Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import fitz
|
3 |
+
import gradio as gr
|
4 |
+
import re
|
5 |
+
from summarizer import Summarizer
|
6 |
+
|
7 |
+
def preprocess(text):
|
8 |
+
text = text.replace('\n', ' ')
|
9 |
+
text = re.sub('\s+', ' ', text)
|
10 |
+
return text
|
11 |
+
|
12 |
+
|
13 |
+
def pdf_to_text(path, start_page=1, end_page=None):
|
14 |
+
doc = fitz.open(path)
|
15 |
+
total_pages = doc.page_count
|
16 |
+
|
17 |
+
if end_page is None:
|
18 |
+
end_page = total_pages
|
19 |
+
|
20 |
+
text_list = []
|
21 |
+
|
22 |
+
for i in range(start_page-1, end_page):
|
23 |
+
text = doc.load_page(i).get_text("text")
|
24 |
+
text = preprocess(text)
|
25 |
+
text_list.append(text)
|
26 |
+
|
27 |
+
doc.close()
|
28 |
+
return ' '.join(text_list)
|
29 |
+
|
30 |
+
|
31 |
+
def generate_summary(text, model='bert-base-uncased', ratio=0.2):
|
32 |
+
model = Summarizer(model_name=model)
|
33 |
+
summary = model(text, ratio=ratio)
|
34 |
+
return summary
|
35 |
+
|
36 |
+
|
37 |
+
def pdf_summary(file, secret):
|
38 |
+
if secret != os.environ.get('Secret'):
|
39 |
+
return '[Error]: Please provide the correct secret'
|
40 |
+
|
41 |
+
elif file is None:
|
42 |
+
return '[ERROR]: Please upload a PDF file.'
|
43 |
+
|
44 |
+
else:
|
45 |
+
old_file_name = file.name
|
46 |
+
file_name = file.name
|
47 |
+
file_name = file_name[:-12] + file_name[-4:]
|
48 |
+
os.rename(old_file_name, file_name)
|
49 |
+
text = pdf_to_text(file_name)
|
50 |
+
|
51 |
+
if text.strip() == '':
|
52 |
+
return '[ERROR]: The content of PDF is empty.'
|
53 |
+
|
54 |
+
return generate_summary(text)
|
55 |
+
|
56 |
+
|
57 |
+
title = 'PDF Summarizer'
|
58 |
+
description = "A platform for generating summary for a PDF using BERT model"
|
59 |
+
|
60 |
+
with gr.Interface(
|
61 |
+
fn=pdf_summary,
|
62 |
+
inputs=[
|
63 |
+
gr.File(label='PDF', file_types=['.pdf']),
|
64 |
+
gr.Textbox(label='Secret')
|
65 |
+
],
|
66 |
+
outputs=gr.Textbox(label='Summary'),
|
67 |
+
title=title,
|
68 |
+
description=description
|
69 |
+
) as iface:
|
70 |
+
iface.launch()
|