Label-studio: Error syncing tasks from S3 or GCS

[davidgasquez]

Describe the bug
Not able to read tasks from S3 or GCS.

To Reproduce
Steps to reproduce the behavior:

1. Create a new project: label-studio start my_s3_project/ --init --source s3 --source-path raw-data --input-format json --force --log-level DEBUG --source-params '{"use_blob_urls": false, "regex": ".*"}'
2. Upload a JSON file to s3://raw-data

[{"text":"To have faith is to trust yourself to the water"},{"text":"To have faith is to trust yourself to the water"}]

1. Use default _Text Classification_ task.
2. Skip import (doesn't work with S3 or GCS) and sync tasks and completions.

[2020-12-15 09:26:06,066] [INFO] [werkzeug::_log::122] 172.17.0.1 - - [15/Dec/2020 09:26:06] "GET /api/project/?polling=true&rnd=35136 HTTP/1.1" 200 -
Exception in thread Thread-220:
Traceback (most recent call last):
  File "/usr/local/lib/python3.6/threading.py", line 916, in _bootstrap_inner
    self.run()
  File "/usr/local/lib/python3.6/threading.py", line 864, in run
    self._target(*self._args, **self._kwargs)
  File "/label-studio/label_studio/storage/base.py", line 440, in _sync
    id = _get_new_id(key)
  File "/label-studio/label_studio/storage/base.py", line 434, in _get_new_id
    id = new_id
NameError: name 'new_id' is not defined

Expected behavior
The texts on the JSON file should be imported and ready to label.

Environment (please complete the following information):

• Docker: heartexlabs/label-studio:latest
• Label Studio Version: 0.8.2

Additional context
Same error happening with GCS and also with other input formats.

[makseq]

Thank you for your report. We have fixed it already in our latest pre-release 0.9.0.

[davidgasquez]

Thanks @makseq! Is there any release date for 0.9.0?

I'll try on 0.9.0 then. Feel free to close this one.

[makseq]

We are planning to make an Release Candidate on this week.

[makseq]

@davidgasquez could you check the latest master branch of LS? I've fixed this bug there. Or you can try docker == 0.8.2.post0.

[davidgasquez]

Thanks for getting back with this @makseq! Just tried and seems to work. I'm getting some errors related to loading multiple tasks on a JSON file but at least I made it work with a simple task.json file¹.

Let me know if I can test anything else. I'm happy to try things out on this branch if you need any help.

¹ Simple as:

{"text": "hello world"}

[makseq]

It will be cool if you can check 0.9.0rc1 release candidate for it 🙂

[davidgasquez]

Sorry for the delay! Back from holidays.

Just tested with 0.9.0rc1 and seems to work fine too!