# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# TODO: Address all TODOs and remove all explanatory comments
"""TODO: Add a description here."""
import csv
import json
import os
import re
import tempfile
import urllib
import requests
from pathlib import Path
from zipfile import ZipFile
import datasets
# TODO: Add BibTeX citation
# Find for instance the citation on arxiv or on the dataset repo/website
_CITATION = """\
@InProceedings{huggingface:dataset,
title = {A great new dataset},
author={huggingface, Inc.
},
year={2020}
}
"""
# TODO: Add description of the dataset here
# You can copy an official description
_DESCRIPTION = """\
This dataset contains 402 argumentative essays from non-native
"""
# TODO: Add a link to an official homepage for the dataset here
_HOMEPAGE = ""
# TODO: Add the licence for the dataset here if you can find it
_LICENSE = ""
# TODO: Add link to the official dataset URLs here
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
_URLS = {
"tu_darmstadt": "https://tudatalib.ulb.tu-darmstadt.de/bitstream/handle/tudatalib/2422/ArgumentAnnotatedEssays-2.0.zip?sequence=1&isAllowed=y",
}
# TODO: Name of the dataset usually matches the script name with CamelCase instead of snake_case
class NewDataset(datasets.GeneratorBasedBuilder):
"""TODO: Short description of my dataset."""
VERSION = datasets.Version("1.1.0")
temp_dir = tempfile.TemporaryDirectory()
# This is an example of a dataset with multiple configurations.
# If you don't want/need to define several sub-sets in your dataset,
# just remove the BUILDER_CONFIG_CLASS and the BUILDER_CONFIGS attributes.
# If you need to make complex sub-parts in the datasets with configurable options
# You can create your own builder configuration class to store attribute, inheriting from datasets.BuilderConfig
# BUILDER_CONFIG_CLASS = MyBuilderConfig
# You will be able to load one or the other configurations in the following list with
# data = datasets.load_dataset('my_dataset', 'first_domain')
BUILDER_CONFIGS = [
datasets.BuilderConfig(
name="full_labels",
version=VERSION,
description="get all the data conveyed by the labels, O, B-Claim, I-Claim, etc.",
),
datasets.BuilderConfig(
name="spans",
version=VERSION,
description="get the spans, O, B-Span, I-Span.",
),
datasets.BuilderConfig(
name="simple",
version=VERSION,
description="get the labels without B/I, O, MajorClaim, Claim, Premise",
),
datasets.BuilderConfig(
name="sep_tok",
version=VERSION,
description="get the labels without B/I, meaning O, Claim, Premise"
+ ", etc.\n insert seperator tokens ... ",
),
datasets.BuilderConfig(
name="sep_tok_full_labels",
version=VERSION,
description="get the labels with B/I, meaning O, I-Claim, I-Premise"
+ ", etc.\n insert seperator tokens ... ",
),
]
DEFAULT_CONFIG_NAME = "full_labels" # It's not mandatory to have a default configuration. Just use one if it make sense.
def _info(self):
# This method specifies the datasets.DatasetInfo object which contains informations and typings for the dataset
if (
self.config.name == "full_labels"
): # This is the name of the configuration selected in BUILDER_CONFIGS above
features = datasets.Features(
{
"id": datasets.Value("int16"),
"tokens": datasets.Sequence(datasets.Value("string")),
"ner_tags": datasets.Sequence(
datasets.ClassLabel(
names=[
"O",
"B-MajorClaim",
"I-MajorClaim",
"B-Claim",
"I-Claim",
"B-Premise",
"I-Premise",
]
)
),
"text": datasets.Value("string"),
"span_begins": datasets.Sequence(datasets.Value("int16")),
"span_ends": datasets.Sequence(datasets.Value("int16")),
}
)
elif (
self.config.name == "spans"
): # This is an example to show how to have different features for "first_domain" and "second_domain"
features = datasets.Features(
{
"id": datasets.Value("int16"),
"tokens": datasets.Sequence(datasets.Value("string")),
"ner_tags": datasets.Sequence(
datasets.ClassLabel(
names=[
"O",
"B",
"I",
]
)
),
"text": datasets.Value("string"),
"span_begins": datasets.Sequence(datasets.Value("int16")),
"span_ends": datasets.Sequence(datasets.Value("int16")),
}
)
elif (
