unique_constraint_with_null_cols_supported(self):
test_values = {
'10.1': False,
'10.4.99': False,
'10.5': True,
'10.5.1': True
}
for version, supported in six.iteritems(test_values):
mock_dialect = mock.MagicMock()
mock_dialect.dbms_ver = version
self.assertEqual(
supported,
ibmdb2.is_unique_constraint_with_null_columns_supported(
mock_dialect),
'Assertion failed on version: %s' % version)

24
migrate/tests/changeset/test_changeset.py
View File

@ -231,18 +231,18 @@ class TestAddDropColumn(fixture.DB):
col.drop()
@fixture.usedb(not_supported='mysql')
@fixture.usedb(not_supported=['mysql'])
def test_check(self):
"""Can create columns with check constraint"""
col = Column('data',
col = Column('foo',
Integer,
sqlalchemy.schema.CheckConstraint('data > 4'))
sqlalchemy.schema.CheckConstraint('foo > 4'))
col.create(self.table)
# check if constraint was added (cannot test on objects)
self.table.insert(values={'data': 5}).execute()
self.table.insert(values={'foo': 5}).execute()
try:
self.table.insert(values={'data': 3}).execute()
self.table.insert(values={'foo': 3}).execute()
except (sqlalchemy.exc.IntegrityError,
sqlalchemy.exc.ProgrammingError):
pass
@ -272,10 +272,11 @@ class TestAddDropColumn(fixture.DB):
col.drop(self.table)
# TODO: remove already attached columns with uniques, pks, fks ..
@fixture.usedb(not_supported='postgresql')
@fixture.usedb(not_supported=['ibm_db_sa', 'postgresql'])
def test_drop_column_of_composite_index(self):
# NOTE(rpodolyaka): postgresql automatically drops a composite index
# if one of its columns is dropped
# NOTE(mriedem): DB2 does the same.
self.table_idx.c.b.drop()
reflected = Table(self.table_idx.name, MetaData(), autoload=True,
@ -450,10 +451,15 @@ class TestAddDropColumn(fixture.DB):
self.table.drop()
self.meta.clear()
# NOTE(mriedem): DB2 does not currently support unique constraints
# on nullable columns, so the columns that are used to create the
# foreign keys here need to be non-nullable for testing with DB2
# to work.
# create FK's target
reftable = Table('tmp_ref', self.meta,
Column('id', Integer, primary_key=True),
Column('jd', Integer),
Column('jd', Integer, nullable=False),
UniqueConstraint('id','jd')
)
if self.engine.has_table(reftable.name):
@ -464,8 +470,8 @@ class TestAddDropColumn(fixture.DB):
self.table = Table(
self.table_name, self.meta,
Column('id', Integer, primary_key=True),
Column('r1', Integer),
Column('r2', Integer),
Column('r1', Integer, nullable=False),
Column('r2', Integer, nullable=False),
ForeignKeyConstraint(['r1','r2'],
[reftable.c.id,reftable.c.jd],
name='test_fk')

4
migrate/tests/versioning/test_genmodel.py
View File

@ -39,7 +39,9 @@ class TestSchemaDiff(fixture.DB):
diff = schemadiff.getDiffOfModelAgainstDatabase(self.meta, self.engine, excludeTables=['migrate_version'])
genmodel.ModelGenerator(diff,self.engine).runB2A()
@fixture.usedb()
# NOTE(mriedem): DB2 handles UnicodeText as LONG VARGRAPHIC
# so the schema diffs on the columns don't work with this test.
@fixture.usedb(not_supported='ibm_db_sa')
def test_functional(self):
def assertDiff(isDiff, tablesMissingInDatabase, tablesMissingInModel, tablesWithDiff):

5
migrate/tests/versioning/test_schemadiff.py
View File

@ -157,7 +157,10 @@ class Test_getDiffOfModelAgainstDatabase(SchemaDiffBase):
Column('data', Float()),
)
@fixture.usedb()
# NOTE(mriedem): The ibm_db_sa driver handles the Float() as a DOUBLE()
# which extends Numeric() but isn't defined in sqlalchemy.types, so we
# can't check for it as a special case like is done in schemadiff.ColDiff.
@fixture.usedb(not_supported='ibm_db_sa')
def test_float_vs_numeric(self):
self._assert_diff(
Column('data', Float()),

9
migrate/tests/versioning/test_version.py
View File

@ -160,6 +160,15 @@ class TestVersion(fixture.Pathed):
ver = Version(1, self.temp_usable_dir, [])
self.assertRaises(ScriptError, ver.add_script, '123.sql')
# tests bad ibm_db_sa filename
ver = Version(123, self.temp_usable_dir, [])
self.assertRaises(ScriptError, ver.add_script,
'123_ibm_db_sa_upgrade.sql')
# tests that the name is ok but the script doesn't exist
self.assertRaises(InvalidScriptError, ver.add_script,
'123_test_ibm_db_sa_upgrade.sql')
pyscript = os.path.join(self.temp_usable_dir, 'bla.py')
open(pyscript, 'w')
ver.add_script(pyscript)

10
migrate/versioning/version.py
View File

@ -221,6 +221,16 @@ class Version(object):
"(needs to be ###_description_database_operation.sql)")
version = parts[0]
op = parts[-1]
# NOTE(mriedem): check for ibm_db_sa as the database in the name
if 'ibm_db_sa' in basename:
if len(parts) == 6:
dbms = '_'.join(parts[-4: -1])
else:
raise exceptions.ScriptError(
"Invalid ibm_db_sa SQL script name '%s'; "
"(needs to be "
"###_description_ibm_db_sa_operation.sql)" % basename)
else:
dbms = parts[-2]
else:
raise exceptions.ScriptError(

1
requirements.txt
View File

@ -5,4 +5,5 @@ pbr>=0.5.21,<1.0
# against all the versions we are compatible with.
SQLAlchemy>=0.7.8
decorator
six>=1.4.1
Tempita >= 0.4

2
test-requirements.txt
View File

@ -8,6 +8,8 @@ coverage>=3.6
discover
feedparser
fixtures>=0.3.14
ibm_db_sa>=0.3.0
mock>=1.0
mox>=0.5.3
MySQL-python
psycopg2

1
test_db.cfg
View File

@ -12,3 +12,4 @@ postgresql://openstack_citest:openstack_citest@localhost/openstack_citest
mysql://openstack_citest:openstack_citest@localhost/openstack_citest
#oracle://scott:tiger@localhost
#firebird://scott:tiger@localhost//var/lib/firebird/databases/test_migrate
#ibm_db_sa://migrate:migrate@localhost:50000/migrate