import subprocess
import sys
import os

# Check if running in a standard environment (not Colab/Jupyter)
# and install packages if needed
if not os.path.exists("/.dockerenv") and not os.path.exists("/kaggle"):
    try:
        import spacy
        import matplotlib
        import gradio
    except ImportError:
        print("Installing required packages...")
        subprocess.check_call([sys.executable, "-m", "pip", "install",
                              "spacy", "matplotlib", "gradio"])
        # Download spaCy model
        subprocess.check_call([sys.executable, "-m", "spacy", "download", "en_core_web_md"])

import spacy
import numpy as np
import matplotlib.pyplot as plt
import gradio as gr
import re
from collections import Counter

print("Setting up spaCy-based emotion analysis model...")

# Load spaCy model
print("Loading spaCy model (this takes just a moment)...")
nlp = spacy.load("en_core_web_md")

# Enhanced emotion categories with carefully selected keywords
EMOTION_CATEGORIES = {
    'joy': [
        'happy', 'joyful', 'delighted', 'excited', 'cheerful',
        'glad', 'elated', 'jubilant', 'overjoyed', 'pleased',
        'ecstatic', 'thrilled', 'euphoric', 'content', 'blissful'
    ],
    'sadness': [
        'sad', 'unhappy', 'depressed', 'disappointed', 'sorrowful',
        'heartbroken', 'melancholy', 'grief', 'somber', 'mournful',
        'gloomy', 'despondent', 'downcast', 'miserable', 'devastated'
    ],
    'anger': [
        'angry', 'furious', 'enraged', 'irritated', 'annoyed',
        'outraged', 'hostile', 'mad', 'infuriated', 'indignant',
        'livid', 'irate', 'fuming', 'seething', 'resentful'
    ],
    'fear': [
        'afraid', 'scared', 'frightened', 'terrified', 'anxious',
        'worried', 'nervous', 'panicked', 'horrified', 'apprehensive',
        'fearful', 'uneasy', 'alarmed', 'dread', 'paranoid'
    ],
    'surprise': [
        'surprised', 'amazed', 'astonished', 'shocked', 'stunned',
        'startled', 'astounded', 'bewildered', 'unexpected', 'awestruck',
        'flabbergasted', 'dumbfounded', 'incredulous', 'perplexed', 'thunderstruck'
    ],
    'love': [
        'loving', 'affectionate', 'fond', 'adoring', 'caring',
        'devoted', 'passionate', 'tender', 'compassionate', 'cherishing',
        'enamored', 'smitten', 'infatuated', 'admiring', 'doting'
    ],
    'sarcasm': [
        'sarcastic', 'ironic', 'mocking', 'cynical', 'satirical',
        'sardonic', 'facetious', 'contemptuous', 'caustic', 'biting',
        'scornful', 'derisive', 'snide', 'taunting', 'wry'
    ],
    'disgust': [
        'disgusted', 'revolted', 'nauseated', 'repulsed', 'sickened',
        'appalled', 'repelled', 'abhorred', 'loathing', 'distaste',
        'aversion', 'revulsion', 'repugnance', 'horrified', 'offended'
    ],
    'anticipation': [
        'anticipating', 'expecting', 'awaiting', 'looking forward', 'hopeful',
        'eager', 'excited', 'impatient', 'prepared', 'ready',
        'vigilant', 'attentive', 'watchful', 'alert', 'expectant'
    ],
    'trust': [
        'trusting', 'confident', 'assured', 'secure', 'certain',
        'reliant', 'faithful', 'believing', 'dependable', 'reliable',
        'credible', 'trustworthy', 'honest', 'loyal', 'devoted'
    ]
}

# Define emotion colors for visualization
EMOTION_COLORS = {
    'joy': '#F1C40F',       # Yellow
    'sadness': '#3498DB',   # Blue
    'anger': '#E74C3C',     # Red
    'fear': '#7D3C98',      # Purple
    'surprise': '#2ECC71',  # Green
    'love': '#E91E63',      # Pink
    'sarcasm': '#FF7F50',   # Coral
    'disgust': '#8E44AD',   # Dark Purple
    'anticipation': '#F39C12', # Orange
    'trust': '#16A085'      # Teal
}

