Questions from AWS customer about amazon-redshift-udfs HOT 5 CLOSED

Can we get the error that was thrown why trying to install the wheel file?

from amazon-redshift-udfs.

Here are the steps followed and the errors encountered by the customer.

I've been having some trouble trying to load a particular Python package (https://github.com/selwin/python-user-agents) into Redshift as a library.

Things I've tried:
(1) From the link in the Github repo, I zipped the python-user-agents/user_agents folder, uploaded it to s3 and created the library using a CREATE LIBRARY statement. Once I did this, I wrote a small UDF just importing the relevant class from the user_agents module I wanted to use. However, I got an import error for a different module (ua_parser), which user_agents depends on.
(2) Given the outcome of (1), I figured I needed to have all module dependencies present in the zip file (is this information documented somewhere?). I ended up finding this nifty looking import script from you guys here (https://github.com/aws-samples/amazon-redshift-udfs/blob/master/bin/PipLibraryInstaller/installPipModuleAsRedshiftLibrary.sh), which builds wheel files for a python module and all its dependencies, zips them together and uploads the zip file to s3. After making a couple of modifications (namely, those in: https://github.com/aws-samples/amazon-redshift-udfs/pull/16/files), I uploaded the new user_agents.zip file to s3 and recreated the library in Redshift (along with the test UDF). I got another import error, but this time for the user_agents module itself.

I'd like to be able to use the module import script in the future, so could you tell me what I'm doing wrong? I'm attaching both versions of the python package (from (1) and (2), but I'm attaching them with different names, to be less confusing for you). The CREATE LIBRARY statement I used was:
"""
CREATE OR REPLACE LIBRARY user_agents
LANGUAGE plpythonu
FROM 's3://<s3_dir>/user_agents.zip'
CREDENTIALS '<redshift_iam_role_creds>'
;
"""
(NOTE: again, the zip files actually had the name user_agents.zip at the time of uploading).

The test UDF I wrote was:
"""
CREATE OR REPLACE FUNCTION f_user_agent_test (ua_string VARCHAR)
RETURNS VARCHAR
IMMUTABLE
AS $$
from user_agents import parse
return ua_string
$$ LANGUAGE plpythonu
;
"""

And the query I used to validate was:
"""
SELECT f_user_agent_test('blah');

The error messages were as follows (numbered as in "Things I've tried"):

(1) -----
message: "ImportError: No module named ua_parser"
traceback: """
line 2, in f_user_agent_test
File "init.py", line 3, in
File "parsers.py", line 3, in
"""

(2) -----
message: "ImportError: No module named user_agents"
traceback: "line 2, in f_user_agent_test"

from amazon-redshift-udfs.

Did you ever resolve this? I'm having the same issue.

from amazon-redshift-udfs.

See latest libraryInstaller.sh this is resolved.

from amazon-redshift-udfs.

Fix included here https://github.com/aws-samples/amazon-redshift-udfs/blob/master/bin/libraryInstaller.sh

from amazon-redshift-udfs.