Albert Villanova del Moral
commited on
Add dataset loading script
Browse files- open_access.py +210 -0
open_access.py
ADDED
|
@@ -0,0 +1,210 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""PMC Open Access Subset."""
|
| 16 |
+
|
| 17 |
+
import datetime
|
| 18 |
+
|
| 19 |
+
import pandas as pd
|
| 20 |
+
|
| 21 |
+
import datasets
|
| 22 |
+
from datasets.tasks import LanguageModeling
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
# TODO: Add BibTeX citation
|
| 26 |
+
# Find for instance the citation on arxiv or on the dataset repo/website
|
| 27 |
+
_CITATION = """\
|
| 28 |
+
@InProceedings{huggingface:dataset,
|
| 29 |
+
title = {A great new dataset},
|
| 30 |
+
author={huggingface, Inc.
|
| 31 |
+
},
|
| 32 |
+
year={2020}
|
| 33 |
+
}
|
| 34 |
+
"""
|
| 35 |
+
|
| 36 |
+
_DESCRIPTION = """\
|
| 37 |
+
The PMC Open Access Subset includes more than 3.4 million journal articles and preprints that are made available under
|
| 38 |
+
license terms that allow reuse.
|
| 39 |
+
|
| 40 |
+
Not all articles in PMC are available for text mining and other reuse, many have copyright protection, however articles
|
| 41 |
+
in the PMC Open Access Subset are made available under Creative Commons or similar licenses that generally allow more
|
| 42 |
+
liberal redistribution and reuse than a traditional copyrighted work.
|
| 43 |
+
|
| 44 |
+
The PMC Open Access Subset is one part of the PMC Article Datasets
|
| 45 |
+
"""
|
| 46 |
+
|
| 47 |
+
_HOMEPAGE = "https://www.ncbi.nlm.nih.gov/pmc/tools/openftlist/"
|
| 48 |
+
|
| 49 |
+
# TODO: Add the licence for the dataset here if you can find it
|
| 50 |
+
_LICENSE = ""
|
| 51 |
+
|
| 52 |
+
_URL = "https://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_bulk/{subset}/txt/"
|
| 53 |
+
_SUBSETS = {
|
| 54 |
+
"commercial": "oa_comm",
|
| 55 |
+
"non_commercial": "oa_noncomm",
|
| 56 |
+
"other": "oa_other",
|
| 57 |
+
}
|
| 58 |
+
_BASELINE_DATE = "2021-12-17"
|
| 59 |
+
|
| 60 |
+
|
| 61 |
+
class OpenAccessConfig(datasets.BuilderConfig):
|
| 62 |
+
"""BuilderConfig for the PMC Open Access Subset."""
|
| 63 |
+
|
| 64 |
+
def __init__(self, subsets=None, **kwargs):
|
| 65 |
+
"""BuilderConfig for the PMC Open Access Subset.
|
| 66 |
+
|
| 67 |
+
Args:
|
| 68 |
+
subsets (:obj:`List[str]`): List of subsets/groups to load.
|
| 69 |
+
**kwargs: Keyword arguments forwarded to super.
|
| 70 |
+
"""
|
| 71 |
+
subsets = [subsets] if isinstance(subsets, str) else subsets
|
| 72 |
+
super().__init__(
|
| 73 |
+
name="+".join(subsets), **kwargs,
|
| 74 |
+
)
|
| 75 |
+
self.subsets = subsets if self.name != "all" else list(_SUBSETS.keys())
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
class OpenAccess(datasets.GeneratorBasedBuilder):
|
| 79 |
+
"""PMC Open Access Subset."""
