java-8m-methods-v2.py

import os
import sys 
import csv
import datasets
csv.field_size_limit(sys.maxsize)

_DESCRIPTION = "Java 8M Methods :: A collection 8 million java methods (Version 2) [PROBE COMPATIBLE]"
_CITATION    = "NOT AVAILABLE"
_HOMEPAGE    = "NOT AVAILABLE"
_LICENSE     = "MIT"


_BASE_TRAIN_FILE_URL = "https://huggingface.co/datasets/anjandash/java-8m-methods-v2/resolve/main/train.csv"
_BASE_VALID_FILE_URL = "https://huggingface.co/datasets/anjandash/java-8m-methods-v2/resolve/main/valid.csv"
_BASE_TEST_FILE_URL  = "https://huggingface.co/datasets/anjandash/java-8m-methods-v2/resolve/main/test.csv"

_URLS = {
    "train": _BASE_TRAIN_FILE_URL,
    "valid": _BASE_VALID_FILE_URL,
    "test":  _BASE_TEST_FILE_URL             
}

class Java8mMethodsV2(datasets.GeneratorBasedBuilder): # or datasets.DatasetBuilder
    """Java 8M Methods (Version 2) [PROBE COMPATIBLE]"""

    def _info(self):
        """Returns Info"""
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "id": datasets.Value("string"),
                    "text": datasets.Value("string"),
                }
            ),
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators"""
        data_file = dl_manager.download(_URLS)
        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN,      gen_kwargs={"filepath": data_file["train"]}),
            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": data_file["valid"]}),
            datasets.SplitGenerator(name=datasets.Split.TEST,       gen_kwargs={"filepath": data_file["test"] }),
        ]     


    def _generate_examples(self, filepath):
        """Yields Examples"""
        with open(filepath, encoding="utf-8") as f:
            reader = csv.reader(f)
            for id_, row in enumerate(reader):
                if id_ == 0:
                    continue
                yield id_, {
                    "id": row[0],
                    "text": row[1],
                }