NameError: name 'ujson' is not defined

[najiya-12]

I listed the files from S3
urls = ['s3://' + f for f in fs.glob("s3://noaa-goes16/ABI-L2-ACMC/2022/116/*/*.nc")]
given around
len(urls)
272 files
and then generated the jsons files and created in the ./jsons/ directory, however, I run into an issue with gen_json() function
NameError: name 'ujson' is not defined. I followed @lsterzinger tutorial in
(https://medium.com/pangeo/fake-it-until-you-make-it-reading-goes-netcdf4-data-on-aws-s3-as-zarr-for-rapid-data-access-61e33f8fe685)

[peterm790]

Try installing and importing the ujson package: https://pypi.org/project/ujson/ . I believe it is functionally the same as the normal json encoder just a bit faster

[keewis]

I think this is a bug in the blog post: it was updated to use json instead of ujson, but it seems like that particular instance was forgotten.

[martindurant]

Yep, originally we just ran with stock builtin JSON, but as reference files got bigger, we decided to switch to ujson. If installing kerchunk doesn't grab this, then we need to add it to the deps.

[lsterzinger]

Thanks for catching this typo @najiya-12 - I corrected the import to import ujson at the beginning of the tutorial. As @martindurant said, it should be installed alongside Kerchunk since it's listed in the dependencies

[najiya-12]

Thanks for catching this typo @najiya-12 - I corrected the import to import ujson at the beginning of the tutorial. As @martindurant said, it should be installed alongside Kerchunk since it's listed in the dependencies

Thank you! It worked

[keewis]

there's still a typo:

m_list = []
for js in tqdm(json_list):
    with open(js) as f:
        m_list.append(fsspec.get_mapper("reference://", 
            fo=json.load(f), remote_protocol="s3",
            remote_options={"anon": True}))

json.load → ujson.load

[lsterzinger]

Good catch, thanks!