tools_engine.py

"""
tools_engine.py - Improved perform_web_search to return structured results with URLs and snippets,
and canonical intent detection unchanged.
"""

from duckduckgo_search import DDGS
from transformers import pipeline
import re

print(">>> Tools: Loading Intent Classification Model...")
intent_classifier = pipeline("zero-shot-classification", model="typeform/distilbert-base-uncased-mnli")

def analyze_intent(user_text):
    if not user_text:
        return "general"
    text_lower = user_text.lower().strip()
    direct_chat_triggers = [
        "hi","hello","hey","hlo","namaste",
        "what is your name", "who are you", "your name"
    ]
    if text_lower in direct_chat_triggers or any(text_lower.startswith(t + " ") for t in direct_chat_triggers):
        return "general"

    candidate_labels = ["internet search","general conversation","coding request","checking time"]
    try:
        result = intent_classifier(user_text, candidate_labels)
        top_label = result['labels'][0]
        confidence = result['scores'][0]
        mapping = {
            "internet search": "internet_search",
            "general conversation": "general",
            "coding request": "coding_request",
            "checking time": "checking_time"
        }
        if confidence > 0.45:
            return mapping.get(top_label, "general")
    except Exception:
        pass
    return "general"

def perform_web_search(user_text, max_results=4):
    """
    Return structured results:
    {
      "query": "...",
      "results": [
         {"title": "...", "snippet": "...", "url": "..."},
         ...
      ]
    }
    """
    try:
        query = user_text
        # sanitize small verbs
        remove_phrases = ["search for","find","google","look up","lookup","what is","tell me"]
        q = query.lower()
        for p in remove_phrases:
            q = q.replace(p, "")
        q = q.strip() or query

        results = list(DDGS().text(q, max_results=max_results))
        structured = {"query": q, "results": []}
        for r in results:
            title = r.get("title","").strip()
            body = re.sub(r'\s+',' ', r.get("body","").strip())
            url = r.get("href") or r.get("url") or r.get("link") or ""
            # short snippet
            snippet = body[:320]
            structured["results"].append({"title": title, "snippet": snippet, "url": url})
        return structured
    except Exception as e:
        print(f"Search error: {e}")
        return {"query": user_text, "results": []}