self.config.name == "simple"
): # This is an example to show how to have different features for "first_domain" and "second_domain"
features = datasets.Features(
{
"id": datasets.Value("int16"),
"tokens": datasets.Sequence(datasets.Value("string")),
"ner_tags": datasets.Sequence(
datasets.ClassLabel(
names=[
"O",
"X_placeholder_X",
"MajorClaim",
"Claim",
"Premise",
]
)
),
"text": datasets.Value("string"),
"span_begins": datasets.Sequence(datasets.Value("int16")),
"span_ends": datasets.Sequence(datasets.Value("int16")),
}
)
elif self.config.name == "sep_tok":
features = datasets.Features(
{
"id": datasets.Value("int16"),
"tokens": datasets.Sequence(datasets.Value("string")),
"ner_tags": datasets.Sequence(
datasets.ClassLabel(
names=[
"O",
"X_placeholder_X",
"MajorClaim",
"Claim",
"Premise",
]
)
),
"text": datasets.Value("string"),
"span_begins": datasets.Sequence(datasets.Value("int16")),
"span_ends": datasets.Sequence(datasets.Value("int16")),
}
)
elif self.config.name == "sep_tok_full_labels":
features = datasets.Features(
{
"id": datasets.Value("int16"),
"tokens": datasets.Sequence(datasets.Value("string")),
"ner_tags": datasets.Sequence(
datasets.ClassLabel(
names=[
"O",
"B-MajorClaim",
"I-MajorClaim",
"B-Claim",
"I-Claim",
"B-Premise",
"I-Premise",
]
)
),
"text": datasets.Value("string"),
"span_begins": datasets.Sequence(datasets.Value("int16")),
"span_ends": datasets.Sequence(datasets.Value("int16")),
}
)
return datasets.DatasetInfo(
# This is the description that will appear on the datasets page.
description=_DESCRIPTION,
# This defines the different columns of the dataset and their types
features=features, # Here we define them above because they are different between the two configurations
# If there's a common (input, target) tuple from the features, uncomment supervised_keys line below and
# specify them. They'll be used if as_supervised=True in builder.as_dataset.
# supervised_keys=("sentence", "label"),
# Homepage of the dataset for documentation
homepage=_HOMEPAGE,
# License for the dataset if available
license=_LICENSE,
# Citation for the dataset
citation=_CITATION,
)
def __load_data(self):
# set up paths
save_dir = Path(self.temp_dir.name)
save_file = Path("essays.zip")
# get url to data
url = _URLS["tu_darmstadt"]
# download data
r = requests.get(url, stream=True)
# save data to temporary dir
with open(save_dir / save_file, 'wb') as fd:
for chunk in r.iter_content(chunk_size=128):
fd.write(chunk)
# recursively unzip files
for glob_path in save_dir.rglob("*.zip"):
with ZipFile(glob_path, 'r') as zip_ref:
zip_ref.extractall(glob_path.parent)
return save_dir
def __range_generator(self, train=0.8, test=0.2):
"""
returns three range objects to access the list of essays
these are the train, test, and validate range, where the size of the
validation range is dictated by the other two ranges
"""
# START RANGE AT 1!!!
return (
range(1, int(403 * train)), # train
range(int(403 * train), int(403 * (train + test))), # test
range(int(403 * (train + test)), 403), # validate
)
def _split_generators(self, _):
data_dir = self.__load_data()
# this dataset will return a "train" split only, allowing for
# 5-fold cross-validation
train, test, validate = self.__range_generator(1, 0)
# essays = self._get_essay_list()
if len(validate) > 0 and len(test) > 0:
return [
datasets.SplitGenerator(
name=datasets.Split.TRAIN,
# These kwargs will be passed to _generate_examples
gen_kwargs={
"data_dir": data_dir,
"id_range": train,
},
),
datasets.SplitGenerator(
name=datasets.Split.VALIDATION,
# These kwargs will be passed to _generate_examples
gen_kwargs={
"data_dir": data_dir,
"id_range": validate,
},
),
datasets.SplitGenerator(
name=datasets.Split.TEST,
# These kwargs will be passed to _generate_examples
gen_kwargs={
"data_dir": data_dir,
"id_range": test,
},
),
]
elif len(test) > 0:
return [
datasets.SplitGenerator(
name=datasets.Split.TRAIN,
# These kwargs will be passed to _generate_examples
gen_kwargs={
"data_dir": data_dir,
"id_range": train,
},
),
datasets.SplitGenerator(
name=datasets.Split.TEST,
