Fixed a bug in the dependency checking for Connnection.itertableas.

oradelete now has a new option to use truncate table instead of delete from.

Added a new script oradelete that can be used to delete all records from all tables and to reset all sequences.

Connection has a new method itersequences.

Fixed a bug in the generated SQl code for triggers (the name always included the name of the original schema).

ll.orasql now requires cx_Oracle 5.0 compiled in Unicode mode (i.e. with WITH_UNICODE=1). Lots of unicode handling stuff has been rewritten to take advantage of Unicode mode.

orafind has a new option --encoding to decode the searchstring on the commandline.

The Pool constructor now supports the additional arguments getmode and homogeneous from cx_Oracle 5.0.

Fix a typo in Privilege.grantddl.

Procedures and functions with timestamp arguments can now be called.

Procedures and functions now should handle arguments of type BLOB correctly.

Record has a new method get which works like the dictionary method get.

orafind.py now has an additional options readlobs (defaulting to false). If this option is set, the value of LOBs in the records found, will be printed.

A new script has been added: orafind.py will search for a specified string in all columns of all tables in a schema.

Fixed two bugs in Callable._calcargs and Connection.getobject.

Connection.getobject, Procedure and Function now support functions and procedures in packages.

Added __repr__ to the exception classes.

All database scripts now have an additional option encoding that specifies the encoding for the output script.

Fixed a regression in the scripts oracreate.py, oradrop.py and oragrant.py.

When calling functions/procedures, arguments are now wrapped in variable objects for their real type instead of ones for the type the function or procedure expects.

Added a __contains__ to Record for checking the existence of a field.

Calling procedures and functions has been rewritten: ll.orasql will only pass those parameters to the procedure/function that are passed to the call (or variables for out parameters). Internally this is handled by executing the call as a parameterized query calling the procedure/function with named arguments.

FetchRecord has been renamed to Record (and is used for the result of procedure and function calls now, which required some internal changes to FetchRecord). The former Record has been renamed to Args as its only use now is collecting arguments for procedure/function calls. (The method fromdata has been dropped.)

The __repr__ output of Argument objects now shows the datatype.

Added a new method _getobject to Connection that does what getobject does, but is case sensitive (This is used internally by Synonym.getobject).

The methods xfetchone, xfetchmany, xfetchall, xfetch, xexecute and xexecutemany have been dropped again. Fetch result objects are now of type FetchRecord. Field access is available via index (i.e. row[0]), key (row["name"]) and attribute (row.name). These result objects are generated via the rowfactory attribute (which was added in cx_Oracle 4.3.2). All fetch and execute methods support unicode values.

Updated the scripts to work with the new execute methods.

Connection has a new method getobject, which returns the schema object with a specified name.

Synonym has a new method getobject, that returns the object for which the Synonym object is a synonym.

The name of Procedure and Function objects now is case sensitive when calling the procedure or function.

The fancy fetch methods have been renamed to xfetchone, xfetchmany, xfetchall and xfetch. __iter__ no longer gets overwritten. New methods xexecute and xexecutemany have been added, that support passing unicode parameters.

All docstrings use ReST now.

Updated the docstrings to XIST 3.0.

Added ReST versions of the documentation.

Fixed a bug in the error handling of wrong arguments when calling functions or procedures.

The threshold for string length for procedure and function arguments has been reduced to 4000.

BLOB arguments for procedures and functions are always passed as variables now.

Arguments for procedures and functions that are longer that 32000 characters are passed as variables now (the threshold was 32768 before and didn't work).

Fix an inverted logic bug in Record.fromdata that surfaced in unicode mode: BLOBs were treated as string and CLOBs as binary data.

The readlobs and unicode parameters are now honored when calling procedures and functions via Procedure and Function objects.

A parameter unicode has been added to various constructors and methods. This parameter can be used to get strings (i.e. VARCHAR2 and CLOBs) as unicode object instead of str objects.

Fixed an output bug in oradiff.py when running in full output mode.

A parameter readlobs has been added to various constructors and methods that can be used to get small (or all) LOB values as strings in cursor fetch calls.

A new method iterprivileges has been added to Connection.

A script oragrant.py has been added for copying privileges.

Two new methods (itertables and iterfks) have been added to Connection. They yield all table definitions or all foreign keys respectively.

A new method isenabled has been added to ForeignKey.

A __str__ method has been added to Object.

A bug in oramerge.py has been fixed: In certain situations oramerge.py used merging actions that were meant to be used for the preceeding object.

Fixed a bug that showed up when an index and a foreign key of the same name existed.

Fixed a bug in Index.__xattrs__.

Function objects are now callable too. They return the return value and a Record containing the modified input parameters.

Fixed a bug in Column.modifyddl.

The class Column has gained a few new methods: datatype, default, nullable and comment.

Calling a procedure will now raise a SQLObjectNotFoundError error, if the procedure doesn't exist.

The classes Proc and LLProc have been removed. The functionality of Proc has been merged into ProcedureDefinition (with has been renamed to Procedure). Information about the procedure arguments is provided by the iteraguments method.

All other subclasses of Definition have been renamed to remove the "Definition" for the name to reduce typing. (Methods have been renamed accordingly too.)</li>

oramerge.main and oradiff.main now accept option arrays as arguments.

oradiff.py has finally been fixed.

Fixed a bug in oradiff.py.

Fixed a bug in oramerge.py.

Fixed a bug in TableDefinition.iterdefinitions.

Fixed a bug in oracreate.py.

Dependencies involving MaterializedViewDefinition and IndexDefinition objects generated by constraints work properly now, so that iterating all definitions in create order really results in a working SQL script.

