loader.py

from pathlib import Path
from typing import List

import datasets
import pdf2image

logger = datasets.logging.get_logger(__name__)

_DESCRIPTION = "A generic pdf folder"


class PdfFolder(datasets.GeneratorBasedBuilder):
    def _info(self):

        folder=None
        if isinstance(self.config.data_files, str):
            folder = self.config.data_files
        elif isinstance(self.config.data_files, dict):
            folder = self.config.data_files.get('train', None)

        if folder is None:
            raise RuntimeError()

        classes = sorted([x.name.lower() for x in Path(folder).glob('*/**')]).unique()
        
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "file": datasets.Sequence(datasets.Image()),
                    "labels": datasets.features.ClassLabel(names=classes)
                }
            ),
            task_templates=None,
        )

    def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:

        data_files = self.config.data_files

        if isinstance(data_files, str):
            return [datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={'archive_path': data_files})]

        splits = []
        for split_name, folder in data_files.items():
            splits.append(datasets.SplitGenerator(name=split_name, gen_kwargs={'archive_path': folder}))

        return splits

    def _generate_examples(self, archive_path):
        labels = self.info.features['labels']
        logger.info("generating examples from = %s", archive_path)
        extensions = {'.pdf'}
        for i, path in enumerate(Path(archive_path).glob('**/*')):
            if path.suffix in extensions:

                images = pdf2image.convert_from_bytes(path.posix())

                #convert PDF to list of images
                yield i, {'file': images, 'labels': labels.encode_example(path.parent.name.lower())}