schemapile.py

import datasets
import json

class SchemapileConfig(datasets.BuilderConfig):
    def __init__(self, **kwargs):
        super().__init__(**kwargs)

class Schemapile(datasets.GeneratorBasedBuilder):
    BUILDER_CONFIGS = [
        SchemapileConfig(name="default", version=datasets.Version("1.0.0"), description="Schemapile dataset"),
    ]

    def _info(self):
        return datasets.DatasetInfo(
            description="Schemapile dataset",
            features=datasets.Features({
            "INFO": {
                "URL": datasets.Value("string"),
                "LICENSE": datasets.Value("string"),
                "PERMISSIVE": datasets.Value("bool"),
                "ID": datasets.Value("string"),
                "FILENAME": datasets.Value("string"),
            },
            "LICENSE": datasets.Value("string"),
            "PERMISSIVE": datasets.Value("bool"),
            "TABLES": datasets.Sequence({
                "TABLE_NAME": datasets.Value("string"),
                "COLUMNS": datasets.Sequence({
                "NAME": datasets.Value("string"),
                "TYPE": datasets.Value("string"),
                "NULLABLE": datasets.Value("bool"),
                "UNIQUE": datasets.Value("bool"),
                "DEFAULT": datasets.Value("string"),
                "CHECKS": datasets.Sequence(datasets.Value("string")),
                "IS_PRIMARY": datasets.Value("bool"),
                "IS_INDEX": datasets.Value("bool"),
                "VALUES": datasets.Sequence(datasets.Value("string")),
                }),
                "PRIMARY_KEYS": datasets.Sequence(datasets.Value("string")),
                "FOREIGN_KEYS": datasets.Sequence({
                "COLUMNS": datasets.Sequence(datasets.Value("string")),
                "FOREIGN_TABLE": datasets.Value("string"),
                "REFERRED_COLUMNS": datasets.Sequence(datasets.Value("string")),
                "ON_DELETE": datasets.Value("string"),
                "ON_UPDATE": datasets.Value("string"),
                }),
                "CHECKS": datasets.Sequence(datasets.Value("string")),
                "INDEXES": datasets.Sequence(datasets.Value("string")),
            }),
            }),
            supervised_keys=None,
        )

    def _split_generators(self, dl_manager):
        data_files = getattr(self.config, "data_files", None)
        data_path = dl_manager.download_and_extract(data_files or "data.jsonl")

        return [
            datasets.SplitGenerator(
                name="full",   # or any non-reserved name
                gen_kwargs={"filepath": data_path},
            )
        ]

    def _generate_examples(self, filepath):
        with open(filepath, encoding="utf-8") as f:
            for idx, line in enumerate(f):
                data = json.loads(line)
                yield idx, data