fix: run java data generation tests via surefire overrides

Author: PsiACE

tools/generate_serialization_test_data.py

@@ -24,6 +24,7 @@
 import argparse
 from pathlib import Path
 
+
 def check_command_installed(command):
     """Checks if a command is available in the system path."""
     if shutil.which(command) is None:
@@ -73,8 +74,24 @@ def generate_java_files(workspace_dir, project_dir):
     repo_url = "https://github.com/apache/datasketches-java.git"
     run_command(["git", "clone", repo_url, str(temp_dir)])
 
+    # Upstream surefire excludes the generate_java_files group by default; drop that exclusion in the temp clone.
+    pom_path = temp_dir / "pom.xml"
+    pom_text = pom_path.read_text()
+    pom_text = pom_text.replace(
+        "${testng.generate-java-files},${testng.check-cpp-files},${testng.check-go-files},\n${testng.check-cpp-historical-files}",
+        "${testng.check-cpp-files},${testng.check-go-files},${testng.check-cpp-historical-files}",
+    )
+    pom_path.write_text(pom_text)
+
     # 5. Run Maven to generate files
-    mvn_cmd = ["mvn", "test", "-P", "generate-java-files"]
+    mvn_cmd = [
+        "mvn",
+        "test",
+        "-P",
+        "generate-java-files",
+        "-Dsurefire.groups=generate_java_files",
+        "-Dsurefire.excludedGroups=",
+    ]
     use_shell = False
     if os.name == 'nt': # Windows
         mvn_cmd[0] = "mvn.cmd"