fix ! tuple/list (#33)

src/hope_dedup_engine/apps/faces/celery_tasks.py

@@ -15,7 +15,7 @@ def deduplicate(
     self: Task,
     filenames: tuple[str],
     ignore_pairs: tuple[tuple[str, str], ...] = tuple(),
-) -> tuple[tuple[str, ...], ...]:
+) -> list[list[str]]:
     """
     Deduplicate a set of filenames, ignoring any specified pairs of filenames.
 
@@ -25,7 +25,7 @@ def deduplicate(
                                         a pair of filenames to be ignored in the duplication check.
 
     Returns:
-        tuple[tuple[str]]: A tuple of tuples, where each inner tuple represents a group of duplicates.
+        list[list[str]]: A list of lists, where each inner list represents a group of duplicates.
     """
     try:
         dd = DuplicationDetector(filenames, ignore_pairs)

src/hope_dedup_engine/apps/faces/services/duplication_detector.py

@@ -81,12 +81,12 @@ def _load_encodings_all(self) -> dict[str, list[np.ndarray[np.float32, Any]]]:
             raise e
         return data
 
-    def find_duplicates(self) -> tuple[tuple[str, ...], ...]:
+    def find_duplicates(self) -> list[list[str]]:
         """
         Find and return a list of duplicate images based on face encodings.
 
         Returns:
-            tuple[tuple[str, ...], ...]: A tuple of tuples, where each inner tuple contains
+            list[list[str]]: A list of lists, where each inner list contains
                                         the filenames of duplicate images.
         """
         try:

src/hope_dedup_engine/apps/faces/utils/duplicate_groups_builder.py

@@ -5,15 +5,15 @@
 
 class DuplicateGroupsBuilder:
     @staticmethod
-    def build(checked: set[tuple[str, str, float]]) -> tuple[tuple[str, ...], ...]:
+    def build(checked: set[tuple[str, str, float]]) -> list[list[str]]:
         """
         Transform a set of tuples with distances into a tuple of grouped duplicate paths.
 
         Args:
             checked (set[tuple[str, str, float]]): A set of tuples containing the paths and their distances.
 
         Returns:
-            tuple[tuple[str, ...], ...]: A tuple of grouped duplicate paths.
+            list[list[str]]: A list of grouped duplicate paths.
         """
         # Dictionary to store connections between paths where distances are less than the threshold
         groups = []
@@ -41,4 +41,4 @@ def build(checked: set[tuple[str, str, float]]) -> tuple[tuple[str, ...], ...]:
                         )
                 # Add the newly formed group to the list of groups
                 groups.append(new_group)
-        return tuple(map(tuple, groups))
+        return list(map(list, groups))

src/hope_dedup_engine/apps/faces/validators.py

@@ -7,24 +7,23 @@ def validate(ignore: tuple[tuple[str, str], ...]) -> set[tuple[str, str]]:
         if not ignore:
             return set()
         if not (
-            isinstance(ignore, tuple)
+            isinstance(ignore, list)
             and all(
                 all(
                     (
-                        isinstance(pair, tuple),
+                        isinstance(pair, list),
                         len(pair) == 2,
                         all(isinstance(item, str) and item for item in pair),
                     )
                 )
                 for pair in ignore
             )
         ):
-            raise ValidationError(
-                "Invalid format. Expected a tuple of tuples, each containing exactly two strings."
-            )
+            raise ValidationError("Invalid format for ignore pairs.")
 
         result_set = set()
         for pair in ignore:
+            pair = tuple(pair)
             result_set.add(pair)
             result_set.add((pair[1], pair[0]))
         return result_set

tests/faces/faces_const.py

@@ -4,10 +4,10 @@
 FILENAME_ENCODED: Final[str] = "test_file.jpg.npy"
 FILENAME_ENCODED_FORMAT: Final[str] = "{}.npy"
 FILENAMES: Final[list[str]] = ["test_file.jpg", "test_file2.jpg", "test_file3.jpg"]
-IGNORE_PAIRS: Final[tuple[tuple[str, str]]] = (
-    ("ignore_file.jpg", "ignore_file2.jpg"),
-    ("ignore_file4.jpg", "ignore_file3.jpg"),
-)
+IGNORE_PAIRS: Final[list[list[str, str]]] = [
+    ["ignore_file.jpg", "ignore_file2.jpg"],
+    ["ignore_file4.jpg", "ignore_file3.jpg"],
+]
 
 CELERY_TASK_NAME: Final[str] = "Deduplicate"
 CELERY_TASK_TTL: Final[int] = 1 * 60 * 60

tests/faces/test_duplication_detector.py

@@ -42,17 +42,19 @@ def test_init_successful(mock_dd):
 @pytest.mark.parametrize(
     "ignore_input, expected_output",
     [
-        (tuple(), set()),
+        (list(), set()),
         (
-            (("file1.jpg", "file2.jpg"),),
+            [
+                ["file1.jpg", "file2.jpg"],
+            ],
             {("file1.jpg", "file2.jpg"), ("file2.jpg", "file1.jpg")},
         ),
         (
-            (("file1.jpg", "file2.jpg"), ("file2.jpg", "file1.jpg")),
+            [["file1.jpg", "file2.jpg"], ["file2.jpg", "file1.jpg"]],
             {("file1.jpg", "file2.jpg"), ("file2.jpg", "file1.jpg")},
         ),
         (
-            (("file1.jpg", "file3.jpg"), ("file2.jpg", "file3.jpg")),
+            [["file1.jpg", "file3.jpg"], ["file2.jpg", "file3.jpg"]],
             {
                 ("file1.jpg", "file3.jpg"),
                 ("file3.jpg", "file1.jpg"),