|
import json |
|
|
|
import datasets |
|
|
|
|
|
logger = datasets.logging.get_logger(__name__) |
|
|
|
|
|
_DESCRIPTION = """\ |
|
Ukrainian Multi30k |
|
""" |
|
|
|
_CITATION = """\ |
|
|
|
""" |
|
|
|
_URLS = { |
|
"train" : "train.json", |
|
"flickr_2016" : "test_2016_flickr.json", |
|
"flickr_2017" : "test_2017_flickr.json", |
|
"flickr_2018" : "test_2018_flickr.json", |
|
"mscoco_2017" : "test_2017_mscoco.json" |
|
} |
|
|
|
|
|
class UkrainianMulti30k(datasets.GeneratorBasedBuilder): |
|
"""Ukrainian Multi30k Dataset""" |
|
VERSION = datasets.Version("0.0.1") |
|
DEFAULT_CONFIG_NAME = "multi30k" |
|
BUILDER_CONFIGS = [ |
|
datasets.BuilderConfig(name="multi30k", version=VERSION, description=""), |
|
datasets.BuilderConfig(name="flickr_2016", version=VERSION, description=""), |
|
datasets.BuilderConfig(name="flickr_2017", version=VERSION, description=""), |
|
datasets.BuilderConfig(name="flickr_2018", version=VERSION, description=""), |
|
datasets.BuilderConfig(name="mscoco_2017", version=VERSION, description=""), |
|
] |
|
|
|
def _info(self): |
|
return datasets.DatasetInfo( |
|
description=_DESCRIPTION, |
|
features=datasets.Features( |
|
{ |
|
"en": datasets.Value("string"), |
|
"uk": datasets.Value("string") |
|
} |
|
), |
|
citation=_CITATION, |
|
) |
|
|
|
|
|
def _split_generators(self, dl_manager): |
|
downloaded_files = dl_manager.download(_URLS) |
|
if self.config.name == "multi30k": |
|
return [ |
|
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": downloaded_files["train"]}) |
|
] |
|
elif self.config.name == "flickr_2016": |
|
return [ |
|
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": downloaded_files["flickr_2016"]}) |
|
] |
|
elif self.config.name == "flickr_2017": |
|
return [ |
|
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": downloaded_files["flickr_2017"]}) |
|
] |
|
elif self.config.name == "flickr_2018": |
|
return [ |
|
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": downloaded_files["flickr_2018"]}) |
|
] |
|
elif self.config.name == "mscoco_2017": |
|
return [ |
|
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": downloaded_files["mscoco_2017"]}) |
|
] |
|
|
|
def _generate_examples(self, filepaths): |
|
|
|
with open(filepaths, encoding="utf-8") as f: |
|
for num, rows_str in enumerate(f): |
|
rows = json.loads(rows_str) |
|
yield num, rows |