"""Push HEMATO configs (cls, all, malaria) to one Hugging Face repo as Parquet; HF_TOKEN from env."""
import csv, os, sys
from datasets import Dataset, Features, Value, ClassLabel, Image

REPO = 'KeremTT/hemato'
R = list(csv.DictReader(open('meta/manifest.csv', encoding='utf-8-sig')))
for cfg in sys.argv[1:] or ['cls', 'all', 'malaria']:
    rs = [r for r in R if r['config'] == cfg]
    names = sorted({r['label'] for r in rs})
    feats = Features({'image': Image(), 'label': ClassLabel(names=names), 'label_tr': Value('string'), 'source': Value('string'),
                      'original_label': Value('string'), 'patient_id': Value('string'), 'source_license': Value('string'), 'file': Value('string')})
    def gen(split):
        for r in rs:
            if r['split'] == split:
                yield {'image': os.path.join('dataset', r['file']), 'label': r['label'], 'label_tr': r['label_tr'], 'source': r['source'],
                       'original_label': r['original_label'], 'patient_id': r['patient_id'], 'source_license': r['source_license'],
                       'file': r['file'].split('/', 1)[1]}
    for hf_split, split in (('train', 'train'), ('validation', 'val'), ('test', 'test')):
        ds = Dataset.from_generator(gen, gen_kwargs={'split': split}, features=feats, cache_dir='hf_cache')
        ds.push_to_hub(REPO, config_name=cfg, split=hf_split, private=False, max_shard_size='500MB',
                       commit_message=f'HEMATO v1.0: {cfg}/{hf_split}')
        print('PUSHED', cfg, hf_split, len(ds), flush=True)