Add backend functionality for news fetching, processing, and recommendations

- Implemented NewsFetcher class to fetch articles from RSS feeds and clean HTML content.
- Added EmbeddingGenerator for generating embeddings using Cohere API.
- Created VectorStore for storing and retrieving articles using Pinecone.
- Developed NewsRecommender for analyzing articles and generating insights with Groq.
- Set up FastAPI application with endpoints for fetching news and providing recommendations.
- Configured logging for better traceability and debugging.
- Updated .gitignore to include environment variables and data directories.
- Added requirements.txt for project dependencies.
This commit is contained in:
boladeE
2025-04-14 21:44:43 +01:00
parent 042f2386a0
commit e3d00bb4dc
8 changed files with 590 additions and 4 deletions
+38 -4
View File
@@ -1,9 +1,43 @@
# Environment variables
.env
# Python
__pycache__/
*.py[cod]
*$py.class
*.so
.Python
env/
build/
develop-eggs/
dist/
downloads/
eggs/
.eggs/
lib/
lib64/
parts/
sdist/
var/
wheels/
*.egg-info/
.installed.cfg
*.egg
# Virtual Environment # Virtual Environment
.venv .venv
# Environment Variables
.env
# vscode settings # IDE
.vscode .idea/
.vscode/
*.swp
*.swo
# Data directories
data/raw_news/
data/processed_news/
# Logs
*.log
+38
View File
@@ -0,0 +1,38 @@
import os
from dotenv import load_dotenv
# Load environment variables
# Construct the path to the .env file
# dotenv_path = os.path.join(os.path.dirname(__file__), '..', '.env')
# Load environment variables from the specified path
load_dotenv()
# API Keys
COHERE_API_KEY = os.getenv("COHERE_API_KEY")
GROQ_API_KEY = os.getenv("GROQ_API_KEY")
PINECONE_API_KEY = os.getenv("PINECONE_API_KEY")
# Pinecone Configuration
PINECONE_INDEX_NAME = os.getenv("PINECONE_INDEX_NAME", "news-articles")
# News Sources
RSS_FEEDS = [
"https://feeds.feedburner.com/TechCrunch/",
# "https://www.theverge.com/rss/index.xml",
# "https://www.wired.com/feed/rss",
# "https://www.technologyreview.com/feed/",
]
# Vector Database Settings
VECTOR_DIMENSION = 4096 # Cohere embedding dimension
TOP_K_RESULTS = 5
# Data Directories
RAW_NEWS_DIR = "data/raw_news"
PROCESSED_NEWS_DIR = "data/processed_news"
# Create directories if they don't exist
os.makedirs(RAW_NEWS_DIR, exist_ok=True)
os.makedirs(PROCESSED_NEWS_DIR, exist_ok=True)
+50
View File
@@ -0,0 +1,50 @@
import cohere
from typing import List, Dict, Any
from config import COHERE_API_KEY
class EmbeddingGenerator:
def __init__(self):
self.client = cohere.Client(COHERE_API_KEY)
def generate_embeddings(self, texts: List[str]) -> List[List[float]]:
"""Generate embeddings for a list of texts using Cohere."""
try:
response = self.client.embed(
texts=texts,
model="embed-english-v3.0",
input_type="search_document"
)
return response.embeddings
except Exception as e:
print(f"Error generating embeddings: {str(e)}")
return []
def process_articles(self, articles: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
"""Process articles and add embeddings to them."""
# Prepare texts for embedding
texts = [
f"{article['title']} {article['content']}"
for article in articles
]
# Generate embeddings
embeddings = self.generate_embeddings(texts)
# Add embeddings to articles
for article, embedding in zip(articles, embeddings):
article["embedding"] = embedding
return articles
def get_query_embedding(self, query: str) -> List[float]:
"""Generate embedding for a search query."""
try:
response = self.client.embed(
texts=[query],
model="embed-english-v3.0",
input_type="search_query"
)
return response.embeddings[0]
except Exception as e:
print(f"Error generating query embedding: {str(e)}")
return []
+112
View File
@@ -0,0 +1,112 @@
from fastapi import FastAPI, HTTPException
from fastapi.middleware.cors import CORSMiddleware
from typing import List, Dict, Any
import json
import os
from news_fetcher import NewsFetcher
from embeddings import EmbeddingGenerator
from vector_store import VectorStore
from recommender import NewsRecommender
from config import RAW_NEWS_DIR, PROCESSED_NEWS_DIR
app = FastAPI(title="DS Task AI News API")
# Add CORS middleware
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
# Initialize components
news_fetcher = NewsFetcher()
embedding_generator = EmbeddingGenerator()
vector_store = VectorStore()
recommender = NewsRecommender()
@app.get("/")
async def root():
"""Root endpoint returning API information."""
