Support for the Oracle database.
Oracle version 8 through current (11g at the time of this writing) are supported.
For information on connecting via specific drivers, see the documentation for that driver.
The dialect supports several create_engine() arguments which affect the behavior of the dialect regardless of driver in use.
SQLAlchemy Table objects which include integer primary keys are usually assumed to have “autoincrementing” behavior, meaning they can generate their own primary key values upon INSERT. Since Oracle has no “autoincrement” feature, SQLAlchemy relies upon sequences to produce these values. With the Oracle dialect, a sequence must always be explicitly specified to enable autoincrement. This is divergent with the majority of documentation examples which assume the usage of an autoincrement-capable database. To specify sequences, use the sqlalchemy.schema.Sequence object which is passed to a Column construct:
t = Table('mytable', metadata,
Column('id', Integer, Sequence('id_seq'), primary_key=True),
Column(...), ...
)
This step is also required when using table reflection, i.e. autoload=True:
t = Table('mytable', metadata,
Column('id', Integer, Sequence('id_seq'), primary_key=True),
autoload=True
)
In Oracle, the data dictionary represents all case insensitive identifier names using UPPERCASE text. SQLAlchemy on the other hand considers an all-lower case identifier name to be case insensitive. The Oracle dialect converts all case insensitive identifiers to and from those two formats during schema level communication, such as reflection of tables and indexes. Using an UPPERCASE name on the SQLAlchemy side indicates a case sensitive identifier, and SQLAlchemy will quote the name - this will cause mismatches against data dictionary data received from Oracle, so unless identifier names have been truly created as case sensitive (i.e. using quoted names), all lowercase names should be used on the SQLAlchemy side.
SQLAlchemy 0.6 uses the “native unicode” mode provided as of cx_oracle 5. cx_oracle 5.0.2 or greater is recommended for support of NCLOB. If not using cx_oracle 5, the NLS_LANG environment variable needs to be set in order for the oracle client library to use proper encoding, such as “AMERICAN_AMERICA.UTF8”.
Also note that Oracle supports unicode data through the NVARCHAR and NCLOB data types. When using the SQLAlchemy Unicode and UnicodeText types, these DDL types will be used within CREATE TABLE statements. Usage of VARCHAR2 and CLOB with unicode text still requires NLS_LANG to be set.
Oracle has no support for the LIMIT or OFFSET keywords. Whereas previous versions of SQLAlchemy used the “ROW NUMBER OVER...” construct to simulate LIMIT/OFFSET, SQLAlchemy 0.5 now uses a wrapped subquery approach in conjunction with ROWNUM. The exact methodology is taken from http://www.oracle.com/technology/oramag/oracle/06-sep/o56asktom.html . Note that the “FIRST ROWS()” optimization keyword mentioned is not used by default, as the user community felt this was stepping into the bounds of optimization that is better left on the DBA side, but this prefix can be added by enabling the optimize_limits=True flag on create_engine().
Oracle doesn’t have native ON UPDATE CASCADE functionality. A trigger based solution is available at http://asktom.oracle.com/tkyte/update_cascade/index.html .
When using the SQLAlchemy ORM, the ORM has limited ability to manually issue cascading updates - specify ForeignKey objects using the “deferrable=True, initially=’deferred’” keyword arguments, and specify “passive_updates=False” on each relation().
When using Oracle 8, a “use_ansi=False” flag is available which converts all JOIN phrases into the WHERE clause, and in the case of LEFT OUTER JOIN makes use of Oracle’s (+) operator.
When using reflection with Table objects, the dialect can optionally search for tables indicated by synonyms that reference DBLINK-ed tables by passing the flag oracle_resolve_synonyms=True as a keyword argument to the Table construct. If DBLINK is not in use this flag should be left off.
Support for the Oracle database via the cx_oracle driver.
The Oracle dialect uses the cx_oracle driver, available at http://cx-oracle.sourceforge.net/ . The dialect has several behaviors which are specifically tailored towards compatibility with this module.
Connecting with create_engine() uses the standard URL approach of oracle://user:pass@host:port/dbname[?key=value&key=value...]. If dbname is present, the host, port, and dbname tokens are converted to a TNS name using the cx_oracle makedsn() function. Otherwise, the host token is taken directly as a TNS name.
Additional arguments which may be specified either as query string arguments on the URL, or as keyword arguments to create_engine() are:
As of cx_oracle 5, Python unicode objects can be bound directly to statements, and it appears that cx_oracle can handle these even without NLS_LANG being set. SQLAlchemy tests for version 5 and will pass unicode objects straight to cx_oracle if this is the case. For older versions of cx_oracle, SQLAlchemy will encode bind parameters normally using dialect.encoding as the encoding.
cx_oracle presents some challenges when fetching LOB objects. A LOB object in a result set is presented by cx_oracle as a cx_oracle.LOB object which has a read() method. By default, SQLAlchemy converts these LOB objects into Python strings. This is for two reasons. First, the LOB object requires an active cursor association, meaning if you were to fetch many rows at once such that cx_oracle had to go back to the database and fetch a new batch of rows, the LOB objects in the already-fetched rows are now unreadable and will raise an error. SQLA “pre-reads” all LOBs so that their data is fetched before further rows are read. The size of a “batch of rows” is controlled by the cursor.arraysize value, which SQLAlchemy defaults to 50 (cx_oracle normally defaults this to one).
Secondly, the LOB object is not a standard DBAPI return value so SQLAlchemy seeks to “normalize” the results to look more like that of other DBAPIs.
The conversion of LOB objects by this dialect is unique in SQLAlchemy in that it takes place for all statement executions, even plain string-based statements for which SQLA has no awareness of result typing. This is so that calls like fetchmany() and fetchall() can work in all cases without raising cursor errors. The conversion of LOB in all cases, as well as the “prefetch” of LOB objects, can be disabled using auto_convert_lobs=False.
Two Phase transactions are implemented using XA transactions. Success has been reported with this feature but it should be regarded as experimental.
Support for the Oracle database via the zxjdbc JDBC connector.
The official Oracle JDBC driver is at http://www.oracle.com/technology/software/tech/java/sqlj_jdbc/index.html.