import os
import datasets


logger = datasets.logging.get_logger(__name__)

ID_POOL = ()
URL = "https://huggingface.co/datasets/thewall/DeepBindWeight/resolve/main"


class DeepBindWeightConfig(datasets.BuilderConfig):
    pass


class DeepBindWeight(datasets.GeneratorBasedBuilder):
    BUILDER_CONFIGS = [
        DeepBindWeightConfig(name=key) for key in ID_POOL
    ]

    DEFAULT_CONFIG_NAME = "params"

    def _info(self):
        return datasets.DatasetInfo(
            features=datasets.Features(
                {
                    "config": datasets.Value("string"),
                    "existed": datasets.Value("bool"),
                    "selex": datasets.Value("string"),
                    "tf": datasets.Value("string")
                }
            ),
            homepage="http://tools.genes.toronto.edu/deepbind",
        )

    def _split_generators(self, dl_manager):
        param_url = f"{URL}/params.tar.gz"
        selex_url = f"{URL}/ERP001824-deepbind.xlsx"
        tf_url = f"{URL}/ERP001824-UniprotKB.xlsx"
        aptani2_url = f"{URL}/aptani2_config.tar.gz"
        downloaded_files = [os.path.join(f"{dl_manager.download_and_extract(param_url)}", "params")]
        downloaded_files.extend(dl_manager.download([selex_url, tf_url]))
        downloaded_files.append(f"{dl_manager.download_and_extract(aptani2_url)}")
        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloaded_files}),
        ]

    def _generate_examples(self, filepath):
        """This function returns the examples in the raw (text) form."""
        logger.info("generating examples from = %s", filepath)
        flag = True
        for file in filepath:
            flag = flag and os.path.exists(file)
        
        yield 0, {"config": filepath[0],
                  "existed": flag,
                  "selex": filepath[1],
                  'tf': filepath[2]}


if __name__=="__main__":
    from datasets import load_dataset
    dataset = load_dataset("thewall/deepbindweight", split="all")
    # dataset.push_to_hub("thewall/DeepBindWeight")