Simplify and enhance discover_targeted_packages (#42740)

* simplfy discover_targeted_packages and add a test for it
* we need this to work by default for unfiltered input, instead of forcing default of **

Author: scbedd

eng/tools/azure-sdk-tools/ci_tools/functions.py

@@ -163,20 +163,26 @@ def glob_packages(glob_string: str, target_root_dir: str) -> List[str]:
     collected_top_level_directories = []
 
     for glob_string in individual_globs:
-        globbed = glob.glob(os.path.join(target_root_dir, glob_string, "setup.py")) + glob.glob(
+        globbed = glob.glob(os.path.join(target_root_dir, glob_string, "setup.py"), recursive=True) + glob.glob(
             os.path.join(target_root_dir, "sdk/*/", glob_string, "setup.py")
         )
         collected_top_level_directories.extend([os.path.dirname(p) for p in globbed])
 
     # handle pyproject.toml separately, as we need to filter them by the presence of a `[project]` section
     for glob_string in individual_globs:
-        globbed = glob.glob(os.path.join(target_root_dir, glob_string, "pyproject.toml")) + glob.glob(
+        globbed = glob.glob(os.path.join(target_root_dir, glob_string, "pyproject.toml"), recursive=True) + glob.glob(
             os.path.join(target_root_dir, "sdk/*/", glob_string, "pyproject.toml")
         )
         for p in globbed:
             if get_pyproject(os.path.dirname(p)):
                 collected_top_level_directories.append(os.path.dirname(p))
 
+    # drop any packages that exist within a tests or test directory
+    collected_top_level_directories = [
+        p for p in collected_top_level_directories
+        if not any(part in ("test", "tests") for part in p.split(os.sep))
+    ]
+
     # deduplicate, in case we have double coverage from the glob strings. Example: "azure-mgmt-keyvault,azure-mgmt-*"
     return list(set(collected_top_level_directories))
 

eng/tools/azure-sdk-tools/tests/integration/test_package_discovery.py

@@ -5,8 +5,9 @@
 
 
 repo_root = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..", "..", "..", ".."))
-core_service_root = os.path.join(repo_root, "sdk", "core")
-storage_service_root = os.path.join(repo_root, "sdk", "storage")
+sdk_root = os.path.join(repo_root, "sdk")
+core_service_root = os.path.join(sdk_root, "core")
+storage_service_root = os.path.join(sdk_root, "storage")
 
 
 def test_discovery():
@@ -19,6 +20,20 @@ def test_discovery():
     assert len(non_empty_results) == 1
 
 
+def test_discovery_against_sdk():
+    package_directories = discover_targeted_packages("**", sdk_root)
+
+    # ensure we didn't accidentally pick up a couple known packages from within a tests directory
+
+    known_mgmt_test_setup = os.path.join(sdk_root, "netapp", "azure-mgmt-netapp", "tests", "setup.py")
+    known_test_core_setup = os.path.join(repo_root, "sdk", "core", "azure-core", "tests", "testserver_tests", "coretestserver", "setup.py")
+
+    assert known_test_core_setup not in package_directories
+    assert known_mgmt_test_setup not in package_directories
+
+    # this is effectively checking to ensure we don't crash on any of the packages
+    assert len(package_directories) > 0
+
 def test_discovery_omit_mgmt():
     results = discover_targeted_packages("azure*", storage_service_root, filter_type="Omit_management")