evaluation/emotion_statistics.py

# -*- coding: utf-8 -*-
"""
Emotion Statistics - Complete overview of all emotions managed by the system

Shows:
- Total emotion count
- Emoji mapping
- Recognition accuracy from latest evaluation
"""

import sys
import os
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))

from avatar.sentiment_emoji_map import EmojiMapper


def get_emotion_stats():
    """Generate complete emotion statistics table"""
    
    mapper = EmojiMapper()
    all_emojis = mapper.get_all_emojis()
    
    # Remove legacy/system mappings for counting
    legacy_keys = ["very_negative", "negative", "slightly_negative", 
                   "slightly_positive", "positive", "very_positive",
                   "default", "ready"]
    
    emotion_emojis = {k: v for k, v in all_emojis.items() if k not in legacy_keys}
    
    # Recognition accuracy from latest evaluation (Multi-Emotion V2)
    # Data from evaluation/reports/comparison_multi-v2_2025-12-01_18-46-46.md
    recognition_accuracy = {
        # 100% accuracy
        "excitement": 100.0, "thrill": 100.0, "delight": 100.0, "cheerfulness": 100.0,
        "enthusiasm": 100.0, "calm": 100.0, "pleasure": 100.0, "love": 100.0,
        "affection": 100.0, "tenderness": 100.0, "caring": 100.0, "compassion": 100.0,
        "empathy": 100.0, "gratitude": 100.0, "wonder": 100.0, "awe": 100.0,
        "amazement": 100.0, "surprise": 100.0, "thinking": 100.0, "sadness": 100.0,
        "grief": 100.0, "disappointment": 100.0, "loneliness": 100.0, "hurt": 100.0,
        "fear": 100.0, "terror": 100.0, "horror": 100.0, "dread": 100.0,
        "anxiety": 100.0, "worry": 100.0, "nervousness": 100.0, "panic": 100.0,
        "anger": 100.0, "fury": 100.0, "irritation": 100.0, "hostility": 100.0,
        "disgust": 100.0, "contempt": 100.0, "disdain": 100.0, "shame": 100.0,
        "embarrassment": 100.0, "guilt": 100.0, "regret": 100.0, "remorse": 100.0,
        "tiredness": 100.0, "jealousy": 100.0, "sympathy": 100.0, "nostalgia": 100.0,
        "pessimism": 100.0, "yearning": 100.0, "determination": 100.0,
        "anticipation": 100.0, "trust": 100.0,
        # 90%+
        "happiness": 90.0, "silly": 87.5, "joy": 85.7, "pride": 85.7,
        "uncertain": 85.7, "confused": 85.7, "playful": 85.7, "hope": 85.7,
        # 80%+
        "satisfaction": 83.3, "relaxed": 83.3, "relief": 83.3, "interest": 83.3,
        "shock": 83.3, "annoyance": 83.3, "frustration": 83.3, "boredom": 83.3,
        "optimism": 83.3, "elation": 80.0, "euphoria": 80.0, "serenity": 80.0,
        "adoration": 80.0, "fascination": 80.0, "intrigue": 80.0, "astonishment": 80.0,
        "puzzled": 80.0, "perplexed": 80.0, "bewildered": 80.0, "baffled": 80.0,
        "sorrow": 80.0, "melancholy": 80.0, "despair": 80.0, "misery": 80.0,
        "apprehension": 80.0, "rage": 80.0, "resentment": 80.0, "bitterness": 80.0,
        "revulsion": 80.0, "scorn": 80.0, "humiliation": 80.0, "exhaustion": 80.0,
        "fatigue": 80.0, "weariness": 80.0, "inspiration": 80.0, "sarcasm": 80.0,
        # 70%+
        "neutral": 76.9, "amusement": 72.7, "ecstasy": 71.4, "longing": 71.4,
        # 60%+
        "envy": 66.7, "mischievous": 66.7, "acceptance": 60.0,
        # 50%+
        "contentment": 57.1, "curiosity": 57.1, "confidence": 50.0,
    }
    
    # Categories
    categories = {
        "POSITIVE (High Arousal)": ["ecstasy", "joy", "happiness", "delight", "elation", 
                                    "euphoria", "excitement", "thrill", "enthusiasm", "cheerfulness"],
        "POSITIVE (Medium Arousal)": ["contentment", "satisfaction", "pleasure", "relief", 
                                       "serenity", "calm", "relaxed", "pride", "confidence", "triumph"],
        "LOVE & AFFECTION": ["love", "adoration", "affection", "tenderness", "caring", 
                            "compassion", "empathy", "gratitude", "thankful"],
        "INTEREST & CURIOSITY": ["curiosity", "interest", "fascination", "wonder", "awe", 
                                 "amazement", "intrigue"],
        "SURPRISE": ["surprise", "astonishment", "shock", "startled"],
        "NEUTRAL / THINKING": ["neutral", "thinking", "contemplative", "pensive", "reflective", 
                               "uncertain", "ambivalent", "indifferent"],
        "CONFUSION": ["confused", "confusion", "puzzled", "perplexed", "bewildered", "baffled"],
        "SADNESS": ["sadness", "sorrow", "grief", "melancholy", "disappointment", "dejection", 
                   "despair", "hopelessness", "loneliness", "hurt", "misery"],
        "FEAR & ANXIETY": ["fear", "terror", "horror", "dread", "anxiety", "worry", 
                          "nervousness", "apprehension", "panic"],
        "ANGER & FRUSTRATION": ["anger", "rage", "fury", "irritation", "annoyance", "frustration", 
                                "exasperation", "resentment", "hostility", "bitterness"],
        "DISGUST & CONTEMPT": ["disgust", "revulsion", "contempt", "disdain", "loathing", "scorn"],
        "SHAME & EMBARRASSMENT": ["shame", "embarrassment", "guilt", "regret", "remorse", "humiliation"],
        "BOREDOM & TIREDNESS": ["boredom", "tiredness", "exhaustion", "fatigue", "weariness", "sleepy"],
        "ENVY & JEALOUSY": ["envy", "jealousy"],
        "PLAYFUL & SILLY": ["playful", "silly", "mischievous", "teasing", "sarcastic", "witty", 
                           "amusement", "funny"],
        "SPECIAL STATES": ["sympathy", "nostalgia", "hope", "optimism", "pessimism", "longing", 
                          "yearning", "determination", "inspiration", "anticipation", "trust", 
                          "acceptance", "sarcasm"],
    }
    
