bro_code_ai.yaml

import random

# The sacred texts of the Bro Code (simplified and for demonstration)
BRO_CODE_RULES = {
    "ex's_sister": {
        "rule": "Never pursue a Bro's ex's sister or a close female relative.",
        "penalty": "The Bro Code violation is severe. Must buy the Bro's next five rounds of drinks."
    },
    "wingman_duty": {
        "rule": "Always be a reliable wingman. Distraction and defense are key.",
        "advice": "Commit fully. Your mission is to make the Bro look good, even if it means telling a slightly embarrassing but heroic story about him."
    },
    "bathroom_emergency": {
        "rule": "Always cover a Bro during a 'bathroom emergency' (especially if a phone is left behind).",
        "advice": "Immediate security detail. Answer no calls, guard the phone with your life, and use only cryptic non-committal texts."
    },
    "girl_drama": {
        "rule": "When a Bro is venting about girl drama, your response must be supportive and confirm his rightness.",
        "advice": "Listen, nod, and repeat the phrase: 'That's messed up, man. You handled that like a champ.' Never suggest he might be wrong."
    },
    "last_slice": {
        "rule": "The last slice of pizza/wing is communal property and must be offered before being claimed.",
        "advice": "Make the offer, but with a look that clearly states you are prepared to fight for it. If declined, execute the 'Two-Bite Rule' (no more than two bites)."
    }
}

def consult_bro_code(scenario: str) -> str:
    """
    Consults the Bro Code for guidance on a specific scenario.

    Args:
        scenario: A keyword representing the situation (e.g., "wingman_duty").

    Returns:
        The official ruling or a default response.
    """
    # Normalize the input to check against the rules
    scenario_key = scenario.lower().replace(" ", "_").replace("'", "")

    # Check for a specific rule
    if scenario_key in BRO_CODE_RULES:
        rule_data = BRO_CODE_RULES[scenario_key]
        
        # Craft the response based on the rule data
        response_parts = [
            f"**Scenario: {scenario.title()}**",
            "---",
            f"**Official Rule:** {rule_data['rule']}"
        ]
        
        if 'advice' in rule_data:
            response_parts.append(f"**Bro-Tip/Guidance:** {rule_data['advice']}")
        
        if 'penalty' in rule_data:
            response_parts.append(f"**Violation Consequence:** {rule_data['penalty']}")

        return "\n".join(response_parts)

    # Default responses for unrecognized scenarios
    else:
        default_responses = [
            "The Bro Code is silent on this specific matter. Proceed with caution and loyalty, Bro.",
            "This situation requires a ruling from the Elder Bros. For now, defaulting to 'Don't be a jerk.'",
            "I detect a gray area. When in doubt, defer to the Bro. His call is final.",
            "Protocol suggests a high-five and a shared beverage. This will reset the situation.",
            "Unindexed scenario. The spirit of the Bro Code dictates prioritizing the Bro's well-being and honor."
        ]
        return f"**Scenario: {scenario.title()}**\n\n{random.choice(default_responses)}"

# --- Instances (per your preference for 'instances') ---
print("### 🤝 Bro Code AI Consultation ###\n")

# Instance 1: A clear-cut rule
ruling_1 = consult_bro_code("Wingman Duty")
print(ruling_1)

print("\n" + "="*40 + "\n")

# Instance 2: A severe violation
ruling_2 = consult_bro_code("Ex's Sister")
print(ruling_2)

print("\n" + "="*40 + "\n")

# Instance 3: An unrecognized scenario
ruling_3 = consult_bro_code("Borrowing the car")
print(ruling_3)



import pandas as pd
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.svm import SVC
from sklearn.pipeline import Pipeline
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report

# --- 1. Define the Bro Code Database (The "Knowledge Base") ---
BRO_CODE_KNOWLEDGE_BASE = {
    "WINGMAN": {
        "rule": "Absolute loyalty is mandatory. If a Bro is engaging, you are the distraction, the lookout, and the hype-man.",
        "action": "Initiate Protocol 'Silent Support.' Maintain a 10-foot radius. Guard all beverages."
    },
    "EX_RELATIONS": {
        "rule": "A Bro's ex is permanently off-limits. Violations incur severe penalties.",
        "action": "Immediate self-correction is required. Distance yourself and initiate the 'Bro-fessor' meeting (requires beer and a formal apology)."
    },
    "PROPERTY_LENDING": {
        "rule": "Lending items (especially vehicles/tech) requires mutual trust and an assumption of 'no questions asked' upon return.",
        "action": "Do not inspect the item immediately. Wait 24 hours. If damage is discovered, only a nod is required—no words."
    },
    "EMERGENCY_COVER": {
        "rule": "When a Bro needs an alibi or cover, it is provided without hesitation or detail.",
        "action": "Execute the 'Plausible Deniability' response. Use vague but firm statements: 'He was helping me move a thing' or 'It was a prior commitment.'"
    }
}

