Merge pull request #2 from ACCESS-NRI/cice5-profiling

Add cice5 profiling parser

Author: micaeljtoliveira

src/access/profiling/__init__.py

@@ -14,9 +14,6 @@
 from access.profiling.parser import ProfilingParser
 from access.profiling.fms_parser import FMSProfilingParser
 from access.profiling.payujson_parser import PayuJSONProfilingParser
+from access.profiling.cice5_parser import CICE5ProfilingParser
 
-__all__ = [
-    "ProfilingParser",
-    "FMSProfilingParser",
-    "PayuJSONProfilingParser",
-]
+__all__ = ["ProfilingParser", "FMSProfilingParser", "PayuJSONProfilingParser", "CICE5ProfilingParser"]

src/access/profiling/cice5_parser.py

@@ -0,0 +1,79 @@
+# Copyright 2025 ACCESS-NRI and contributors. See the top-level COPYRIGHT file for details.
+# SPDX-License-Identifier: Apache-2.0
+
+"""Parser for CICE5 profiling data.
+The data to be parsed is written in the following form, where block stats are discarded:
+
+Timer   1:     Total    8133.37 seconds
+  Timer stats (node): min =     8133.36 seconds
+                      max =     8133.37 seconds
+                      mean=     8133.36 seconds
+  Timer stats(block): min =        0.00 seconds
+                      max =        0.00 seconds
+                      mean=        0.00 seconds
+Timer   2:  TimeLoop    8133.00 seconds
+  Timer stats (node): min =     8132.99 seconds
+                      max =     8133.00 seconds
+                      mean=     8132.99 seconds
+  Timer stats(block): min =        0.00 seconds
+                      max =        0.00 seconds
+                      mean=        0.00 seconds
+
+These timers are printed at the end of the CICE5 run and can be an arbitrary number of timers.
+For example, ESM1.6 has 17 timers printed at the end of ice_diag.d output log.
+"""
+
+from access.profiling.parser import ProfilingParser
+import re
+
+
+class CICE5ProfilingParser(ProfilingParser):
+    """CICE5 profiling output parser."""
+
+    def __init__(self):
+        super().__init__()
+        self._metrics = ["min", "max", "mean"]
+
+    @property
+    def metrics(self) -> list:
+        """Implements "metrics" abstract method/property.
+
+        Returns:
+            list: the metric names captured by this parser.
+        """
+        return self._metrics
+
+    def read(self, stream: str) -> dict:
+        """Implements "read" abstract method to parse profiling data in CICE5 log output.
+
+        Args:
+            stream (str): String containing the CICE5 log to be parsed.
+
+        Returns:
+            dict: Parsed timing information.
+
+        Raises:
+            ValueError: If matching timings aren't found.
+        """
+        # Initialize result dictionary
+        result = {"region": [], "min": [], "max": [], "mean": []}
+
+        # Regex pattern to match timer blocks
+        # This captures the region name and the three node timing values
+        pattern = r"Timer\s+\d+:\s+(\w+)\s+[\d.]+\s+seconds\s+Timer stats \(node\): min =\s+([\d.]+) seconds\s+max =\s+([\d.]+) seconds\s+mean=\s+([\d.]+) seconds"
+
+        # Find all matches
+        matches = re.findall(pattern, stream, re.MULTILINE | re.DOTALL)
+
+        if not matches:
+            raise ValueError("No CICE5 profiling data found")
+
+        # Extract data from matches
+        for match in matches:
+            region, min_time, max_time, mean_time = match
+            result["region"].append(region)
+            result["min"].append(float(min_time))
+            result["max"].append(float(max_time))
+            result["mean"].append(float(mean_time))
+
+        return result

tests/test_cice5_profiling.py

@@ -0,0 +1,96 @@
+# Copyright 2025 ACCESS-NRI and contributors. See the top-level COPYRIGHT file for details.
+# SPDX-License-Identifier: Apache-2.0
+
+import pytest
+
+from access.profiling import CICE5ProfilingParser
+
+
+@pytest.fixture(scope="module")
+def cice5_required_metrics():
+    return ("min", "max", "mean")
+
+
+@pytest.fixture(scope="module")
+def cice5_parser():
+    """Fixture instantiating the CICE5 parser."""
+    return CICE5ProfilingParser()
+
+
+@pytest.fixture(scope="module")
+def cice5_profiling():
+    """Fixture returning a dict holding the parsed CICE5 timing content."""
+    return {
+        "region": ["Total", "TimeLoop"],
+        "min": [16197.42, 16197.14],
+        "max": [16197.47, 16197.19],
+        "mean": [16197.44, 16197.16],
+    }
+
+
+@pytest.fixture(scope="module")
+def cice5_log_file():
+    """Fixture returning the CICE5 timing content."""
+    return """ --------------------------------
+   CICE model diagnostic output  
+ --------------------------------
+  
+  Document ice_in namelist parameters:
+  ==================================== 
+  
+  runtype                   = continue
+ Restart read/written        17520   58159382400.0000     
+  0.000000000000000E+000
+
+Timing information:
+
+Timer   1:     Total   16197.47 seconds
+  Timer stats (node): min =    16197.42 seconds
+                      max =    16197.47 seconds
+                      mean=    16197.44 seconds
+  Timer stats(block): min =        0.00 seconds
+                      max =        0.00 seconds
+                      mean=        0.00 seconds
+Timer   2:  TimeLoop   16197.19 seconds
+  Timer stats (node): min =    16197.14 seconds
+                      max =    16197.19 seconds
+                      mean=    16197.16 seconds
+  Timer stats(block): min =        0.00 seconds
+                      max =        0.00 seconds
+                      mean=        0.00 seconds
+"""
+
+
+@pytest.fixture(scope="module")
+def cice5_incorrect_log_file():
+    """Fixture returning an incorrect CICE5 timing content."""
+    return """Timer stats (node): min =    16197.42 seconds
+                      max =    16197.47 seconds
+                      mean=    16197.44 seconds
+  Timer stats(block): min =        0.00 seconds
+                      max =        0.00 seconds
+                      mean=        0.00 seconds"""
+
+
+def test_cice5_profiling(cice5_required_metrics, cice5_parser, cice5_log_file, cice5_profiling):
+    """Test the correct parsing of CICE5 timing information."""
+    parsed_log = cice5_parser.read(cice5_log_file)
+
+    # check metrics are present in parser and parsed output
+    for metric in cice5_required_metrics:
+        assert metric in cice5_parser.metrics, f"{metric} metric not found in CICE5 parser metrics."
+        assert metric in parsed_log, f"{metric} metric not found in CICE5 parsed log."
+
+    # check content for each metric is correct
+    for idx, region in enumerate(cice5_profiling["region"]):
+        assert region in parsed_log["region"], f"{region} not found in CICE5 parsed log"
+        for metric in cice5_required_metrics:
+            assert (
+                cice5_profiling[metric][idx] == parsed_log[metric][idx]
+            ), f"Incorrect {metric} for region {region} (idx: {idx})."
+
+
+def test_cice5_incorrect_profiling(cice5_parser, cice5_incorrect_log_file):
+    """Test the parsing of incirrect CICE5 timing information."""
+    with pytest.raises(ValueError):
+        cice5_parser.read(cice5_incorrect_log_file)