tmp-one-config.py

import csv

import datasets


_URL = "data.csv"


class TmpOneConfig(datasets.GeneratorBasedBuilder):
    BUILDER_CONFIGS = [datasets.BuilderConfig(name="main")]

    def _info(self):
        return datasets.DatasetInfo(
            features=datasets.Features({"text": datasets.Value("string"), "label": datasets.Value("int32")}),
        )

    def _split_generators(self, dl_manager):
        data_path = dl_manager.download(_URL)
        return [datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"data_path": data_path})]

    def _generate_examples(self, data_path):
        with open(data_path, newline="", encoding="utf-8") as f:
            for idx, row in enumerate(csv.DictReader(f)):
                yield idx, row