# Common sentiment phrases and expressions for improved detection
EMOTION_PHRASES = {
    'joy': [
        'over the moon', 'on cloud nine', 'couldn\'t be happier',
        'best day ever', 'made my day', 'feeling great',
        'absolutely thrilled', 'jumping for joy', 'bursting with happiness',
        'walking on sunshine', 'flying high', 'tickled pink'
    ],
    'sadness': [
        'broke my heart', 'in tears', 'feel like crying',
        'deeply saddened', 'lost all hope', 'feel empty',
        'devastating news', 'hit hard', 'feel down', 'soul-crushing',
        'falling apart', 'world is ending', 'deeply hurt'
    ],
    'anger': [
        'makes my blood boil', 'fed up with', 'had it with',
        'sick and tired of', 'drives me crazy', 'lost my temper',
        'absolutely furious', 'beyond frustrated', 'driving me up the wall',
        'at my wit\'s end', 'through the roof', 'blow a gasket', 'see red'
    ],
    'fear': [
        'scared to death', 'freaking out', 'keeps me up at night',
        'terrified of', 'living in fear', 'panic attack',
        'nervous wreck', 'can\'t stop worrying', 'break out in a cold sweat',
        'shaking like a leaf', 'scared stiff', 'frozen with fear'
    ],
    'surprise': [
        'can\'t believe', 'took me by surprise', 'out of nowhere',
        'never expected', 'caught off guard', 'mind blown',
        'plot twist', 'jaw dropped', 'knocked my socks off',
        'took my breath away', 'blew me away', 'speechless'
    ],
    'love': [
        'deeply in love', 'means the world to me', 'treasure every moment',
        'hold dear', 'close to my heart', 'forever grateful',
        'truly blessed', 'never felt this way', 'head over heels',
        'madly in love', 'heart skips a beat', 'love with all my heart'
    ],
    'sarcasm': [
        'just what I needed', 'couldn\'t get any better', 'how wonderful',
        'oh great', 'lucky me', 'my favorite part',
        'thrilled to bits', 'way to go', 'thanks for nothing',
        'brilliant job', 'story of my life', 'what a surprise'
    ],
    'disgust': [
        'makes me sick', 'turn my stomach', 'can\'t stand',
        'absolutely disgusting', 'utterly repulsive', 'gross',
        'revolting sight', 'nauseating', 'skin crawl',
        'makes me want to vomit', 'repulsed by', 'can hardly look at'
    ],
    'anticipation': [
        'looking forward to', 'can\'t wait for', 'counting down the days',
        'eagerly awaiting', 'excited about', 'in anticipation of',
        'on the edge of my seat', 'can hardly wait', 'dying to see',
        'marked on my calendar', 'preparing for', 'gearing up for'
    ],
    'trust': [
        'rely on completely', 'trust with my life', 'put my faith in',
        'never let me down', 'count on', 'believe in',
        'have confidence in', 'trustworthy', 'dependable',
        'true to their word', 'rock solid', 'through thick and thin'
    ]
}

# Contextual emotion indicators for better analysis
CONTEXTUAL_INDICATORS = {
    'intensifiers': ['very', 'extremely', 'incredibly', 'absolutely', 'totally', 'completely', 'utterly'],
    'negators': ['not', 'never', 'no', 'none', 'neither', 'nor', 'hardly', 'barely'],
    'hedges': ['somewhat', 'kind of', 'sort of', 'a bit', 'slightly', 'perhaps', 'maybe'],
    'boosters': ['definitely', 'certainly', 'absolutely', 'undoubtedly', 'surely', 'clearly'],
    'punctuation': {'!': 'emphasis', '?': 'question', '...': 'hesitation'}
}

