Add trace_analysis/src/trace_analysis/core.py

trace_analysis/src/trace_analysis/core.py

@@ -0,0 +1,75 @@
+import statistics
+from typing import Dict, Any, List
+
+
+def analyze_trace(trace_data: Dict[str, Any]) -> Dict[str, Any]:
+    """Analysiert Trace-Daten auf CPU-Migrationen, Reorder-Score und potenzielle Anomalien.
+
+    Args:
+        trace_data (dict): Enthält Trace-Events, die jeweils timestamp, cpu_path, reorder_score
+            und mixed_snapshot_signature_count umfassen.
+
+    Returns:
+        dict: Analyseergebnis mit Statistiken und Identifikation möglicher Anomalien.
+    """
+    events: List[Dict[str, Any]] = trace_data.get("events", [])
+    
+    if not events:
+        return {
+            "cpu_switch_count": 0,
+            "reorder_score": {"mean": None, "stdev": None, "max": None},
+            "mixed_snapshot_sum": 0,
+            "anomalies": []
+        }
+
+    cpu_switch_count = 0
+    reorder_scores: List[float] = []
+    mixed_snapshot_sum = 0
+    anomalies: List[Dict[str, Any]] = []
+
+    last_cpu = None
+    
+    for ev in events:
+        cpu = ev.get("cpu_path")
+        reorder = ev.get("reorder_score", 0.0)
+        mixed_sig = ev.get("mixed_snapshot_signature_count", 0)
+
+        if last_cpu is not None and cpu != last_cpu:
+            cpu_switch_count += 1
+        last_cpu = cpu
+
+        if isinstance(reorder, (int, float)):
+            reorder_scores.append(reorder)
+            if reorder > 0.8:  # heuristische Anomalieschwelle
+                anomalies.append({
+                    "timestamp": ev.get("timestamp"),
+                    "issue": "High reorder score",
+                    "score": reorder
+                })
+
+        if isinstance(mixed_sig, int):
+            mixed_snapshot_sum += mixed_sig
+            if mixed_sig > 5:
+                anomalies.append({
+                    "timestamp": ev.get("timestamp"),
+                    "issue": "High mixed snapshot signature count",
+                    "count": mixed_sig
+                })
+
+    if reorder_scores:
+        mean_r = statistics.mean(reorder_scores)
+        stdev_r = statistics.stdev(reorder_scores) if len(reorder_scores) > 1 else 0.0
+        max_r = max(reorder_scores)
+    else:
+        mean_r = stdev_r = max_r = None
+
+    return {
+        "cpu_switch_count": cpu_switch_count,
+        "reorder_score": {
+            "mean": mean_r,
+            "stdev": stdev_r,
+            "max": max_r
+        },
+        "mixed_snapshot_sum": mixed_snapshot_sum,
+        "anomalies": anomalies
+    }