generator.py

# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# 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.


import csv
from pathlib import Path

import datasets
import pandas as pd


# Add BibTeX citation
# Find for instance the citation on arxiv or on the dataset repo/website
_CITATION = """
@inproceedings{fersini2022semeval,
    title={SemEval-2022 Task 5: Multimedia automatic misogyny identification},
    author={Fersini, Elisabetta and
            Gasparini, Francesca and
            Rizzi, Giulia and
            Saibene, Aurora and
            Chulvi, Berta and
            Rosso, Paolo and
            Lees, Alyssa and
            Sorensen, Jeffrey},
    booktitle={Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022)},
    pages={533--549},
    year={2022}
}
"""

# Add description of the dataset here
# You can copy an official description
_DESCRIPTION = """These are the datasets for Multimodal Misogyny Detection (MAMI), Task 5 of SemEval-2022."""

# Add a link to an official homepage for the dataset here
_HOMEPAGE = "https://competitions.codalab.org/competitions/34175"

# Add the licence for the dataset here if you can find it
_LICENSE = ""

# Add link to the official dataset URLs here
# The HuggingFace dataset library don't host the datasets but only point to the original files
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
_URLs = ""

_HEADER = ["file_name", "misogynous", "shaming", "stereotype", "objectification", "violence", "Text Transcription"]


class SemEval2022Task5(datasets.GeneratorBasedBuilder):
    """These are the datasets for Multimodal Misogyny Detection (MAMI), Task 5 of SemEval-2022."""

    def __init__(self, training_dir=None, test_dir=None, validation_dir=None, **kwargs):
        super().__init__(**kwargs)
        assert training_dir is not None, "Training directory must be specified"
        assert test_dir is not None, "Test directory must be specified"

        self.training_dir = training_dir
        self.test_dir = test_dir
        self.validation_dir = validation_dir

        # Ensure that labels are correctly set up
        train_csv = pd.read_csv(Path(self.training_dir, "training.csv"), delimiter="\t", encoding="utf-8-sig")
        assert train_csv.columns.tolist() == _HEADER, (
            f"Training header is not correct. Expected: {_HEADER}, got: {train_csv.columns.tolist()}"
        )

        try:
            test_csv = pd.read_csv(Path(self.test_dir, "test_with_labels.csv"), delimiter="\t", encoding="utf-8-sig")
            assert test_csv.columns.tolist() == _HEADER, (
                f"Test header is not correct. Expected: {_HEADER}, got: {test_csv.columns.tolist()}"
            )

        except FileNotFoundError:
            test_csv = pd.read_csv(Path(self.test_dir, "test.csv"), delimiter="\t", encoding="utf-8-sig")
            assert test_csv.columns.tolist() == _HEADER[:1] + _HEADER[-1:], (
                f"Test Header is not correct. Expected: {_HEADER[:1] + _HEADER[-1:]}, got: {test_csv.columns.tolist()}"
            )

            labels = pd.read_csv(Path(self.test_dir, "test_labels.txt"), delimiter="\t", header=None)
            labels.columns = _HEADER[:-1]

            test_with_labels = pd.merge(labels, test_csv, on=_HEADER[0])

            assert len(test_with_labels) == len(test_csv)
            assert test_with_labels.columns.tolist() == _HEADER

            test_with_labels.to_csv(Path(self.test_dir, "test_with_labels.csv"), index=False, sep="\t")

        if self.validation_dir is not None:
            validation_csv = pd.read_csv(Path(self.validation_dir, "trial.csv"), delimiter="\t", encoding="utf-8-sig")
            assert validation_csv.columns.tolist() == _HEADER, (
                f"Validation header is not correct. Expected: {_HEADER}, got: {validation_csv.columns.tolist()}"
            )

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "image_path": datasets.Value("string"),
                    "image_name": datasets.Value("string"),
                    "misogynous": datasets.Value("int32"),
                    "shaming": datasets.Value("int32"),
                    "stereotype": datasets.Value("int32"),
                    "objectification": datasets.Value("int32"),
                    "violence": datasets.Value("int32"),
                    "text": datasets.Value("string"),
                },
            ),
        )

    def _split_generators(self, dl_manager):
        splits = [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "images_dir": self.training_dir,
                    "metadata": Path(self.training_dir, "training.csv").resolve().as_posix(),
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={
                    "images_dir": self.test_dir,
                    "metadata": Path(self.test_dir, "test_with_labels.csv").resolve().as_posix(),
                },
            ),
        ]
        if self.validation_dir is not None:
            splits.append(
                datasets.SplitGenerator(
                    name=datasets.Split.VALIDATION,
                    gen_kwargs={
                        "images_dir": self.validation_dir,
                        "metadata": Path(self.validation_dir, "trial.csv").resolve().as_posix(),
                    },
                )
            )
        return splits

    def _generate_examples(self, images_dir, metadata):
        with open(metadata, "r", encoding="utf-8-sig") as f:
            reader = csv.reader(f, delimiter="\t")
            next(reader)  # skip header
            for id_, row in enumerate(reader):
                yield (
                    id_,
                    {
                        "image_path": Path(images_dir, row[0]).resolve().as_posix(),
                        "image_name": row[0],
                        "misogynous": row[1],
                        "shaming": row[2],
                        "stereotype": row[3],
                        "objectification": row[4],
                        "violence": row[5],
                        "text": row[6],
                    },
                )