Dataset JSON is incorrect

[severo]

Describe the bug

The JSON file generated for https://github.com/huggingface/datasets/blob/573f3d35081cee239d1b962878206e9abe6cde91/datasets/journalists_questions/journalists_questions.py is https://github.com/huggingface/datasets/blob/573f3d35081cee239d1b962878206e9abe6cde91/datasets/journalists_questions/dataset_infos.json.

The only config should be plain_text, but the first key in the JSON is journalists_questions (the dataset id) instead.

{
  "journalists_questions": {
    "description": "The journalists_questions corpus (version 1.0) is a collection of 10K human-written Arabic\ntweets manually labeled for question identification over Arabic tweets posted by journalists.\n",
    ...

Steps to reproduce the bug

Look at the files.

Expected results

The first key should be plain_text:

{
  "plain_text": {
    "description": "The journalists_questions corpus (version 1.0) is a collection of 10K human-written Arabic\ntweets manually labeled for question identification over Arabic tweets posted by journalists.\n",
    ...

Actual results

{
  "journalists_questions": {
    "description": "The journalists_questions corpus (version 1.0) is a collection of 10K human-written Arabic\ntweets manually labeled for question identification over Arabic tweets posted by journalists.\n",
    ...

[albertvillanova]

As discussed, the metadata JSON files must be regenerated because the keys were nor properly generated and they will not be read by the builder:

Indeed there is some problem/bug while reading the datasets_info.json file: there is a mismatch with the config.name keys in the file...
In the meanwhile, in order to be able to use the datasets_info.json file content, you can create the builder without passing the name :

In [25]: builder = datasets.load_dataset_builder("journalists_questions")
In [26]: builder.info.splits
Out[26]: {'train': SplitInfo(name='train', num_bytes=342296, num_examples=10077, dataset_name='journalists_questions')}

After regenerating the metadata JSON file for this dataset, I get the right key:

{"plain_text": {"description": "The journalists_questions corpus (

[severo]

Thanks!