import csv

def analyze_tail(filename):
    durations = []
    with open(filename, 'r') as f:
        reader = csv.reader(f)
        next(reader)
        for row in reader:
            if not row or row[0].startswith('#'):
                continue
            try:
                durations.append(int(row[1]))
            except:
                pass
    
    durations.sort()
    n = len(durations)
    
    print(f"\n=== TAIL ANALYSIS ({filename}) ===")
    print(f"Total: {n}")
    print(f"P50: {durations[n//2]}µs")
    print(f"P90: {durations[int(n*0.90)]}µs")
    print(f"P95: {durations[int(n*0.95)]}µs")
    print(f"P99: {durations[int(n*0.99)]}µs")
    print(f"P99.5: {durations[int(n*0.995)]}µs")
    print(f"P99.9: {durations[int(n*0.999)]}µs")
    print(f"Max: {durations[-1]}µs")
    
    # Count stutters
    for threshold in [500, 1000, 2000, 3000, 5000]:
        count = sum(1 for d in durations if d > threshold)
        print(f"Paths > {threshold}µs: {count} ({100*count/n:.2f}%)")
    
    # Distribution buckets
    print("\n=== DISTRIBUTION ===")
    buckets = [(0, 50), (50, 100), (100, 200), (200, 500), (500, 1000), (1000, 2000), (2000, 5000), (5000, float('inf'))]
    for lo, hi in buckets:
        count = sum(1 for d in durations if lo <= d < hi)
        print(f"{lo}-{hi if hi != float('inf') else '∞'}µs: {count} ({100*count/n:.1f}%)")

analyze_tail("/home/popertots/bench/pathfinding_benchmark_baseline_release.csv")
analyze_tail("pathfinding_benchmark_current.csv")