A method table has been added to PKDefinition, FKDefinition, UniqueDefinition and IndexDefinition. This method returns the TableDefinition to object belongs to.

A method pk has been added to FKDefinition. It returns the primary key that this foreign key references.

Indexes and constraints belonging to skipped tables are now skipped too in oracreate.py.

Arguments other than sys.argv[1:] can now be passed to the oracreate.py and oradrop.py main functions.

ll.orasql can now handle objects name that are not in uppercase.

Connection.iterobjects has been renamed to iterdefinitions.

Each Definition subclass has a new classmethod iterdefinitions that iterates through all definitions of this type in a schema (or all schemas).

Each Definition subclass has new methods iterreferences and iterreferencedby that iterate through related definitions. The methods iterreferencesall and iterreferencedbyall do this recursively. The method iterdependent is gone now.

The method iterschema of Connection now has an additional parameter schema. Passing "all" for schema will give you statistics for the complete database not just one schema.

A new definition class MaterializedViewDefinition has been added that handles materialized views. Handling of create options is rudimentary though. Patches are welcome.

TableDefinition has a three new methods: ismview returns whether the table is a materialized view; itercomments iterates through comments and iterconstraints iterates through primary keys, foreign keys and unique constraints.

The method getcursor will now raise a TypeError if it can't get a cursor.

RAW fields in tables are now output properly in TableDefinition.createddl.

A class PackageBodyDefinition has been added. oracreate.py will output package body definitions and oradrop.py will drop them.

Duplicate code in the scripts has been removed.

Fixed a bug in oramerge.py: If the source to be diffed was long enough the call to diff3 deadlocked.

The method iterobjects has been moved from Cursor to Connection.

The method itercolumns has been moved from Cursor to TableDefinition.

LLProc now recognizes the c_out parameter used by ll.toxic 0.8.

Support for positional arguments has been added for Proc and LLProc. Error messages for calling procedures have been enhanced.

SequenceDefinition now has a new method createddlcopy that returns code that copies the sequence value. oracreate.py has a new option -s/--seqcopy that uses this feature.

setuptools is now supported for installation.

Added a SessionPool (a subclass of SessionPool in cx_Oracle) whose acquire method returns ll.orasql.Connection objects.

Added a class IndexDefinition for indexes. oracreate.py will now issue create statements for indexes.

Fixed a bug with empty lines in procedure sources.

Remove spurious spaces at the start of procedure and function definitions.

Fixed a bug that the DDL output of Java source.

Trailing whitespace in each line of procedures, functions etc. is now stripped.

Fixed a bug that resulted in omitted field lengths.

The option -m/--mode has been dropped from the script oramerge.py.

A new class ColumnDefinition has been added to ll.orasql. The Cursor class has a new method itercolumns that iterates the ColumnDefinition objects of a table.

oramerge.py now doesn't output a merged create table statement, but the appropriate alter table statements.

Added an option -i to oracreate.py and oradrop.py to ignore errors.

The argument all of the cursor method iterobjects is now named schema and may have three values: "own", "dep" and "all".

Added an script oramerge.py that does a three way merge of three database schemas and outputs the resulting script.

DB links are now copied over in SynonymDefinition objects.

Added a argument to createddl and dropddl to specify if terminated or unterminated DDL is wanted (i.e. add ; or / or not).

CommentsDefinition has been renamed to CommentDefinition and holds the comment for one field only.

JavaSourceDefinition has been added.

The scripts oracreate.py, oradrop.py and oradiff.py now skip objects with "$" in their name by default. This can be changed with the -k option (but this will lead to unexecutable scripts).

oradiff.py has a new options -b: This allows you to specify how whitespace should be treated.

Added an option -x to oracreate.py to make it possible to directly execute the DDL in another database.

Fixed a bug in SequenceDefinition when the CACHE field was 0.

A script oradiff.py has been added which can be used for diffing Oracle schemas.

Definition classes now have two new methods cdate and udate that give the creation and modification time of the schema object (if available).

A "flat" iteration mode has been added to Cursor.iterobjects that returns objects unordered.

Connection has a new method connectstring.

A class LibraryDefinition has been added.

CommentsDefinition.createddl returns "" instead of "\n" now if there are no comments.

SQLObjectNotfoundError has been renamed to SQLObjectNotFoundError.

ll.orasql requires version 1.0 of the core package now.

A new generator method iterobjects has been added to the Cursor class. This generator returns "definition objects" for all the objects in a schema in topological order (i.e. if the name of an object (e.g. a table) is generated it will only depend on objects whose name has been yielded before). SQL for recreating and deleting these SQL objects can be generated from the definition objects.

Two scripts (oracreate.py and oradrop.py) have been added, that create SQL scripts for recreating or deleting the content of an Oracle schema.

The commands generated by iterdrop no longer have a terminating ;, as this seems to confuse Oracle/cx_Oracle.

Two new functions have been added: iterdrop is a generator that yields information about how to clear the schema (i.e. drop all table, sequences, etc.). itercreate yields information about how to recreate a schema.

Date values are now supported as OUT parameters.

Added a note about the package init file to the installation documentation.

ll.orasql now requires ll-core.

Procedures can now be called with string arguments longer that 32768 characters. In this case the argument will be converted to a variable before the call. The procedure argument must be a CLOB in this case.

Creating Record instances from database data is now done by the class method Record.fromdata. This means it's now possible to use any other class as long as it provides this method.

ll.orasql requires cx_Oracle 4.1 now.

Fixed a regression bug in Proc._calcrealargs as cursors will now always return Record objects.

Now generating Record object is done automatically in a subclass of cx_Oracle.Cursor. So now it's possible to use ll.orasql as an extended cx_Oracle.

Initial release.