# Emotional verdict categories for intelligently classifying mixed emotions
EMOTION_VERDICT_CATEGORIES = {
    # Single dominant emotions (when over 35%)
    'purely_joyful': {'conditions': [('joy', 0.35)], 'label': 'Purely Joyful', 'description': 'Expressing happiness and positive emotions'},
    'deeply_sad': {'conditions': [('sadness', 0.35)], 'label': 'Deeply Sad', 'description': 'Expressing sadness and negative emotions'},
    'intensely_angry': {'conditions': [('anger', 0.35)], 'label': 'Intensely Angry', 'description': 'Expressing anger and frustration'},
    'primarily_fearful': {'conditions': [('fear', 0.35)], 'label': 'Primarily Fearful', 'description': 'Expressing fear and anxiety'},
    'genuinely_surprised': {'conditions': [('surprise', 0.35)], 'label': 'Genuinely Surprised', 'description': 'Expressing surprise and astonishment'},
    'deeply_loving': {'conditions': [('love', 0.35)], 'label': 'Deeply Loving', 'description': 'Expressing love and affection'},
    'clearly_sarcastic': {'conditions': [('sarcasm', 0.35)], 'label': 'Clearly Sarcastic', 'description': 'Expressing sarcasm and irony'},
    'utterly_disgusted': {'conditions': [('disgust', 0.35)], 'label': 'Utterly Disgusted', 'description': 'Expressing disgust and revulsion'},
    'eagerly_anticipating': {'conditions': [('anticipation', 0.35)], 'label': 'Eagerly Anticipating', 'description': 'Expressing anticipation and eagerness'},
    'firmly_trusting': {'conditions': [('trust', 0.35)], 'label': 'Firmly Trusting', 'description': 'Expressing trust and confidence'},

    # Common emotional combinations
    'bitter_sweet': {
        'conditions': [('joy', 0.2), ('sadness', 0.2)],
        'label': 'Bittersweet',
        'description': 'Mixed feelings of happiness and sadness'
    },
    'anxious_excitement': {
        'conditions': [('anticipation', 0.2), ('fear', 0.2)],
        'label': 'Anxious Excitement',
        'description': 'Mixture of excitement and nervousness'
    },
    'angry_disappointment': {
        'conditions': [('anger', 0.2), ('sadness', 0.2)],
        'label': 'Angry Disappointment',
        'description': 'Disappointment expressed through anger'
    },
    'ironic_amusement': {
        'conditions': [('sarcasm', 0.2), ('joy', 0.15)],
        'label': 'Ironic Amusement',
        'description': 'Finding humor through irony or sarcasm'
    },
    'fearful_anticipation': {
        'conditions': [('fear', 0.2), ('anticipation', 0.2)],
        'label': 'Fearful Anticipation',
        'description': 'Anxiously awaiting something'
    },
    'relieved_surprise': {
        'conditions': [('surprise', 0.2), ('joy', 0.15)],
        'label': 'Relieved Surprise',
        'description': 'Surprise with positive outcome'
    },
    'shocked_disappointment': {
        'conditions': [('surprise', 0.2), ('sadness', 0.15)],
        'label': 'Shocked Disappointment',
        'description': 'Unexpectedly negative outcome'
    },
    'disgusted_anger': {
        'conditions': [('disgust', 0.2), ('anger', 0.2)],
        'label': 'Disgusted Anger',
        'description': 'Angry response to something repulsive'
    },
    'loving_trust': {
        'conditions': [('love', 0.2), ('trust', 0.2)],
        'label': 'Loving Trust',
        'description': 'Deep affection with confidence'
    },
    'sarcastic_frustration': {
        'conditions': [('sarcasm', 0.2), ('anger', 0.15)],
        'label': 'Sarcastic Frustration',
        'description': 'Using sarcasm to express frustration'
    },
    'confused_surprise': {
        'conditions': [('surprise', 0.2), ('fear', 0.15)],
        'label': 'Confused Surprise',
        'description': 'Startled with uncertainty'
    },
    'hopeful_joy': {
        'conditions': [('joy', 0.2), ('anticipation', 0.2)],
        'label': 'Hopeful Joy',
        'description': 'Happy anticipation of something positive'
    },
    'betrayed_trust': {
        'conditions': [('sadness', 0.2), ('trust', 0.15), ('anger', 0.15)],
        'label': 'Betrayed Trust',
        'description': 'Sadness from broken trust'
    },
    'fearful_disgust': {
        'conditions': [('fear', 0.2), ('disgust', 0.2)],
        'label': 'Fearful Disgust',
        'description': 'Fear of something repulsive'
    },

