Using third party modules like pandas in python udx

krmvsingh

Hello everyone,

Although this question has been asked twice earlier but I did not understand procedure of it. Once it remained unanswered and one it was suggested to install pandas in vertica but how?

My question is, if someone wants to create a udx and want to import pandas just like vertica_sdk in python udx then how to achieve that goal. How will vertica understand third parties from inside code?

SruthiA

First, install python and its corresponding pip module on your vertica node. Now use pip to install required packages. In your case, it is pandas. Create a directory named "dependencies" in the location where your UDx exists. Now copy the pandas and other libraries folders from site-packages folder (it is created, when you install python) to dependencies folder. Now create library with DEPENDS clause and specify the path of dependencies directory

CREATE LIBRARY myLib AS '/home/dbadmin/myprogram.py'
DEPENDS '/home/dbadmin/dependencies/*' LANGUAGE 'PYTHON';
https://www.vertica.com/docs/9.2.x/HTML/Content/Authoring/SQLReferenceManual/Statements/CREATELIBRARY.htm