fix: updated the unit tests workflow to simplify job names and read test paths from a separate file (#29472)

scripts/gha_unit_tests_collector.py

@@ -2,24 +2,25 @@ import sys
 import os
 import yaml
 import argparse
+import json
 
 
-def get_all_unit_test_modules():
-    unit_tests_yml = f'{os.getcwd()}/.github/workflows/unit-tests.yml'
-    with open(unit_tests_yml) as file:
-        unit_test_workflow_yaml = yaml.safe_load(file)
+def get_all_unit_test_shards():
+    unit_tests_json = f'{os.getcwd()}/.github/workflows/unit-test-shards.json'
+    with open(unit_tests_json) as file:
+        unit_test_workflow_shards = json.loads(file.read())
 
-    return unit_test_workflow_yaml['jobs']['run-tests']['strategy']['matrix']['test_module']
+    return unit_test_workflow_shards
 
 
 def get_modules_except_cms():
-    all_unit_test_modules = get_all_unit_test_modules()
-    return [module for module in all_unit_test_modules if not module.startswith('cms')]
+    all_unit_test_shards = get_all_unit_test_shards()
+    return [paths for shard_name, paths in all_unit_test_shards.items() if not paths.startswith('cms')]
 
 
 def get_cms_modules():
-    all_unit_test_modules = get_all_unit_test_modules()
-    return [module for module in all_unit_test_modules if module.startswith('cms')]
+    all_unit_test_shards = get_all_unit_test_shards()
+    return [paths for shard_name, paths in all_unit_test_shards.items() if paths.startswith('cms')]
 
 
 if __name__ == "__main__":

scripts/unit_test_shards_parser.py

@@ -0,0 +1,27 @@
+import sys
+import os
+import argparse
+import json
+
+
+def get_test_paths_for_shard(shard_name):
+    unit_tests_json = f'{os.getcwd()}/.github/workflows/unit-test-shards.json'
+    with open(unit_tests_json) as file:
+        unit_test_workflow_shards = json.loads(file.read())
+
+    if shard_name not in unit_test_workflow_shards:
+        sys.stdout.write("Error, invalid shard name provided. please provide a valid shard name as specified in unit-test-shards.json")
+
+    return unit_test_workflow_shards.get(shard_name)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--shard-name", action="store", default="")
+    argument = parser.parse_args()
+
+    if not argument.shard_name:
+        sys.stdout.write("Error, no shard name provided. please provide a valid shard name as specified in unit-test-shards.json")
+
+    unit_test_paths = get_test_paths_for_shard(argument.shard_name)
+    sys.stdout.write(unit_test_paths)