mtg_text.py

"""Loading script"""
import datasets

import json

_DESCRIPTION = "Magic the Gathering card text dataset"
"""We have two .json files with the train and validation cards. The
json is basically a list of dictionaries with the following keys:

"card_name": string
"type_line": string
"oracle_text": string
"""

_FILES = {
    "train": "mtg_text/train_cards.json",
    "validation": "mtg_text/val_cards.json",
}


class MTGCardText(datasets.GeneratorBasedBuilder):
    """BuilderConfig for MTGCardText"""

    def __init__(self, **kwargs):
        """BuilderConfig for MTGCardText.
        Args:
          **kwargs: keyword arguments forwarded to super.
        """
        super(MTGCardText, self).__init__(**kwargs)

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features({
                "card_name": datasets.Value("string"),
                "type_line": datasets.Value("string"),
                "oracle_text": datasets.Value("string")
            }),
            supervised_keys=None,
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators.
        Args:
            dl_manager: DownloadManager object used to download and extract the
                dataset files.
        Returns:
            `dict<str, SplitGenerator>`.
        """
        downloaded_files = dl_manager.download_and_extract(_FILES)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepath": downloaded_files["train"],
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={
                    "filepath": downloaded_files["validation"],
                },
            ),
        ]

    def _generate_examples(self, filepath):
        """Yields examples.
        Args:
            filepath: a string
        Yields:
            dictionaries containing "card_name", "type_line" and "oracle_text"
        """
        with open(filepath, encoding="utf-8") as f:
            data = json.load(f)
            for id_, row in enumerate(data):
                yield id_, {
                    "card_name": row["card_name"],
                    "type_line": row["type_line"],
                    "oracle_text": row["oracle_text"]
                }