holylovenia commited on
Commit
019ffec
1 Parent(s): d3f68ca

Upload glotstorybook.py with huggingface_hub

Browse files
Files changed (1) hide show
  1. glotstorybook.py +143 -0
glotstorybook.py ADDED
@@ -0,0 +1,143 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from pathlib import Path
2
+ from typing import Dict, List, Tuple
3
+
4
+ import datasets
5
+ import pandas as pd
6
+
7
+ from seacrowd.utils.configs import SEACrowdConfig
8
+ from seacrowd.utils.constants import Tasks, Licenses, TASK_TO_SCHEMA, SCHEMA_TO_FEATURES
9
+
10
+ _CITATION = """\
11
+ @inproceedings{kargaran2023glotlid,
12
+ title = {{GlotLID: Language Identification for Low-Resource Languages}},
13
+ author = {Kargaran, Amir Hossein and Imani, Ayyoob and Yvon, Fran{\c{c}}ois and Sch{\"u}tze, Hinrich},
14
+ year = 2023,
15
+ booktitle = {The 2023 Conference on Empirical Methods in Natural Language Processing},
16
+ url = {https://openreview.net/forum?id=dl4e3EBz5j}
17
+ }
18
+ """
19
+
20
+ _DATASETNAME = "glotstorybook"
21
+ _DESCRIPTION = """\
22
+ The GlotStoryBook dataset is a compilation of children's storybooks from the Global
23
+ Storybooks project, encompassing 174 languages organized for machine translation tasks. It
24
+ features rows containing the text segment (text number), the language code, and the file
25
+ name, which corresponds to the specific book and story segment. This structure allows for
26
+ the comparison of texts across different languages by matching file names and text numbers
27
+ between rows.
28
+ """
29
+
30
+ _HOMEPAGE = "https://huggingface.co/datasets/cis-lmu/GlotStoryBook"
31
+ _LICENSE = f"""{Licenses.OTHERS.value} | \
32
+ We do not own any of the text from which these data has been extracted. All the files are
33
+ collected from the repository located at https://github.com/global-asp/. The source
34
+ repository for each text and file is stored in the dataset. Each file in the dataset is
35
+ associated with one license from the CC family. The licenses include 'CC BY', 'CC BY-NC',
36
+ 'CC BY-NC-SA', 'CC-BY', 'CC-BY-NC', and 'Public Domain'. We also license the code, actual
37
+ packaging and the metadata of these data under the cc0-1.0.
38
+ """
39
+
40
+ _LOCAL=False
41
+ _LANGUAGES = ["khg", "khm", "mya", "tet", "tha", "vie"]
42
+
43
+ _URLS = "https://huggingface.co/datasets/cis-lmu/GlotStoryBook/resolve/main/GlotStoryBook.csv"
44
+
45
+ _SUPPORTED_TASKS = [Tasks.SELF_SUPERVISED_PRETRAINING]
46
+ _SOURCE_VERSION = "1.0.0"
47
+ _SEACROWD_VERSION = "2024.06.20"
48
+
49
+
50
+ class GlotStoryBookDataset(datasets.GeneratorBasedBuilder):
51
+ """Compilation of storybooks from the Global Storybooks project"""
52
+
53
+ SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
54
+ SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
55
+
56
+ SEACROWD_SCHEMA_NAME = TASK_TO_SCHEMA[_SUPPORTED_TASKS[0]].lower()
57
+
58
+ BUILDER_CONFIGS = [
59
+ SEACrowdConfig(
60
+ name=f"{_DATASETNAME}_source",
61
+ version=SOURCE_VERSION,
62
+ description=f"{_DATASETNAME} source schema",
63
+ schema="source",
64
+ subset_id=f"{_DATASETNAME}",
65
+ ),
66
+ SEACrowdConfig(
67
+ name=f"{_DATASETNAME}_seacrowd_{SEACROWD_SCHEMA_NAME}",
68
+ version=SEACROWD_VERSION,
69
+ description=f"{_DATASETNAME} SEACrowd schema",
70
+ schema=f"seacrowd_{SEACROWD_SCHEMA_NAME}",
71
+ subset_id=f"{_DATASETNAME}",
72
+ ),
73
+ ]
74
+
75
+ DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
76
+
77
+ def _info(self) -> datasets.DatasetInfo:
78
+ if self.config.schema == "source":
79
+ features = datasets.Features(
80
+ {
81
+ "text": datasets.Value("string"),
82
+ "text_number": datasets.Value("int64"),
83
+ "license": datasets.Value("string"),
84
+ "text_by": datasets.Value("string"),
85
+ "translation_by": datasets.Value("string"),
86
+ "language": datasets.Value("string"),
87
+ "file_name": datasets.Value("string"),
88
+ "source": datasets.Value("string"),
89
+ "iso639-3": datasets.Value("string"),
90
+ "script": datasets.Value("string"),
91
+ }
92
+ )
93
+ elif self.config.schema == f"seacrowd_{self.SEACROWD_SCHEMA_NAME}":
94
+ features = SCHEMA_TO_FEATURES[self.SEACROWD_SCHEMA_NAME.upper()]
95
+
96
+ return datasets.DatasetInfo(
97
+ description=_DESCRIPTION,
98
+ features=features,
99
+ homepage=_HOMEPAGE,
100
+ license=_LICENSE,
101
+ citation=_CITATION,
102
+ )
103
+
104
+ def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
105
+ data_path = Path(dl_manager.download_and_extract(_URLS))
106
+
107
+ return [
108
+ datasets.SplitGenerator(
109
+ name=datasets.Split.TRAIN,
110
+ gen_kwargs={
111
+ "filepath": data_path,
112
+ "split": "train",
113
+ },
114
+ ),
115
+ ]
116
+
117
+ def _generate_examples(self, filepath: Path, split: str) -> Tuple[int, Dict]:
118
+ """Yields examples as (key, example) tuples."""
119
+ df = pd.read_csv(filepath)
120
+ df = df[df["ISO639-3"].isin(_LANGUAGES)]
121
+
122
+ if self.config.schema == "source":
123
+ for i, row in df.iterrows():
124
+ yield i, {
125
+ "text": row["Text"],
126
+ "text_number": row["Text Number"],
127
+ "license": row["License"],
128
+ "text_by": row["Text By"],
129
+ "translation_by": row["Translation By"],
130
+ "language": row["Language"],
131
+ "file_name": row["File Name"],
132
+ "source": row["Source"],
133
+ "iso639-3": row["ISO639-3"],
134
+ "script": row["Script"],
135
+ }
136
+ elif self.config.schema == f"seacrowd_{self.SEACROWD_SCHEMA_NAME}":
137
+ df = df.sort_values(by=["ISO639-3", "Source", "File Name", "Text Number"])
138
+ df = df.groupby(["ISO639-3", "Source", "File Name"]).agg({"Text": " ".join}).reset_index()
139
+ for i, row in df.iterrows():
140
+ yield i, {
141
+ "id": str(i),
142
+ "text": row["Text"],
143
+ }