cicero-bonorum-et-malorum.py

import os
import os.path

from datasets import DatasetInfo, GeneratorBasedBuilder, SplitGenerator, Features, Value


FILE_PATHS = {
    "book1": "fin1.txt",
    "book2": "fin2.txt",
    "book3": "fin3.txt",
    "book4": "fin4.txt",
    "book5": "fin5.txt"
}

class CiceroDataset(GeneratorBasedBuilder):
    
    def _info(self):
        return DatasetInfo(
            features=Features({
                'book': Value('string'),
                'line_number': Value('int32'),
                'text': Value('string'),
            })
        )
    
    def _generate_examples(self, filepath):
        book_name = os.path.splitext(os.path.basename(filepath))[0]  # e.g., 'book1'
        with open(filepath, encoding='utf-8') as f:
            for line_number, line in enumerate(f, start=1):
                yield f'{book_name}_{line_number}', {
                    'book': book_name,
                    'line_number': line_number,
                    'text': line.strip()
                }
    
    def _split_generators(self, dl_manager):
        return [
            SplitGenerator(
                name=key,
                gen_kwargs={"filepath": os.path.join(self.base_path, FILE_PATHS[key])}
            ) for key in FILE_PATHS
        ]