crackerjack 0.32.0__py3-none-any.whl → 0.33.1__py3-none-any.whl

crackerjack/services/performance_monitor.py

@@ -0,0 +1,565 @@
+import json
+import typing as t
+from collections import defaultdict, deque
+from dataclasses import dataclass, field
+from datetime import datetime
+from pathlib import Path
+from threading import Lock
+from typing import Any
+
+from crackerjack.services.logging import get_logger
+from crackerjack.services.memory_optimizer import MemoryOptimizer
+from crackerjack.services.performance_cache import get_performance_cache
+
+
+@dataclass
+class PerformanceMetric:
+    name: str
+    value: float
+    unit: str
+    timestamp: datetime = field(default_factory=datetime.now)
+    metadata: dict[str, Any] = field(default_factory=dict[str, t.Any])
+
+
+@dataclass
+class PhasePerformance:
+    phase_name: str
+    start_time: datetime
+    end_time: datetime | None = None
+    duration_seconds: float = 0.0
+    memory_start_mb: float = 0.0
+    memory_peak_mb: float = 0.0
+    memory_end_mb: float = 0.0
+    cache_hits: int = 0
+    cache_misses: int = 0
+    parallel_operations: int = 0
+    sequential_operations: int = 0
+    success: bool = True
+    metrics: list[PerformanceMetric] = field(default_factory=list)
+
+    def finalize(self, end_time: datetime | None = None) -> None:
+        self.end_time = end_time or datetime.now()
+        self.duration_seconds = (self.end_time - self.start_time).total_seconds()
+
+
+@dataclass
+class WorkflowPerformance:
+    workflow_id: str
+    start_time: datetime
+    end_time: datetime | None = None
+    total_duration_seconds: float = 0.0
+    phases: list[PhasePerformance] = field(default_factory=list)
+    overall_success: bool = True
+    performance_score: float = 0.0
+
+    def add_phase(self, phase: PhasePerformance) -> None:
+        self.phases.append(phase)
+
+    def finalize(self, success: bool = True) -> None:
+        self.end_time = datetime.now()
+        self.total_duration_seconds = (self.end_time - self.start_time).total_seconds()
+        self.overall_success = success
+        self.performance_score = self._calculate_performance_score()
+
+    def _calculate_performance_score(self) -> float:
+        if not self.phases:
+            return 0.0
+
+        duration_score = max(0, 100 - (self.total_duration_seconds / 10))
+
+        total_hits = sum(p.cache_hits for p in self.phases)
+        total_misses = sum(p.cache_misses for p in self.phases)
+        cache_ratio = (
+            total_hits / (total_hits + total_misses)
+            if total_hits + total_misses > 0
+            else 0
+        )
+        cache_score = cache_ratio * 20
+
+        total_parallel = sum(p.parallel_operations for p in self.phases)
+        total_sequential = sum(p.sequential_operations for p in self.phases)
+        parallel_ratio = (
+            total_parallel / (total_parallel + total_sequential)
+            if total_parallel + total_sequential > 0
+            else 0
+        )
+        parallel_score = parallel_ratio * 15
+
+        max_memory = max((p.memory_peak_mb for p in self.phases), default=0)
+        memory_score = max(0, 15 - (max_memory / 50))
+
+        success_score = 10 if self.overall_success else 0
+
+        return min(
+            100,
+            duration_score
+            + cache_score
+            + parallel_score
+            + memory_score
+            + success_score,
+        )
+
+
+@dataclass
+class PerformanceBenchmark:
+    operation_name: str
+    baseline_duration_seconds: float
+    current_duration_seconds: float
+    improvement_percentage: float = 0.0
+    regression: bool = False
+
+    def __post_init__(self) -> None:
+        if self.baseline_duration_seconds > 0:
+            self.improvement_percentage = (
+                (self.baseline_duration_seconds - self.current_duration_seconds)
+                / self.baseline_duration_seconds
+                * 100
+            )
+            self.regression = self.improvement_percentage < 0
+
+
+class PerformanceMonitor:
+    def __init__(
+        self,
+        data_retention_days: int = 30,
+        benchmark_history_size: int = 100,
+    ):
