The primary purpose of this is to have a working dialect for Apache Drill that can be used with Apache Superset.
https://superset.incubator.apache.org
Obviously, a working, robust dialect for Drill serves other purposes as well, but most of the iterative planning for this REPO will be based on working with Superset. Other changes will gladly be incorporated, as long as it doesn't hurt Superset integration.
Installing the dialect is straightforward. Simply:
pip install sqlalchemy-drill
Alternatively, you can download the latest release from github and install from here:
python3 -m pip install git+https://github.com/JohnOmernik/sqlalchemy-drill.git
Drill's REST API can execute queries with results streamed to JSON returned over chunked HTTP for Drill >= 1.19, otherwise with results buffered and then returned in a conventional HTTP response. A SQLAlchemy URL to connect to Drill over REST looks like the following.
drill+sadrill://<username>:<password>@<host>:<port>/<storage_plugin>?use_ssl=True
To connect to Drill running on a local machine running in embedded mode you can use the following connection string.
drill+sadrill://localhost:8047/dfs?use_ssl=False
| Parameter | Type | Description |
|---|---|---|
| use_ssl | boolean | Whether to connect to Drill using HTTPS |
| verify_ssl | boolean | Whether to verify the server's TLS certificate |
| impersonation_target[1] | string | Username of a Drill user to be impersonated by this connection |
[1] Requires a build of Drill that incorporates the fix for DRILL-8168.
Query result metadata returned by the Drill REST API is stored in the result_md field of the DB-API Cursor object. Note that any trailing metadata, i.e. metadata which comes after result row data, will only be populated after you have iterated through all of the returned rows. If you need this trailing metadata you can make the cursor object reachable after it has been completely iterated by obtaining a reference to it beforehand, as follows.
r = engine.execute('select current_timestamp')
r.cursor.result_md # access metadata, but only leading metadata
cur = r.cursor # obtain a reference for use later
r.fetchall() # iterate through all result data
cur.result_md # access metadata, including trailing metadata
del cur # optionally delete the reference when done
In versions of Drill earlier than 1.19, all data values are serialised to JSON strings and column type metadata comes after the data itself. As a result, for these versions of Drill, the drill+sadrill dialect returns every data value as a string. To convert non-string data to its native type you need to typecast it yourself.
In Drill 1.19 the REST API began making use of numeric types in JSON for numbers and times, the latter via a UNIX time representation while column type metadata was moved ahead of the result data. Because of this, the drill+sadrill dialect is able to return appropriate types for numbers and times when used with Drill >= 1.19.
Connecting to Drill via JDBC is a little more complicated than a local installation and complete instructions can be found on the Drill documentation here: https://drill.apache.org/docs/using-the-jdbc-driver/.
In order to configure SQLAlchemy to work with Drill via JDBC you must:
DRILL_JDBC_DRIVER_PATH to the full path of your driver locationDRILL_JDBC_JAR_NAME to the name of the .jar file for the Drill driver.Additionally, you will need to install JayDeBeApi as well as jPype version 0.6.3.
These modules are listed as optional dependencies and will not be installed by the default installer.
If the JDBC driver is not available, the dialect will throw errors when trying to connect. In addition, sqlalchemy-drill will not launch a JVM for you so you need to do this yourself with a call to JPype like the following. See the file test-jdbc.py in this repo for a working example.
jpype.startJVM("-ea", classpath="lib/*")
drill+jdbc://<username>:<passsword>@<host>:<port>
For a simple installation, this might look like:
drill+jdbc://admin:password@localhost:31010
In order to configure SQLAlchemy to work with Drill via ODBC you must:
unixODBC package for RedHat-based systems).pyodbc Python package.
This module is listed as an optional dependency and will not be installed by the default installer.To connect to Drill with SQLAlchemy use the following connection string:
drill+odbc:///?<ODBC connection parameters>
Connection properties are available in the official documentation: https://drill.apache.org/docs/odbc-configuration-reference/
For a simple installation, this might look like:
drill+odbc:///?Driver=/opt/mapr/drill/lib/64/libdrillodbc_sb64.so&ConnectionType=Direct&HOST=localhost&PORT=31010&AuthenticationType=Plain&UID=admin&PWD=password
or for the case when you have DSN configured in odbc.ini:
drill+odbc:///?DSN=drill_dsn_name
Note: it's better to avoid using connection string with hostname:port or username/password, like 'drill+odbc://admin:password@localhost:31010/' but use only ODBC properties instead to avoid any misinterpretation between these parameters.
For a complete tutorial on how to use Superset with Drill, read the tutorial on @cgivre's blog available here: http://thedataist.com/visualize-anything-with-superset-and-drill/.
Currently we can connect to drill, and issue queries for most visualizations and get results. We also enumerate table columns for some times of tables. Here are things that are working as some larger issues to work out. (Individual issues are tracked under issues)
to drillpy and pydrill for code used in creating the original drilldbapi.py code for connecting!
It is recommended to extend the official Docker image to include this Apache Drill driver:
FROM apache/superset
# Switching to root to install the required packages
USER root
RUN pip install git+https://github.com/JohnOmernik/sqlalchemy-drill.git
# Switching back to using the `superset` user
USER superset
FAQs
Apache Drill for SQLAlchemy
We found that sqlalchemy-drill demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
PyPI confirms no security flaws were exploited in the Ultralytics supply chain attack and highlights improvements for safer package publishing.
Security News
Research
The Socket Research Team breaks down a malicious wrapper package that uses obfuscation to harvest credentials and exfiltrate sensitive data.
Research
Security News
Attackers used a malicious npm package typosquatting a popular ESLint plugin to steal sensitive data, execute commands, and exploit developer systems.