jordiae commited on
Commit
3d1b00c
1 Parent(s): abca939

Create new file

Browse files
Files changed (1) hide show
  1. escorpius.py +147 -0
escorpius.py ADDED
@@ -0,0 +1,147 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ # Copyright 2022 esCorpius authors
3
+ # The code required to produce and load this dataset is licensed under MIT License.
4
+ # The code samples included in this dataset keep their own licenses, which can be retrieved via their metadata.
5
+ # Unless required by applicable law or agreed to in writing, software
6
+ # distributed under the License is distributed on an "AS IS" BASIS,
7
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
8
+ # See the License for the specific language governing permissions and
9
+ # limitations under the License.
10
+
11
+ # Please note that the dataset release is still work in progress.
12
+
13
+ """The esCorpius dataset."""
14
+
15
+ import json
16
+
17
+ import datasets
18
+
19
+ from pathlib import Path
20
+
21
+
22
+ _CITATION = """\
23
+ @misc{TODO
24
+ }
25
+ """
26
+
27
+ _DESCRIPTION = """\
28
+ Spanish dataset
29
+ """ # TODO: expand
30
+
31
+ _HOMEPAGE = "https://huggingface.co/datasets/LHF/escorpius"
32
+
33
+ _LICENSE = "CC BY-NC-ND 4.0"
34
+
35
+ _URL = ""
36
+
37
+
38
+ _FEATURES = datasets.Features(
39
+ {
40
+ "id": datasets.Value("string"),
41
+ "text": datasets.Value("string"),
42
+ "url_warc": datasets.Value("string"),
43
+ "url": datasets.Value("string")
44
+
45
+ }
46
+ )
47
+
48
+
49
+ class EsCorpiusConfig(datasets.BuilderConfig):
50
+ """BuilderConfig for esCorpius."""
51
+
52
+ def __init__(self, *args, **kwargs):
53
+ """BuilderConfig for The Pile.
54
+ Args:
55
+ **kwargs: keyword arguments forwarded to super.
56
+ """
57
+ super().__init__(
58
+ *args,
59
+ **kwargs,
60
+ )
61
+
62
+
63
+ class EsCorpius(datasets.GeneratorBasedBuilder):
64
+ """The esCorpius dataset."""
65
+
66
+ BUILDER_CONFIGS = [
67
+ EsCorpiusConfig(
68
+ name="esCorpius",
69
+ version=datasets.Version("1.0.1"),
70
+ description="Spanish dataset"
71
+ ),
72
+ ]
73
+
74
+ def _info(self):
75
+ """Give information and typings for the dataset."""
76
+ return datasets.DatasetInfo(
77
+ # This is the description that will appear on the datasets page.
78
+ description=_DESCRIPTION,
79
+ # This defines the different columns of the dataset and their types
80
+ features=_FEATURES,
81
+ # If there's a common (input, target) tuple from the features,
82
+ # specify them here. They'll be used if as_supervised=True in
83
+ # builder.as_dataset.
84
+ supervised_keys=None,
85
+ # Homepage of the dataset for documentation
86
+ homepage=_HOMEPAGE,
87
+ # License for the dataset if available
88
+ license=_LICENSE,
89
+ # Citation for the dataset
90
+ citation=_CITATION,
91
+ )
92
+
93
+ def _split_generators(self, dl_manager):
94
+ """Returns SplitGenerators."""
95
+ urls_to_download = [
96
+ 'es_corpus.jsonl.aa',
97
+ 'es_corpus.jsonl.ab',
98
+ 'es_corpus.jsonl.ac',
99
+ 'es_corpus.jsonl.ad',
100
+ 'es_corpus.jsonl.ae',
101
+ 'es_corpus.jsonl.af',
102
+ 'es_corpus.jsonl.ag',
103
+ 'es_corpus.jsonl.ah',
104
+ 'es_corpus.jsonl.ai',
105
+ 'es_corpus.jsonl.aj',
106
+ 'es_corpus.jsonl.ak',
107
+ 'es_corpus.jsonl.al',
108
+ 'es_corpus.jsonl.am',
109
+ 'es_corpus.jsonl.an',
110
+ 'es_corpus.jsonl.ao',
111
+ 'es_corpus.jsonl.ap',
112
+ 'es_corpus.jsonl.aq',
113
+ 'es_corpus.jsonl.ar',
114
+ 'es_corpus.jsonl.as',
115
+ 'es_corpus.jsonl.at',
116
+ 'es_corpus.jsonl.au',
117
+ 'es_corpus.jsonl.av',
118
+ 'es_corpus.jsonl.aw',
119
+ 'es_corpus.jsonl.ax',
120
+ 'es_corpus.jsonl.ay',
121
+ 'es_corpus.jsonl.az',
122
+ 'es_corpus.jsonl.ba',
123
+ 'es_corpus.jsonl.bb',
124
+ 'es_corpus.jsonl.bc',
125
+ 'es_corpus.jsonl.bd',
126
+ 'es_corpus.jsonl.be',
127
+ 'es_corpus.jsonl.bf',
128
+ 'es_corpus.jsonl.bg'
129
+ ]
130
+ downloaded_files = dl_manager.download_and_extract(urls_to_download)
131
+
132
+ return [
133
+
134
+ datasets.SplitGenerator(name='train',
135
+ gen_kwargs={"files": downloaded_files}),
136
+ ]
137
+
138
+ def _generate_examples(self, files):
139
+ """Yield examples as (key, example) tuples."""
140
+ key = 0
141
+
142
+ for path in sorted(Path(files).rglob('*.jsonl*')):
143
+ with open(open(path, "r"), encoding="utf-8") as f:
144
+ for row in f:
145
+ data = json.loads(row)
146
+ yield key, data
147
+ key += 1