import json
import datasets
import pandas as pd

logger = datasets.logging.get_logger(__name__)

labels = pd.read_csv('/Users/Atharv/All scripts/Project/Multi label classification/train_classes.csv')

_URL = "https://huggingface.co/datasets/subhuatharva/amazon_from_space/resolve/main/train-jpg.tar.gz?download=true"


class amazon_from_space(datasets.GeneratorBasedBuilder):
    """Planet Dataset: Understanding Amazon Rainforest from Space"""

    def _info(self):
        return datasets.DatasetInfo(
            description = _DESCRIPTION,
            features = datasets.Features(
                {
                    "image": datasets.Image(),
                    "labels": datasets.Value("string"),
                }
            ),
            supervised_keys = None,
            homepage = "https://huggingface.co/datasets/subhuatharva/amazon_from_space",
        )

    def _split_generators(self, dl_manager):
        path = dl_manager.download_and_extract(_URL)
        image_iters = dl_manager.iter_archive(path)
        
        return [
            datasets.SplitGenerator(
                name = datasets.Split.TRAIN,
                gen_kwargs = {
                    "images": image_iters
                }
            ),
        ]
    def _generate_examples(self, images):
        """This function returns the examples in the raw (text) form."""

        idx = 0

        # iterate through images
        for filepath, image in images:
            yield idx, {
                "image": {"filepath": filepath, "image": image.read()},
                "text": labels['tags'][idx],
            }
            
            idx+=1