app/utils/mahalla_matcher.py

# app/utils/mahalla_matcher.py - YAXSHILANGAN (THRESHOLD 0.35)

"""

Mahalla Matcher - Noto'g'ri yozilgan mahalla nomlarini topish

Fuzzy matching + substring + word matching

"""

import logging
from typing import Optional, List
from difflib import SequenceMatcher

from app.services.location_validator import get_mahallas_by_district

logger = logging.getLogger(__name__)


def normalize_mahalla_text(text: str) -> str:
    """

    Mahalla nomini normalizatsiya qilish (KENGAYTIRILGAN)

    O'zbek tili qo'shimchalarini olib tashlaydi

    

    Args:

        text: Asl matn

        

    Returns:

        Normalized matn

    """
    if not text:
        return ""
    
    # Kichik harf
    text = text.lower().strip()
    
    # Ko'p probellarni bitta probelga
    text = " ".join(text.split())
    
    # "mahallasi", "mahalla" so'zlarini olib tashlash
    text = text.replace(' mahallasi', '').replace(' mahalla', '')
    text = text.replace('mahallasi', '').replace('mahalla', '')
    
    # O'zbek tili kelishik qo'shimchalarini olib tashlash (KENGAYTIRILGAN)
    suffixes = [
        "ni", "ga", "da", "dan", "ning", "niki", 
        "dagi", "dagina", "gacha", "dan", "dek",
        "lar", "larni", "larga", "larda", "lardan"
    ]
    
    words = text.split()
    
    if words:
        last_word = words[-1]
        for suffix in suffixes:
            if last_word.endswith(suffix) and len(last_word) > len(suffix) + 2:  # Juda qisqa so'zlarni saqlab qolish
                words[-1] = last_word[:-len(suffix)]
                break
    
    text = " ".join(words)
    
    # Oxiridagi tinish belgilarini olib tashlash
    text = text.strip('.,!? ')
    
    return text.strip()


def similarity_score(str1: str, str2: str) -> float:
    """

    Ikki string orasidagi o'xshashlik (0.0 - 1.0)

    

    Args:

        str1: Birinchi string

        str2: Ikkinchi string

        

    Returns:

        Similarity score (1.0 = 100% o'xshash)

    """
    return SequenceMatcher(None, str1, str2).ratio()


def word_similarity(str1: str, str2: str) -> float:
    """

    So'zma-so'z o'xshashlik (word-level matching)

    

    Args:

        str1: Birinchi matn

        str2: Ikkinchi matn

        

    Returns:

        Word overlap score (0.0 - 1.0)

    """
    words1 = set(str1.split())
    words2 = set(str2.split())
    
    if not words1 or not words2:
        return 0.0
    
    intersection = words1.intersection(words2)
    union = words1.union(words2)
    
    return len(intersection) / len(union) if union else 0.0


def find_mahalla_fuzzy(district_name: str, user_text: str, threshold: float = 0.35) -> Optional[str]:
    """

    Noto'g'ri yozilgan mahalla nomini topish (YAXSHILANGAN - THRESHOLD 0.35)

    

    Args:

        district_name: Tuman nomi

        user_text: Bemorning kiritgan matni (masalan: "katta chilonzor" yoki "besh qorgon")

        threshold: Minimal o'xshashlik darajasi (0.35 = 35%) ← PASAYTIRILDI!

        

    Returns:

        Mahalla nomi (masalan: "Katta Chilonzor-1 mahallasi") yoki None

    """
    try:
        if not user_text or not district_name:
            return None
        
        # Matnni normalizatsiya qilish
        normalized_input = normalize_mahalla_text(user_text)
        logger.info(f"🏘️ Mahalla qidirilmoqda: '{user_text}' → '{normalized_input}' ({district_name})")
        
        if len(normalized_input) < 2:
            logger.warning("⚠️ Matn juda qisqa")
            return None
        
        # Tuman bo'yicha mahallalarni olish
        mahallas = get_mahallas_by_district(district_name)
        
        if not mahallas:
            logger.warning(f"⚠️ {district_name} uchun mahallalar topilmadi")
            return None
        
        # Eng yaxshi moslikni topish
        best_match = None
        best_score = 0.0
        scoring_details = []
        
        for mahalla in mahallas:
            normalized_mahalla = normalize_mahalla_text(mahalla)
            
            # 1. To'liq fuzzy match (SequenceMatcher)
            fuzzy_score = similarity_score(normalized_input, normalized_mahalla)
            
            # 2. Substring match (KATTA BONUS)
            substring_score = 0.0
            if normalized_input in normalized_mahalla:
                substring_score = 0.9  # 90% match
                logger.debug(f"  ✓ Substring (input in mahalla): '{normalized_input}' in '{normalized_mahalla}'")
            elif normalized_mahalla in normalized_input:
                substring_score = 0.85  # 85% match
                logger.debug(f"  ✓ Substring (mahalla in input): '{normalized_mahalla}' in '{normalized_input}'")
            
