Spaces:
Build error
Build error
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import fitz # PyMuPDF
|
3 |
+
|
4 |
+
def pdf_to_markdown(pdf_file):
|
5 |
+
"""Extract text from a PDF and format it into markdown."""
|
6 |
+
if pdf_file is None:
|
7 |
+
return "No file uploaded."
|
8 |
+
|
9 |
+
doc = fitz.open(pdf_file.name)
|
10 |
+
markdown_text = []
|
11 |
+
|
12 |
+
for i, page in enumerate(doc):
|
13 |
+
text = page.get_text("text")
|
14 |
+
markdown_text.append(f"=== Page {i + 1}\n\n{text.strip()}")
|
15 |
+
|
16 |
+
return "\n\n".join(markdown_text)
|
17 |
+
|
18 |
+
# Gradio Interface
|
19 |
+
iface = gr.Interface(
|
20 |
+
fn=pdf_to_markdown,
|
21 |
+
inputs=gr.File(type="file"),
|
22 |
+
outputs=gr.Textbox(label="Markdown Output", lines=15),
|
23 |
+
title="PDF to Markdown Extractor",
|
24 |
+
description="Upload a PDF and get a copyable markdown output.",
|
25 |
+
)
|
26 |
+
|
27 |
+
if __name__ == "__main__":
|
28 |
+
iface.launch()
|