jhotdraw.py

"""Jhotdraw"""

import datasets
import pandas as pd
from collections import deque

_CITATION = """

@inproceedings{li2018helpful,

  title={Helpful or Not? An investigation on the feasibility of identifier splitting via CNN-BiLSTM-CRF.},

  author={Li, Jiechu and Du, Qingfeng and Shi, Kun and He, Yu and Wang, Xin and Xu, Jincheng},

  booktitle={SEKE},

  pages={175--174},

  year={2018}

}

"""

_DESCRIPTION = """

In programming languages, identifiers are tokens (also called symbols) which name language entities.

Some of the kinds of entities an identifier might denote include variables, types, labels, subroutines, and packages.



Jhotdraw is a dataset for identifier segmentation, 

i.e. the task of adding spaces between the words on a identifier.

"""
_URL = "https://raw.githubusercontent.com/ruanchaves/hashformers/master/datasets/jhotdraw.txt"

class Jhotdraw(datasets.GeneratorBasedBuilder):

    VERSION = datasets.Version("1.0.0")

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "index": datasets.Value("int32"),
                    "identifier": datasets.Value("string"),
                    "segmentation": datasets.Value("string")
                }
            ),
            supervised_keys=None,
            homepage="",
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        downloaded_files = dl_manager.download(_URL)
        return [
            datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files}),
        ]

    def _generate_examples(self, filepath):

        with open(filepath, "r") as f:

            for idx, line in enumerate(f):
                fields = line.split(":")
                identifier = fields[0].strip()
                segmentation = fields[1].strip()
                yield idx, {
                    "index": idx,
                    "identifier": identifier,
                    "segmentation": segmentation
                }