|
| 80 |
+
|
| 81 |
+
VERSION = datasets.Version("1.0.0")
|
| 82 |
+
BUILDER_CONFIG_CLASS = OpenAccessConfig
|
| 83 |
+
BUILDER_CONFIGS = [OpenAccessConfig(subsets="all")] + [OpenAccessConfig(subsets=subset) for subset in _SUBSETS]
|
| 84 |
+
DEFAULT_CONFIG_NAME = "all"
|
| 85 |
+
|
| 86 |
+
def _info(self):
|
| 87 |
+
return datasets.DatasetInfo(
|
| 88 |
+
description=_DESCRIPTION,
|
| 89 |
+
features=datasets.Features(
|
| 90 |
+
{
|
| 91 |
+
"text": datasets.Value("string"),
|
| 92 |
+
"pmid": datasets.Value("string"),
|
| 93 |
+
"accession_id": datasets.Value("string"),
|
| 94 |
+
"license": datasets.Value("string"),
|
| 95 |
+
"last_updated": datasets.Value("string"),
|
| 96 |
+
"retracted": datasets.Value("string"),
|
| 97 |
+
"citation": datasets.Value("string"),
|
| 98 |
+
}
|
| 99 |
+
),
|
| 100 |
+
homepage=_HOMEPAGE,
|
| 101 |
+
license=_LICENSE,
|
| 102 |
+
citation=_CITATION,
|
| 103 |
+
task_templates=[LanguageModeling(text_column="text")],
|
| 104 |
+
)
|
| 105 |
+
|
| 106 |
+
def _split_generators(self, dl_manager):
|
| 107 |
+
for subset in self.config.subsets:
|
| 108 |
+
url = _URL.format(subset=_SUBSETS[subset])
|
| 109 |
+
basename = f"{_SUBSETS[subset]}_txt."
|
| 110 |
+
# Baselines
|
| 111 |
+
baselines = [f"PMC00{i}xxxxxx.baseline.{_BASELINE_DATE}" for i in range(9)]
|
| 112 |
+
# baseline_urls = {
|
| 113 |
+
# "baseline_file_lists": [f"{url}{basename}{baseline}.filelist.csv" for baseline in baselines],
|
| 114 |
+
# "baseline_archives": [f"{url}{basename}{baseline}.tar.gz" for baseline in baselines],
|
| 115 |
+
# }
|
| 116 |
+
# baseline_paths = dl_manager.download(baseline_urls)
|
| 117 |
+
baseline_file_lists = []
|
| 118 |
+
baseline_archives = []
|
| 119 |
+
for baseline in baselines:
|
| 120 |
+
baseline_file_list_url = f"{url}{basename}{baseline}.filelist.csv"
|
| 121 |
+
try:
|
| 122 |
+
baseline_file_list = dl_manager.download(baseline_file_list_url)
|
| 123 |
+
except FileNotFoundError: # non-commercial PMC000xxxxxx baseline does not exist
|
| 124 |
+
continue
|
| 125 |
+
baseline_archive_url = f"{url}{basename}{baseline}.tar.gz"
|
| 126 |
+
try:
|
| 127 |
+
baseline_archive = dl_manager.download(baseline_archive_url)
|
| 128 |
+
except FileNotFoundError:
|
| 129 |
+
continue
|
| 130 |
+
baseline_file_lists.append(baseline_file_list)
|
| 131 |
+
baseline_archives.append(baseline_archive)
|
| 132 |
+
# Incremental
|
| 133 |
+
date_delta = datetime.date.today() - datetime.date.fromisoformat(_BASELINE_DATE)
|
| 134 |
+
incremental_dates = [
|
| 135 |
+
(datetime.date.fromisoformat(_BASELINE_DATE) + datetime.timedelta(days=i + 1)).isoformat()
|
| 136 |
+
for i in range(date_delta.days)
|
| 137 |
+
]
|
| 138 |
+
incrementals = [f"incr.{date}" for date in incremental_dates]
|
| 139 |
+
incremental_urls = {
|
| 140 |
+
"incremental_file_lists": [
|
| 141 |
+
f"{url}{basename}{incremental}.filelist.csv" for incremental in incrementals
|
| 142 |
+
],
|
| 143 |
+
