🛠️ Real-Time Sports News Summarizer with OpenAI Function Calling + AI Recommendations via Dappier

import os
from getpass import getpass

# Prompt for the Dappier API key securely
dappier_api_key = getpass('Enter your API key: ')
os.environ["DAPPIER_API_KEY"] = dappier_api_key

# Prompt for the OpenAI API key securely
openai_api_key = getpass('Enter your API key: ')
os.environ["OPENAI_API_KEY"] = openai_api_key

from openai import OpenAI
from dappier import Dappier

# Initialize clients
openai_client = OpenAI()
dappier_client = Dappier()

def get_latest_sports_news(
    query: str,
    similarity_top_k: int = 9,
    ref: str = "",
    num_articles_ref: int = 1,
    search_algorithm: str = "most_recent"
) -> str:
    response = dappier_client.get_ai_recommendations(
        query=query,
        data_model_id="dm_01j0pb465keqmatq9k83dthx34",
        similarity_top_k=similarity_top_k,
        ref=ref,
        num_articles_ref=num_articles_ref,
        search_algorithm=search_algorithm
    )

    if not response or not response.response:
        return "No relevant articles found."

    articles = []
    for article in response.response.results:
        formatted_article = (
            f"Title: {article.title}\n"
            f"Author: {article.author}\n"
            f"Published: {article.pubdate}\n"
            f"Source: {article.site} ({article.site_domain})\n"
            f"Preview: {article.preview_content}\n"
            f"URL: {article.url}\n"
        )
        articles.append(formatted_article)

    return "\n---\n".join(articles)

user_prompt = """
Summarize the most recent sports news using live content recommendations. Follow these steps:

1. Fetch AI-Curated News:
Retrieve the latest articles related to sports using the AI Recommendations API.

2. Summarize Content:
Review the titles and preview content of each article. Create a concise summary covering the key highlights across all articles.

3. Cite Sources:
For each summary point, include the article title and source link.
"""

tools = [{
    "type": "function",
    "function": {
        "name": "get_latest_sports_news",
        "description": "Fetches AI-recommended sports news articles and returns their details.",
        "parameters": {
            "type": "object",
            "properties": {
                "query": {
                    "type": "string",
                    "description": "Search query for the type of sports news to retrieve (e.g., 'latest sports news')"
                },
                "similarity_top_k": {
                    "type": "integer",
                    "description": "Number of top similar articles to retrieve"
                },
                "ref": {
                    "type": "string",
                    "description": "Preferred domain for article sources (e.g., 'boundingintosports.com')"
                },
                "num_articles_ref": {
                    "type": "integer",
                    "description": "Number of articles to fetch specifically from the preferred domain"
                },
                "search_algorithm": {
                    "type": "string",
                    "enum": ["semantic", "most_recent"],
                    "description": "Search algorithm to use: 'semantic' for contextual match or 'most_recent' for latest news"
                }
            },
            "required": ["query"]
        }
    }
}]

import json

def run_conversation(user_prompt: str):
    messages = [{"role": "user", "content": user_prompt}]

    # Step 1: Let OpenAI decide on needed functions
    response = openai_client.chat.completions.create(
        model="gpt-4o-mini",
        messages=messages,
        tools=tools,
        tool_choice="auto",
        temperature=0
    )

    response_message = response.choices[0].message

    # Step 2: Call the Dappier tool if requested
    if tool_calls := response_message.tool_calls:
        messages.append(response_message)

        for tool_call in tool_calls:
            function_args = json.loads(tool_call.function.arguments)
            tool_output = get_latest_sports_news(**function_args)

            messages.append({
                "tool_call_id": tool_call.id,
                "role": "tool",
                "name": tool_call.function.name,
                "content": tool_output,
            })

    # Step 3: Final call to OpenAI with all gathered data
    final_response = openai_client.chat.completions.create(
        model="gpt-4o-mini",
        messages=messages,
        temperature=0,
        stream=True
    )

    return final_response

if __name__ == "__main__":
    print("Fetching and summarizing sports news...\n")

    response = run_conversation(user_prompt)

    print("\n📰 Sports News Summary:\n")
    for chunk in response:
        print(chunk.choices[0].delta.content or "", end='', flush=True)