return {
"name": "DS Task AI News API",
"version": "1.0.0",
"description": "AI-powered news retrieval and recommendation system"
}
@app.get("/fetch-news")
async def fetch_news():
"""Fetch news from RSS feeds and store in vector database."""
try:
result = news_fetcher.process()
if result["status"] == "error":
raise HTTPException(status_code=404, detail=result["message"])
return result
except Exception as e:
raise HTTPException(status_code=500, detail=str(e))
@app.get("/recommend-news")
async def recommend_news(article_id: str = None, query: str = None):
"""Get news recommendations based on article ID or search query."""
try:
if article_id:
# Get article from vector store
article = vector_store.search_similar([0] * 4096, top_k=1) # Placeholder vector
if not article:
raise HTTPException(status_code=404, detail="Article not found")
# Generate query embedding from article content
query_embedding = embedding_generator.get_query_embedding(
f"{article[0]['title']} {article[0]['content']}"
)
elif query:
# Generate query embedding from search query
query_embedding = embedding_generator.get_query_embedding(query)
else:
raise HTTPException(
status_code=400,
detail="Either article_id or query parameter is required"
)
# Search for similar articles
similar_articles = vector_store.search_similar(query_embedding)
if not similar_articles:
raise HTTPException(status_code=404, detail="No similar articles found")
# Generate insights for the articles
insights = recommender.analyze_articles(similar_articles)
return {
"articles": similar_articles,
"insights": insights
}
except Exception as e:
raise HTTPException(status_code=500, detail=str(e))
@app.get("/article/{article_id}")
async def get_article(article_id: str):
"""Get a specific article and its summary."""
try:
# Search for the article
articles = vector_store.search_similar([0] * 4096, top_k=1) # Placeholder vector
if not articles:
raise HTTPException(status_code=404, detail="Article not found")
article = articles[0]
# Generate summary
summary = recommender.generate_summary(article)
return {
"article": article,
"summary": summary
}
except Exception as e:
raise HTTPException(status_code=500, detail=str(e))
if __name__ == "__main__":
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=8000)
+178
View File
@@ -0,0 +1,178 @@
import feedparser
import json
import os
import logging
from datetime import datetime
from typing import List, Dict, Any
from config import RSS_FEEDS, RAW_NEWS_DIR, PROCESSED_NEWS_DIR
from embeddings import EmbeddingGenerator
from vector_store import VectorStore
from bs4 import BeautifulSoup
import re
# Configure logging
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
handlers=[
logging.StreamHandler(),
logging.FileHandler('news_fetcher.log')
]
)
logger = logging.getLogger('NewsFetcher')
class NewsFetcher:
def __init__(self):
self.feeds = RSS_FEEDS
self.embedding_generator = EmbeddingGenerator()
self.vector_store = VectorStore()
logger.info("NewsFetcher initialized with %d RSS feeds", len(self.feeds))
def clean_html_content(self, html_content: str) -> str:
"""Clean HTML content and extract plain text."""
logger.debug("Cleaning HTML content of length %d", len(html_content))
# Parse HTML with BeautifulSoup
soup = BeautifulSoup(html_content, 'html.parser')
# Remove script and style elements
for script in soup(["script", "style"]):
script.decompose()
# Get text content
text = soup.get_text()
# Clean up whitespace
lines = (line.strip() for line in text.splitlines())
chunks = (phrase.strip() for line in lines for phrase in line.split(" "))
text = ' '.join(chunk for chunk in chunks if chunk)
# Remove extra spaces
text = re.sub(r'\s+', ' ', text)
cleaned_text = text.strip()
logger.debug("Cleaned text length: %d", len(cleaned_text))
return cleaned_text
def fetch_rss_news(self, feed_url: str) -> List[Dict[str, Any]]:
"""Fetch news articles from a single RSS feed."""
