"""
Quick script to generate telecom dataset and place it where the backend expects it.
Run: python generate_data.py
"""
import sys
from pathlib import Path

# Add backend to path
sys.path.insert(0, str(Path(__file__).parent))

import pandas as pd
import numpy as np
from datetime import datetime, timedelta
import random

np.random.seed(42)
random.seed(42)

CITIES = [
    ("Lagos", "Nigeria", 6.5244, 3.3792),
    ("Abuja", "Nigeria", 9.0765, 7.3986),
    ("Kano", "Nigeria", 12.0022, 8.5920),
    ("Port Harcourt", "Nigeria", 4.8156, 7.0498),
    ("Ibadan", "Nigeria", 7.3775, 3.9470),
    ("Benin City", "Nigeria", 6.3350, 5.6037),
    ("Kaduna", "Nigeria", 10.5222, 7.4383),
    ("Enugu", "Nigeria", 6.4402, 7.4943),
    ("Jos", "Nigeria", 9.8965, 8.8583),
    ("Ilorin", "Nigeria", 8.4966, 4.5421),
    ("New York", "USA", 40.7128, -74.0060),
    ("Los Angeles", "USA", 34.0522, -118.2437),
    ("Chicago", "USA", 41.8781, -87.6298),
    ("Houston", "USA", 29.7604, -95.3698),
    ("London", "UK", 51.5074, -0.1278),
    ("Paris", "France", 48.8566, 2.3522),
    ("Berlin", "Germany", 52.5200, 13.4050),
    ("Amsterdam", "Netherlands", 52.3676, 4.9041),
    ("Tokyo", "Japan", 35.6762, 139.6503),
    ("Seoul", "South Korea", 37.5665, 126.9780),
    ("Mumbai", "India", 19.0760, 72.8777),
    ("Dubai", "UAE", 25.2048, 55.2708),
    ("Singapore", "Singapore", 1.3521, 103.8198),
    ("Sydney", "Australia", -33.8688, 151.2093),
    ("Toronto", "Canada", 43.6532, -79.3832),
]

NUM_CELLS = 50
NUM_RECORDS = 20000

print("Generating telecom dataset...")

cells = []
# Assign cities with varying "health profiles" for realistic map colors
CITY_PROFILES = {
    # Good zones (low SLA risk → green on map)
    "Lagos": "good", "Abuja": "good", "Toronto": "good",
    "Singapore": "good", "Sydney": "good", "Amsterdam": "good",
    # Medium zones (moderate SLA risk → yellow on map)
    "London": "medium", "Paris": "medium", "Tokyo": "medium",
    "Seoul": "medium", "Dubai": "medium", "Mumbai": "medium",
    "Berlin": "medium", "Houston": "medium",
    # Poor zones (high SLA risk → orange/red on map)
    "Kano": "poor", "Port Harcourt": "poor", "Ibadan": "poor",
    "Benin City": "poor", "Kaduna": "poor", "Enugu": "poor",
    "Jos": "poor", "Ilorin": "poor",
    # Critical zones
    "New York": "critical", "Los Angeles": "critical", "Chicago": "critical",
}

for i in range(NUM_CELLS):
    city_data = CITIES[i % len(CITIES)]
    cells.append({
        "cell_id": f"CELL_{i+1:03d}",
        "city": city_data[0],
        "country": city_data[1],
        "lat": city_data[2] + np.random.uniform(-0.1, 0.1),
        "lon": city_data[3] + np.random.uniform(-0.1, 0.1),
        "profile": CITY_PROFILES.get(city_data[0], "medium"),
    })

end_time = datetime.now()
start_time = end_time - timedelta(hours=72)

records = []
for i in range(NUM_RECORDS):
    cell = cells[i % NUM_CELLS]
    ts = start_time + timedelta(minutes=i * (72*60/NUM_RECORDS))
    hour = ts.hour
    is_busy = hour in range(17, 22)
    profile = cell.get("profile", "medium")