            # 3. So'zma-so'z match (Word overlap)
            word_score = word_similarity(normalized_input, normalized_mahalla)
            
            # 4. So'z boshi match (First word matching)
            input_words = normalized_input.split()
            mahalla_words = normalized_mahalla.split()
            first_word_score = 0.0
            if input_words and mahalla_words:
                if input_words[0] == mahalla_words[0]:
                    first_word_score = 0.7  # Birinchi so'z mos kelsa - 70%
                    logger.debug(f"  ✓ First word match: '{input_words[0]}'")
            
            # 5. FINAL SCORE (eng yuqori ball)
            final_score = max(fuzzy_score, substring_score, word_score, first_word_score)
            
            scoring_details.append({
                "mahalla": mahalla,
                "fuzzy": fuzzy_score,
                "substring": substring_score,
                "word": word_score,
                "first_word": first_word_score,
                "final": final_score
            })
            
            if final_score > best_score:
                best_score = final_score
                best_match = mahalla
        
        # Debug: Top 3 natijalar
        scoring_details.sort(key=lambda x: x['final'], reverse=True)
        logger.debug(f"  Top 3 matches:")
        for i, detail in enumerate(scoring_details[:3], 1):
            logger.debug(f"    {i}. {detail['mahalla']}: {detail['final']:.2f} "
                        f"(fuzzy={detail['fuzzy']:.2f}, sub={detail['substring']:.2f}, "
                        f"word={detail['word']:.2f}, first={detail['first_word']:.2f})")
        
        # Threshold tekshirish
        if best_score >= threshold:
            logger.info(f"✅ Mahalla topildi: '{best_match}' (score: {best_score:.2f})")
            return best_match
        else:
            logger.warning(f"⚠️ Mahalla topilmadi (best score: {best_score:.2f} < {threshold})")
            return None
            
    except Exception as e:
        logger.error(f"❌ Mahalla matching xatoligi: {e}", exc_info=True)
        return None


def get_mahalla_display_name(mahalla_name: str) -> str:
    """

    Mahalla nomini to'liq formatda qaytarish

    

    Args:

        mahalla_name: "Beltepa" yoki "Beltepa mahallasi"

        

    Returns:

        "Beltepa mahallasi"

    """
    if not mahalla_name:
        return ""
    
    if 'mahallasi' not in mahalla_name.lower():
        return f"{mahalla_name} mahallasi"
    
    return mahalla_name


def suggest_mahallas(district_name: str, user_text: str, top_n: int = 3) -> List[tuple]:
    """

    Mahallalar tavsiyalari (top N eng o'xshash)

    

    Args:

        district_name: Tuman nomi

        user_text: Bemorning matni

        top_n: Ko'rsatiladigan maksimal tavsiyalar soni

        

    Returns:

        [(mahalla_name, score), ...] - eng yaxshilaridan boshlab

    """
    try:
        normalized_input = normalize_mahalla_text(user_text)
        mahallas = get_mahallas_by_district(district_name)
        
        if not mahallas:
            return []
        
        results = []
        
        for mahalla in mahallas:
            normalized_mahalla = normalize_mahalla_text(mahalla)
            
            # Barcha scoringlar
            fuzzy = similarity_score(normalized_input, normalized_mahalla)
            substring = 0.9 if normalized_input in normalized_mahalla else 0.0
            word = word_similarity(normalized_input, normalized_mahalla)
            
            final_score = max(fuzzy, substring, word)
            
            results.append((mahalla, final_score))
        
        # Scorecard bo'yicha tartiblash
        results.sort(key=lambda x: x[1], reverse=True)
        
        return results[:top_n]
        
    except Exception as e:
        logger.error(f"❌ Mahalla tavsiyalari xatoligi: {e}")
        return []


# ==================== TESTING HELPER ====================

def test_mahalla_matching(district_name: str, test_inputs: List[str]):
    """

    Mahalla matching'ni test qilish uchun helper

    

    Args:

        district_name: Test qilinadigan tuman

        test_inputs: Test qilinadigan inputlar ro'yxati

    """
    logger.info(f"\n{'='*60}")
    logger.info(f"TEST: {district_name}")
    logger.info(f"{'='*60}")
    
    for test_input in test_inputs:
        logger.info(f"\nTest input: '{test_input}'")
        
        result = find_mahalla_fuzzy(district_name, test_input, threshold=0.35)
        
        if result:
            logger.info(f"  ✅ MATCH: {result}")
        else:
            logger.warning(f"  ❌ NO MATCH")
            suggestions = suggest_mahallas(district_name, test_input, top_n=3)
            if suggestions:
                logger.info(f"  💡 Suggestions:")
                for mahalla, score in suggestions:
                    logger.info(f"    - {mahalla} ({score:.2f})")
    
    logger.info(f"{'='*60}\n")


# ==================== AUTO-TEST ON IMPORT (DEBUG MODE) ====================

if __name__ == "__main__":
    # Test uchun
    logging.basicConfig(level=logging.DEBUG)
    
    test_cases = [
        ("Chilonzor tumani", ["katta chilonzor", "beltepa", "beshqozon", "qorgon"]),
        ("Bektemir tumani", ["abay", "bektemir", "binokor"]),
        ("Shayxontohur tumani", ["kamolon", "shayx"]),
    ]
    
    for district, inputs in test_cases:
        test_mahalla_matching(district, inputs)