[metrics] Add custom thresholds for metrics calculations

README.md

@@ -57,9 +57,12 @@ grimoirelab-metrics spdx.xml \
   --opensearch-url https://admin:[email protected]:9200 \
   --opensearch-index events \
   --from-date 2024-01-01 --to-date 2025-01-01 \
-  --repository-timeout 3600
+  --repository-timeout 3600 \
   --code-file-pattern "\.py$|\.js$" \
   --binary-file-pattern "\.exe$|\.tar$" \
+  --pony-threshold 0.5 \
+  --elephant-threshold 0.5 \
+  --dev-categories-thresholds 0.8 0.95 \
   --output metrics.json
 ```
 

grimoirelab_metrics/cli.py

@@ -80,6 +80,17 @@
 @click.option("--verbose", is_flag=True, default=False, help="Increase output verbosity")
 @click.option("--code-file-pattern", help="Regular expression to match code file types")
 @click.option("--binary-file-pattern", help="Regular expression to match binary file types")
+@click.option("--pony-threshold", type=click.FloatRange(0, 1), show_default=True, help="Pony factor threshold", default=0.5)
+@click.option(
+    "--elephant-threshold", type=click.FloatRange(0, 1), show_default=True, help="Elephant factor threshold", default=0.5
+)
+@click.option(
+    "--dev-categories-thresholds",
+    type=(click.FloatRange(0, 1), click.FloatRange(0, 1)),
+    show_default=True,
+    help="Developer categories thresholds",
+    default=(0.8, 0.95),
+)
 def grimoirelab_metrics(
     filename: str,
     grimoirelab_url: str,
@@ -95,6 +106,9 @@ def grimoirelab_metrics(
     verbose: bool = False,
     code_file_pattern: str | None = None,
     binary_file_pattern: str | None = None,
+    pony_threshold: float = 0.5,
+    elephant_threshold: float = 0.5,
+    dev_categories_thresholds: tuple[float, float] = (0.8, 0.95),
 ) -> None:
     """Calculate metrics using GrimoireLab.
 
@@ -136,6 +150,9 @@ def grimoirelab_metrics(
             timeout=repository_timeout,
             code_file_pattern=code_file_pattern,
             binary_file_pattern=binary_file_pattern,
+            pony_threshold=pony_threshold,
+            elephant_threshold=elephant_threshold,
+            dev_categories_thresholds=dev_categories_thresholds,
         )
 
         package_metrics = {"packages": {}}
@@ -212,6 +229,9 @@ def generate_metrics_when_ready(
     timeout: int = 3600,
     code_file_pattern: str | None = None,
     binary_file_pattern: str | None = None,
+    pony_threshold: float = 0.5,
+    elephant_threshold: float = 0.5,
+    dev_categories_thresholds: tuple[float, float] = (0.8, 0.95),
 ) -> dict[str:Any]:
     """Generate metrics once the repositories have finished the collection.
 
@@ -225,6 +245,9 @@ def generate_metrics_when_ready(
     :param timeout: Seconds to wait before failing getting metrics
     :param code_file_pattern: Regular expression to match code file types.
     :param binary_file_pattern: Regular expression to match binary file types.
+    :param pony_threshold: Pony Factor threshold.
+    :param elephant_threshold: Elephant Factor threshold.
+    :param dev_categories_thresholds: Developer Categories thresholds.
     """
     logging.info("Generating metrics")
 
@@ -247,6 +270,9 @@ def generate_metrics_when_ready(
                     verify_certs=verify_certs,
                     code_file_pattern=code_file_pattern,
                     binary_file_pattern=binary_file_pattern,
+                    pony_threshold=pony_threshold,
+                    elephant_threshold=elephant_threshold,
+                    dev_categories_thresholds=dev_categories_thresholds,
                 )
                 processed.add(repository)
 

grimoirelab_metrics/metrics.py

@@ -51,7 +51,14 @@
 
 
 class GitEventsAnalyzer:
-    def __init__(self, code_file_pattern: str | None = None, binary_file_pattern: str | None = None):
+    def __init__(
+        self,
+        code_file_pattern: str | None = None,
+        binary_file_pattern: str | None = None,
+        pony_threshold: float = 0.5,
+        elephant_threshold: float = 0.5,
+        dev_categories_thresholds: tuple[float, float] = (0.8, 0.95),
+    ):
         self.total_commits: int = 0
         self.contributors: Counter = Counter()
         self.companies: Counter = Counter()
@@ -61,6 +68,9 @@ def __init__(self, code_file_pattern: str | None = None, binary_file_pattern: st
         self.messages_sizes: list = []
         self.re_code_pattern = re.compile(code_file_pattern or FILE_TYPE_CODE)
         self.re_binary_pattern = re.compile(binary_file_pattern or FILE_TYPE_BINARY)
+        self.pony_threshold = pony_threshold
+        self.elephant_threshold = elephant_threshold
+        self.dev_categories_thresholds = dev_categories_thresholds
 
     def process_events(self, events: iter(dict[str, Any])):
         for event in events:
@@ -93,7 +103,7 @@ def get_pony_factor(self):
         for _, contributions in self.contributors.most_common():
             partial_contributions += contributions
