Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import streamlit as st
|
| 2 |
+
from wordcloud import WordCloud
|
| 3 |
+
import requests
|
| 4 |
+
import xml.etree.ElementTree as ET
|
| 5 |
+
from io import BytesIO
|
| 6 |
+
from datetime import datetime, timedelta
|
| 7 |
+
|
| 8 |
+
# Set page configuration
|
| 9 |
+
st.set_page_config(page_title="Word Cloud from News Headlines", layout="wide")
|
| 10 |
+
|
| 11 |
+
# Custom CSS for unique design and to remove the white bar on top
|
| 12 |
+
st.markdown("""
|
| 13 |
+
<style>
|
| 14 |
+
.stApp {
|
| 15 |
+
background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
|
| 16 |
+
color: white;
|
| 17 |
+
}
|
| 18 |
+
.css-18e3th9 {
|
| 19 |
+
background-color: rgba(255, 255, 255, 0.1);
|
| 20 |
+
padding: 20px;
|
| 21 |
+
border-radius: 10px;
|
| 22 |
+
}
|
| 23 |
+
.css-1d391kg {
|
| 24 |
+
background-color: rgba(255, 255, 255, 0.1);
|
| 25 |
+
padding: 20px;
|
| 26 |
+
border-radius: 10px;
|
| 27 |
+
}
|
| 28 |
+
.st-emotion-cache-18ni7ap {
|
| 29 |
+
display: none;
|
| 30 |
+
}
|
| 31 |
+
.stButton button {
|
| 32 |
+
color: black !important;
|
| 33 |
+
background-color: #667eea;
|
| 34 |
+
border-radius: 10px;
|
| 35 |
+
padding: 10px 20px;
|
| 36 |
+
font-weight: bold;
|
| 37 |
+
}
|
| 38 |
+
.headline-container {
|
| 39 |
+
background-color: rgba(255, 255, 255, 0.1);
|
| 40 |
+
padding: 20px;
|
| 41 |
+
border-radius: 10px;
|
| 42 |
+
margin-top: 20px;
|
| 43 |
+
}
|
| 44 |
+
.st-emotion-cache-7ym5gk {
|
| 45 |
+
color: black !important;
|
| 46 |
+
}
|
| 47 |
+
</style>
|
| 48 |
+
""", unsafe_allow_html=True)
|
| 49 |
+
|
| 50 |
+
# Title and description
|
| 51 |
+
st.title("Word Cloud from News Headlines")
|
| 52 |
+
st.markdown("### Generating a word cloud from live news headlines")
|
| 53 |
+
|
| 54 |
+
# Sidebar for user inputs
|
| 55 |
+
st.sidebar.title("Customize Your Word Cloud")
|
| 56 |
+
bg_color = st.sidebar.color_picker("Background Color", "#ffffff")
|
| 57 |
+
max_words = st.sidebar.slider("Maximum Number of Words", 10, 200, 100)
|
| 58 |
+
keyword = st.sidebar.text_input("Search Keyword (Optional)")
|
| 59 |
+
date_range = st.sidebar.date_input("Date Range", [datetime.now() - timedelta(days=7), datetime.now()])
|
| 60 |
+
|
| 61 |
+
# Function to fetch news headlines from Google News RSS feed
|
| 62 |
+
def fetch_news_headlines(keyword=None, date_range=None):
|
| 63 |
+
url = "https://news.google.com/rss"
|
| 64 |
+
if keyword:
|
| 65 |
+
url += f"/search?q={keyword}"
|
| 66 |
+
response = requests.get(url)
|
| 67 |
+
root = ET.fromstring(response.content)
|
| 68 |
+
headlines = [item.find('title').text for item in root.findall('./channel/item')]
|
| 69 |
+
return headlines
|
| 70 |
+
|
| 71 |
+
# Generate word cloud
|
| 72 |
+
headlines = fetch_news_headlines(keyword, date_range)
|
| 73 |
+
if headlines:
|
| 74 |
+
wordcloud_text = ' '.join(headlines)
|
| 75 |
+
wordcloud = WordCloud(width=800, height=400, background_color=bg_color, max_words=max_words).generate(wordcloud_text)
|
| 76 |
+
st.image(wordcloud.to_array(), use_column_width=True)
|
| 77 |
+
|
| 78 |
+
# Download option
|
| 79 |
+
img = BytesIO()
|
| 80 |
+
wordcloud.to_image().save(img, format='PNG')
|
| 81 |
+
st.download_button(label="Download Word Cloud", data=img, file_name="wordcloud.png", mime="image/png")
|
| 82 |
+
|
| 83 |
+
# Display headlines
|
| 84 |
+
st.markdown("### Fetched Headlines")
|
| 85 |
+
with st.expander("Show Headlines"):
|
| 86 |
+
st.markdown('<div class="headline-container">', unsafe_allow_html=True)
|
| 87 |
+
for headline in headlines:
|
| 88 |
+
st.markdown(f"- {headline}")
|
| 89 |
+
st.markdown('</div>', unsafe_allow_html=True)
|
| 90 |
+
else:
|
| 91 |
+
st.warning("No headlines fetched. Please try again later.")
|
| 92 |
+
|
| 93 |
+
# Footer
|
| 94 |
+
st.markdown("""
|
| 95 |
+
<hr>
|
| 96 |
+
<div style="text-align: center;">
|
| 97 |
+
Created by randeom
|
| 98 |
+
</div>
|
| 99 |
+
""", unsafe_allow_html=True)
|