Spaces:
Build error
Build error
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import gradio as gr
|
3 |
+
import shutil
|
4 |
+
import main
|
5 |
+
|
6 |
+
|
7 |
+
def predict_from_pdf(pdf_file):
|
8 |
+
upload_dir = "./catalogue/"
|
9 |
+
os.makedirs(upload_dir, exist_ok=True)
|
10 |
+
|
11 |
+
try:
|
12 |
+
dest_path = os.path.join(upload_dir, pdf_file.name)
|
13 |
+
with open(dest_path, "wb") as f:
|
14 |
+
f.write(pdf_file.read())
|
15 |
+
|
16 |
+
df, response = main_oss.process_pdf_catalog(dest_path)
|
17 |
+
return df, response
|
18 |
+
except Exception as e:
|
19 |
+
return None, f"Error: {str(e)}"
|
20 |
+
|
21 |
+
pdf_examples = [
|
22 |
+
["examples/flexpocket.pdf"],
|
23 |
+
["examples/ASICS_Catalog.pdf"],
|
24 |
+
]
|
25 |
+
|
26 |
+
demo = gr.Interface(
|
27 |
+
fn=predict_from_pdf,
|
28 |
+
inputs=gr.File(label="Upload PDF Catalog"),
|
29 |
+
outputs=["json", "text"],
|
30 |
+
examples=pdf_examples,
|
31 |
+
title="Open Source PDF Catalog Parser",
|
32 |
+
description="Efficient PDF catalog processing using MinerU and OpenLLM",
|
33 |
+
article="Uses MinerU for layout analysis and DeepSeek-1.3B for structured extraction"
|
34 |
+
)
|
35 |
+
|
36 |
+
if __name__ == "__main__":
|
37 |
+
demo.queue().launch(server_name="0.0.0.0", server_port=7860)
|