Add marker_analysis/src/marker_analysis/core.py

marker_analysis/src/marker_analysis/core.py

@@ -0,0 +1,63 @@
+from __future__ import annotations
+import logging
+from typing import List, Dict
+from pydantic import BaseModel, Field, ValidationError
+
+logger = logging.getLogger(__name__)
+
+
+class MarkerData(BaseModel):
+    """Repräsentiert ein einzelnes Marker-Ereignis aus der eBPF-Instrumentierung."""
+
+    marker_id: str = Field(..., description="Eindeutige ID des Marker-Ereignisses.")
+    timestamp: float = Field(..., description="Zeitstempel des Ereignisses in Sekunden.")
+    status: str = Field(..., description="Status, z. B. 'write_pre', 'write_post', 'read_between_steps'.")
+
+
+class MarkerValidationError(Exception):
+    """Wird ausgelöst, wenn Marker-Daten ungültig sind."""
+    pass
+
+
+def validate_markers(marker_data: List[MarkerData]) -> Dict[str, int | float]:
+    """Validiert Marker-Daten auf Konsistenz (write_pre/write_post Paare)."""
+
+    assert isinstance(marker_data, list), "marker_data muss eine Liste von MarkerData sein"
+    
+    try:
+        validated_markers = [MarkerData(**m.dict()) if isinstance(m, MarkerData) else MarkerData(**m) for m in marker_data]
+    except ValidationError as e:
+        logger.error("Validierung der MarkerData fehlgeschlagen: %s", e)
+        raise MarkerValidationError(f"Ungültige Marker-Daten: {e}") from e
+
+    write_pre = {}
+    write_post = set()
+    retry_free_reads = 0
+
+    for m in validated_markers:
+        match m.status:
+            case 'write_pre':
+                write_pre[m.marker_id] = m.timestamp
+            case 'write_post':
+                write_post.add(m.marker_id)
+            case 'read_between_steps':
+                retry_free_reads += 1
+            case _:
+                logger.debug("Unbekannter Markerstatus ignoriert: %s", m.status)
+
+    valid_pairs = sum(1 for mid in write_pre if mid in write_post)
+    missing_post = sum(1 for mid in write_pre if mid not in write_post)
+
+    total_events = len(validated_markers)
+    ratio_complete = valid_pairs / max(len(write_pre), 1)
+
+    result = {
+        'total_events': total_events,
+        'valid_write_pairs': valid_pairs,
+        'missing_post_markers': missing_post,
+        'retry_free_reads': retry_free_reads,
+        'pair_completion_ratio': round(ratio_complete, 3),
+    }
+
+    logger.info("Marker-Validierung abgeschlossen: %s", result)
+    return result