+        self.data_retention_days = data_retention_days
+        self.benchmark_history_size = benchmark_history_size
+        self._initialize_data_structures(benchmark_history_size)
+        self._initialize_services()
+        self._initialize_thresholds()
+
+    def _initialize_data_structures(self, history_size: int) -> None:
+        self._active_workflows: dict[str, WorkflowPerformance] = {}
+        self._active_phases: dict[str, PhasePerformance] = {}
+        self._completed_workflows: deque[WorkflowPerformance] = deque(
+            maxlen=history_size
+        )
+        self._benchmarks: dict[str, deque[float]] = defaultdict(
+            lambda: deque(maxlen=history_size)  # type: ignore[arg-type]
+        )
+
+    def _initialize_services(self) -> None:
+        self._lock = Lock()
+        self._logger = get_logger("crackerjack.performance_monitor")
+        self._memory_optimizer = MemoryOptimizer.get_instance()
+        self._cache = get_performance_cache()
+
+    def _initialize_thresholds(self) -> None:
+        self._warning_thresholds = {
+            "duration_seconds": 30.0,
+            "memory_mb": 100.0,
+            "cache_hit_ratio": 0.5,
+        }
+
+    def start_workflow(self, workflow_id: str) -> None:
+        with self._lock:
+            if workflow_id in self._active_workflows:
+                self._logger.warning(f"Workflow {workflow_id} already being monitored")
+                return
+
+            workflow = WorkflowPerformance(
+                workflow_id=workflow_id,
+                start_time=datetime.now(),
+            )
+
+            self._active_workflows[workflow_id] = workflow
+            self._logger.debug(f"Started monitoring workflow: {workflow_id}")
+
+            self._memory_optimizer.start_profiling()
+
+    def end_workflow(
+        self, workflow_id: str, success: bool = True
+    ) -> WorkflowPerformance:
+        with self._lock:
+            if workflow_id not in self._active_workflows:
+                self._logger.warning(f"Workflow {workflow_id} not found for ending")
+                return WorkflowPerformance(
+                    workflow_id=workflow_id, start_time=datetime.now()
+                )
+
+            workflow = self._active_workflows.pop(workflow_id)
+            workflow.finalize(success)
+
+            self._completed_workflows.append(workflow)
+
+            self._logger.info(
+                f"Completed workflow {workflow_id}: "
+                f"{workflow.total_duration_seconds: .2f}s, "
+                f"score: {workflow.performance_score: .1f}, "
+                f"phases: {len(workflow.phases)}"
+            )
+
+            self._check_performance_warnings(workflow)
+
+            return workflow
+
+    def start_phase(self, workflow_id: str, phase_name: str) -> None:
+        phase_key = f"{workflow_id}: {phase_name}"
+
+        with self._lock:
+            if phase_key in self._active_phases:
+                self._logger.warning(f"Phase {phase_key} already being monitored")
+                return
+
+            memory_mb = self._memory_optimizer.record_checkpoint(f"{phase_name}_start")
+
+            phase = PhasePerformance(
+                phase_name=phase_name,
+                start_time=datetime.now(),
+                memory_start_mb=memory_mb,
+            )
+
+            self._active_phases[phase_key] = phase
+            self._logger.debug(f"Started monitoring phase: {phase_key}")
+
+    def end_phase(
+        self, workflow_id: str, phase_name: str, success: bool = True
+    ) -> PhasePerformance:
+        phase_key = f"{workflow_id}: {phase_name}"
+
+        with self._lock:
+            if phase_key not in self._active_phases:
+                self._logger.warning(f"Phase {phase_key} not found for ending")
+                return PhasePerformance(
+                    phase_name=phase_name, start_time=datetime.now()
+                )
+
+            phase = self._active_phases.pop(phase_key)
+            phase.success = success
+
+            phase.memory_end_mb = self._memory_optimizer.record_checkpoint(
+                f"{phase_name}_end"
+            )
+
+            cache_stats = self._cache.get_stats()
