Fix incomplete unit tests for scan_directory

python/src/magika/magika.py

@@ -182,7 +182,11 @@ def scan_directory(
         collected_paths: List[Union[str, os.PathLike]] = []
 
         # Use rglob('*') for recursive scan, glob('*') for single directory
-        glob_pattern = path_obj.rglob("*") if recursive_scan else path_obj.glob("*")
+        glob_pattern = (
+            sorted(path_obj.rglob("*"))
+            if recursive_scan
+            else sorted(path_obj.glob("*"))
+        )
 
         for item in glob_pattern:
             # We only want files, not sub-directories themselves

python/tests/test_magika_python_module.py

@@ -74,10 +74,20 @@ def test_magika_module_with_explicit_model_dir() -> None:
 
 
 def test_magika_module_with_basic_tests_by_directory() -> None:
-    tests_paths = utils.get_directory_test_dir()
+    tests_paths = utils.get_directory_tests_files_dir()
 
     m = Magika()
-    _ = m.scan_directory(tests_paths)
+
+    # Only scan direct children of tests_data/directory.
+    # Expected output is "directory" content type.
+    results = m.scan_directory(tests_paths)
+    direct_children = sorted([p for p in tests_paths.glob("*")])
+    check_results_vs_expected_results(direct_children, results)
+
+    # Scan all files recursively. Expected output is content type of each file.
+    results = m.scan_directory(tests_paths, recursive_scan=True)
+    all_files = sorted([p for p in tests_paths.rglob("*") if p.is_file()])
+    check_results_vs_expected_results(all_files, results)
 
 
 def test_magika_module_with_basic_tests_by_paths() -> None:

python/tests/utils.py

@@ -42,8 +42,8 @@ def get_basic_tests_files_dir() -> Path:
     return tests_files_dir
 
 
-def get_directory_test_dir() -> Path:
-    tests_files_dir = get_tests_data_dir() / "scan_directory"
+def get_directory_tests_files_dir() -> Path:
+    tests_files_dir = get_tests_data_dir() / "directory"
     assert tests_files_dir.is_dir()
     return tests_files_dir
 

tests_data/directory/txt/complex-sentence.txt

@@ -0,0 +1 @@
+This is yet another simple test, it includes one simple sentence, but it is not as trivial as other simpler tests.