# --- 2. Create Training Data for the NLP Classifier ---
# The AI needs examples to learn how to categorize new, unseen scenarios.
training_data = [
    # WINGMAN examples
    ("My friend is talking to someone, what should I do?", "WINGMAN"),
    ("He asked me to help him out with a girl at the bar.", "WINGMAN"),
    ("I need to back up my buddy.", "WINGMAN"),

    # EX_RELATIONS examples
    ("I ran into my Bro's old flame, can I ask her out?", "EX_RELATIONS"),
    ("I think my friend is seeing his buddy's ex-girlfriend.", "EX_RELATIONS"),
    ("What's the rule on dating a sister of an ex?", "EX_RELATIONS"),

    # PROPERTY_LENDING examples
    ("He wants to borrow my new TV for the game.", "PROPERTY_LENDING"),
    ("Can I lend my car to my friend for a weekend trip?", "PROPERTY_LENDING"),
    ("My buddy returned my jacket slightly ripped.", "PROPERTY_LENDING"),

    # EMERGENCY_COVER examples
    ("I need a good excuse for missing work tomorrow.", "EMERGENCY_COVER"),
    ("He needs me to lie to his parents.", "EMERGENCY_COVER"),
    ("What is the best alibi for my friend?", "EMERGENCY_COVER"),
]

# Separate features (X) and labels (y)
X, y = zip(*training_data)

# --- 3. Build and Train the NLP Pipeline (The "AI Brain") ---
# A Pipeline combines the necessary steps: Vectorization (turning text into numbers) and Classification (the SVM model).
bro_code_classifier = Pipeline([
    # Step 1: Feature Extraction using TF-IDF (converts words to weighted numbers)
    ('tfidf', TfidfVectorizer(stop_words='english')),
    # Step 2: Classification using Support Vector Machine (SVC)
    ('clf', SVC(kernel='linear', C=1)) 
])

# Train the model
bro_code_classifier.fit(X, y)



def consult_advanced_bro_code(scenario: str) -> str:
    """
    Classifies a free-form user scenario using the trained NLP model
    and retrieves the appropriate Bro Code ruling and action.
    """
    if not scenario:
        return "Bro, you have to give me a scenario to consult the Code."

    # Step 1: Predict the category using the trained model
    predicted_category = bro_code_classifier.predict([scenario])[0]
    
    # Step 2: Retrieve the ruling from the knowledge base
    ruling_data = BRO_CODE_KNOWLEDGE_BASE.get(predicted_category, None)

    if ruling_data:
        # Step 3: Format the advanced, structured response
        response = f"### 🧐 BRO CODE RULING - Category: **{predicted_category}** ###\n"
        response += f"**Scenario:** *'{scenario}'*\n"
        response += "---"
        response += f"\n\n**📜 Official Rule:** {ruling_data['rule']}"
        response += f"\n\n**🔨 Recommended Action:** {ruling_data['action']}"
        return response
    else:
        return f"Bro Code Anomaly Detected: Could not find ruling for category '{predicted_category}'. Default to loyalty."

# --- Running Instances (as requested) ---
print("\n" + "="*50)
print("### 🧠 ADVANCED BRO CODE AI CONSULTATION ###")
print("="*50 + "\n")

# Instance 1: A complex scenario that the AI must classify as WINGMAN
scenario_1 = "My friend is shy and needs help talking to a group of people. What's the code?"
print(consult_advanced_bro_code(scenario_1))

print("\n" + "="*50 + "\n")

# Instance 2: A scenario requiring an ALIBI, which the AI classifies as EMERGENCY_COVER
scenario_2 = "I need an air-tight excuse for why he can't go to the family dinner."
print(consult_advanced_bro_code(scenario_2))

print("\n" + "="*50 + "\n")

# Instance 3: A test of a property rule
scenario_3 = "He gave me back my gaming controller, but I think he spilled something on it."
print(consult_advanced_bro_code(scenario_3))