Merge pull request intel#161 from edwarddavidbaker/fix-inefficient-regex

create_perf_json: Fix inefficient regex warning

.github/workflows/create-perf-json.yml

@@ -29,7 +29,7 @@ jobs:
 
     - name: Run unittests
       working-directory: ./scripts/unittesting
-      run: python metric_test.py
+      run: python -m unittest metric_test.py test_create_perf_json.py
 
     - name: Create perf json files
       working-directory: ./scripts

scripts/create_perf_json.py

@@ -653,8 +653,15 @@ def smi_json(self) -> metric.MetricGroup:
         ])
 
     @staticmethod
-    def extract_pebs_formula(formula):
-        MIN_MAX_PEBS = re.compile(r"([A-Za-z0-9\_\.@]+)\*(min|max)\(\s*(\$PEBS)\s*,((\s*([^\s\)])\s*)+)\)")
+    def extract_pebs_formula(formula: str) -> str:
+        """
+        Convert metric formulas using $PEBS.
+
+        Example:
+            Input:  MEM_INST_RETIRED.STLB_HIT_LOADS*min($PEBS, 7) / tma_info_thread_clks + tma_load_stlb_miss
+            Return: MEM_INST_RETIRED.STLB_HIT_LOADS * min(MEM_INST_RETIRED.STLB_HIT_LOADS:R, 7) / tma_info_thread_clks + tma_load_stlb_miss
+        """
+        MIN_MAX_PEBS = re.compile(r"([A-Za-z0-9_.@]+)\*(min|max)\( *(\$PEBS) *,([a-z0-9_ */+-]+)\)")
         NON_REL_OPS = r"(?<!##)(?<!##2)/|[\(\)]+|[\+\-,]|\*(?![\$])|(?<!##)\?| if not | if | else |min\(|max\(| and | or | in | not "
         REL_OPS = r"[<>]=?|=="
         STR_OPS = r"'[A-Z\-]+'"
@@ -666,11 +673,10 @@ def extract_pebs_formula(formula):
             for m in re.finditer(MIN_MAX_PEBS, formula):
                 main_event = m.group(1).strip()
                 min_max = m.group(2)
-                pebs = m.group(3)
                 alternative = m.group(4).strip()
 
                 mod = 'R' if '@' in main_event else ':R'
-                new_string = f' {main_event} * {min_max}({main_event}{mod}, {alternative}) '
+                new_string = f'{main_event} * {min_max}({main_event}{mod}, {alternative})'
                 new_formula = re.sub(m.re, new_string, new_formula, count=1)
 
         formula_list = re.split(OPS, new_formula)
@@ -681,6 +687,7 @@ def extract_pebs_formula(formula):
                 mod = 'R' if '@' in event_name else ':R'
                 new_element = f'( {event_name} * {event_name}{mod} )'
                 new_formula = new_formula.replace(element, new_element)
+
         return new_formula
 
     def extract_tma_metrics(self, csvfile: TextIO, pmu_prefix: str,

scripts/unittesting/test_create_perf_json.py

@@ -0,0 +1,57 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: BSD-3-Clause
+
+import unittest
+import sys
+from pathlib import Path
+
+# Add create_perf_json.py directory to the path before importing.
+_script_dir = Path(__file__).resolve().parent
+sys.path.append(str(_script_dir.parent))
+
+from create_perf_json import Model
+
+
+class TestModel(unittest.TestCase):
+
+    def test_extract_pebs_formula(self):
+        """Test formulas which use $PEBS but do not include min() or max()."""
+        tests = [
+            (
+                'EVENT.A*$PEBS',
+                '( EVENT.A * EVENT.A:R )',
+            ),
+            (
+                'EVENT.A + [email protected]@*$PEBS',
+                'EVENT.A + ( [email protected]@ * [email protected]@R )',
+            ),
+        ]
+
+        for input, expected_result in tests:
+            with self.subTest(input=input, expected_result=expected_result):
+                self.assertEqual(expected_result, Model.extract_pebs_formula(input))
+
+    def test_extract_pebs_formula_with_min_max(self):
+        """Test formulas which use $PEBS and also min() or max()."""
+        tests = [
+            (
+                'EVENT.A*min( $PEBS, 4) / EVENT.B',
+                'EVENT.A * min(EVENT.A:R, 4) / EVENT.B',
+            ),
+            (
+                'EVENT.A*min($PEBS, 2) / (1 + EVENT.B*max($PEBS, 8))',
+                'EVENT.A * min(EVENT.A:R, 2) / (1 + EVENT.B * max(EVENT.B:R, 8))',
+            ),
+            (
+                'EVENT.A*min($PEBS, 9 * test_info) * (1 + ([email protected]@ / [email protected]@) / 2) / test_info_2',
+                'EVENT.A * min(EVENT.A:R, 9 * test_info) * (1 + ([email protected]@ / [email protected]@) / 2) / test_info_2',
+            ),
+            (
+                '([email protected]@*min($PEBS, 24 * test_info) + [email protected]@*min($PEBS, 24 - test_info) * (1 - ([email protected]@ / ([email protected]@ + [email protected]@)))) * 5',
+                '([email protected]@ * min([email protected]@R, 24 * test_info) + [email protected]@ * min([email protected]@R, 24 - test_info) * (1 - ([email protected]@ / ([email protected]@ + [email protected]@)))) * 5',
+            )
+        ]
+
+        for input, expected_result in tests:
+            with self.subTest(input=input, expected_result=expected_result):
+                self.assertEqual(expected_result, Model.extract_pebs_formula(input))