Spaces:
Running
Running
slimshadow
commited on
Commit
•
37264eb
1
Parent(s):
ba92e06
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import fitz # PyMuPDF
|
3 |
+
from docx import Document
|
4 |
+
from io import BytesIO
|
5 |
+
|
6 |
+
def convert_pdf_to_word(pdf_file):
|
7 |
+
# Open the PDF file
|
8 |
+
pdf_document = fitz.open(stream=pdf_file.read(), filetype="pdf")
|
9 |
+
doc = Document()
|
10 |
+
|
11 |
+
# Extract text from each page
|
12 |
+
for page_num in range(len(pdf_document)):
|
13 |
+
page = pdf_document.load_page(page_num)
|
14 |
+
text = page.get_text()
|
15 |
+
doc.add_paragraph(text)
|
16 |
+
|
17 |
+
return doc
|
18 |
+
|
19 |
+
def main():
|
20 |
+
st.title("PDF to Word Converter")
|
21 |
+
|
22 |
+
uploaded_file = st.file_uploader("Choose a PDF file", type="pdf")
|
23 |
+
|
24 |
+
if uploaded_file is not None:
|
25 |
+
with st.spinner('Converting...'):
|
26 |
+
word_doc = convert_pdf_to_word(uploaded_file)
|
27 |
+
buffer = BytesIO()
|
28 |
+
word_doc.save(buffer)
|
29 |
+
buffer.seek(0)
|
30 |
+
|
31 |
+
st.success('Conversion successful!')
|
32 |
+
st.download_button(
|
33 |
+
label="Download Word document",
|
34 |
+
data=buffer,
|
35 |
+
file_name="converted_document.docx",
|
36 |
+
mime="application/vnd.openxmlformats-officedocument.wordprocessingml.document"
|
37 |
+
)
|
38 |
+
|
39 |
+
if __name__ == "__main__":
|
40 |
+
main()
|