api/benchmarks_hub.py

"""
backend/api/benchmarks_hub.py — Gap 3: router per i 3 benchmark standalone

Espone come endpoint HTTP le classi benchmark finora orfane:
  POST /api/debug/benchmark/advanced   ← AdvancedBenchmark (concorrenza)
  POST /api/debug/benchmark/fabric     ← FabricBenchmark   (token rotation)
  POST /api/debug/benchmark/extended   ← ExtendedBenchmark (multi-scenario)

Tutti usano mock interni (asyncio.sleep + dict simulati) — sicuri da eseguire
in produzione senza side-effect su provider reali.
"""
import asyncio, logging, time
from fastapi import APIRouter
from fastapi.responses import JSONResponse

router  = APIRouter()
_logger = logging.getLogger("api.benchmarks_hub")


@router.post("/api/debug/benchmark/advanced")
async def run_advanced() -> JSONResponse:
    """Stress test concorrenza: 3 scenari × N task simulati."""
    t0 = time.monotonic()
    try:
        from api.advanced_complex_benchmark import AdvancedBenchmark
        b = AdvancedBenchmark()
        await b.setup()
        await asyncio.gather(
            b.run_complex_task("complex_reasoning",  "reasoning", count=5),
            b.run_complex_task("data_heavy_sync",    "memory",    count=5),
            b.run_complex_task("multi_provider_chain","compute",  count=5),
        )
        elapsed = round((time.monotonic() - t0) * 1000)
        total   = b.metrics["successes"] + b.metrics["failures"]
        return JSONResponse({
            "ok":           True,
            "type":         "advanced",
            "elapsed_ms":   elapsed,
            "success_rate": round(b.metrics["successes"] / max(1, total) * 100, 1),
            "total_calls":  total,
            "successes":    b.metrics["successes"],
            "failures":     b.metrics["failures"],
            "latencies": {
                k: {"avg_ms": round(sum(v) / len(v), 1), "max_ms": round(max(v), 1), "n": len(v)}
                for k, v in b.metrics.items() if isinstance(v, list) and v
            },
        })
    except Exception as exc:
        _logger.exception("[advanced_bench]")
        return JSONResponse({"ok": False, "type": "advanced", "error": str(exc)}, status_code=500)


@router.post("/api/debug/benchmark/fabric")
async def run_fabric() -> JSONResponse:
    """Benchmark rotazione token e fallback provider."""
    t0 = time.monotonic()
    try:
        from api.fabric_benchmark import FabricBenchmark
        b = FabricBenchmark()
        await b.setup_simulated_environment()
        await b.run_benchmark()
        elapsed = round((time.monotonic() - t0) * 1000)
        lats    = b.stats.get("latencies", [])
        return JSONResponse({
            "ok":            True,
            "type":          "fabric",
            "elapsed_ms":    elapsed,
            "rotations":     b.stats.get("rotations", 0),
            "fallbacks":     b.stats.get("fallbacks", 0),
            "errors":        len(b.stats.get("errors", [])),
            "avg_latency_ms": round(sum(lats) / len(lats), 1) if lats else None,
        })
    except Exception as exc:
        _logger.exception("[fabric_bench]")
        return JSONResponse({"ok": False, "type": "fabric", "error": str(exc)}, status_code=500)


@router.post("/api/debug/benchmark/extended")
async def run_extended() -> JSONResponse:
    """Multi-scenario: normal load + cascading failure + oracle stress."""
    t0 = time.monotonic()
    try:
        from api.extended_benchmark import ExtendedBenchmark
        b = ExtendedBenchmark()
        await b.setup()
        await asyncio.gather(
            b.scenario_normal_load(),
            b.scenario_cascading_failure(),
            b.scenario_oracle_stress(),
        )
        elapsed = round((time.monotonic() - t0) * 1000)
        return JSONResponse({
            "ok":        True,
            "type":      "extended",
            "elapsed_ms": elapsed,
            "scenarios": b.results,
        })
    except Exception as exc:
        _logger.exception("[extended_bench]")
        return JSONResponse({"ok": False, "type": "extended", "error": str(exc)}, status_code=500)