open-license-corpus.py

import datasets
import os
from datasets import load_dataset
import gzip
import json

logger = datasets.logging.get_logger(__name__)


CITATION = """
"""

DESCRIPTION = """
The Open License Corpus
"""

OLC_SUBSET_NAMES = [
    "ccby_law",
    "ccby_s2orc",
    "ccby_stackexchange",
    "ccby_stackoverflow",
    "ccby_wikinews",
    "ccby_wikipedia",
    "pd_arxiv_abstracts",
    "pd_books",
    "pd_law",
    "pd_news",
    "pd_s2orc",
    "sw_amps_math",
    "sw_dm_math",
    "sw_github",
    "sw_hackernews",
    "sw_ubuntu_irc"
] 

URL = "https://huggingface.co/datasets/kernelmachine/open-license-corpus/"



N_SHARDS_PER_SPLIT = {
    "ccby_s2orc": {"train": 5000},
    "ccby_law": {"train": 50},
    "ccby_stackexchange": {"train": 1500},
    "ccby_stackoverflow": {"train": 750},
    "ccby_wikinews": {"train": 42},
    "ccby_wikipedia": {"train": 3000},
    "pd_arxiv_abstracts": {"train": 1},
    "pd_books": {"train": 150},
    "pd_law": {"train": 2000},
    "pd_news": {"train": 10},
    "pd_s2orc": {"train": 30},
    "sw_amps_math": {"train": 2},
    "sw_dm_math": {"train": 239},
    "sw_github": {"train": 2500},
    "sw_hackernews": {"train": 16},
    "sw_ubuntu_irc": {"train": 27}
}

#DATA_URL = 'https://huggingface.co/datasets/kernelmachine/open-license-corpus/blob/main/data/{name}/{split}-{index:05d}-of-{n_shards:05d}.jsonl.gz'
DATA_URL = 'https://huggingface.co/datasets/kernelmachine/open-license-corpus/resolve/main/data/{name}/{split}-{index:05d}-of-{n_shards:05d}.jsonl.gz'

class OpenLicenseCorpusConfig(datasets.BuilderConfig):
    def __init__(self, features, citation, **kwargs):
        super().__init__(**kwargs)


class OpenLicenseCorpus(datasets.GeneratorBasedBuilder):

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name=name)
        for name in OLC_SUBSET_NAMES
    ]

    def _info(self):
        return datasets.DatasetInfo(
            description=DESCRIPTION,
            features=datasets.Features(
                {
                    "text": datasets.Value("string"),
                }
            ),
            supervised_keys=None,
            homepage=URL,
            citation=CITATION,
        )

    def _split_generators(self, dl_manager):
        data_urls = {}
        for split in ["train"]:
            n_shards = N_SHARDS_PER_SPLIT[self.config.name][split] - 1
            data_urls[split] = [
                DATA_URL.format(name=self.config.name, split=split, index=index, n_shards=n_shards)
                for index in range(n_shards)
            ]

        train_downloaded_files = dl_manager.download(data_urls["train"])
        
        return [datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": train_downloaded_files})]

    def _generate_examples(self, filepaths):
        """This function returns the examples in the raw (text) form by iterating on all the files."""
        id_ = 0
        for filepath in filepaths:
            logger.info("generating examples from = %s", filepath)
            with gzip.open(open(filepath, "rb"), "rt", encoding="utf-8") as f:
                for line in f:
                    if line:
                        example = json.loads(line)
                        yield id_, example
                        id_ += 1