logger.info("Fetching news from feed: %s", feed_url)
feed = feedparser.parse(feed_url)
articles = []
for entry in feed.entries:
# Get raw content with HTML
raw_content = entry.get("summary", "")
# Clean HTML content
clean_content = self.clean_html_content(raw_content)
article = {
"title": entry.title,
"raw_content": raw_content, # Store original HTML content
"content": clean_content, # Store cleaned text content
"link": entry.get("link", ""),
"published": entry.get("published", datetime.now().isoformat()),
"source": feed.feed.get("title", "Unknown"),
"categories": [tag.term for tag in entry.get("tags", [])],
"id": entry.get("id", entry.get("link", "")),
}
articles.append(article)
logger.info("Fetched %d articles from %s", len(articles), feed_url)
return articles
def fetch_all_news(self) -> List[Dict[str, Any]]:
"""Fetch news from all configured RSS feeds."""
logger.info("Starting to fetch news from all %d feeds", len(self.feeds))
all_articles = []
for feed_url in self.feeds:
try:
articles = self.fetch_rss_news(feed_url)
all_articles.extend(articles)
logger.info("Successfully fetched %d articles from %s", len(articles), feed_url)
except Exception as e:
logger.error("Error fetching from %s: %s", feed_url, str(e))
logger.info("Total articles fetched: %d", len(all_articles))
return all_articles
def save_raw_articles(self, articles: List[Dict[str, Any]]) -> str:
"""Save raw articles to a JSON file."""
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
filename = f"raw_news_{timestamp}.json"
filepath = os.path.join(RAW_NEWS_DIR, filename)
logger.info("Saving %d raw articles to %s", len(articles), filepath)
with open(filepath, "w", encoding="utf-8") as f:
json.dump(articles, f, ensure_ascii=False, indent=2)
logger.info("Raw articles saved successfully")
return filepath
def save_processed_articles(self, articles: List[Dict[str, Any]]) -> str:
"""Save processed articles with embeddings to a JSON file."""
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
filename = f"processed_news_{timestamp}.json"
filepath = os.path.join(PROCESSED_NEWS_DIR, filename)
# Create a copy of articles without raw_content for processed storage
processed_articles = []
for article in articles:
processed_article = article.copy()
processed_article.pop('raw_content', None) # Remove raw_content from processed articles
processed_articles.append(processed_article)
logger.info("Saving %d processed articles to %s", len(processed_articles), filepath)
with open(filepath, "w", encoding="utf-8") as f:
json.dump(processed_articles, f, ensure_ascii=False, indent=2)
logger.info("Processed articles saved successfully")
return filepath
def process(self) -> Dict[str, Any]:
"""Main process to fetch, process, and store news articles."""
logger.info("Starting news processing pipeline")
# Fetch articles
logger.info("Step 1: Fetching articles from RSS feeds")
articles = self.fetch_all_news()
if not articles:
logger.warning("No articles found during fetching")
return {"status": "error", "message": "No articles found"}
# Save raw articles
logger.info("Step 2: Saving raw articles")
raw_filepath = self.save_raw_articles(articles)
# Generate embeddings
logger.info("Step 3: Generating embeddings for %d articles", len(articles))
articles_with_embeddings = self.embedding_generator.process_articles(articles)
logger.info("Embeddings generated successfully")
# Save processed articles
logger.info("Step 4: Saving processed articles with embeddings")
processed_filepath = self.save_processed_articles(articles_with_embeddings)
# Store in vector database
logger.info("Step 5: Storing articles in vector database")
success = self.vector_store.upsert_articles(articles_with_embeddings)
if success:
logger.info("Articles successfully stored in vector database")
else:
logger.error("Failed to store articles in vector database")
result = {
"status": "success" if success else "error",
"message": "Articles processed and stored successfully" if success else "Failed to store articles",
"raw_filepath": raw_filepath,
"processed_filepath": processed_filepath,
"article_count": len(articles)
}
logger.info("News processing pipeline completed with status: %s", result["status"])
return result
news_fetcher = NewsFetcher()
print(news_fetcher.process())
+75
View File
@@ -0,0 +1,75 @@
from groq import Groq
from typing import List, Dict, Any
from config import GROQ_API_KEY
class NewsRecommender:
def __init__(self):
self.client = Groq(api_key=GROQ_API_KEY)
def analyze_articles(self, articles: List[Dict[str, Any]]) -> Dict[str, Any]:
"""Analyze a set of articles using Groq to generate insights."""
