pyserveX/benchmarks/bench_routing.py

#!/usr/bin/env python3
"""
Benchmark script for routing performance comparison.

Compares:
- Pure Python implementation with standard re (_routing_py)
- Cython implementation with PCRE2 JIT (_routing)

Usage:
    python benchmarks/bench_routing.py
"""

import re
import time
import statistics
from typing import Callable, Tuple

from pyserve._routing_py import (
    FastRouter as PyFastRouter,
    FastRouteMatch as PyFastRouteMatch,
)

try:
    from pyserve._routing import (
        FastRouter as CyFastRouter,
        FastRouteMatch as CyFastRouteMatch,
    )
    CYTHON_AVAILABLE = True
except ImportError:
    CYTHON_AVAILABLE = False
    print("Cython module not compiled. Run: poetry run python scripts/build_cython.py\n")


def benchmark(func: Callable, iterations: int = 100000) -> Tuple[float, float]:
    """Benchmark a function and return mean/stdev in nanoseconds."""
    times = []
    
    # Warmup
    for _ in range(1000):
        func()
    
    # Actual benchmark
    for _ in range(iterations):
        start = time.perf_counter_ns()
        func()
        end = time.perf_counter_ns()
        times.append(end - start)
    
    return statistics.mean(times), statistics.stdev(times)


def format_time(ns: float) -> str:
    """Format time in nanoseconds to human readable format."""
    if ns < 1000:
        return f"{ns:.1f} ns"
    elif ns < 1_000_000:
        return f"{ns/1000:.2f} µs"
    else:
        return f"{ns/1_000_000:.2f} ms"


def setup_router(router_class):
    """Setup a router with typical routes."""
    router = router_class()
    
    # Exact routes
    router.add_route("=/health", {"return": "200 OK"})
    router.add_route("=/api/status", {"return": "200 OK"})
    router.add_route("=/favicon.ico", {"return": "204"})
    
    # Regex routes
    router.add_route("~^/api/v1/users/(?P<user_id>\\d+)$", {"proxy_pass": "http://users-service"})
    router.add_route("~^/api/v1/posts/(?P<post_id>\\d+)$", {"proxy_pass": "http://posts-service"})
    router.add_route("~\\.(css|js|png|jpg|gif|svg|woff2?)$", {"root": "./static"})
    router.add_route("~^/api/", {"proxy_pass": "http://api-gateway"})
    
    # Default route
    router.add_route("__default__", {"spa_fallback": True, "root": "./dist"})
    
    return router


def run_benchmarks():
    print("=" * 70)
    print("ROUTING BENCHMARK")
    print("=" * 70)
    print()
    
    # Test paths with different matching scenarios
    test_cases = [
        ("/health", "Exact match (first)"),
        ("/api/status", "Exact match (middle)"),
        ("/api/v1/users/12345", "Regex match with groups"),
        ("/static/app.js", "Regex match (file extension)"),
        ("/api/v2/other", "Regex match (simple prefix)"),
        ("/some/random/path", "Default route (fallback)"),
        ("/nonexistent", "Default route (fallback)"),
    ]
    
    iterations = 100000
    
    print(f"Iterations: {iterations:,}")
    print()
    
    # Setup routers
    py_router = setup_router(PyFastRouter)
    cy_router = setup_router(CyFastRouter) if CYTHON_AVAILABLE else None
    
    results = {}
    
    for path, description in test_cases:
        print(f"Path: {path}")
        print(f"  {description}")
        
        # Python implementation (standard re)
        py_mean, py_std = benchmark(lambda p=path: py_router.match(p), iterations)
        results[(path, "Python (re)")] = py_mean
        print(f"    Python (re):     {format_time(py_mean):>12} ± {format_time(py_std)}")
        
        # Cython implementation (PCRE2 JIT)
        if CYTHON_AVAILABLE and cy_router:
            cy_mean, cy_std = benchmark(lambda p=path: cy_router.match(p), iterations)
            results[(path, "Cython (PCRE2)")] = cy_mean
            speedup = py_mean / cy_mean if cy_mean > 0 else 0
            print(f"    Cython (PCRE2):  {format_time(cy_mean):>12} ± {format_time(cy_std)}  ({speedup:.2f}x faster)")
        
        print()
    
    # Summary
    if CYTHON_AVAILABLE:
        print("=" * 70)
        print("SUMMARY")
        print("=" * 70)
        
        py_total = sum(v for k, v in results.items() if k[1] == "Python (re)")
        cy_total = sum(v for k, v in results.items() if k[1] == "Cython (PCRE2)")
        
        print(f"  Python (re) total:     {format_time(py_total)}")
        print(f"  Cython (PCRE2) total:  {format_time(cy_total)}")
        print(f"  Overall speedup: {py_total / cy_total:.2f}x")
        
        # Show JIT compilation status
        print()
        print("PCRE2 JIT Status:")
        for route in cy_router.list_routes(): # type: ignore False linter error
            if route["type"] == "regex":
                jit = route.get("jit_compiled", False)
                status = "✓ JIT" if jit else "✗ No JIT"
                print(f"  {status}: {route['pattern']}")


if __name__ == "__main__":
    run_benchmarks()