    # Special cases for multiple emotions
    'emotionally_complex': {
        'conditions': ['multiple_over_15'],
        'label': 'Emotionally Complex',
        'description': 'Multiple competing emotions'
    },
    'mildly_emotional': {
        'conditions': ['all_under_20'],
        'label': 'Mildly Emotional',
        'description': 'Low intensity emotional content'
    },
    'predominantly_neutral': {
        'conditions': ['all_under_15'],
        'label': 'Predominantly Neutral',
        'description': 'No strong emotional signals detected'
    }
}

# Sarcasm patterns with refined detection logic
SARCASM_PATTERNS = [
    # Exaggerated positive with negative context
    r'(?i)\b(?:so+|really|absolutely|totally|completely)\s+(?:thrilled|excited|happy|delighted)\s+(?:about|with|by)\b.*?(?:terrible|awful|worst|bad)',

    # Classic sarcastic phrases
    r'(?i)(?:^|\W)just\s+what\s+(?:I|we)\s+(?:need|wanted|hoped for)\b',
    r'(?i)(?:^|\W)how\s+(?:wonderful|nice|great|lovely|exciting)\b.*?(?:\!|\?{2,})',

    # Thanks for nothing pattern
    r'(?i)(?:^|\W)thanks\s+for\s+(?:nothing|that|pointing|stating)\b',

    # Quotation marks around positive words (scare quotes)
    r'(?i)"(?:great|wonderful|excellent|perfect|amazing)"',

    # Typical sarcastic responses
    r'(?i)^(?:yeah|sure|right|oh)\s+(?:right|sure|okay|ok)(?:\W|$)',

    # Exaggerated praise in negative context
    r'(?i)\b(?:brilliant|genius|impressive)\b.*?(?:disaster|failure|mess)',

    # Obvious understatements
    r'(?i)\b(?:slightly|bit|little)\s+(?:catastrophic|disastrous|terrible|awful)\b',

    # Oh great patterns
    r'(?i)(?:^|\W)oh\s+(?:great|wonderful|perfect|fantastic|awesome)(?:\W|$)'
]

def tokenize_and_clean(text):
    """Tokenize text using spaCy"""
    doc = nlp(text.lower().strip())
    # Return only alphabetic tokens
    return [token.text for token in doc if token.is_alpha]

def detect_phrases(text, emotion_phrases):
    """Detect emotion-specific phrases in text"""
    text_lower = text.lower()
    detected_phrases = {}

    for emotion, phrases in emotion_phrases.items():
        found_phrases = []
        for phrase in phrases:
            if phrase.lower() in text_lower:
                found_phrases.append(phrase)

        if found_phrases:
            detected_phrases[emotion] = found_phrases

    return detected_phrases

def detect_contextual_features(text):
    """Detect contextual features in text that may influence emotion"""
    features = {
        'intensifiers': 0,
        'negators': 0,
        'hedges': 0,
        'boosters': 0,
        'exclamations': text.count('!'),
        'questions': text.count('?'),
        'ellipses': text.count('...'),
        'capitalized_words': len(re.findall(r'\b[A-Z]{2,}\b', text))
    }

    doc = nlp(text.lower())

    # Get tokens for counting
    tokens = [token.text for token in doc]

    # Count contextual indicators
    for indicator_type, words in CONTEXTUAL_INDICATORS.items():
        if indicator_type != 'punctuation':
            for word in words:
                if ' ' in word:  # Multi-word phrase
                    if word in text.lower():
                        features[indicator_type] += 1
                else:  # Single word
                    features[indicator_type] += tokens.count(word)

    return features

def detect_sarcasm_patterns(text):
    """Detect linguistic patterns of sarcasm in text with context awareness"""
    # Match sarcasm patterns
    matches = 0
    pattern_matches = []

    for pattern in SARCASM_PATTERNS:
        if re.search(pattern, text):
            matches += 1
            pattern_matches.append(pattern)

