| from typing import List |
|
|
| import requests |
| from bs4 import BeautifulSoup |
| from langchain_core.tools import tool |
|
|
|
|
| DUCKDUCKGO_HTML_URL = "https://html.duckduckgo.com/html/" |
| MAX_RESULTS = 5 |
| REQUEST_TIMEOUT_SECONDS = 10 |
| FALLBACK_MESSAGE = ( |
| "External search returned no results or is unavailable. " |
| "Answer using the available medical knowledge and clearly state " |
| "that no external source was retrieved." |
| ) |
|
|
|
|
| def _search_duckduckgo(query: str) -> str: |
| response = requests.get( |
| DUCKDUCKGO_HTML_URL, |
| params={"q": query, "kl": "us-en", "kp": "-1"}, |
| headers={"User-Agent": "dmChatbot/1.0"}, |
| timeout=REQUEST_TIMEOUT_SECONDS, |
| ) |
| response.raise_for_status() |
|
|
| soup = BeautifulSoup(response.text, "html.parser") |
| results: List[str] = [] |
| for result in soup.select(".result")[:MAX_RESULTS]: |
| link = result.select_one(".result__a") |
| if not link: |
| continue |
|
|
| title = link.get_text(" ", strip=True) |
| url = link.get("href", "").strip() |
| snippet_element = result.select_one(".result__snippet") |
| snippet = snippet_element.get_text(" ", strip=True) if snippet_element else "" |
| details = [title] |
| if snippet: |
| details.append(snippet) |
| if url: |
| details.append(url) |
| results.append("\n".join(details)) |
|
|
| if not results: |
| raise ValueError("DuckDuckGo returned no search results") |
| return "\n\n".join(results) |
|
|
|
|
| @tool |
| def web_search_tool(query: str) -> str: |
| """Search the web for medical information or nutritional data.""" |
| try: |
| return _search_duckduckgo(query) |
| except Exception: |
| return FALLBACK_MESSAGE |
|
|
|
|
| def search_web(query: str) -> str: |
| """Useful for searching the web for medical information or nutritional data.""" |
| return web_search_tool.invoke({"query": query}) |
