floors-cities-targz.py

# Lint as: python3
"""Residential Floorplans and City Scapes dataset for Urban planning"""


import json
import datasets


logger = datasets.logging.get_logger(__name__)


_CITATION = """\
@article{2016arXiv160605250R,
       author = {{Rajpurkar}, Pranav and {Zhang}, Jian and {Lopyrev},
                 Konstantin and {Liang}, Percy},
        title = "{SQuAD: 100,000+ Questions for Machine Comprehension of Text}",
      journal = {arXiv e-prints},
         year = 2016,
          eid = {arXiv:1606.05250},
        pages = {arXiv:1606.05250},
archivePrefix = {arXiv},
       eprint = {1606.05250},
}
"""

_DESCRIPTION = """\
    Text-to-image dataset, fine tuned specifically for architectural planning
"""

_URL = "https://github.com/RamyaRamachandran-driod/HF_datasets/raw/main/dataset.tar.gz"



class FloorplansTargz(datasets.GeneratorBasedBuilder):

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                # Option to use any Apache arrow feature other than "string"
                {
                    "id": datasets.Value("string"),
                    "image": datasets.Value("string"),
                    "text": datasets.Value("string"),
                    # "prompt": datasets.Value("string"),   (optional)
                }
            ),
            # No default supervised_keys (as we have to pass both question
            # and context as input).
            supervised_keys=None,
            homepage="https://github.com/RamyaRamachandran-driod/HF_datasets/",
            citation=_CITATION,
            
        )

    def _split_generators(self, dl_manager):
        #download manager - hf utility
        path = dl_manager.download_and_extract(_URL)

        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN
                                    , gen_kwargs={"filepath": path+"/train.jsonl"}
                                    ),
        ]

    def _generate_examples(self, filepath):
        """This function returns the examples in the raw (text) form."""
        idx = 0
        #open the file and read lines
        with open(filepath, encoding="utf-8") as fp:
            for line in fp:
                #load json line
                obj = json.loads(line)
                yield idx, obj
                idx += 1