OCR-IDL.py

# coding=utf-8
# Copyright 2020 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# Lint as: python3
"""OCR-IDL: OCR annotations for the Industry Document Library."""


import json

import datasets


logger = datasets.logging.get_logger(__name__)


_CITATION = """\
    @article{biten2022ocr,
      title         = {OCR-IDL: Ocr annotations for industry document library dataset},
      author        = {{Biten}, Ali Furkan and {Tito}, Ruben and {Gomez}, Lluis and {Valveny}, Ernest and {Karatzas}, Dimosthenis},
      journal       = {arXiv preprint arXiv:2202.12985},
      year          = 2022,
      eid           = {arXiv:2202.12985},
      pages         = {arXiv:2202.12985},
      archivePrefix = {arXiv},
      eprint        = {2202.12985},
    }
"""

_DESCRIPTION = """\
    The OCR-IDL Dataset contains the OCR annotations of 26M pages of theIndustry Document Library (IDL).\
    It is specially intended to be used for text-layout self-supervised tasks such as Masked Language Modeling or Text De-noising.\
    However, we also include the url to the documents so that can be downloaded for image-text alignment tasks.
"""

_URL = "http://datasets.cvc.uab.es/UCSF_IDL/"
_PROJECT_URL = "https://github.com/furkanbiten/idl_data"
_URLS = {
    "train": _URL + "train-v1.1.json",
    "dev": _URL + "dev-v1.1.json",
}


class OCRIDLConfig(datasets.BuilderConfig):
    """BuilderConfig for OCR-IDL."""

    def __init__(self, **kwargs):
        """BuilderConfig for OCR-IDL.
        Args:
          **kwargs: keyword arguments forwarded to super.
        """
        super(OCRIDLConfig, self).__init__(**kwargs)


class OCR_IDL(datasets.GeneratorBasedBuilder):
    """SQUAD: The Stanford Question Answering Dataset. Version 1.1."""

    BUILDER_CONFIGS = [
        OCRIDLConfig(
            name="OCR-IDL",
            version=datasets.Version("1.0.0", ""),
            description=_DESCRIPTION,  # This should be the description of the version. Since we have only one, use the same as the global dataset description.
        ),
    ]

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "document_id": datasets.Value("string"),
                    "document_url": datasets.Value("string"),
                    "page_id": datasets.Value("string"),
                    "page_height": datasets.Value("int32"),
                    "page_width": datasets.Value("int32"),
                    "words": [],
                    "boxes": [],
                    "word_lines_id": [],
                    "text_types": [],
                    "recog_conf": []
                }
            ),
            # No default supervised_keys (as we have to pass both question and context as input).
            supervised_keys=None,
            homepage=_PROJECT_URL,
            citation=_CITATION,

        )

    def _split_generators(self, dl_manager):
        data_dir = dl_manager.download('https://huggingface.co/datasets/rubentito/OCR-IDL/resolve/main/val.csv')
#         data_dir = dl_manager.download_and_extract('http://datasets.cvc.uab.es/UCSF_IDL/Samples/imdb_sample_v2.tar.gz')

        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"data": downloaded_files[0]}),
            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloaded_files[0]}),
        ]

    def _generate_examples(self, filepath):
        """This function returns the examples in the raw (text) form."""
        logger.info("generating examples from = %s", filepath)
        key = 0
        with open(filepath, encoding="utf-8") as f:
            squad = json.load(f)
            for article in squad["data"]:
                title = article.get("title", "")
                for paragraph in article["paragraphs"]:
                    context = paragraph["context"]  # do not strip leading blank spaces GH-2585
                    for qa in paragraph["qas"]:
                        answer_starts = [answer["answer_start"] for answer in qa["answers"]]
                        answers = [answer["text"] for answer in qa["answers"]]
                        # Features currently used are "context", "question", and "answers".
                        # Others are extracted here for the ease of future expansions.
                        yield key, {
                            "title": title,
                            "context": context,
                            "question": qa["question"],
                            "id": qa["id"],
                            "answers": {
                                "answer_start": answer_starts,
                                "text": answers,
                            },
                        }
                        key += 1