    return emotion_emojis, recognition_accuracy, categories


def print_emotion_table():
    """Print formatted emotion statistics table"""
    
    emotion_emojis, recognition_accuracy, categories = get_emotion_stats()
    
    print("=" * 80)
    print("EMOJI AI AVATAR - EMOTION RECOGNITION STATISTICS")
    print("=" * 80)
    print()
    print(f"📊 Total Emotions Mapped: {len(emotion_emojis)}")
    print(f"📊 Emotions with Recognition Data: {len(recognition_accuracy)}")
    print(f"📊 Overall Recognition Accuracy: 89.0%")
    print()
    
    # Print by category
    for category, emotions in categories.items():
        print(f"\n{'─' * 80}")
        print(f"📁 {category}")
        print(f"{'─' * 80}")
        print(f"{'Emotion':<20} {'Emoji':<6} {'Accuracy':<10} {'Status'}")
        print(f"{'─' * 20} {'─' * 5} {'─' * 9} {'─' * 10}")
        
        for emotion in emotions:
            emoji = emotion_emojis.get(emotion, "❓")
            accuracy = recognition_accuracy.get(emotion, None)
            
            if accuracy is not None:
                if accuracy >= 80:
                    status = "✅ PASS"
                elif accuracy >= 50:
                    status = "⚠️ OK"
                else:
                    status = "❌ FAIL"
                acc_str = f"{accuracy:.1f}%"
            else:
                status = "🔹 N/T"  # Not tested
                acc_str = "N/A"
            
            print(f"{emotion:<20} {emoji:<6} {acc_str:<10} {status}")
    
    # Summary statistics
    print()
    print("=" * 80)
    print("SUMMARY STATISTICS")
    print("=" * 80)
    
    tested = [acc for acc in recognition_accuracy.values()]
    
    perfect = sum(1 for acc in tested if acc == 100.0)
    high = sum(1 for acc in tested if 80.0 <= acc < 100.0)
    medium = sum(1 for acc in tested if 50.0 <= acc < 80.0)
    low = sum(1 for acc in tested if acc < 50.0)
    
    print(f"✅ 100% Accuracy: {perfect} emotions")
    print(f"✅ 80-99% Accuracy: {high} emotions")
    print(f"⚠️ 50-79% Accuracy: {medium} emotions")
    print(f"❌ <50% Accuracy: {low} emotions")
    print()
    print(f"Average Accuracy: {sum(tested)/len(tested):.1f}%")
    print()


def generate_markdown_table():
    """Generate markdown table for documentation"""
    
    emotion_emojis, recognition_accuracy, categories = get_emotion_stats()
    
    lines = [
        "# Emoji AI Avatar - Complete Emotion Table",
        "",
        f"**Total Emotions:** {len(emotion_emojis)}",
        f"**Overall Accuracy:** 89.0%",
        f"**Last Updated:** 2025-12-01",
        "",
        "## Recognition Accuracy by Category",
        "",
    ]
    
    for category, emotions in categories.items():
        lines.append(f"### {category}")
        lines.append("")
        lines.append("| Emotion | Emoji | Accuracy | Status |")
        lines.append("|---------|-------|----------|--------|")
        
        for emotion in emotions:
            emoji = emotion_emojis.get(emotion, "❓")
            accuracy = recognition_accuracy.get(emotion, None)
            
            if accuracy is not None:
                if accuracy >= 80:
                    status = "✅ PASS"
                elif accuracy >= 50:
                    status = "⚠️ OK"
                else:
                    status = "❌ FAIL"
                acc_str = f"{accuracy:.1f}%"
            else:
                status = "🔹 N/T"
                acc_str = "N/A"
            
            lines.append(f"| {emotion} | {emoji} | {acc_str} | {status} |")
        
        lines.append("")
    
    # Add summary
    tested = list(recognition_accuracy.values())
    perfect = sum(1 for acc in tested if acc == 100.0)
    high = sum(1 for acc in tested if 80.0 <= acc < 100.0)
    medium = sum(1 for acc in tested if 50.0 <= acc < 80.0)
    low = sum(1 for acc in tested if acc < 50.0)
    
    lines.extend([
        "## Summary",
        "",
        "| Accuracy Range | Count |",
        "|----------------|-------|",
        f"| 100% (Perfect) | {perfect} |",
        f"| 80-99% (High) | {high} |",
        f"| 50-79% (Medium) | {medium} |",
        f"| <50% (Low) | {low} |",
        "",
        f"**Average Accuracy:** {sum(tested)/len(tested):.1f}%",
        "",
    ])
    
    return "\n".join(lines)


if __name__ == "__main__":
    print_emotion_table()
    
    # Also save markdown version
    md_content = generate_markdown_table()
    output_path = os.path.join(os.path.dirname(__file__), "reports", "emotion_statistics.md")
    with open(output_path, "w", encoding="utf-8") as f:
        f.write(md_content)
    
    print(f"\n📄 Markdown table saved to: {output_path}")