unit-test_PDFfolder.py

import os
from typing import List

import datasets
import pdf2image

logger = datasets.logging.get_logger(__name__)

_DESCRIPTION = "A generic pdf folder"

_CLASSES = ["categoryA", "categoryB"]  # define in advance

_URL = "https://huggingface.co/datasets/jordyvl/unit-test_PDFfolder/resolve/main/data/data.tar.gz"

# folder
#   train
#       categoryA
#           file1
#   test
# ...


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

        archive_path = dl_manager.download(_URL)

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "archive_iterator": dl_manager.iter_archive(archive_path),
                    "supposed_labelset": "train",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={
                    "archive_iterator": dl_manager.iter_archive(archive_path),
                    "supposed_labelset": "test",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={
                    "archive_iterator": dl_manager.iter_archive(archive_path),
                    "supposed_labelset": "val",
                },
            ),
        ]

    def _generate_examples(self, archive_iterator, supposed_labelset):

        extensions = {"pdf", "PDF"}
        for file_path, file_obj in archive_iterator:

            if file_path.split(".")[-1] not in extensions:  # metadata.jsonlines
                continue

            folder, labelset, label, filename = file_path.split("/")
            if labelset != supposed_labelset:
                continue

            images = pdf2image.convert_from_bytes(file_obj.read())

            yield file_path, {"file": images, "labels": label}