Upload PAIR.py
Browse files
PAIR.py
CHANGED
@@ -16,6 +16,7 @@
|
|
16 |
import json
|
17 |
|
18 |
import datasets
|
|
|
19 |
|
20 |
# TODO: Add BibTeX citation
|
21 |
# Find for instance the citation on arxiv or on the dataset repo/website
|
@@ -49,19 +50,43 @@ _URLS = {
|
|
49 |
|
50 |
|
51 |
# TODO: Name of the dataset usually matches the script name with CamelCase instead of snake_case
|
52 |
-
annotation2type = {
|
53 |
-
|
54 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
55 |
|
56 |
class PAIRDataset(datasets.GeneratorBasedBuilder):
|
57 |
"""PAIRDataset."""
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
|
|
|
|
|
|
|
|
|
|
63 |
def _info(self):
|
64 |
"""_info."""
|
|
|
|
|
65 |
return datasets.DatasetInfo(
|
66 |
description="My custom dataset.",
|
67 |
features=datasets.Features(
|
@@ -73,8 +98,8 @@ class PAIRDataset(datasets.GeneratorBasedBuilder):
|
|
73 |
),
|
74 |
supervised_keys=None,
|
75 |
)
|
76 |
-
|
77 |
-
|
78 |
|
79 |
def _split_generators(self, dl_manager):
|
80 |
"""_split_generators.
|
@@ -86,6 +111,7 @@ class PAIRDataset(datasets.GeneratorBasedBuilder):
|
|
86 |
"""
|
87 |
# Implement logic to download and extract data files
|
88 |
# For simplicity, assume data_files is a dict with paths to your data
|
|
|
89 |
data_files = {
|
90 |
"train": "train.json",
|
91 |
"test": "test.json",
|
@@ -110,26 +136,26 @@ class PAIRDataset(datasets.GeneratorBasedBuilder):
|
|
110 |
filepath
|
111 |
"""
|
112 |
# Implement your data reading logic here
|
|
|
113 |
with open(filepath, encoding="utf-8") as f:
|
114 |
data = json.load(f)
|
115 |
counter = 0
|
116 |
for idx, annotation_type in enumerate(data.keys()):
|
117 |
-
|
|
|
|
|
118 |
# Parse your line into the appropriate fields
|
119 |
samples = data[annotation_type]
|
120 |
for idx_2, elem in enumerate(samples):
|
121 |
# example = parse_line_to_example(line)
|
122 |
if elem["content"] != [None]:
|
123 |
-
unique_id = f"{elem['pid']}_{idx}"
|
124 |
content = elem["content"][0]
|
125 |
# print(literal_eval(content), "done")
|
126 |
-
yield
|
127 |
"sequence": elem["seq"],
|
128 |
"pid": elem["pid"],
|
129 |
annotation_type: content,
|
130 |
}
|
131 |
counter += 1
|
132 |
|
133 |
-
|
134 |
-
#"annotation_type": annotation_type,
|
135 |
-
|
|
|
16 |
import json
|
17 |
|
18 |
import datasets
|
19 |
+
from datasets import BuilderConfig
|
20 |
|
21 |
# TODO: Add BibTeX citation
|
22 |
# Find for instance the citation on arxiv or on the dataset repo/website
|
|
|
50 |
|
51 |
|
52 |
# TODO: Name of the dataset usually matches the script name with CamelCase instead of snake_case
|
53 |
+
annotation2type = {
|
54 |
+
"names": datasets.Value("string"),
|
55 |
+
"EC": datasets.Sequence(datasets.Value("string")),
|
56 |
+
}
|
57 |
+
|
58 |
+
|
59 |
+
class CustomConfig(datasets.BuilderConfig):
|
60 |
+
"""CustomConfig."""
|
61 |
+
|
62 |
+
def __init__(self, **kwargs):
|
63 |
+
"""__init__.
|
64 |
+
|
65 |
+
Parameters
|
66 |
+
----------
|
67 |
+
kwargs :
|
68 |
+
kwargs
|
69 |
+
"""
|
70 |
+
self.annotation_type = kwargs.pop("annotation_type", "function")
|
71 |
+
super(CustomConfig, self).__init__(**kwargs)
|
72 |
+
|
73 |
|
74 |
class PAIRDataset(datasets.GeneratorBasedBuilder):
|
75 |
"""PAIRDataset."""
|
76 |
+
|
77 |
+
BUILDER_CONFIGS = [
|
78 |
+
CustomConfig(
|
79 |
+
name="custom_config",
|
80 |
+
version="1.0.0",
|
81 |
+
description="your description",
|
82 |
+
),
|
83 |
+
] # Configs initialization
|
84 |
+
BUILDER_CONFIG_CLASS = CustomConfig # Must specify this to use custom config
|
85 |
+
|
86 |
def _info(self):
|
87 |
"""_info."""
|
88 |
+
self.annotation_type = self.config_kwargs["annotation_type"]
|
89 |
+
# Confirm annotation_type is set before continuing
|
90 |
return datasets.DatasetInfo(
|
91 |
description="My custom dataset.",
|
92 |
features=datasets.Features(
|
|
|
98 |
),
|
99 |
supervised_keys=None,
|
100 |
)
|
101 |
+
# "annotation_type": datasets.Value("string"),
|
102 |
+
# "annotation": datasets.Value("string"),
|
103 |
|
104 |
def _split_generators(self, dl_manager):
|
105 |
"""_split_generators.
|
|
|
111 |
"""
|
112 |
# Implement logic to download and extract data files
|
113 |
# For simplicity, assume data_files is a dict with paths to your data
|
114 |
+
print("in generator self.annotation", self.annotation_type)
|
115 |
data_files = {
|
116 |
"train": "train.json",
|
117 |
"test": "test.json",
|
|
|
136 |
filepath
|
137 |
"""
|
138 |
# Implement your data reading logic here
|
139 |
+
print("in generator 2 self.annotation", self.annotation_type)
|
140 |
with open(filepath, encoding="utf-8") as f:
|
141 |
data = json.load(f)
|
142 |
counter = 0
|
143 |
for idx, annotation_type in enumerate(data.keys()):
|
144 |
+
print(annotation_type, self.annotation_type)
|
145 |
+
if annotation_type != self.annotation_type:
|
146 |
+
continue
|
147 |
# Parse your line into the appropriate fields
|
148 |
samples = data[annotation_type]
|
149 |
for idx_2, elem in enumerate(samples):
|
150 |
# example = parse_line_to_example(line)
|
151 |
if elem["content"] != [None]:
|
|
|
152 |
content = elem["content"][0]
|
153 |
# print(literal_eval(content), "done")
|
154 |
+
yield counter, {
|
155 |
"sequence": elem["seq"],
|
156 |
"pid": elem["pid"],
|
157 |
annotation_type: content,
|
158 |
}
|
159 |
counter += 1
|
160 |
|
161 |
+
# "annotation_type": annotation_type,
|
|
|
|