dataset/dataset.py

import json
import os
import datasets

# Define the different configurations (languages)
_LANGUAGES = ["java", "kt", "py"]

class BugLocalizationConfig(datasets.BuilderConfig):
    """BuilderConfig for BugLocalization dataset"""
    def __init__(self, **kwargs):
        """BuilderConfig for BugLocalization.
        Args:
            **kwargs: keyword arguments forwarded to super.
        """
        super(BugLocalizationConfig, self).__init__(**kwargs)


class BugLocalization(datasets.GeneratorBasedBuilder):
    """BugLocalization dataset"""
    BUILDER_CONFIGS = [
        BugLocalizationConfig(name=lang, description=f"BugLocalization dataset for {lang}.")
        for lang in _LANGUAGES
    ]
    def _info(self):
        sample_file = os.path.join(os.path.abspath(os.path.dirname(__file__)), f"{self.config.name}.jsonl")
        
        # Infer schema from a sample
        schema = self._infer_schema(sample_file)
        
        return datasets.DatasetInfo(
            features=datasets.Features(schema),
        )

    def _infer_schema(self, sample_file, num_samples=100):
        """Infer schema from a sample of JSONL data."""
        schema = {}
        with open(sample_file, encoding="utf-8") as f:
            for line in f:
                data = json.loads(line)
                for key, value in data.items():
                    if key not in schema:
                        schema[key] = datasets.Value('string')  # Assuming all values are strings, tweak as needed
        return schema

    def _split_generators(self, dl_manager):
        # Path where the dataset files are located
        data_dir = os.path.abspath(os.path.dirname(__file__))

        # Define path to each of the JSONL files
        paths = {
            "python": os.path.join(data_dir, "python.jsonl"),
            "java": os.path.join(data_dir, "java.jsonl"),
            "kt": os.path.join(data_dir, "kotlin.jsonl"),
        }

        # Return the split with the appropriate file path
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepath": paths[self.config.name]
                },
            )
        ]

    def _generate_examples(self, filepath):
        """Generate examples from a JSONL file."""
        with open(filepath, encoding="utf-8") as f:
            for id_, line in enumerate(f):
                data = json.loads(line)
                yield id_, data