Spaces:
Sleeping
Sleeping
File size: 5,212 Bytes
49cd06a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 |
from smolagents import tool
import os
from newsapi import NewsApiClient
import datetime
@tool
def get_news_headlines(topic: str, count: int = 5) -> str:
"""Get latest news headlines for a specific topic
Args:
topic: Topic to search news for
count: Number of headlines to return (default: 5)
"""
API_KEY = os.getenv("NEWSAPI_KEY")
if not API_KEY:
return "Error: NewsAPI key not found in environment variables"
newsapi = NewsApiClient(api_key=API_KEY)
try:
# Define search strategies with different parameters
search_strategies = [
{
'query': f'"{topic}"', # Exact phrase match
'relevance': 'high'
},
{
'query': topic, # Normal search
'relevance': 'high'
},
{
'query': f"{topic} latest", # Latest news
'relevance': 'medium'
}
]
relevant_articles = [] # Store only relevant articles
seen_titles = set()
required_keywords = set(topic.lower().split())
# Function to check article relevance
def is_relevant(article, required_words, relevance_level):
title = article['title'].lower()
description = (article.get('description') or '').lower()
content = (article.get('content') or '').lower()
# Count how many required words appear in the article
title_matches = sum(1 for word in required_words if word in title)
desc_matches = sum(1 for word in required_words if word in description)
content_matches = sum(1 for word in required_words if word in content)
# Calculate relevance score
total_score = (title_matches * 3) + (desc_matches * 2) + content_matches
# For exact phrase matching
if relevance_level == 'high':
# Check if the exact topic phrase appears
if topic.lower() in title or topic.lower() in description:
return True
return total_score >= len(required_words) * 2
else:
return total_score >= len(required_words)
for strategy in search_strategies:
if len(relevant_articles) >= count:
break
# Calculate how many more articles we need
remaining_count = count - len(relevant_articles)
try:
news = newsapi.get_everything(
q=strategy['query'],
language='en',
sort_by='relevancy', # Changed to relevancy sort
page_size=min(50, remaining_count * 5) # Request more articles to filter through
)
if news['articles']:
for article in news['articles']:
# Skip if we've seen this title or have enough articles
if article['title'] in seen_titles:
continue
# Check if article is relevant enough
if is_relevant(article, required_keywords, strategy['relevance']):
seen_titles.add(article['title'])
pub_date = datetime.datetime.strptime(article['publishedAt'], '%Y-%m-%dT%H:%M:%SZ')
relevant_articles.append({
'title': article['title'],
'source': article['source']['name'],
'date': pub_date,
'url': article['url'],
'relevance': strategy['relevance']
})
# Break if we have enough relevant articles
if len(relevant_articles) >= count:
break
except Exception as e:
continue # If one strategy fails, try the next one
# Sort by date (newest first)
relevant_articles.sort(key=lambda x: x['date'], reverse=True)
if relevant_articles:
headlines = []
for idx, article in enumerate(relevant_articles, 1):
date_str = article['date'].strftime('%Y-%m-%d %H:%M UTC')
relevance_indicator = "🎯" if article['relevance'] == 'high' else "✓"
headlines.append(f"{idx}. {relevance_indicator} [{date_str}] {article['title']} ({article['source']})")
# Add a summary of how many relevant articles were found
found_count = len(relevant_articles)
summary = f"Found {found_count} relevant {'article' if found_count == 1 else 'articles'} out of {count} requested.\n\n"
return summary + "\n".join(headlines)
return f"No relevant news found for topic: {topic}"
except Exception as e:
return f"Error fetching news: {str(e)}" |