    # Get contextual features
    features = detect_contextual_features(text)

    # Check for phrases specific to sarcasm
    phrases = detect_phrases(text, {'sarcasm': EMOTION_PHRASES['sarcasm']})
    sarcasm_phrases = len(phrases.get('sarcasm', []))

    # Calculate raw score based on pattern matches and features
    raw_score = (matches * 0.15) + (sarcasm_phrases * 0.2)

    # Adjust based on contextual features
    if features['exclamations'] > 1:
        raw_score += min(features['exclamations'] * 0.05, 0.2)

    if features['capitalized_words'] > 0:
        raw_score += min(features['capitalized_words'] * 0.1, 0.3)

    # Detect positive-negative contrasts
    pos_neg_contrast = 0
    emotion_phrases = detect_phrases(text, {
        'positive': EMOTION_PHRASES['joy'] + EMOTION_PHRASES['love'],
        'negative': EMOTION_PHRASES['sadness'] + EMOTION_PHRASES['anger']
    })

    if emotion_phrases.get('positive') and emotion_phrases.get('negative'):
        pos_neg_contrast = 0.3

    # Add contrast score
    raw_score += pos_neg_contrast

    # Normalize to [0, 1]
    return min(raw_score, 1.0), pattern_matches

def calculate_emotion_similarity(text, emotion_keywords):
    """Calculate similarity between text and emotion keywords using spaCy"""
    if not text.strip():
        return 0.0

    # Process the input text
    doc = nlp(text.lower())

    # Get average similarity with emotion keywords
    keyword_scores = []

    # Use a subset of keywords for efficiency
    for keyword in emotion_keywords[:6]:  # Use first 6 keywords for each emotion
        keyword_doc = nlp(keyword)
        # Calculate maximum similarity between any token in text and the keyword
        max_similarity = 0
        for token in doc:
            if token.is_alpha and not token.is_stop:
                for keyword_token in keyword_doc:
                    similarity = token.similarity(keyword_token)
                    max_similarity = max(max_similarity, similarity)

        keyword_scores.append(max_similarity)

    # Return average of top 3 similarities if we have at least 3 scores
    if len(keyword_scores) >= 3:
        return sum(sorted(keyword_scores, reverse=True)[:3]) / 3
    # Otherwise return average of all scores
    elif keyword_scores:
        return sum(keyword_scores) / len(keyword_scores)
    else:
        return 0.0

def get_emotion_score(text, emotion, keywords):
    """Calculate emotion score based on similarity, context, and phrase detection"""
    # Get emotion score using spaCy word vectors
    similarity_score = calculate_emotion_similarity(text, keywords)

    # Check for emotion-specific phrases
    detected_phrases = detect_phrases(text, {emotion: EMOTION_PHRASES[emotion]})
    phrase_count = len(detected_phrases.get(emotion, []))
    phrase_score = min(phrase_count * 0.2, 0.6)  # Cap at 0.6

    # Get contextual features
    features = detect_contextual_features(text)

    # Calculate feature-based adjustment
    feature_adjustment = 0

    # Search for direct emotion mentions in text
    doc = nlp(text.lower())
    direct_mention_score = 0

    for token in doc:
        if token.lemma_ in keywords:
            direct_mention_score += 0.2  # Direct mention of emotion word
            break

    # Adjust score based on emotional context
    if emotion in ['joy', 'love', 'surprise'] and features['exclamations'] > 0:
        feature_adjustment += min(features['exclamations'] * 0.05, 0.2)

    if emotion in ['anger', 'sadness'] and features['negators'] > 0:
        feature_adjustment += min(features['negators'] * 0.05, 0.2)

    if emotion == 'fear' and features['intensifiers'] > 0:
        feature_adjustment += min(features['intensifiers'] * 0.05, 0.2)

