Hindi_Trec_Covid.py

import json
import datasets

_CITATION = """\
@inproceedings{your_citation,
  title={Hindi_Trec_Covid},
  author={Arkadeep Acharya},
  year={2025},
  url={https://huggingface.co/datasets/AIhnIndicRag/Hindi_Trec_Covid}
}
"""

_DESCRIPTION = """\
Your dataset description goes here.
"""

_HOMEPAGE = "https://your-dataset-homepage"

# URLs for the dataset files
_URLs = {
    "default": "https://huggingface.co/datasets/AIhnIndicRag/Hindi_Trec_Covid/resolve/main/qrels/test.tsv",
    "corpus": "https://huggingface.co/datasets/AIhnIndicRag/Hindi_Trec_Covid/resolve/main/corpus.jsonl",
    "queries": "https://huggingface.co/datasets/AIhnIndicRag/Hindi_Trec_Covid/resolve/main/queries.jsonl",
}

class MyDataset(datasets.GeneratorBasedBuilder):
    """My custom dataset."""

    VERSION = datasets.Version("1.0.0")

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name="default", version=VERSION, description="Default configuration"),
        datasets.BuilderConfig(name="corpus", version=VERSION, description="Corpus configuration"),
        datasets.BuilderConfig(name="queries", version=VERSION, description="Queries configuration"),
    ]

    DEFAULT_CONFIG_NAME = "default"

    def _info(self):
        if self.config.name == "default":
            features = datasets.Features({
                "query-id": datasets.Value("string"),
                "corpus-id": datasets.Value("string"),
                "score": datasets.Value("float64"),
            })
        elif self.config.name == "queries":
            features = datasets.Features({
                "_id": datasets.Value("string"),
                "text": datasets.Value("string"),
            })
        elif self.config.name == "corpus":
            features = datasets.Features({
                "_id": datasets.Value("string"),
                "title": datasets.Value("string"),
                "text": datasets.Value("string"),
                "metadata": {
                    "url": datasets.Value("string"),
                    "pubmed_id": datasets.Value("string"),
                },
            })
        else:
            raise ValueError(f"Unknown config name: {self.config.name}")

        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            supervised_keys=None,
            homepage=_HOMEPAGE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        if self.config.name not in _URLs:
            raise ValueError(f"Unknown config name: {self.config.name}")

        # Download the data file
        data_file = dl_manager.download_and_extract(_URLs[self.config.name])

        # Define custom split names
        split_name = datasets.Split.TEST if self.config.name == "default" else self.config.name

        return [
            datasets.SplitGenerator(
                name=split_name,
                gen_kwargs={"file_path": data_file},
            ),
        ]

    def _generate_examples(self, file_path):
        if self.config.name == "default":
            with open(file_path, encoding="utf-8") as f:
                next(f)
                for id_, line in enumerate(f):
                    query_id, corpus_id, score = line.strip().split("\t")
                    yield id_, {
                        "query-id": query_id,
                        "corpus-id": corpus_id,
                        "score": float(score),
                    }
        else:
            with open(file_path, encoding="utf-8") as f:
                for id_, line in enumerate(f):
                    data = json.loads(line)
                    if self.config.name == "corpus":
                        yield id_, {
                            "_id": data["_id"],
                            "title": data["title"],
                            "text": data["text"],
                            "metadata": {
                                "url": data.get("metadata", {}).get("url", ""),
                                "pubmed_id": data.get("metadata", {}).get("pubmed_id", ""),
                            },
                        }
                    elif self.config.name == "queries":
                        yield id_, {
                            "_id": data["_id"],
                            "text": data["text"],
                        }