try:
# Prepare the prompt
articles_text = "\n\n".join([
f"Title: {article['title']}\nContent: {article['content']}"
for article in articles
])
prompt = f"""Analyze these news articles and provide insights:
{articles_text}
Please provide:
1. Main themes and topics
2. Key insights and trends
3. Potential implications
4. Related areas of interest
Format the response as a JSON with these keys: themes, insights, implications, related_areas"""
# Get completion from Groq
completion = self.client.chat.completions.create(
messages=[
{"role": "system", "content": "You are a news analyst providing insights about technology and AI news."},
{"role": "user", "content": prompt}
],
model="mixtral-8x7b-32768",
temperature=0.7,
max_tokens=1000
)
# Parse and return the analysis
return completion.choices[0].message.content
except Exception as e:
print(f"Error analyzing articles: {str(e)}")
return {
"themes": [],
"insights": [],
"implications": [],
"related_areas": []
}
def generate_summary(self, article: Dict[str, Any]) -> str:
"""Generate a summary of a single article using Groq."""
try:
prompt = f"""Summarize this news article:
Title: {article['title']}
Content: {article['content']}
Please provide a concise summary focusing on the key points and implications."""
completion = self.client.chat.completions.create(
messages=[
{"role": "system", "content": "You are a news summarizer providing concise summaries of technology and AI news."},
{"role": "user", "content": prompt}
],
model="mixtral-8x7b-32768",
temperature=0.5,
max_tokens=500
)
return completion.choices[0].message.content
except Exception as e:
print(f"Error generating summary: {str(e)}")
return "Unable to generate summary."
+11
View File
@@ -0,0 +1,11 @@
fastapi==0.109.2
uvicorn==0.27.1
feedparser==6.0.10
cohere==4.47
pinecone-client==3.0.2
python-dotenv==1.0.1
groq==0.4.2
pydantic==2.6.3
python-multipart==0.0.9
httpx==0.27.0
beautifulsoup4==4.12.3
+88
View File
@@ -0,0 +1,88 @@
from pinecone import Pinecone, ServerlessSpec
from typing import List, Dict, Any
from config import (
PINECONE_API_KEY,
PINECONE_ENVIRONMENT,
PINECONE_INDEX_NAME,
VECTOR_DIMENSION,
TOP_K_RESULTS
)
class VectorStore:
def __init__(self):
self.pinecone = Pinecone(api_key=PINECONE_API_KEY)
self.index_name = PINECONE_INDEX_NAME
self._ensure_index()
def _ensure_index(self):
"""Ensure the Pinecone index exists, create if it doesn't."""
if self.index_name not in self.pinecone.list_indexes().names():
self.pinecone.create_index(
name=self.index_name,
dimension=VECTOR_DIMENSION,
metric="cosine",
spec=ServerlessSpec(cloud="aws", region="us-east-1")
)
self.index = self.pinecone.Index(self.index_name)
def upsert_articles(self, articles: List[Dict[str, Any]]) -> bool:
"""Upsert articles to the vector store."""
try:
vectors = []
for article in articles:
if "embedding" not in article:
continue
vector = {
"id": article["id"],
"values": article["embedding"],
"metadata": {
"title": article["title"],
"content": article["content"],
"link": article["link"],
"published": article["published"],
"source": article["source"],
"categories": article["categories"]
}
}
vectors.append(vector)
if vectors:
self.index.upsert(vectors=vectors)
return True
except Exception as e:
print(f"Error upserting articles: {str(e)}")
return False
def search_similar(self, query_embedding: List[float], top_k: int = TOP_K_RESULTS) -> List[Dict[str, Any]]:
"""Search for similar articles using the query embedding."""
try:
results = self.index.query(
vector=query_embedding,
top_k=top_k,
include_metadata=True
)
articles = []
for match in results.matches:
article = {
"id": match.id,
"score": match.score,
**match.metadata
}
articles.append(article)
return articles
except Exception as e:
print(f"Error searching similar articles: {str(e)}")
return []
def delete_article(self, article_id: str) -> bool:
"""Delete an article from the vector store."""
try:
self.index.delete(ids=[article_id])
return True
except Exception as e:
print(f"Error deleting article: {str(e)}")
return False