Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,21 +1,24 @@
|
|
1 |
import gradio as gr
|
2 |
import pandas as pd
|
3 |
import requests
|
4 |
-
import
|
5 |
|
6 |
-
@spaces.GPU
|
7 |
def convert_parquet_to_jsonl(parquet_file=None, parquet_url=None):
|
8 |
if parquet_file is not None:
|
9 |
df = pd.read_parquet(parquet_file.name)
|
10 |
elif parquet_url is not None:
|
11 |
response = requests.get(parquet_url)
|
12 |
-
|
|
|
13 |
else:
|
14 |
raise ValueError("Either parquet_file or parquet_url must be provided")
|
|
|
15 |
jsonl_data = df.to_json(orient='records', lines=True)
|
16 |
-
|
|
|
17 |
f.write(jsonl_data)
|
18 |
-
|
|
|
19 |
|
20 |
demo = gr.Interface(
|
21 |
fn=convert_parquet_to_jsonl,
|
@@ -26,4 +29,4 @@ demo = gr.Interface(
|
|
26 |
)
|
27 |
|
28 |
if __name__ == "__main__":
|
29 |
-
demo.launch()
|
|
|
1 |
import gradio as gr
|
2 |
import pandas as pd
|
3 |
import requests
|
4 |
+
from io import BytesIO
|
5 |
|
|
|
6 |
def convert_parquet_to_jsonl(parquet_file=None, parquet_url=None):
|
7 |
if parquet_file is not None:
|
8 |
df = pd.read_parquet(parquet_file.name)
|
9 |
elif parquet_url is not None:
|
10 |
response = requests.get(parquet_url)
|
11 |
+
response.raise_for_status() # Ensure the request was successful
|
12 |
+
df = pd.read_parquet(BytesIO(response.content))
|
13 |
else:
|
14 |
raise ValueError("Either parquet_file or parquet_url must be provided")
|
15 |
+
|
16 |
jsonl_data = df.to_json(orient='records', lines=True)
|
17 |
+
output_file_path = "output.jsonl"
|
18 |
+
with open(output_file_path, "w") as f:
|
19 |
f.write(jsonl_data)
|
20 |
+
|
21 |
+
return output_file_path
|
22 |
|
23 |
demo = gr.Interface(
|
24 |
fn=convert_parquet_to_jsonl,
|
|
|
29 |
)
|
30 |
|
31 |
if __name__ == "__main__":
|
32 |
+
demo.launch()
|