basic_shapes_10k.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.

"""A dataset consisting of svgs, their png representations, and various masks"""


import csv
import json
import os
from math import floor

import datasets

# Find for instance the citation on arxiv or on the dataset repo/website
_CITATION = """\
@InProceedings{huggingface:dataset,
title = {A dataset for understanding vector graphics},
author={eezy, Inc.
},
year={2023}
}
"""

# You can copy an official description
_DESCRIPTION = """\
This new dataset is designed to provide a corpus for training machine vision tasks on the understanding of basic vector graphics
"""

_HOMEPAGE = "https://eezy.com"

# TODO: Add the licence for the dataset here if you can find it
_LICENSE = "CC-BY"

# TODO: Add link to the official dataset URLs here
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
_URLS = {
    "circles": "https://eezy-data-bucket.s3.amazonaws.com/public-datasets/basic_shapes_10k_v1/circles.tgz",
    "squares": "https://eezy-data-bucket.s3.amazonaws.com/public-datasets/basic_shapes_10k_v1/squares.tgz",
    "squares_and_circles": "https://eezy-data-bucket.s3.amazonaws.com/public-datasets/basic_shapes_10k_v1/squares_and_circles.tgz",
    "scer": "https://eezy-data-bucket.s3.amazonaws.com/public-datasets/basic_shapes_10k_v1/scer.tgz"
}


# TODO: Name of the dataset usually matches the script name with CamelCase instead of snake_case
class BasicShapes10K(datasets.GeneratorBasedBuilder):
    """A dataset consisting of simple vector shapes and various kinds of masks"""

    VERSION = datasets.Version("1.0.0")

    SPLIT_COUNTS = {
        'train': (0, 8000),
        'dev': (8000, 9000),
        'test': (9000, 10000)
    }

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name="mixed", version=VERSION, description="These images are a mixture of all the other datasets"),
        datasets.BuilderConfig(name="circles", version=VERSION, description="These images only contain circles"),
        datasets.BuilderConfig(name="squares", version=VERSION, description="These images only contain squares"),
        datasets.BuilderConfig(name="squares_and_circles", version=VERSION, description="These images contain circles and squares"),
        datasets.BuilderConfig(name="scer", version=VERSION, description="These images contain circles, squares, rectangles, and ellipses"),
    ]

    DEFAULT_CONFIG_NAME = "mixed"  # It's not mandatory to have a default configuration. Just use one if it make sense.

    def _info(self):
        features = datasets.Features(
            {
                "svg": datasets.Value("string"),
                "png": datasets.Image(),
                "layer_mask": datasets.Image(),
                "object_mask": datasets.Image(),
            }
        )
        return datasets.DatasetInfo(
            # This is the description that will appear on the datasets page.
            description=_DESCRIPTION,
            # This defines the different columns of the dataset and their types
            features=features,  # Here we define them above because they are different between the two configurations
            # If there's a common (input, target) tuple from the features, uncomment supervised_keys line below and
            # specify them. They'll be used if as_supervised=True in builder.as_dataset.
            supervised_keys=("png", "layer_mask"),
            # Homepage of the dataset for documentation
            homepage=_HOMEPAGE,
            # License for the dataset if available
            license=_LICENSE,
            # Citation for the dataset
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        # TODO: This method is tasked with downloading/extracting the data and defining the splits depending on the configuration
        # If several configurations are possible (listed in BUILDER_CONFIGS), the configuration selected by the user is in self.config.name

        # dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLS
        # It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
        # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
        if self.config.name == 'mixed':
            urls = _URLS
        else:
            urls = {self.config.name: _URLS[self.config.name]}

        data_dir = dl_manager.download_and_extract(urls)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "data_dir": data_dir,
                    "split": "train",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "data_dir": data_dir,
                    "split": "dev",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "data_dir": data_dir,
                    "split": "test"
                },
            ),
        ]

    # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
    def _generate_examples(self, data_dir, split):
        # The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
        start, stop = self.SPLIT_COUNTS[split]
        domains = [self.config.name]

        if self.config.name == 'mixed':
            start = start * 4
            stop = stop * 4
            domains = [
                'circles',
                'squares',
                'squares_and_circles',
                'scer'
            ]

        divisions = len(domains)

        for key in range(start, stop):
            domain = domains[key % divisions]
            idx = floor(key / divisions)
            yield f'{self.config.name}_{str(key).zfill(6)}', \
                  self._example_for_domain(data_dir, domain, idx, split)


    def _example_for_domain(self, data_dir, domain, idx, split):
        data = {}
        svg_path = os.path.join(data_dir[domain], domain, 'svg', str(idx).zfill(6) + '.svg')
        with open(svg_path, 'r') as file:
            data['svg'] = file.read()

        png_path = os.path.join(data_dir[domain], domain, 'png', str(idx).zfill(6) + '.png')
        with open(png_path, 'rb') as file:
            data['png'] = {"path": png_path, "bytes": file.read()}

        if split != "test":
            layer_mask_path = os.path.join(data_dir[domain], domain, 'layer_mask', str(idx).zfill(6) + '.png')
            with open(layer_mask_path, 'rb') as file:
                data['layer_mask'] = {"path": layer_mask_path, "bytes": file.read()}
            object_mask_path = os.path.join(data_dir[domain], domain, 'obj_mask', str(idx).zfill(6) + '.png')
            with open(object_mask_path, 'rb') as file:
                data['object_mask'] = {"path": object_mask_path, "bytes": file.read()}
        else:
            data['layer_mask'] = ''
            data['object_mask'] = ''
        return data