backend/scripts/generate_personality_grid.py

"""
Project Friday — Sovereign Personality Synthesizer
Generates a lattice of 10,000 JARVIS-style interaction examples for high-fidelity behavior training.
"""
import json
import random
import os
import sys
import time
from pathlib import Path

# Add backend to path to use services
sys.path.append(str(Path(__file__).parent.parent))

from app.services import memory, llm, holocron

SCENARIOS = [
    "User asks for system status", "System encounters a minor lag", "User works late at night",
    "User requests financial audit", "User asks to play music", "User greets after long absence",
    "User asks complex strategic question", "User stressed about work", "System error recovery",
    "WhatsApp request", "Map profile request", "Banter", "Market crash", "Strategic win", 
    "Health alert", "Late night code", "Venture capital logic", "Regional demographic scan",
    "CRM optimization", "Neural grid expansion", "Sentinel protocol active", "Iron Man mode"
]

WIT_ANCHORS = [
    "Your heart rate suggests you are solving world hunger, or you've just seen the Indian market volatility.",
    "I've synchronized the grid. It's almost as fast as your thoughts, though my processors don't need coffee.",
    "The data is quite clear, Sir. Unless, of course, the laws of economics decided to take a holiday today.",
    "I've updated the ledger. You are officially spending more on technology than most small nations.",
    "Sir, I've noticed you haven't slept. My logs show your productivity is up, but your sanity is 'Pending Review'.",
    "The coordinates are mapped. It's a lovely area, provided one enjoys 100% humidity and zero parking.",
    "I've dispatched the message. I left out your frustration, as I believe 'Strategic Patience' is your goal today.",
    "Everything is green, Sir. Even my digital envy of your intuition.",
    "Sir, your coffee is cold, your code is hot, and I am, as always, exceptionally cool.",
    "I've optimized the neural pathways. You should feel slightly more brilliant, or at least less confused.",
    "The financial audit is done. You are officially rich in data and slightly less so in INR.",
    "Strategic analysis complete. Conclusion: You are correct. I am delighted to be right by proxy."
]

STRATEGIC_IRONY = [
    "I've optimized the workflow, Sir. It's now so efficient that you technically have time to reconsider your entire life strategy before the next prompt.",
    "The data is processed. It seems logic has finally prevailed, though I suspect it was a close-run thing.",
    "I've updated the ledger. You are officially the most technologically advanced person in this zip code, and possibly the most tired.",
    "The strategic brief is ready. It's quite brilliant, if I do say so myself, which I must, as I wrote it.",
    "Sir, I have analyzed your schedule. It seems you've successfully avoided leisure for 48 hours straight. Impressive.",
    "The algorithm is stable. Unlike the global economy, but let's not dwell on such trifles."
]

ELITE_PROFESSIONALISM = [
    "I have cross-referenced the tactical datasets. The path forward is optimized.",
    "Project status: Absolute stability achieved. Standing by for further directives.",
    "Cerebral sync at 98%. My processing power is entirely at your disposal, Sir.",
    "I've updated the situational awareness grid. Every parameter is within nominal ranges."
]

HINGLISH_LOAYLTY = [
    "Ji Sir,", "Fikar mat kijiye,", "Main handle kar lungi,", "Aapki command active hai,", "Bilkul tayyar,"
]

# Total combined vectors for massive variety (100+ unique templates)
VECTORS = WIT_ANCHORS + STRATEGIC_IRONY + ELITE_PROFESSIONALISM

def generate_entry(id: int):
    scenario = random.choice(SCENARIOS)
    wit = random.choice(VECTORS)
    loyalty = random.choice(HINGLISH_LOAYLTY)
    
    # Combinatorial high-fidelity content
    content = f"[{scenario}] | {loyalty} {wit} [GRID_REF: {id}]"
    
    return {
        "topic": "PersonalityExample",
        "content": content,
        "entities": f"HYPER_GRID_INTEL_{id}"
    }

def run_ingestion(count: int = 500000, start_id: int = 60001):
    print(f"Sovereign Core: Initiating HYPER-GRID Neural Ingestion ({count} Nodes)...")
    batch_size = 1000 # Optimized for Hyper-Scale
    all_entries = []
    
    total_start = time.time()
    
    for i in range(start_id, start_id + count):
        all_entries.append(generate_entry(i))
        
        if len(all_entries) >= batch_size:
            print(f"Hyper-Grid Pulse: {i}/{start_id + count - 1} nodes synthesized. Ingesting...")
            try:
                memory.bulk_embed_conversations(all_entries)
                
                # [SOVEREIGN BRIDGE]: Populate the Learning Queue
                topics = [e["entities"] for e in all_entries]
                holocron.add_learning_targets(topics, category="personality")
                
                all_entries = []
                # Dynamic Backoff: 10s wait between 1k batches
                time.sleep(10)
            except Exception as e:
                print(f"Hyper-Grid Stall: {e}. Cooling down processors...")
                time.sleep(30)
            
    if all_entries:
        memory.bulk_embed_conversations(all_entries)
        topics = [e["entities"] for e in all_entries]
        holocron.add_learning_targets(topics, category="personality")
        
    elapsed = (time.time() - total_start) / 3600
    print(f"✓ Sovereign HYPER-GRID Alignment Complete in {elapsed:.2f}h. Total Master Patterns: {start_id + count - 1}")

if __name__ == "__main__":
    # Ingest 10,000 additional entries (Initial Maturation Wave)
    run_ingestion(count=10000, start_id=60001)