"""
Signal Tier Classifier
Classifies signals into Tier-1 (tradeable alone), Tier-2 (needs confirmation), or Ignore
"""
import pandas as pd
from utils.logger import logger


class SignalTierClassifier:
    """Service for classifying signals into tiers based on badge combinations"""
    
    def __init__(self):
        self.tier1_min_premium = 500000  # $500K minimum for Tier-1
    
    def classify_tier(self, row) -> str:
        """
        Classify signal tier based on badge combinations and premium
        
        Tier-1: 🟢/🔴 + 💎 + ⭐ + premium > 500K + direction aligned
        Tier-2: 🟢 + 💎 (no ⭐) OR ⭐ without 💎
        Ignore: OTM-only, mixed signals, low volume/OI ratio
        """
        # Use proper pandas Series access with fallback
        def safe_get(row, key, default=None):
            try:
                val = row.get(key, default) if hasattr(row, 'get') else (row[key] if key in row.index else default)
                return default if pd.isna(val) else val
            except (KeyError, IndexError):
                return default
        
        badge_round = safe_get(row, 'badge_round', '') or ''
        badge_more = safe_get(row, 'badge_more', '') or ''
        premium = float(safe_get(row, 'premium_num', 0) or 0)
        direction = safe_get(row, 'direction', '') or ''
        bull_total = float(safe_get(row, 'bull_total', 0) or 0)
        bear_total = float(safe_get(row, 'bear_total', 0) or 0)
        
        has_diamond = '💎' in str(badge_more)
        has_star = '⭐' in str(badge_more)
        has_money = '💰' in str(badge_more)
        
        # Check if OTM-only (no ITM premium)
        prem_cb_itm = float(safe_get(row, 'prem_cb_itm', 0) or 0)
        prem_ps_itm = float(safe_get(row, 'prem_ps_itm', 0) or 0)
        prem_cs_itm = float(safe_get(row, 'prem_cs_itm', 0) or 0)
        prem_pb_itm = float(safe_get(row, 'prem_pb_itm', 0) or 0)
        
        bull_prem_itm = prem_cb_itm + prem_ps_itm
        bear_prem_itm = prem_cs_itm + prem_pb_itm
        
        # Ignore: OTM-only prints (no ITM premium)
        if bull_prem_itm == 0 and bear_prem_itm == 0:
            return 'IGNORE'
        
        # Ignore: Mixed 🟢/🔴 with no net premium edge
        if not badge_round or badge_round == '':
            return 'IGNORE'
        
        # Ignore: Big premium but volume << OI (likely rolls/hedges)
        vol_num = float(safe_get(row, 'vol_num', 0) or 0)
        oi_num = float(safe_get(row, 'oi_num', 0) or 0)
        if premium > 500000 and vol_num > 0 and oi_num > 0:
            vol_oi_ratio = vol_num / oi_num if oi_num > 0 else 0
            if vol_oi_ratio < 0.3:  # Volume is less than 30% of OI
                return 'IGNORE'
        
        # Tier-1 conditions: 🟢/🔴 + 💎 + ⭐ + premium > 500K + direction aligned
        if (badge_round in ['🟢', '🔴'] and 
            has_diamond and 
            has_star and 
            premium >= self.tier1_min_premium):
            
            # Check direction alignment
            net_premium = bull_total - bear_total
            if badge_round == '🟢' and direction == 'BULL' and net_premium > 0:
                return 'TIER_1'
            elif badge_round == '🔴' and direction == 'BEAR' and net_premium < 0:
                return 'TIER_1'
        
        # Tier-2 conditions: 🟢 + 💎 (no ⭐ yet) OR ⭐ without 💎
        if badge_round == '🟢' and has_diamond and not has_star:
            return 'TIER_2'
        
        if has_star and not has_diamond:
            return 'TIER_2'
        
        # If we have direction and diamond but missing star, it's Tier-2
        if badge_round in ['🟢', '🔴'] and has_diamond and not has_star:
            return 'TIER_2'
        
        # Default to ignore if doesn't meet criteria
        return 'IGNORE'
    
    def classify_tiers(self, df: pd.DataFrame) -> pd.DataFrame:
        """Classify tiers for all rows in DataFrame"""
        df = df.copy()
        
        if df.empty:
            df['signal_tier'] = pd.Series(dtype=str)
            df['is_tradeable'] = pd.Series(dtype=bool)
            return df
        
        logger.info(f"Classifying signal tiers for {len(df)} rows...")
        
        # Apply classification and ensure it returns a Series
        signal_tiers = df.apply(self.classify_tier, axis=1)
        
        # Ensure it's a Series, not a DataFrame
        if isinstance(signal_tiers, pd.DataFrame):
            # If somehow it's a DataFrame, take the first column
            signal_tiers = signal_tiers.iloc[:, 0]
        
        df['signal_tier'] = signal_tiers
        
        # Add is_tradeable flag (Tier-1 only)
        df['is_tradeable'] = df['signal_tier'] == 'TIER_1'
        
        # Log tier distribution
        tier_counts = df['signal_tier'].value_counts()
        logger.info(f"Signal tier distribution: {tier_counts.to_dict()}")
        logger.info(f"Tradeable signals (Tier-1): {df['is_tradeable'].sum()}")
        
        return df