|
from datasets import DatasetBuilder, SplitGenerator, DownloadConfig |
|
from pathlib import Path |
|
import json |
|
|
|
|
|
URL = "ukrainian-stackexchange.json" |
|
|
|
|
|
class UkrainianStackExchange(DatasetBuilder): |
|
|
|
VERSION = "1.0.0" |
|
BUILDER_CONFIGS = [ |
|
{ |
|
"name": "ukrainian-stackexchange", |
|
"version": VERSION, |
|
"description": "Ukrainian StackExchange Dataset collected from https://ukrainian.stackexchange.com/" |
|
} |
|
] |
|
|
|
def _info(self): |
|
return datasets.DatasetInfo( |
|
description=_DESCRIPTION, |
|
features=datasets.Features({ |
|
"question": { |
|
"id": datasets.Value("string"), |
|
"title": datasets.Value("string"), |
|
"text": datasets.Value("string"), |
|
"upvoteCount": datasets.Value("int32"), |
|
"tags": datasets.Sequence(datasets.Value("string")), |
|
}, |
|
"answers": datasets.Sequence({ |
|
"id": datasets.Value("string"), |
|
"text": datasets.Value("string"), |
|
"upvoteCount": datasets.Value("int32"), |
|
"isAccepted": datasets.Value("bool"), |
|
}), |
|
"url": datasets.Value("string"), |
|
}), |
|
supervised_keys=None, |
|
) |
|
|
|
def _split_generators(self, dl_manager): |
|
path_to_file = Path(URL) |
|
return [ |
|
SplitGenerator( |
|
name=datasets.Split.TRAIN, |
|
gen_kwargs={ |
|
"filepath": path_to_file, |
|
}, |
|
), |
|
] |
|
|
|
def _generate_examples(self, filepath): |
|
with open(filepath, "r", encoding="utf-8") as f: |
|
json_data = json.load(f) |
|
|
|
for idx, data in enumerate(json_data): |
|
yield idx, data |
|
|