+            phase.cache_hits = cache_stats.hits
+            phase.cache_misses = cache_stats.misses
+
+            phase.finalize()
+
+            if workflow_id in self._active_workflows:
+                self._active_workflows[workflow_id].add_phase(phase)
+
+            self._logger.debug(
+                f"Completed phase {phase_key}: {phase.duration_seconds: .2f}s"
+            )
+
+            return phase
+
+    def record_metric(
+        self,
+        workflow_id: str,
+        phase_name: str,
+        metric_name: str,
+        value: float,
+        unit: str = "",
+        metadata: dict[str, t.Any] | None = None,
+    ) -> None:
+        metric = PerformanceMetric(
+            name=metric_name,
+            value=value,
+            unit=unit,
+            metadata=metadata or {},
+        )
+
+        phase_key = f"{workflow_id}: {phase_name}"
+
+        with self._lock:
+            if phase_key in self._active_phases:
+                self._active_phases[phase_key].metrics.append(metric)
+            else:
+                self._logger.warning(
+                    f"Phase {phase_key} not found for metric {metric_name}"
+                )
+
+    def record_parallel_operation(self, workflow_id: str, phase_name: str) -> None:
+        phase_key = f"{workflow_id}: {phase_name}"
+
+        with self._lock:
+            if phase_key in self._active_phases:
+                self._active_phases[phase_key].parallel_operations += 1
+
+    def record_sequential_operation(self, workflow_id: str, phase_name: str) -> None:
+        phase_key = f"{workflow_id}: {phase_name}"
+
+        with self._lock:
+            if phase_key in self._active_phases:
+                self._active_phases[phase_key].sequential_operations += 1
+
+    def benchmark_operation(
+        self, operation_name: str, duration_seconds: float
+    ) -> PerformanceBenchmark:
+        with self._lock:
+            history = self._benchmarks[operation_name]
+            history.append(duration_seconds)
+
+            if len(history) > 1:
+                sorted_history = sorted(history)
+                baseline = sorted_history[len(sorted_history) // 2]
+
+                return PerformanceBenchmark(
+                    operation_name=operation_name,
+                    baseline_duration_seconds=baseline,
+                    current_duration_seconds=duration_seconds,
+                )
+            else:
+                return PerformanceBenchmark(
+                    operation_name=operation_name,
+                    baseline_duration_seconds=duration_seconds,
+                    current_duration_seconds=duration_seconds,
+                )
+
+    def get_performance_summary(self, last_n_workflows: int = 10) -> dict[str, Any]:
+        with self._lock:
+            recent_workflows = list[t.Any](self._completed_workflows)[
+                -last_n_workflows:
+            ]
+
+            if not recent_workflows:
+                return {"message": "No completed workflows to analyze"}
+
+            basic_stats = self._calculate_basic_workflow_stats(recent_workflows)
+            cache_stats = self._calculate_cache_statistics(recent_workflows)
+            parallel_stats = self._calculate_parallelization_statistics(
+                recent_workflows
+            )
+
+            return (
+                {
+                    "workflows_analyzed": len(recent_workflows),
+                }
+                | basic_stats
+                | cache_stats
+                | parallel_stats
+                | {}
+            )
+
+    def _calculate_basic_workflow_stats(
+        self, workflows: list[WorkflowPerformance]
+    ) -> dict[str, Any]:
+        total_duration = sum(w.total_duration_seconds for w in workflows)
+        avg_duration = total_duration / len(workflows)
+        avg_score = sum(w.performance_score for w in workflows) / len(workflows)
+        success_rate = sum(1 for w in workflows if w.overall_success) / len(workflows)
+
+        return {
+            "avg_duration_seconds": round(avg_duration, 2),
+            "avg_performance_score": round(avg_score, 1),
+            "success_rate": round(success_rate, 2),
