File size: 2,428 Bytes
0ae3740 8af70ea 0ae3740 8af70ea 0ae3740 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 |
import datasets
import pyarrow as pa
import pyarrow.parquet as pq
logger = datasets.utils.logging.get_logger(__name__)
_URLS = { "train": "https://huggingface.co/datasets/moska/test_parquet/resolve/main/data/example.parquet" }
class ParquetDatasetConfig(datasets.BuilderConfig):
"""BuilderConfig """
def __init__(self, **kwargs):
"""BuilderConfig.
Args:
**kwargs: keyword arguments forwarded to super.
"""
super(ParquetDatasetConfig, self).__init__(**kwargs)
class ParquetDataset(datasets.ArrowBasedBuilder):
BUILDER_CONFIGS = [
ParquetDatasetConfig(
name="parquet",
version=datasets.Version(version="0.0.1"),
description=f"test_parquet dataset.",
)
]
def _info(self):
return datasets.DatasetInfo(
# This is the description that will appear on the datasets page.
description="reading parquet format.",
# This defines the different columns of the dataset and their types
features=datasets.Features(
{ "pop_est": datasets.Value(dtype="float64"),
"continent": datasets.Value(dtype="string"),
"name": datasets.Value(dtype="string"),
"iso_a3": datasets.Value(dtype="string"),
"gdp_md_est": datasets.Value(dtype="int64"),
"geometry": datasets.Value("binary"),
# These are the features of your dataset like images, labels ...
}
),
)
def _split_generators(self, dl_manager: datasets.download.DownloadManager):
files = _URLS[self.config.name]
downloaded_files = dl_manager.download(files)
return [
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={'filepath': downloaded_files['train']})
]
# method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
def _generate_tables(self, filepaths: list[str]):
for file_idx, filepath in enumerate(filepaths):
with open(filepath, mode="rb") as f:
parquet_file = pq.ParquetFile(source=filepath)
for batch_idx, record_batch in enumerate(parquet_file.iter_batches()):
pa_table = pa.Table.from_batches([record_batch])
yield f"{file_idx}_{batch_idx}", pa_table
|