    # Adjust scenario probability based on city profile
    if profile == "good":
        r = random.random() * 0.55  # mostly normal (0-0.55 → normal)
    elif profile == "poor":
        r = 0.40 + random.random() * 0.60  # mostly issues
    elif profile == "critical":
        r = 0.55 + random.random() * 0.45  # mostly critical issues
    else:  # medium
        r = random.random()
    if r < 0.40:
        scenario = "Normal"
        rsrp = np.random.uniform(-90, -70)
        sinr = np.random.uniform(15, 25)
        prb_dl = np.random.uniform(30, 55)
        prb_ul = np.random.uniform(25, 50)
        thr_dl = np.random.uniform(50, 150)
        thr_ul = np.random.uniform(10, 30)
        lat = np.random.uniform(10, 35)
        mos = np.random.uniform(3.8, 4.5)
        cdr = np.random.uniform(0, 0.8)
        ho = np.random.uniform(95, 99.5)
        bler = np.random.uniform(0.5, 2)
        cqi = np.random.uniform(12, 15)
        rlc = np.random.uniform(0.5, 2)
        sev = "NONE"
        sla_risk = np.random.uniform(0.0, 0.15)   # LOW risk → green on map
        capex = np.random.uniform(5, 25)
    elif r < 0.55:
        scenario = "Capacity Problem"
        rsrp = np.random.uniform(-90, -75)
        sinr = np.random.uniform(10, 18)
        prb_dl = np.random.uniform(87, 100)
        prb_ul = np.random.uniform(85, 100)
        thr_dl = np.random.uniform(10, 30)
        thr_ul = np.random.uniform(2, 8)
        lat = np.random.uniform(40, 90)
        mos = np.random.uniform(2.5, 3.5)
        cdr = np.random.uniform(1.5, 4)
        ho = np.random.uniform(85, 92)
        bler = np.random.uniform(3, 8)
        cqi = np.random.uniform(6, 10)
        rlc = np.random.uniform(3, 8)
        sev = "CRITICAL" if is_busy else "HIGH"
        sla_risk = np.random.uniform(0.6, 0.9)
        capex = np.random.uniform(70, 95)
    elif r < 0.67:
        scenario = "High Latency"
        rsrp = np.random.uniform(-95, -75)
        sinr = np.random.uniform(10, 18)
        prb_dl = np.random.uniform(55, 80)
        prb_ul = np.random.uniform(50, 75)
        thr_dl = np.random.uniform(30, 60)
        thr_ul = np.random.uniform(5, 15)
        lat = np.random.uniform(80, 200)
        mos = np.random.uniform(2.8, 3.5)
        cdr = np.random.uniform(0.5, 2)
        ho = np.random.uniform(90, 96)
        bler = np.random.uniform(2, 5)
        cqi = np.random.uniform(8, 12)
        rlc = np.random.uniform(10, 25)
        sev = "MEDIUM"
        sla_risk = np.random.uniform(0.4, 0.7)
        capex = np.random.uniform(40, 65)
    elif r < 0.77:
        scenario = "Call Drop"
        rsrp = np.random.uniform(-115, -105)
        sinr = np.random.uniform(2, 8)
        prb_dl = np.random.uniform(40, 70)
        prb_ul = np.random.uniform(35, 65)
        thr_dl = np.random.uniform(5, 20)
        thr_ul = np.random.uniform(1, 5)
        lat = np.random.uniform(50, 120)
        mos = np.random.uniform(1.5, 2.5)
        cdr = np.random.uniform(5, 12)
        ho = np.random.uniform(70, 84)
        bler = np.random.uniform(5, 12)
        cqi = np.random.uniform(3, 7)
        rlc = np.random.uniform(5, 15)
        sev = "HIGH"
        sla_risk = np.random.uniform(0.7, 0.95)
        capex = np.random.uniform(60, 85)
    elif r < 0.87:
        scenario = "Handover Failure"
        rsrp = np.random.uniform(-98, -80)
        sinr = np.random.uniform(8, 15)
        prb_dl = np.random.uniform(40, 65)
        prb_ul = np.random.uniform(35, 60)
        thr_dl = np.random.uniform(25, 55)
        thr_ul = np.random.uniform(5, 15)
        lat = np.random.uniform(30, 60)
        mos = np.random.uniform(2.5, 3.5)
        cdr = np.random.uniform(2, 6)
        ho = np.random.uniform(65, 80)
        bler = np.random.uniform(2, 6)
        cqi = np.random.uniform(7, 11)
        rlc = np.random.uniform(3, 7)
        sev = "HIGH"
        sla_risk = np.random.uniform(0.5, 0.75)
        capex = np.random.uniform(45, 70)
    else:
        scenario = "Low MOS"
        rsrp = np.random.uniform(-100, -85)
        sinr = np.random.uniform(5, 12)
        prb_dl = np.random.uniform(60, 85)
        prb_ul = np.random.uniform(55, 80)
        thr_dl = np.random.uniform(15, 40)
        thr_ul = np.random.uniform(3, 10)
        lat = np.random.uniform(40, 80)
        mos = np.random.uniform(1.5, 2.8)
        cdr = np.random.uniform(1, 4)
        ho = np.random.uniform(82, 92)
        bler = np.random.uniform(4, 10)
        cqi = np.random.uniform(5, 9)
        rlc = np.random.uniform(4, 10)
        sev = "MEDIUM"
        sla_risk = np.random.uniform(0.45, 0.70)
        capex = np.random.uniform(35, 60)

