Django DB Backend for SAP HANA

Ready for use.

Prerequisite

1. Python compiled with --enable-unicode=ucs2 or use the python provided with hdbclient package.
2. Django 1.4 (Only tested with this version, 1.5 might work)
3. HANA python driver configured in python path. Only available for windows x86_64 and linux x86_64
4. Tested with HDB version 1.00.45.371235 (NewDB100_REL).

Setup

1. install the python package via setup.py

   python setup.py install

2. The config in the Django project is as follows

   DATABASES = {
   	'default': {
       	'ENGINE': 'django_hana',           # or as per your python path
       	'NAME': '<SCHEMA_NAME>',           # The schema to use. It will be created if doesn't exist
       	'USER': '<USERNAME>',
       	'PASSWORD': '<PASSWORD>',
       	'HOST': '<HOSTNAME>',                      
       	'PORT': '3<INSTANCE_NUMBER>15',               
   	}
   }

3. HANA doesn't support Timezone. Set USE_TZ=False in settings.py.

Config

Column/Row store

Use the column/row-store class decorators to make sure that your models are using the correct HANA engine. If the models are not using any decorators the default behaviour will be a ROW-store column.

from django.db import models
from django_hana import column_store, row_store

@column_store
class ColumnStoreModel(models.Model):
	some_field = models.CharField()
	
@row_store
class RowStoreModel(models.Model):
	some_field = models.CharField()

Log

• On leaving transaction ensuring autocommit is back to enabled
• Fixed bug: transaction were not working. It stayed in autocommit even after requesting for manual commit mode.
• Fixed bug: case insensitive text search
• No showstoppers. Ready for release. Yes!!
• Just found out, sequences gives values based on current session. After calling nextval in the current session, currval will always return the same value that was generated before in the current session irrespective of any other concurrent insert.
• [Fixed]HANA doesn't return id after insert. Currently taking the curval of the sequence after insert. This may cause problems when too many inserts are done simultaneously. Needs rework. May be grab the seq's nextval while constructing insert query and use it in place of id and return it after insert is done.
• Tested with models in official django tutorial and Models References. All queries worked.
• Currently, executes set schema on every cursor creation. Prefixing each table name with schema name is more efficient. Needs rework.
• Select and update works with tested models. Placeholder conversion works!!
• Major problem is the paramstyle hdb supports qmark while django assumes %s, need to write a converter
• Hana doesn't support timestamp with timezone info, stripping timezone info from the string
• Select and update works with tested models
• Simple inserts work
• syncdb creates tables, sequences and indexes
• connects to db
• uses the settings_dict['NAME'] as default schema

TODO

• Make a custom auto field for UUID
• Do more tests
• Try Foreign Keys

Disclaimer

This project is not a part of standard SAP HANA delivery, hence SAP support is not responsible for any queries related to this software. All queries/issues should be reported here.