- This python package allows users to integrate new and existing SQLAlchemy applications with Splice Machine
- SQLAlchemy exposes a uniform API for database-backed applications, which can be easily used with Splice Machine via this adaptor
- Supported on Splice Machine Versions >= 2.8.0.1920
- Requires Python 3.6+
There are two ways to install this Driver. With and without ODBC Driver custom installation. We recommend without. When running these commands, you must either use sudo or have root access.
brew install unixodbc
pip install splicemachinesa
First ensure you have the following packages installed. They are necessary for pyodbc.
Ubuntu:
- build-essential
- libssl-dev
- libffi-dev
- unixodbc
- unixodbc-dev
- python3-dev
- curl
- unzip
- iodbc
sudo apt -y update && sudo apt -y upgrade && sudo apt -y install build-essential libssl-dev libffi-dev unixodbc unixodbc-dev python3-dev curl unzip iodbc
Centos:
- unzip
- gcc
- openssl-devel
- gcc-c++
- unixODBC-devel
- unixODBC
- python3-devel
sudo yum -y update && sudo yum -y install unzip gcc openssl-devel gcc-c++ unixODBC-devel unixODBC python3-devel
Then run pip install splicemachinesa
Note: If you have an error during installation, it is likely with the installation of PyODBC. When searching the error, reference PyODBC in the search.
First, download the appropriate Splice Machine ODBC driver for your system.
Mac OSX 64 bit Driver: Download
Linux 64 bit Driver: Download
Note: Windows is not currently supported for Splice Machine SQLAlchemy ODBC
Then, follow the instructions here to finish the installation, but use the download links above to retrieve the Driver tarball
Once the Driver is installed, you can pip install splicemachinesa.
NOTE: If you are on Mac, you must install unixODBC
brew install unixodbc
pip install splicemachinesa
You can use this package for SqlAlchemy usage or raw ODBC usage
from splicemachinesa.pyodbc import splice_connect
ODBC_CONNECTION = splice_connect(URL=[URL], UID=[UID], PWD=[PWD], SSL=[SSL])
Filling in UID
, PWD
, URL
, and SSL
with the proper values for your database. SSL defaults to 'basic' If you are connecting to the Splice Database inside the same network (ie standalone splice) you will set SSL=None
from splicemachinesa.pyodbc import splice_connect
ODBC_CONNECTION = splice_connect(URL=[URL], JWT_TOKEN=[JWT_TOKEN], JWT_TYPE=[JWT_TYPE], SSL=[SSL])
Filling in JWT_TOKEN
, JWT_TYPE
, URL
, and SSL
with the proper values for your database. SSL defaults to 'basic' If you are connecting to the Splice Database inside the same network (ie standalone splice) you will set SSL=None
You must set the JWT_TYPE to use this format. Available options are ('SPLICE_JWT', 'SPLICE_JWT_PUB', 'OKTA_OAUTH', 'SPLICE_OAUTH')
There are three URL formats that can be used to access SpliceMachine via ODBC through SQLAlchemy. The first will suffice in most cases.
Format: splicemachinesa://[user]:[password]@[host]:[port]/[database]
The Driver Name defaults to SpliceODBCDriver
, which is the default name
when the driver is installed (specified in odbc.ini
). However, this name can be overridden through the
SPLICE_ODBC_DRIVER_NAME
environment variable.
Example:
from sqlalchemy import create_engine
url = 'splicemachinesa://splice:admin@127.0.0.1:1527/splicedb'
engine = create_engine(url)
Format: splicemachinesa://?DRIVER=[driver]&URL=[URL]&PORT=[PORT]&UID=[USER]&PWD=[PASSWORD]
You can use this SQLAlchemy driver without an ODBC configuration (meaning running the installation from Splice Machine docs)
by manually specifying the driver path.
- On OSX (given installation with
sudo
), the Driver is found in/Library/ODBC/SpliceMachine/libsplice_odbc64.dylib
- On OSX, Without
sudo
this driver is located at$HOME/Library/ODBC/SpliceMachine/libsplice_odbc64.dylib
- On Linux (given ODBC Driver installation with
sudo
), this is found in/usr/local/splice/libsplice_odbc.so
- On Linux, Without
sudo
it is found in$HOME/splice
.
We provide a utility function for simplifying the URL building process.
This method is easier for an automated driver configuration (e.g. inside Docker) because you only need to copy to Driver binary.
Note: Error messages will not be rendered properly with this approach.
Example Linux:
from sqlalchemy import create_engine
from splicemachinesa.utilities import url_builder
url = url_builder('/usr/local/splice/lib/libsplice_odbc.so', host='localhost', port=1527,
user='splice', password='admin')
engine = create_engine(url)
Example Mac:
from sqlalchemy import create_engine
from splicemachinesa.utilities import url_builder
url = url_builder('/Library/ODBC/SpliceMachine/libsplice_odbc64.dylib', host='localhost', port=1527,
user='splice', password='admin')
engine = create_engine(url)
Format: splicemachinesa://[dsn]
Splice Machine SQLAlchemy also supports ODBC DSNs for
Driver configuration. This means that rather than explicitly specifying
configuration options in the URL string, they are rendered from the odbc.ini file. This is also the only method that
supports Kerberos Authentication for Splice Clusters. You can see how to use Kerberos
here. The odbc.ini
file is located
at these locations depending on whether or not the installer was run as root.
With/without sudo
installation- Mac OSX: $HOME/.odbc.ini
With sudo
Linux: /etc/odbc.ini
Without sudo
Linux: $HOME/.odbc.ini
. In this file, if you have a key named USER
,
rename this to UID
.
The default DSN is SpliceODBC64
.
Example:
from sqlalchemy import create_engine
url = 'splicemachinesa://SpliceODBC64'
engine = create_engine(url)
- First make sure you have a fresh installation of Splice Machine running (either Standalone or Cloud/Bespoke/On-Prem) with an appropriate SQLAlchemy Splice Machine connection string for accessing it
# register any changes for testing
cd /path/to/splice_sqlalchemy
sudo pip install .
sudo pip install pytest
pytest -vv --db-url 'splicemachinesa://[...]'
- Common Table Expressions
- Indices Reflection
- Check + Unicode Reflection
- Nullable Reflection
- ORDER BY COLLATE
- Empty Set Insertion
- Unicode String Support
- Reflector for Order by Foreign Key
- limit_offset_in_unions_from_alias