    anomaly_flag = 1 if sev in ("CRITICAL", "HIGH") else 0

    # Assign critical_count based on severity for map coloring
    if sev == "CRITICAL":
        critical_count = random.randint(3, 8)
    elif sev == "HIGH":
        critical_count = random.randint(1, 3)
    else:
        critical_count = 0

    records.append({
        "timestamp": ts.strftime("%Y-%m-%d %H:%M:%S"),
        "cell_id": cell["cell_id"],
        "city": cell["city"],
        "country": cell["country"],
        "lat": round(cell["lat"], 6),
        "lon": round(cell["lon"], 6),
        "RSRP": round(rsrp, 2),
        "SINR": round(sinr, 2),
        "PRB_UL": round(prb_ul, 2),
        "PRB_DL": round(prb_dl, 2),
        "throughput_DL": round(thr_dl, 2),
        "throughput_UL": round(thr_ul, 2),
        "latency_ms": round(lat, 2),
        "MOS": round(mos, 3),
        "call_drop_rate": round(cdr, 3),
        "handover_success_rate": round(ho, 2),
        "BLER_UL": round(bler, 2),
        "BLER_DL": round(bler * 0.9, 2),
        "CQI": round(cqi, 1),
        "RLC_retransmissions": round(rlc, 2),
        "hour": hour,
        "issue_type": scenario,
        "severity": sev,
        "anomaly_flag": anomaly_flag,
        "sla_breach_risk": round(sla_risk, 4),
        "capex_score": round(capex, 2),
        "critical_count": critical_count,
        "source_bo": "BO1" if anomaly_flag else ("BO2" if sla_risk > 0.4 else "BO3"),
        "risk_tier": sev if sev != "NONE" else "LOW",
    })

df = pd.DataFrame(records)
# Replace any NaN/inf values with 0
df = df.fillna(0)
df = df.replace([float('inf'), float('-inf')], 0)

# Save to all expected locations
import os

paths = [
    Path(__file__).parent / "data" / "processed" / "network_kpis_full.csv",
    Path(__file__).parent / "unified_telecom_dataset_FILLED.csv",
    Path(__file__).parent.parent / "data" / "unified_telecom_dataset_FILLED.csv",
]

for p in paths:
    p.parent.mkdir(parents=True, exist_ok=True)
    df.to_csv(p, index=False)
    print(f"Saved {len(df)} records to: {p}")

print(f"\nDataset summary:")
print(f"  Records: {len(df):,}")
print(f"  Cells: {df['cell_id'].nunique()}")
print(f"  Cities: {df['city'].nunique()}")
print(f"  Date range: {df['timestamp'].min()} → {df['timestamp'].max()}")
print(f"  Issues: {df['issue_type'].value_counts().to_dict()}")
print(f"  Severities: {df['severity'].value_counts().to_dict()}")
print("\nDone!")