Datasets:
Update from mdebruyn
Browse files
README.md
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
hello
|
mqa.py
ADDED
@@ -0,0 +1,162 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2020 HuggingFace Datasets Authors.
|
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 |
+
|
16 |
+
# Lint as: python3
|
17 |
+
import os
|
18 |
+
import gzip
|
19 |
+
import json
|
20 |
+
|
21 |
+
import datasets
|
22 |
+
|
23 |
+
|
24 |
+
_DESCRIPTION = """ """
|
25 |
+
_HOMEPAGE_URL = ""
|
26 |
+
_CITATION = """"""
|
27 |
+
|
28 |
+
_VERSION = "0.0"
|
29 |
+
_BASE_NAME = ""
|
30 |
+
_BASE_URL = "https://www.clips.uantwerpen.be/~mdebruyn/mqa/data.{}.{}.json.gz"
|
31 |
+
|
32 |
+
|
33 |
+
_LANGUAGES = [
|
34 |
+
"en", "de", "es", "fr",
|
35 |
+
"ru", "ja", "it", "zh", "pt",
|
36 |
+
"nl", "tr", "pl", "vi", "ar",
|
37 |
+
"id", "uk", "ro", "no", "th",
|
38 |
+
"sv", "el", "fi", "he", "da",
|
39 |
+
"cs", "ko", "fa", "hi", "hu",
|
40 |
+
"sk", "lt", "et", "hr", "is",
|
41 |
+
"lv", "ms", "bg", "sr", "ca"
|
42 |
+
]
|
43 |
+
_SCOPES = ["faq", "cqa"]
|
44 |
+
_LEVELS = ["domain", "page", "question"]
|
45 |
+
|
46 |
+
|
47 |
+
class MQAConfig(datasets.BuilderConfig):
|
48 |
+
def __init__(self, *args, language="en", scope="all", level="question", **kwargs):
|
49 |
+
super().__init__(
|
50 |
+
*args,
|
51 |
+
name=f"{language}-{scope}-{level}",
|
52 |
+
**kwargs,
|
53 |
+
)
|
54 |
+
self.language = language
|
55 |
+
self.scope = scope
|
56 |
+
self.level = level
|
57 |
+
|
58 |
+
|
59 |
+
class MQA(datasets.GeneratorBasedBuilder):
|
60 |
+
BUILDER_CONFIGS = []
|
61 |
+
for language in _LANGUAGES:
|
62 |
+
for scope in _SCOPES:
|
63 |
+
for level in _LEVELS:
|
64 |
+
BUILDER_CONFIGS.append(MQAConfig(language=language, scope=scope, level=level))
|
65 |
+
BUILDER_CONFIG_CLASS = MQAConfig
|
66 |
+
|
67 |
+
def _info(self):
|
68 |
+
question = {
|
69 |
+
"id": datasets.Value("string"),
|
70 |
+
"text": datasets.Value("string"),
|
71 |
+
"name": datasets.Value("string"),
|
72 |
+
"answers": [{
|
73 |
+
"text": datasets.Value("string"),
|
74 |
+
"name": datasets.Value("string"),
|
75 |
+
"is_accepted": datasets.Value("bool"),
|
76 |
+
}]
|
77 |
+
}
|
78 |
+
page = {
|
79 |
+
"id": datasets.Value("string"),
|
80 |
+
"bucket": datasets.Value("string"),
|
81 |
+
"domain": datasets.Value("string"),
|
82 |
+
"questions": [question]
|
83 |
+
}
|
84 |
+
domain = {
|
85 |
+
"domain": datasets.Value("string"),
|
86 |
+
"pages": [page]
|
87 |
+
}
|
88 |
+
if self.config.level == "question":
|
89 |
+
features = question
|
90 |
+
elif self.config.level == "page":
|
91 |
+
features = page
|
92 |
+
elif self.config.level == "domain":
|
93 |
+
features = domain
|
94 |
+
else:
|
95 |
+
raise NotImplementedError()
|
96 |
+
return datasets.DatasetInfo(
|
97 |
+
description=_DESCRIPTION,
|
98 |
+
features=datasets.Features(features),
|
99 |
+
supervised_keys=None,
|
100 |
+
homepage=_HOMEPAGE_URL,
|
101 |
+
citation=_CITATION,
|
102 |
+
)
|
103 |
+
|
104 |
+
def _split_generators(self, dl_manager):
|
105 |
+
filenames = []
|
106 |
+
languages = _LANGUAGES if self.config.language == "all" else self.config.language
|
107 |
+
scopes = _SCOPES if self.config.scope == "all" else self.config.scope
|
108 |
+
for language in languages:
|
109 |
+
for scope in scopes:
|
110 |
+
path = dl_manager.download_and_extract(_BASE_URL.format(language, scope))
|
111 |
+
filenames.append(path)
|
112 |
+
return [
|
113 |
+
datasets.SplitGenerator(
|
114 |
+
name=datasets.Split.TRAIN,
|
115 |
+
gen_kwargs={"filenames": filenames},
|
116 |
+
)
|
117 |
+
]
|
118 |
+
|
119 |
+
def _generate_examples(self, filenames):
|
120 |
+
|
121 |
+
def default(e, key, default_value=""):
|
122 |
+
if e[key] is None:
|
123 |
+
return default_value
|
124 |
+
return e[key]
|
125 |
+
|
126 |
+
for filename in filenames:
|
127 |
+
with open(filename, "r") as f:
|
128 |
+
domain = []
|
129 |
+
previous_domain = ''
|
130 |
+
for line in f:
|
131 |
+
page = json.loads(line)
|
132 |
+
questions = [{
|
133 |
+
"text": default(question, "text"),
|
134 |
+
"name": default(question, "name"),
|
135 |
+
"id": question["hash"],
|
136 |
+
"answers": [{
|
137 |
+
"text": default(answer, "text"),
|
138 |
+
"name": default(answer, "name"),
|
139 |
+
"is_accepted": answer["is_accepted"]
|
140 |
+
} for answer in question["answers"]]
|
141 |
+
} for question in page["questions"]]
|
142 |
+
page = {
|
143 |
+
"id": page["page_hash"],
|
144 |
+
"domain": page["domain"],
|
145 |
+
"bucket": page["bucket"],
|
146 |
+
"questions": questions
|
147 |
+
}
|
148 |
+
if self.config.level == "question":
|
149 |
+
for question in questions:
|
150 |
+
question_hash = question.pop("hash")
|
151 |
+
yield question_hash, question
|
152 |
+
if self.config.level == "page":
|
153 |
+
yield page["page_hash"], page
|
154 |
+
if self.config.level == "domain":
|
155 |
+
if page["domain"] == previous_domain or previous_domain == "":
|
156 |
+
domain.append(page)
|
157 |
+
else:
|
158 |
+
yield previous_domain, {
|
159 |
+
"domain": previous_domain,
|
160 |
+
"pages": domain
|
161 |
+
}
|
162 |
+
previous_domain = page["domain"]
|