    # Combine scores with appropriate weights
    final_score = (similarity_score * 0.5) + (phrase_score * 0.3) + (feature_adjustment * 0.1) + (direct_mention_score * 0.1)

    # Normalize to ensure it's in [0, 1]
    return max(0, min(final_score, 1.0)), detected_phrases.get(emotion, [])

def analyze_sarcasm(text):
    """Specialized analysis for sarcasm detection using spaCy and pattern matching"""
    # 1. Keyword similarity for sarcasm words
    sarcasm_keywords = EMOTION_CATEGORIES['sarcasm']
    similarity_score = calculate_emotion_similarity(text, sarcasm_keywords)

    # 2. Linguistic pattern detection
    pattern_score, pattern_matches = detect_sarcasm_patterns(text)

    # 3. Check for semantic incongruity between sentences
    incongruity_score = 0
    sentences = list(nlp(text).sents)

    if len(sentences) > 1:
        # Calculate similarity between adjacent sentences
        similarities = []
        for i in range(len(sentences) - 1):
            sim = sentences[i].similarity(sentences[i+1])
            similarities.append(sim)

        # Low similarity between adjacent sentences might indicate sarcasm
        if similarities and min(similarities) < 0.5:
            incongruity_score = 0.3

    # 4. Check for sarcasm phrases
    detected_phrases = detect_phrases(text, {'sarcasm': EMOTION_PHRASES['sarcasm']})
    phrase_score = min(len(detected_phrases.get('sarcasm', [])) * 0.2, 0.6)

    # 5. Check for emotional contrast
    # (positive words in negative context or vice versa)
    doc = nlp(text.lower())

    # Count positive and negative words
    pos_count = 0
    neg_count = 0

    for token in doc:
        if token.is_alpha and not token.is_stop:
            # Check against positive and negative emotion keywords
            if any(token.similarity(nlp(word)) > 0.7 for word in EMOTION_CATEGORIES['joy'][:5]):
                pos_count += 1
            if any(token.similarity(nlp(word)) > 0.7 for word in EMOTION_CATEGORIES['sadness'][:5] + EMOTION_CATEGORIES['anger'][:5]):
                neg_count += 1

    contrast_score = 0
    if pos_count > 0 and neg_count > 0:
        contrast_score = min(0.3, pos_count * neg_count * 0.05)

    # Weighted combination of all scores
    combined_score = (0.15 * similarity_score) + (0.35 * pattern_score) + \
                    (0.15 * incongruity_score) + (0.25 * phrase_score) + \
                    (0.1 * contrast_score)

    # Normalize to [0, 1]
    return max(0, min(combined_score, 1.0)), detected_phrases.get('sarcasm', []), pattern_matches

def determine_emotional_verdict(emotion_scores):
    """Determine the emotional verdict based on the emotional profile"""
    # Create a sorted list of emotions by score
    sorted_emotions = sorted(emotion_scores.items(), key=lambda x: x[1], reverse=True)

    # Count emotions over different thresholds
    emotions_over_35 = [e for e, s in sorted_emotions if s > 0.35]
    emotions_over_20 = [e for e, s in sorted_emotions if s > 0.20]
    emotions_over_15 = [e for e, s in sorted_emotions if s > 0.15]

    # Check if we have a strong dominant emotion (over 35%)
    if emotions_over_35:
        dominant_emotion = emotions_over_35[0]
        for verdict_key, verdict_info in EMOTION_VERDICT_CATEGORIES.items():
            conditions = verdict_info['conditions']
            # Check single emotion conditions
            if len(conditions) == 1 and isinstance(conditions[0], tuple):
                emotion, threshold = conditions[0]
                if emotion == dominant_emotion and emotion_scores[emotion] >= threshold:
                    return verdict_info['label'], verdict_info['description']

    # Check for emotion combinations
    for verdict_key, verdict_info in EMOTION_VERDICT_CATEGORIES.items():
        conditions = verdict_info['conditions']

        # Skip single emotion conditions we've already checked
        if len(conditions) == 1 and isinstance(conditions[0], tuple):
            continue

