Kaggle / kaggle-api

Official Kaggle API
Apache License 2.0
6.27k stars 1.1k forks source link

Dataset version creation error: Incompatible Dataset Type #255

Open buzibu opened 4 years ago

buzibu commented 4 years ago

I can't update existing dataset. The file is uploaded but this error appeared:

Starting upload for file full_data.csv 100%|███████████████████████████████████████████████████████████████████████████████| 182k/182k [00:05<00:00, 35.2kB/s] Upload successful: full_data.csv (182KB) Dataset version creation error: Incompatible Dataset Type

kevinlu1248 commented 4 years ago

I'm having the same issue uploading a .tar.gz file

neomatrix369 commented 4 years ago

I;m now getting this issue as well!

Starting upload for file googleplaystore.csv
100%|███████████████████████████████████████| 1.30M/1.30M [00:02<00:00, 476kB/s]
Upload successful: googleplaystore.csv (1MB)
Starting upload for file profiled_Translated_Review.csv
100%|██████████████████████████████████████| 10.8M/10.8M [00:02<00:00, 4.27MB/s]
Upload successful: profiled_Translated_Review.csv (11MB)
Starting upload for file googleplaystore_user_reviews.csv
100%|██████████████████████████████████████| 7.31M/7.31M [00:03<00:00, 2.39MB/s]
Upload successful: googleplaystore_user_reviews.csv (7MB)
Dataset version creation error: Incompatible Dataset Typ

Although when I use the API programmatically I get this:

Starting upload for file googleplaystore.csv
100%|██████████| 1.30M/1.30M [00:03<00:00, 427kB/s]
  0%|          | 0.00/10.8M [00:00<?, ?B/s]
Upload successful: googleplaystore.csv (1MB)
Starting upload for file profiled_Translated_Review.csv
100%|██████████| 10.8M/10.8M [00:04<00:00, 2.57MB/s]
  0%|          | 0.00/7.31M [00:00<?, ?B/s]
Upload successful: profiled_Translated_Review.csv (11MB)
Starting upload for file googleplaystore_user_reviews.csv
100%|██████████| 7.31M/7.31M [00:01<00:00, 4.05MB/s]
Upload successful: googleplaystore_user_reviews.csv (7MB)

and then

https://www.kaggle.com/neomatrix369/google-play-store-apps-extended

Although the dataset isn't updated at all.

neomatrix369 commented 4 years ago

@kevinlu1248 @buzibu I think I know what the cause of the error is. It's how you created your Dataset in the first place, there is a bug there but I'm not able to pin it. After I recreated my dataset from scratch and uploaded the csv files manually from my local machine, and then ran the Kaggle API (programmatically and via CLI) they both worked.

Starting upload for file googleplaystore.csv
100%|███████████████████████████████████████| 1.30M/1.30M [00:02<00:00, 675kB/s]
Upload successful: googleplaystore.csv (1MB)
Starting upload for file profiled_Translated_Review.csv
100%|██████████████████████████████████████| 10.8M/10.8M [00:01<00:00, 5.86MB/s]
Upload successful: profiled_Translated_Review.csv (11MB)
Starting upload for file googleplaystore_user_reviews.csv
100%|██████████████████████████████████████| 7.31M/7.31M [00:01<00:00, 4.89MB/s]
Upload successful: googleplaystore_user_reviews.csv (7MB)
Dataset version is being created. Please check progress at https://www.kaggle.com/neomatrix369/google-play-store-apps-extended
neomatrix369 commented 4 years ago

Although I found another issue. If you don't update the dataset-metadata.json in this fashion:

dataset_metadata = kaggle.api.metadata_get(OWNER_SLUG, DATASET_SLUG) # this download a json with two incomplete fields
dataset_metadata['id'] = dataset_metadata["ownerUser"] + "/" + dataset_metadata['datasetSlug']
dataset_metadata['id_no'] = dataset_metadata['datasetId']

You will get other errors that might seem hard to debug in the beginning.