Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -3,6 +3,7 @@ from datasets import load_dataset, concatenate_datasets
|
|
3 |
import json
|
4 |
import os
|
5 |
import base64
|
|
|
6 |
|
7 |
def load_and_combine_datasets():
|
8 |
python_codes_dataset = load_dataset('flytech/python-codes-25k', split='train')
|
@@ -13,10 +14,14 @@ def load_and_combine_datasets():
|
|
13 |
|
14 |
return combined_dataset
|
15 |
|
|
|
|
|
|
|
|
|
16 |
def save_combined_dataset_as_jsonl(combined_dataset, file_path):
|
17 |
with open(file_path, 'w', encoding='utf-8') as f:
|
18 |
for example in combined_dataset:
|
19 |
-
json.dump(example, f, ensure_ascii=False)
|
20 |
f.write('\n')
|
21 |
|
22 |
def main():
|
@@ -42,8 +47,8 @@ def main():
|
|
42 |
file_path = os.path.join(os.getcwd(), f"{output_dataset_name}.jsonl")
|
43 |
save_combined_dataset_as_jsonl(combined_dataset, file_path)
|
44 |
st.write("Download the combined dataset as JSONL file:")
|
45 |
-
with open(file_path, "
|
46 |
-
bytes_data = f.read()
|
47 |
b64 = base64.b64encode(bytes_data).decode()
|
48 |
href = f'<a href="data:file/jsonl;base64,{b64}" download="{output_dataset_name}.jsonl">Download JSONL File</a>'
|
49 |
st.markdown(href, unsafe_allow_html=True)
|
@@ -55,4 +60,4 @@ def main():
|
|
55 |
mime="application/jsonl")
|
56 |
|
57 |
if __name__ == "__main__":
|
58 |
-
main()
|
|
|
3 |
import json
|
4 |
import os
|
5 |
import base64
|
6 |
+
from datetime import datetime
|
7 |
|
8 |
def load_and_combine_datasets():
|
9 |
python_codes_dataset = load_dataset('flytech/python-codes-25k', split='train')
|
|
|
14 |
|
15 |
return combined_dataset
|
16 |
|
17 |
+
def datetime_serializer(o):
|
18 |
+
if isinstance(o, datetime):
|
19 |
+
return o.strftime('%Y-%m-%d %H:%M:%S')
|
20 |
+
|
21 |
def save_combined_dataset_as_jsonl(combined_dataset, file_path):
|
22 |
with open(file_path, 'w', encoding='utf-8') as f:
|
23 |
for example in combined_dataset:
|
24 |
+
json.dump(example, f, ensure_ascii=False, default=datetime_serializer)
|
25 |
f.write('\n')
|
26 |
|
27 |
def main():
|
|
|
47 |
file_path = os.path.join(os.getcwd(), f"{output_dataset_name}.jsonl")
|
48 |
save_combined_dataset_as_jsonl(combined_dataset, file_path)
|
49 |
st.write("Download the combined dataset as JSONL file:")
|
50 |
+
with open(file_path, "rb") as f:
|
51 |
+
bytes_data = f.read()
|
52 |
b64 = base64.b64encode(bytes_data).decode()
|
53 |
href = f'<a href="data:file/jsonl;base64,{b64}" download="{output_dataset_name}.jsonl">Download JSONL File</a>'
|
54 |
st.markdown(href, unsafe_allow_html=True)
|
|
|
60 |
mime="application/jsonl")
|
61 |
|
62 |
if __name__ == "__main__":
|
63 |
+
main()
|