data/recommender.py

"""
data/recommender.py — Rule-Based Worker Recommendation Engine for FixItGo

Given problem text + optional area, returns the best matching workers:
  1. Use AI recommender to detect category
  2. Filter workers by that category
  3. Sort by: area match bonus (local workers first) + rating DESC + total reviews
  4. Return top N with explanation strings.

Usage:
    from data.recommender import recommend_workers
    results = recommend_workers("tap is leaking badly", "Adyar")
    # → list of dicts with worker info + reason string
"""

from data.ai_recommender import recommend_service
from data.database import get_workers


def recommend_workers(problem_text: str, user_area: str = "", top_n: int = 3) -> dict:
    """
    Recommend the best workers for a given problem.

    Args:
        problem_text : Free-text description of the problem.
        user_area    : User's locality (optional, for proximity scoring).
        top_n        : Number of workers to return (default 3).

    Returns:
        {
          'ai_result'   : dict from recommend_service(),
          'workers'     : list of top worker dicts with 'reason' key added,
          'message'     : human-readable summary string,
        }
    """
    ai = recommend_service(problem_text)
    category = ai.get("category")

    if not category:
        return {
            "ai_result": ai,
            "workers":   [],
            "message":   "Could not detect service category. Please describe the problem in more detail.",
        }

    # ── Fetch workers for the suggested category ──────────────────────────
    workers = get_workers(category=category, sort="rating")

    if not workers:
        return {
            "ai_result": ai,
            "workers":   [],
            "message":   f"No {category} workers found in the database yet.",
        }

    # ── Score each worker ─────────────────────────────────────────────────
    area_lower = user_area.lower().strip()

    scored = []
    for w in workers:
        score = 0.0

        # Rating weight (max 5 → contribute up to 50 points)
        score += float(w.get("rating") or 0) * 10

        # Review volume bonus (log scale, up to 10 pts)
        import math
        reviews = int(w.get("total_reviews") or 0)
        if reviews > 0:
            score += min(10, math.log(reviews + 1, 2))

        # Area proximity bonus (up to 20 pts)
        worker_area = (w.get("area") or "").lower()
        if area_lower and area_lower in worker_area:
            score += 20          # exact area match
        elif area_lower and any(part in worker_area for part in area_lower.split()):
            score += 8           # partial match

        scored.append((score, w))

    # Best score first
    scored.sort(key=lambda x: x[0], reverse=True)
    top_workers = scored[:top_n]

    # ── Build result list with explanation ────────────────────────────────
    result_workers = []
    for rank, (score, w) in enumerate(top_workers, start=1):
        reasons = []

        rating = float(w.get("rating") or 0)
        if rating >= 4.7:
            reasons.append("⭐ Top rated")
        elif rating >= 4.0:
            reasons.append("👍 Highly rated")
        else:
            reasons.append("✅ Verified professional")

        # Area match reason
        if area_lower and area_lower in (w.get("area") or "").lower():
            reasons.append(f"📍 Near {user_area}")
        elif w.get("area"):
            reasons.append(f"📍 Based in {w['area']}")

        # Jobs done
        jobs = int(w.get("total_jobs") or 0)
        if jobs >= 30:
            reasons.append(f"🔨 {jobs}+ jobs done")

        w_copy = dict(w)
        w_copy["reason"] = " · ".join(reasons)
        w_copy["rank"]   = rank
        result_workers.append(w_copy)

    urgency = ai.get("urgency", "Normal")
    icon    = ai.get("icon", "🔧")
    price   = ai.get("price_range", "")
    msg = (
        f"We detected this as a **{category}** problem ({icon}). "
        f"Urgency: **{urgency}**. Estimated price: **{price}**. "
        f"Here are the top {len(result_workers)} recommended workers:"
    )

    return {
        "ai_result": ai,
        "workers":   result_workers,
        "message":   msg,
    }