        # Handle special condition types
        if conditions == ['multiple_over_15'] and len(emotions_over_15) >= 3:
            return verdict_info['label'], verdict_info['description']

        if conditions == ['all_under_20'] and not emotions_over_20:
            return verdict_info['label'], verdict_info['description']

        if conditions == ['all_under_15'] and not emotions_over_15:
            return verdict_info['label'], verdict_info['description']

        # Check standard combination conditions
        if all(emotion_scores.get(emotion, 0) >= threshold for emotion, threshold in conditions):
            return verdict_info['label'], verdict_info['description']

    # If we've found nothing specific but have some emotions over 15%
    if emotions_over_15:
        if len(emotions_over_15) == 1:
            # Use the single emotion even though it's not super strong
            emotion = emotions_over_15[0]
            return f"Moderately {emotion.capitalize()}", f"Shows some signs of {emotion}"
        else:
            # Create a custom mixed emotion label
            primary = emotions_over_15[0].capitalize()
            secondary = emotions_over_15[1].capitalize()
            return f"{primary} with {secondary}", f"A mix of {primary.lower()} and {secondary.lower()} emotions"

    # Default fallback
    return "Neutral or Subtle", "No clear emotional signals detected"

def analyze_emotions(text):
    """Analyze emotions in text using spaCy with robust sarcasm detection and emotional verdict"""
    if not text or not text.strip():
        return None, {"error": "Please enter some text to analyze"}

    try:
        # Calculate scores for each emotion with supporting phrases
        emotion_data = {}

        # For each standard emotion category (excluding sarcasm)
        for emotion, keywords in EMOTION_CATEGORIES.items():
            if emotion == 'sarcasm':
                continue

            # Use specialized function to get emotion score and supporting phrases
            score, phrases = get_emotion_score(text, emotion, keywords)
            emotion_data[emotion] = {
                'score': score,
                'phrases': phrases
            }

        # Special handling for sarcasm with multi-method approach
        sarcasm_score, sarcasm_phrases, sarcasm_patterns = analyze_sarcasm(text)
        emotion_data['sarcasm'] = {
            'score': sarcasm_score,
            'phrases': sarcasm_phrases,
            'patterns': sarcasm_patterns
        }

        # Get contextual features for overall analysis
        context_features = detect_contextual_features(text)

        # Apply decision making for final analysis
        # 1. Check for dominant emotions by raw scores
        emotion_scores = {emotion: data['score'] for emotion, data in emotion_data.items()}

        # 2. Adjust based on contextual evidence
        # If we have strong phrase evidence, boost the score slightly
        for emotion, data in emotion_data.items():
            if len(data.get('phrases', [])) > 1:
                emotion_scores[emotion] = min(emotion_scores[emotion] * 1.2, 1.0)

        # 3. Get emotional verdict
        verdict_label, verdict_description = determine_emotional_verdict(emotion_scores)

        # 4. Normalize scores to percentages
        total_score = sum(emotion_scores.values()) or 1  # Avoid division by zero
        emotion_percentages = {emotion: (score / total_score) * 100 for emotion, score in emotion_scores.items()}

        # Sort emotions by percentage for display
        sorted_emotions = sorted(emotion_percentages.items(), key=lambda x: x[1], reverse=True)

        # Prepare result
        result = {
            'emotion_scores': sorted_emotions,
            'emotional_verdict': {
                'label': verdict_label,
                'description': verdict_description
            },
            'top_emotions': sorted_emotions[:3],
            'supporting_evidence': {
                emotion: data.get('phrases', []) for emotion, data in emotion_data.items() if data.get('phrases')
            },
            'context_features': context_features
        }

        return create_visualization(result), result

    except Exception as e:
        import traceback
        error_msg = traceback.format_exc()
        return None, {"error": f"Analysis error: {str(e)}", "details": error_msg}

def create_visualization(result):
    """Create visualization of emotion analysis results"""
    # Create figure and axis
    fig, ax = plt.subplots(figsize=(12, 8))