+        }
+
+    def _calculate_cache_statistics(
+        self, workflows: list[WorkflowPerformance]
+    ) -> dict[str, Any]:
+        total_cache_hits = sum(sum(p.cache_hits for p in w.phases) for w in workflows)
+        total_cache_misses = sum(
+            sum(p.cache_misses for p in w.phases) for w in workflows
+        )
+
+        cache_hit_ratio = (
+            total_cache_hits / (total_cache_hits + total_cache_misses)
+            if total_cache_hits + total_cache_misses > 0
+            else 0
+        )
+
+        return {
+            "cache_hit_ratio": round(cache_hit_ratio, 2),
+            "total_cache_hits": total_cache_hits,
+            "total_cache_misses": total_cache_misses,
+        }
+
+    def _calculate_parallelization_statistics(
+        self, workflows: list[WorkflowPerformance]
+    ) -> dict[str, Any]:
+        total_parallel = sum(
+            sum(p.parallel_operations for p in w.phases) for w in workflows
+        )
+        total_sequential = sum(
+            sum(p.sequential_operations for p in w.phases) for w in workflows
+        )
+
+        parallel_ratio = (
+            total_parallel / (total_parallel + total_sequential)
+            if total_parallel + total_sequential > 0
+            else 0
+        )
+
+        return {
+            "parallel_operation_ratio": round(parallel_ratio, 2),
+            "total_parallel_operations": total_parallel,
+            "total_sequential_operations": total_sequential,
+        }
+
+    def get_benchmark_trends(self) -> dict[str, dict[str, Any]]:
+        trends = {}
+
+        with self._lock:
+            for operation_name, history in self._benchmarks.items():
+                if len(history) < 2:
+                    continue
+
+                history_list = list[t.Any](history)
+                basic_stats = self._calculate_benchmark_basic_stats(history_list)
+                trend_percentage = self._calculate_trend_percentage(history_list)
+
+                trends[operation_name] = basic_stats | {
+                    "trend_percentage": round(trend_percentage, 1),
+                    "sample_count": len(history_list),
+                }
+
+        return trends
+
+    def _calculate_benchmark_basic_stats(
+        self, history_list: list[float]
+    ) -> dict[str, float]:
+        avg_duration = sum(history_list) / len(history_list)
+        min_duration = min(history_list)
+        max_duration = max(history_list)
+
+        return {
+            "avg_duration_seconds": round(avg_duration, 3),
+            "min_duration_seconds": round(min_duration, 3),
+            "max_duration_seconds": round(max_duration, 3),
+        }
+
+    def _calculate_trend_percentage(self, history_list: list[float]) -> float:
+        if len(history_list) < 5:
+            return 0.0
+
+        recent_avg = sum(history_list[-5:]) / 5
+        older_avg = (
+            sum(history_list[:-5]) / len(history_list[:-5])
+            if len(history_list) > 5
+            else recent_avg
+        )
+
+        return ((older_avg - recent_avg) / older_avg * 100) if older_avg > 0 else 0.0
+
+    def export_performance_data(self, output_path: Path) -> None:
+        with self._lock:
+            data = {
+                "export_timestamp": datetime.now().isoformat(),
+                "completed_workflows": [
+                    {
+                        "workflow_id": w.workflow_id,
+                        "start_time": w.start_time.isoformat(),
+                        "end_time": w.end_time.isoformat() if w.end_time else None,
+                        "duration_seconds": w.total_duration_seconds,
+                        "performance_score": w.performance_score,
+                        "success": w.overall_success,
+                        "phases": [
+                            {
+                                "name": p.phase_name,
+                                "duration_seconds": p.duration_seconds,
+                                "memory_peak_mb": p.memory_peak_mb,
+                                "cache_hits": p.cache_hits,
+                                "cache_misses": p.cache_misses,
