Add artifact_3_retry_tail_analysis/src/artifact_3_retry_tail_analysis/core.py

artifact_3_retry_tail_analysis/src/artifact_3_retry_tail_analysis/core.py

@@ -0,0 +1,66 @@
+from __future__ import annotations
+import json
+from pathlib import Path
+from statistics import quantiles
+from typing import Dict, Any
+import pandas as pd
+
+
+class InputValidationError(Exception):
+    """Raised when input data does not meet expected structure."""
+    pass
+
+
+def _validate_input(data: Dict[str, Any]) -> None:
+    """Validate that input data contain retriable latency samples or 'retry_tail_p99'."""
+    if not isinstance(data, dict):
+        raise InputValidationError("Expected a dict input.")
+    if 'retry_tail_p99' in data:
+        val = data['retry_tail_p99']
+        if not isinstance(val, (int, float)):
+            raise InputValidationError("'retry_tail_p99' must be numeric.")
+    elif 'retries' in data:
+        if not isinstance(data['retries'], (list, tuple)):
+            raise InputValidationError("'retries' must be list or tuple of numbers.")
+    else:
+        raise InputValidationError("Input missing 'retry_tail_p99' or 'retries'.")
+
+
+def _extract_tail_p99(data: Dict[str, Any]) -> float:
+    """Extract or compute retry_tail_p99 from the provided data dict."""
+    if 'retry_tail_p99' in data:
+        return float(data['retry_tail_p99'])
+    retries = data.get('retries', [])
+    if not retries:
+        raise InputValidationError("Cannot compute retry_tail_p99 from empty retries list.")
+    series = pd.Series(retries, dtype='float64')
+    return float(series.quantile(0.99))
+
+
+def analyze_retry_tails(data_enforced: Dict[str, Any], data_randomized: Dict[str, Any]) -> Dict[str, float]:
+    """Vergleicht Retry-Tail-Werte zwischen zwei Datenquellen für 'affinity enforced' und 'affinity off'-Modi.
+
+    Args:
+        data_enforced: Retry-Daten im Affinitätsmodus (pinned).
+        data_randomized: Retry-Daten im Randomized-Modus (affinity off).
+
+    Returns:
+        dict: Ergebnisse mit Kennzahlen wie `retry_tail_p99_enforced`, `retry_tail_p99_randomized`,
+              und Differenz `delta_retry_tail_p99`.
+    """
+    _validate_input(data_enforced)
+    _validate_input(data_randomized)
+
+    enforced_p99 = _extract_tail_p99(data_enforced)
+    randomized_p99 = _extract_tail_p99(data_randomized)
+
+    result = {
+        'retry_tail_p99_enforced': enforced_p99,
+        'retry_tail_p99_randomized': randomized_p99,
+        'delta_retry_tail_p99': enforced_p99 - randomized_p99
+    }
+
+    # CI sanity check assertions
+    assert isinstance(result['retry_tail_p99_enforced'], float)
+    assert isinstance(result['retry_tail_p99_randomized'], float)
+    return result