    # Extract emotion data
    emotions = [e[0] for e in result['emotion_scores']]
    scores = [e[1] for e in result['emotion_scores']]

    # Get colors
    colors = [EMOTION_COLORS.get(emotion, '#CCCCCC') for emotion in emotions]

    # Create horizontal bar chart
    bars = ax.barh(emotions, scores, color=colors)

    # Set x-axis to a static 100%
    ax.set_xlim(0, 100)
    ax.set_xticks(range(0, 101, 10))
    ax.set_xticklabels([f'{i}%' for i in range(0, 101, 10)])

    # Add title and labels
    ax.set_title('Emotion Analysis', fontsize=16, fontweight='bold')
    ax.set_ylabel('Emotions', fontsize=12)
    ax.set_xlabel('Score (%)', fontsize=12)

    # Add verdict as text annotation below the chart
    verdict = result['emotional_verdict']['label']
    description = result['emotional_verdict']['description']
    ax.text(0.5, -0.2, f"Verdict: {verdict}", ha='center', transform=ax.transAxes, fontsize=14, fontweight='bold')
    ax.text(0.5, -0.27, f"{description}", ha='center', transform=ax.transAxes, fontsize=12)

    # Add value labels on top of bars
    for bar in bars:
        width = bar.get_width()
        ax.text(width + 0.5, bar.get_y() + bar.get_height() / 2,
                f'{width:.1f}%', ha='left', va='center', fontsize=10)

    # Adjust layout
    plt.tight_layout()
    plt.subplots_adjust(bottom=0.3)  # Make room for the verdict text

    # Return figure
    return fig


def analyze_text(text):
    """Analyze text and return visualization and detailed results"""
    fig, result = analyze_emotions(text)

    if 'error' in result:
        return None, result['error']

    # Format the results for display
    verdict = result['emotional_verdict']['label']
    description = result['emotional_verdict']['description']

    # Create a formatted summary
    summary = f"## Emotional Analysis Results\n\n"
    summary += f"**Verdict:** {verdict}\n\n"
    summary += f"**Description:** {description}\n\n"

    summary += "### Top Emotions:\n"
    for emotion, score in result['top_emotions']:
        summary += f"- {emotion.capitalize()}: {score:.1f}%\n"

    if result['supporting_evidence']:
        summary += "\n### Supporting Evidence:\n"
        for emotion, phrases in result['supporting_evidence'].items():
            if phrases:
                summary += f"- **{emotion.capitalize()}**: {', '.join(phrases)}\n"

    return fig, summary

# Create Gradio interface
def create_interface():
    with gr.Blocks(title="Emotional Analysis Tool") as demo:
        gr.Markdown("# Advanced Emotion Analysis")
        gr.Markdown("Enter text to analyze the emotional content and receive a detailed breakdown.")

        with gr.Row():
            with gr.Column(scale=2):
                text_input = gr.Textbox(
                    label="Text to analyze",
                    placeholder="Enter text here...",
                    lines=10
                )
                analyze_button = gr.Button("Analyze Emotions")

            with gr.Column(scale=3):
                with gr.Tab("Visualization"):
                    plot_output = gr.Plot(label="Emotion Distribution")
                with gr.Tab("Summary"):
                    text_output = gr.Markdown(label="Analysis Summary")

        analyze_button.click(
            fn=analyze_text,
            inputs=text_input,
            outputs=[plot_output, text_output]
        )

        gr.Markdown("""
        ## About This Tool

        This advanced emotion analysis tool uses NLP techniques to detect and analyze emotions in text.
        It can identify:
        - 10 different emotions (joy, sadness, anger, fear, surprise, love, sarcasm, disgust, anticipation, trust)
        - Complex emotional combinations
        - Contextual features that affect emotional interpretation
        - Intelligent emotional verdicts for mixed emotion states

        The model uses word vectors, phrase detection, and contextual analysis for accurate emotion recognition.
        """)

    return demo

# Launch the interface if running directly
if __name__ == "__main__":
    print("Creating Gradio interface...")
    demo = create_interface()
    demo.launch(share=True)
    print("Gradio interface launched!")