Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from newspaper import Article
|
3 |
+
|
4 |
+
# Function to scrape the article from a given URL
|
5 |
+
def scrape_article(url):
|
6 |
+
try:
|
7 |
+
# Create an Article object
|
8 |
+
article = Article(url)
|
9 |
+
|
10 |
+
# Download and parse the article
|
11 |
+
article.download()
|
12 |
+
article.parse()
|
13 |
+
|
14 |
+
# Extract the title and text content
|
15 |
+
title = article.title
|
16 |
+
content = article.text
|
17 |
+
|
18 |
+
if not content:
|
19 |
+
return "Failed to scrape content. Please check the URL."
|
20 |
+
|
21 |
+
return f"Title: {title}\n\nContent:\n{content}"
|
22 |
+
except Exception as e:
|
23 |
+
return f"Error: {str(e)}"
|
24 |
+
|
25 |
+
# Gradio interface
|
26 |
+
def main():
|
27 |
+
# Define the Gradio input-output interface
|
28 |
+
with gr.Blocks() as interface:
|
29 |
+
gr.Markdown("""## Article Scraper using Newspaper3k
|
30 |
+
Enter a URL to scrape the article content.""")
|
31 |
+
|
32 |
+
# Input and output widgets
|
33 |
+
url_input = gr.Textbox(label="Enter the Article URL")
|
34 |
+
output = gr.Textbox(label="Scraped Article", lines=15)
|
35 |
+
|
36 |
+
# Button to trigger scraping
|
37 |
+
scrape_button = gr.Button("Scrape Article")
|
38 |
+
|
39 |
+
# Button functionality
|
40 |
+
scrape_button.click(scrape_article, inputs=[url_input], outputs=[output])
|
41 |
+
|
42 |
+
return interface
|
43 |
+
|
44 |
+
if __name__ == "__main__":
|
45 |
+
interface = main()
|
46 |
+
interface.launch()
|