luist18 commited on
Commit
0ee61b4
1 Parent(s): da6b808

Create ptparl.py

Browse files
Files changed (1) hide show
  1. ptparl.py +117 -0
ptparl.py ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """ptparl dataset."""
2
+
3
+
4
+ import xml.etree.ElementTree as ET
5
+
6
+ import datasets
7
+ import csv
8
+
9
+ _DESCRIPTION = """
10
+ The PTPARL dataset is a dataset containing 5713 interventions in the Portuguese parliament.
11
+ """
12
+
13
+ _HOMEPAGE = "MIT"
14
+
15
+ _LICENSE = ""
16
+
17
+ _URL = "https://gist.githubusercontent.com/luist18/e3730636962a117ef0eb9e0659f1cba8/raw/5e55a95aaec8261c7690c7132037121d99c39279/pt_parl.csv"
18
+
19
+
20
+ group_map = {
21
+ "PS": 0,
22
+ "CDS-PP": 1,
23
+ "PCP": 2,
24
+ "BE": 3,
25
+ "PSD": 4,
26
+ "PEV": 5,
27
+ "PAN": 6,
28
+ "CH": 7,
29
+ "IL": 8,
30
+ "L": 9,
31
+ }
32
+
33
+ wing_map = {
34
+ "LEFT": 0,
35
+ "LEAN_LEFT": 1,
36
+ "CENTER": 2,
37
+ "LEAN_RIGHT": 3,
38
+ "RIGHT": 4,
39
+ }
40
+
41
+
42
+ class PTPARL(datasets.GeneratorBasedBuilder):
43
+ """PTPARL dataset."""
44
+
45
+ VERSION = datasets.Version("1.0.0")
46
+
47
+ BUILDER_CONFIGS = [
48
+ datasets.BuilderConfig(
49
+ name="full",
50
+ version=VERSION,
51
+ description="Full dataset",
52
+ ),
53
+ ]
54
+
55
+ DEFAULT_CONFIG_NAME = "full"
56
+
57
+ def _info(self):
58
+ features = datasets.Features(
59
+ {
60
+ "text": datasets.Value("string"),
61
+ "group": datasets.features.ClassLabel(
62
+ names=[
63
+ "PS",
64
+ "CDS-PP",
65
+ "PCP",
66
+ "BE",
67
+ "PSD",
68
+ "PEV",
69
+ "PAN",
70
+ "CH",
71
+ "IL",
72
+ "L",
73
+ ]
74
+ ),
75
+ "wing": datasets.features.ClassLabel(
76
+ names=[["CENTER", "LEAN_RIGHT", "LEFT", "LEAN_LEFT", "RIGHT"]]
77
+ ),
78
+ }
79
+ )
80
+ return datasets.DatasetInfo(
81
+ description=_DESCRIPTION,
82
+ features=features,
83
+ supervised_keys=None,
84
+ homepage=_HOMEPAGE,
85
+ license=_LICENSE,
86
+ citation=None,
87
+ )
88
+
89
+ def _split_generators(self, dl_manager):
90
+ """Returns SplitGenerators."""
91
+
92
+ data_file = dl_manager.download_and_extract(_URL)
93
+ return [
94
+ datasets.SplitGenerator(
95
+ name=datasets.Split.TRAIN,
96
+ gen_kwargs={
97
+ "filepath": data_file,
98
+ },
99
+ ),
100
+ ]
101
+
102
+ def _generate_examples(self, filepath):
103
+ """Yields examples."""
104
+
105
+ id_ = 0
106
+
107
+ with open(filepath, encoding="utf-8") as tsv_file:
108
+ reader = csv.reader(tsv_file, delimiter="\t")
109
+ for id_, row in enumerate(reader):
110
+ if id_ == 0:
111
+ continue
112
+
113
+ yield id_, {
114
+ "text": row[0],
115
+ "group": group_map[row[1]],
116
+ "wing": wing_map[row[2]],
117
+ }