# These kwargs will be passed to _generate_examples
gen_kwargs={
"data_dir": data_dir,
"id_range": test,
},
),
]
elif len(validate) > 0:
return [
datasets.SplitGenerator(
name=datasets.Split.TRAIN,
# These kwargs will be passed to _generate_examples
gen_kwargs={
"data_dir": data_dir,
"id_range": train,
},
),
datasets.SplitGenerator(
name=datasets.Split.VALIDATION,
# These kwargs will be passed to _generate_examples
gen_kwargs={
"data_dir": data_dir,
"id_range": validate,
},
),
]
else:
return [
datasets.SplitGenerator(
name=datasets.Split.TRAIN,
# These kwargs will be passed to _generate_examples
gen_kwargs={
"data_dir": data_dir,
"id_range": train,
},
),
]
def _get_essay(self, id: int, data_dir: Path):
return data_dir.joinpath(f"essay{str(id).rjust(3, '0')}.txt").read_text(), data_dir.joinpath(f"essay{str(id).rjust(3, '0')}.ann").read_text()
def _parse_raw_ann(self, raw_ann: str):
raw_anns = raw_ann.split("\n")
clean_anns = []
for cur_raw_ann in raw_anns:
matches = re.match(r".+\t(.+) (.+) (.+)\t(.+)", cur_raw_ann)
if matches is not None:
clean_anns.append(
(matches.group(1), int(matches.group(2)), int(matches.group(3)), matches.group(4))
)
# sorting spans by start before returningbefore returning
return sorted(clean_anns, key=lambda x: x[1])
def _tokenise(self, text, clean_anns):
# find spans
previous_end = 0
spans = []
# for every span, add the not span that is before it
for clean_ann in clean_anns:
spans.append(("O", text[previous_end:clean_ann[1]]))
spans.append((clean_ann[0], text[clean_ann[1]:clean_ann[2]]))
previous_end = clean_ann[2]
# add whatever is left over to not spans
spans.append(("O", text[previous_end:]))
tokens = []
labels = []
# tokenise spans
# WARN: the old dataset considered punctuation to be separate tokens, whilst this one doesnt.
# this shouldnt matter however, since this is just pre-tokenisation, which will pre tokenised for the respective model later on.
# i assume that the later tokenisation will create equal results.
for span in spans:
span_tokens = span[1].split()
label = span[0]
first_label = span[0]
if self.config.name == "simple":
# with simple, the token is already correct
pass
elif self.config.name == "sep_tok":
# with sep_tok, the token is correct, but a sep top needs to be inserted
span_tokens.insert(0, "")
span_tokens.append("")
elif self.config.name == "spans":
if label != "O":
first_label = "B"
label = "I"
elif self.config.name == "full_labels":
if label != "O":
first_label = "B-" + label
label = "I-" + label
elif self.config.name == "sep_tok_full_labels":
# ensure I and B
if label != "O":
first_label = "B-" + label
label = "I-" + label
# make sure to include the sep tok!!!
span_tokens.insert(0, "")
span_tokens.append("")
labels.append(first_label)
labels.extend([label] * (len(span_tokens) - 1))
tokens.extend(span_tokens)
return tokens, labels
def _process_essay(self, id, data_dir: Path):
# TODO: get the logic in here. everything else it taken care of i think
text, raw_ann = self._get_essay(id, data_dir)
clean_anns = self._parse_raw_ann(raw_ann)
tokens, labels = self._tokenise(text, clean_anns)
# id = self._get_id(essay)
# # input(id)
# tokens = self._get_tokens(essay)
# # input(tokens)
# label_dict = self._get_label_dict(essay)
# # input(label_dict)
# tokens, labels, begins, ends = self._match_tokens(tokens, label_dict)
# # input(tokens)
# # input(labels)
# text = self._get_text(essay)
# id = 1
# tokens = ["1"]
# labels = [1]
# text = "a"
# begins = [1]
# ends = [2]
return {
"id": id,
"tokens": tokens,
"ner_tags": labels,
"text": text,
"span_begins": [ann[1] for ann in clean_anns],
"span_ends": [ann[2] for ann in clean_anns],
}
# method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
def _generate_examples(self, data_dir: Path, id_range: list):
# This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
# The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
data_dir = data_dir.joinpath("ArgumentAnnotatedEssays-2.0", "brat-project-final")
for id in id_range:
# input(data[id])
yield id, self._process_essay(id, data_dir)