             pony_factor += 1
-            if partial_contributions / self.total_commits > 0.5:
+            if partial_contributions / self.total_commits > self.pony_threshold:
                 break
 
         return pony_factor
@@ -110,7 +120,7 @@ def get_elephant_factor(self):
         for _, contributions in self.companies.most_common():
             partial_contributions += contributions
             elephant_factor += 1
-            if partial_contributions / self.total_commits > 0.5:
+            if partial_contributions / self.total_commits > self.elephant_threshold:
                 break
 
         return elephant_factor
@@ -173,8 +183,8 @@ def get_developer_categories(self):
         core = 0
         regular = 0
         casual = 0
-        regular_threshold = int(0.8 * self.total_commits)
-        casual_threshold = int(0.95 * self.total_commits)
+        regular_threshold = int(self.dev_categories_thresholds[0] * self.total_commits)
+        casual_threshold = int(self.dev_categories_thresholds[1] * self.total_commits)
         acc_commits = 0
         last_core_contribution = 0
 
@@ -244,6 +254,9 @@ def get_repository_metrics(
     verify_certs: bool = True,
     code_file_pattern: str | None = None,
     binary_file_pattern: str | None = None,
+    pony_threshold: float | None = None,
+    elephant_threshold: float | None = None,
+    dev_categories_thresholds: tuple[float, float] = (0.8, 0.95),
 ):
     """
     Get the metrics from a repository.
@@ -256,14 +269,23 @@ def get_repository_metrics(
     :param to_date: End date, by default None
     :param code_file_pattern: Regular expression to match code file types.
     :param binary_file_pattern: Regular expression to match binary file types.
+    :param pony_threshold: Threshold for the pony factor
+    :param elephant_threshold: Threshold for the elephant factor
+    :param dev_categories_thresholds: Threshold for the developer categories
     """
     os_conn = connect_to_opensearch(opensearch_url, verify_certs=verify_certs)
 
     metrics = {"metrics": {}}
 
     events = get_repository_events(os_conn, opensearch_index, repository, from_date, to_date)
 
-    analyzer = GitEventsAnalyzer(code_file_pattern=code_file_pattern, binary_file_pattern=binary_file_pattern)
+    analyzer = GitEventsAnalyzer(
+        code_file_pattern=code_file_pattern,
+        binary_file_pattern=binary_file_pattern,
+        pony_threshold=pony_threshold,
+        elephant_threshold=elephant_threshold,
+        dev_categories_thresholds=dev_categories_thresholds,
+    )
     analyzer.process_events(events)
 
     metrics["metrics"]["total_commits"] = analyzer.get_commit_count()

tests/unit/test_metrics.py

@@ -79,6 +79,14 @@ def test_get_pony_factor(self):
         self.analyzer.process_events(extra_events)
         self.assertEqual(self.analyzer.get_pony_factor(), 2)
 
+    def test_get_pony_factor_custom_threshold(self):
+        """Test the computation of the pony factor is correct with a custom threshold"""
+
+        analyzer = GitEventsAnalyzer(pony_threshold=0.8)
+        analyzer.process_events(self.events)
+
+        self.assertEqual(analyzer.get_pony_factor(), 2)
+
     def test_get_elephant_factor(self):
         """Test the computation of the elephant factor is correct"""
 
@@ -113,6 +121,14 @@ def test_get_elephant_factor(self):
         self.analyzer.process_events(extra_events)
         self.assertEqual(self.analyzer.get_elephant_factor(), 2)
 
+    def test_get_elephant_factor_custom_threshold(self):
+        """Test the computation of the elephant factor is correct with a custom threshold"""
+
+        analyzer = GitEventsAnalyzer(elephant_threshold=0.8)
+        analyzer.process_events(self.events)
+
+        self.assertEqual(analyzer.get_elephant_factor(), 2)
+
     def test_file_type_metrics(self):
         """Test that file type metrics are calculated correctly"""
 
@@ -280,6 +296,21 @@ def test_get_developer_categories_tied(self):
         categories = self.analyzer.get_developer_categories()
         self.assertDictEqual(categories, {"core": 4, "regular": 1, "casual": 0})
 
+    def test_get_developer_categories_custom_threshold(self):
+        """Test if the categories are calculated correctly with a custom threshold"""
+
+        analyzer = GitEventsAnalyzer(dev_categories_thresholds=(0.5, 0.9))
+        analyzer.process_events(self.events)
+
+        categories = analyzer.get_developer_categories()
+        self.assertDictEqual(categories, {"core": 1, "regular": 1, "casual": 1})
+
+        analyzer_2 = GitEventsAnalyzer(dev_categories_thresholds=(0.95, 0.99))
+        analyzer_2.process_events(self.events)
+
+        categories = analyzer_2.get_developer_categories()
+        self.assertDictEqual(categories, {"core": 2, "regular": 0, "casual": 1})
+
 
 if __name__ == "__main__":
     unittest.main()