+                                "parallel_operations": p.parallel_operations,
+                                "sequential_operations": p.sequential_operations,
+                                "success": p.success,
+                            }
+                            for p in w.phases
+                        ],
+                    }
+                    for w in self._completed_workflows
+                ],
+                "benchmarks": {
+                    name: list[t.Any](history)
+                    for name, history in self._benchmarks.items()
+                },
+                "summary": self.get_performance_summary(),
+                "trends": self.get_benchmark_trends(),
+            }
+
+        with output_path.open("w") as f:
+            json.dump(data, f, indent=2)
+
+        self._logger.info(f"Exported performance data to {output_path}")
+
+    def _check_performance_warnings(self, workflow: WorkflowPerformance) -> None:
+        warnings = []
+
+        warnings.extend(self._check_duration_warning(workflow))
+        warnings.extend(self._check_memory_warning(workflow))
+        warnings.extend(self._check_cache_warning(workflow))
+
+        for warning in warnings:
+            self._logger.debug(
+                f"Performance warning for {workflow.workflow_id}: {warning}"
+            )
+
+    def _check_duration_warning(self, workflow: WorkflowPerformance) -> list[str]:
+        if (
+            workflow.total_duration_seconds
+            > self._warning_thresholds["duration_seconds"]
+        ):
+            return [
+                f"Slow workflow duration: {workflow.total_duration_seconds: .1f}s "
+                f"(threshold: {self._warning_thresholds['duration_seconds']}s)"
+            ]
+        return []
+
+    def _check_memory_warning(self, workflow: WorkflowPerformance) -> list[str]:
+        max_memory = max((p.memory_peak_mb for p in workflow.phases), default=0)
+        if max_memory > self._warning_thresholds["memory_mb"]:
+            return [
+                f"High memory usage: {max_memory: .1f}MB "
+                f"(threshold: {self._warning_thresholds['memory_mb']}MB)"
+            ]
+        return []
+
+    def _check_cache_warning(self, workflow: WorkflowPerformance) -> list[str]:
+        total_hits = sum(p.cache_hits for p in workflow.phases)
+        total_misses = sum(p.cache_misses for p in workflow.phases)
+
+        if total_hits + total_misses > 0:
+            hit_ratio = total_hits / (total_hits + total_misses)
+            if hit_ratio < self._warning_thresholds["cache_hit_ratio"]:
+                return [
+                    f"Low cache hit ratio: {hit_ratio: .2f} "
+                    f"(threshold: {self._warning_thresholds['cache_hit_ratio']})"
+                ]
+        return []
+
+
+_global_monitor: PerformanceMonitor | None = None
+_monitor_lock = Lock()
+
+
+def get_performance_monitor() -> PerformanceMonitor:
+    global _global_monitor
+    with _monitor_lock:
+        if _global_monitor is None:
+            _global_monitor = PerformanceMonitor()
+        return _global_monitor
+
+
+class phase_monitor:
+    def __init__(self, workflow_id: str, phase_name: str):
+        self.workflow_id = workflow_id
+        self.phase_name = phase_name
+        self.monitor = get_performance_monitor()
+
+    def __enter__(self) -> "phase_monitor":
+        self.monitor.start_phase(self.workflow_id, self.phase_name)
+        return self
+
+    def __exit__(
+        self,
+        exc_type: type[BaseException] | None,
+        exc_val: BaseException | None,
+        exc_tb: object | None,
+    ) -> None:
+        success = exc_type is None
+        self.monitor.end_phase(self.workflow_id, self.phase_name, success)
+
+    def record_parallel_op(self) -> None:
+        self.monitor.record_parallel_operation(self.workflow_id, self.phase_name)
+
+    def record_sequential_op(self) -> None:
+        self.monitor.record_sequential_operation(self.workflow_id, self.phase_name)
+
+    def record_metric(self, name: str, value: float, unit: str = "") -> None:
+        self.monitor.record_metric(self.workflow_id, self.phase_name, name, value, unit)