Skip to main content

SAP HANA Database Client for Python

Project description

A pure Python client for the SAP HANA Database based on the SAP HANA Database SQL Command Network Protocol.

pyhdb supports Python 2.7, 3.3, 3.4, 3.5 and also PyPy on Linux, OSX and Windows. It implements a large part of the DBAPI Specification v2.0 (PEP 249). We plan to support Python 2.6 again in the future.

Table of contents


Install from Python Package Index:

$ pip install pyhdb

Install from GitHub via pip:

$ pip install git+

You can also install the latest version direct from a cloned git repository.

$ git clone
$ cd pyhdb
$ python install

Getting started

If you do not have access to a SAP HANA server, go to the SAP HANA Developer Center and choose one of the options to get your own trial SAP HANA Server.

For using PyHDB with hanatrial instance, follow this guide.

The basic pyhdb usage is common to database adapters implementing the DBAPI 2.0 interface (PEP 249). The following example shows how easy it’s to use the pyhdb module.

>>> import pyhdb
>>> connection = pyhdb.connect(

>>> cursor = connection.cursor()
>>> cursor.execute("SELECT 'Hello Python World' FROM DUMMY")
>>> cursor.fetchone()
(u"Hello Python World",)

>>> connection.close()

Establish a database connection

The function pyhdb.connect creates a new database session and returns a new Connection instance. Please note that port isn’t the instance number of you SAP HANA database. The SQL port of your SAP HANA is made up of 3<instance-number>15 for example the port of the default instance number 00 is 30015.

Currently pyhdb only supports the user and password authentication method. If you need another authentication method like SAML or Kerberos than please open a GitHub issue. Also there is currently no support of encrypted network communication between client and database.

Cursor object

With the method cursor of your Connection object you create a new Cursor object. This object is able to execute SQL statements and fetch one or multiple rows of the resultset from the database.

Example select

>>> cursor = connection.cursor()

After you executed a statement you can fetch one or multiple rows from the resultset.

>>> cursor.fetchone()
(u'SYS', u'DUMMY')

>>> cursor.fetchmany(3)

You can also fetch all rows from your resultset.

>>> cursor.fetchall()

Example Create table

With the execute method you can also execute DDL statements like CREATE TABLE.

>>> cursor.execute('CREATE TABLE PYHDB_TEST("NAMES" VARCHAR (255) null)')

Example insert

You can also execute DML Statements with the execute method like INSERT or DELETE. The Cursor attribute rowcount contains the number of affected rows by the last statement.

>>> cursor.execute("INSERT INTO PYHDB_TEST VALUES('Hello Python World')")
>>> cursor.rowcount


Three different types of LOBs are supported and corresponding LOB classes have been implemented: * Blob - binary LOB data * Clob - string LOB data containing only ascii characters * NClob - string (unicode for Python 2.x) LOB data containing any valid unicode character

LOB instance provide a file-like interface (similar to StringIO instances) for accessing LOB data. For HANA LOBs lazy loading of the actual data is implemented behind the scenes. An initial select statement for a LOB only loads the first 1024 bytes on the client:

>>> mylob = cursor.execute('select myclob from mytable where id=:1', [some_id]).fetchone()[0]
>>> mylob
<Clob length: 2000 (currently loaded from hana: 1024)>

By calling the read(<num-chars>)-method more data will be loaded from the database once <num-chars> exceeds the number of currently loaded data:

>>>   # -> returns the first 1500 chars, by loading extra 476 chars from the db
>>> mylob
<Clob length: 2000 (currently loaded from hana: 1500)>
>>>   # -> returns the last 500 chars by loading them from the db
>>> mylob
<Clob length: 2000 (currently loaded from hana: 2000)>

Using the seek() methods it is possible to point the file pointer position within the LOB to arbitrary positions. tell() will return the current position.

LOBs can be written to the database via insert or update-statemens with LOB data provided either as strings or LOB instances:

>>> from pyhdb import NClob
>>> nclob_data = u'朱の子ましける日におえつかうまつ'
>>> nclob = NClob(nclob_data)
>>> cursor.execute('update mynclob set nclob_1=:1, nclob_2=:2 where id=:3', [nclob, nclob_data, myid])

Stored Procedures

Rudimentary support for Stored Procedures call, scalar parameters only:

The script shall call the stored procedure PROC_ADD2 (source below):

>>> sql_to_prepare = 'call PROC_ADD2 (?, ?, ?, ?)'
>>> params = {'A':2, 'B':5, 'C':None, 'D': None}
>>> psid = cursor.prepare(sql_to_prepare)
>>> ps = cursor.get_prepared_statement(psid)
>>> cursor.execute_prepared(ps, [params])
>>> result = cursor.fetchall()
>>> for l in result:
>>>     print l

from the stored procedure:

create procedure PROC_ADD2 (in a int, in b int, out c int, out d char)
language sqlscript
reads sql data as
    c := :a + :b;
    d := 'A';

Transaction handling

Please note that all cursors created from the same connection are not isolated. Any change done by one cursor is immediately visible to all other cursors from same connection. Cursors created from different connections are isolated as the connection based on the normal transaction handling.

The connection objects provides to method commit which commit any pending transaction of the connection. The method rollback undo all changes since the last commit.


If you found bugs or have other issues than you are welcome to create a GitHub Issue. If you have questions about usage or something similar please create a Stack Overflow Question with tag pyhdb.

Run tests

pyhdb provides a test suite which covers the most use-cases and protocol parts. To run the test suite you need the pytest and mock package. Afterwards just run py.test inside of the root directory of the repository.

$ pip install pytest mock
$ py.test

You can also test different python version with tox.

$ pip install tox
$ tox


For debugging purposes it is sometimes useful to get detailed tracing information about packages sent to hana and those received from the database. There are two ways to turn on the print out of tracing information:

  1. Set the environment variable HDB_TRACING=1 before starting Python, e.g. (bash-syntax!):

$ HDB_TRACE=1 python
  1. Import the pyhdb module and set pyhdb.tracing = True

Then perform some statements on the database and enjoy the output.

To get tracing information when running pytest provide the -s option:

$ HDB_TRACE=1 py.test -s


  • Allow execution of stored database procedure

  • Support of SELECT FOR UPDATE

  • Authentication methods

    • SAML

    • Kerberos

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pyhdb-0.3.4.tar.gz (45.0 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page