"incremental_archives": [f"{url}{basename}{incremental}.tar.gz" for incremental in incrementals],
|
| 144 |
+
}
|
| 145 |
+
incremental_paths = dl_manager.download(incremental_urls)
|
| 146 |
+
return [
|
| 147 |
+
datasets.SplitGenerator(
|
| 148 |
+
name=datasets.Split.TRAIN,
|
| 149 |
+
gen_kwargs={
|
| 150 |
+
"baseline_file_lists": baseline_file_lists,
|
| 151 |
+
"baseline_archives": [dl_manager.iter_archive(archive) for archive in baseline_archives],
|
| 152 |
+
"incremental_file_lists": incremental_paths["incremental_file_lists"],
|
| 153 |
+
"incremental_archives": [
|
| 154 |
+
dl_manager.iter_archive(archive) for archive in incremental_paths["incremental_archives"]
|
| 155 |
+
],
|
| 156 |
+
},
|
| 157 |
+
),
|
| 158 |
+
]
|
| 159 |
+
|
| 160 |
+
def _generate_examples(self, baseline_file_lists, baseline_archives, incremental_file_lists, incremental_archives):
|
| 161 |
+
key = 0
|
| 162 |
+
# Baselines
|
| 163 |
+
for baseline_file_list, baseline_archive in zip(baseline_file_lists, baseline_archives):
|
| 164 |
+
try:
|
| 165 |
+
baselines = pd.read_csv(baseline_file_list, index_col="Article File").to_dict(orient="index")
|
| 166 |
+
for path, file in baseline_archive:
|
| 167 |
+
data = baselines.pop(path)
|
| 168 |
+
content = file.read()
|
| 169 |
+
try:
|
| 170 |
+
text = content.decode("utf-8").strip()
|
| 171 |
+
except UnicodeDecodeError as e:
|
| 172 |
+
text = content.decode("latin-1").strip()
|
| 173 |
+
data = {
|
| 174 |
+
"text": text,
|
| 175 |
+
"pmid": data["PMID"],
|
| 176 |
+
"accession_id": data["AccessionID"],
|
| 177 |
+
"license": data["License"],
|
| 178 |
+
"last_updated": data["LastUpdated (YYYY-MM-DD HH:MM:SS)"],
|
| 179 |
+
"retracted": data["Retracted"],
|
| 180 |
+
"citation": data["Article Citation"],
|
| 181 |
+
}
|
| 182 |
+
yield key, data
|
| 183 |
+
key += 1
|
| 184 |
+
except FileNotFoundError: # non-commercial PMC000xxxxxx baseline does not exist
|
| 185 |
+
continue
|
| 186 |
+
# Incrementals
|
| 187 |
+
if incremental_file_lists:
|
| 188 |
+
for incremental_file_list, incremental_archive in zip(incremental_file_lists, incremental_archives):
|
| 189 |
+
import pdb
|
| 190 |
+
|
| 191 |
+
pdb.set_trace()
|
| 192 |
+
incrementals = pd.read_csv(incremental_file_list, index_col="Article File").to_dict(orient="index")
|
| 193 |
+
for path, file in incremental_archive:
|
| 194 |
+
data = incrementals.pop(path)
|
| 195 |
+
content = file.read()
|
| 196 |
+
try:
|
| 197 |
+
text = content.decode("utf-8").strip()
|
| 198 |
+
except UnicodeDecodeError as e:
|
| 199 |
+
text = content.decode("latin-1").strip()
|
| 200 |
+
data = {
|
| 201 |
+
"text": text,
|
| 202 |
+
"pmid": data["PMID"],
|
| 203 |
+
"accession_id": data["AccessionID"],
|
| 204 |
+
"license": data["License"],
|
| 205 |
+
"last_updated": data["LastUpdated (YYYY-MM-DD HH:MM:SS)"],
|
| 206 |
+
"retracted": data["Retracted"],
|
| 207 |
+
"citation": data["Article Citation"],
|
| 208 |
+
}
|
| 209 |
+
yield key, data
|
| 210 |
+
key += 1
|