Google Cloud Machine Learning Engine v1 API is not working with filter jobId:rnn*

Question

I started using the Machine Learning Engine API for terminal and python and I discovered some discrepancies between the bash API and python API:

$ gcloud ml-engine jobs list --filter='jobId:eval_*'
JOB_ID                                 STATUS     CREATED
eval_chest_frontal_golden_201903       SUCCEEDED  2019-03-12T14:35:50
... (30 other results)

and in python:

from oauth2client.client import GoogleCredentials
from googleapiclient import discovery
from googleapiclient import errors

ml = discovery.build('ml', 'v1')

request = ml.projects().jobs().list(
  parent="<<<PROJECT_NAME_HERE>>>",
  filter="jobId:eval_*")

response = None
try:
  response = request.execute()
except errors.HttpError as err:
  raise Exception("Request failed!")

print(response)
# Prints: {}

I looked at the API explorer with the same parameters:

https://developers.google.com/apis-explorer/#p/ml/v1/ml.projects.jobs.list

I get the same result as in python: {} but it looks like the URL is not escaping the * in https://.../jobs?filter=jobId%3Aeval_*&key=.... Maybe that is the bug. Any way I can fix this in the python API?

Answer 1

Just figured out you can use this filter instead: "jobId:eval_". Like below:

request = ml.projects().jobs().list(
  parent="<<<PROJECT_NAME_HERE>>>",
  filter="jobId:eval_")

and has the same effect as "jobId:eval_*".

The only concern is that Google will deprecate that feature in the future:

WARNING: --filter : operator evaluation is changing for consistency across Google APIs. jobId:eval_ currently matches but will not match in the near future. Run gcloud topic filters for details.