fix: add type ignores everywhere

pyproject.toml

@@ -56,6 +56,7 @@ faiss-cpu = ">=1.7.3"
 torch = ">=1.11.0,!=1.13.0,<2"
 spacy = "3.7.2"
 en-core-web-sm = { url = "https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-3.7.1/en_core_web_sm-3.7.1-py3-none-any.whl" }
+transformers = ">=4.35.2"
 
 [tool.poetry.group.dev.dependencies]
 wemake-python-styleguide = ">=0.16.1"

src/emma_datasets/augmentations/simbot_augmentators/action_creators.py

@@ -93,7 +93,7 @@ def _create_synthetic_instruction(
                 "attributes": object_attributes.dict(),
             }
         else:
-            object_id = augmentation_instruction.object_id
+            object_id = augmentation_instruction.object_id  # type: ignore[assignment]
             object_attributes = augmentation_instruction.attributes
 
         inventory_object_id = None
@@ -157,12 +157,12 @@ def __init__(self, object_synonyms: dict[str, list[str]]) -> None:
     def __call__(self, augmentation_instruction: AugmentationInstruction) -> dict[str, Any]:
         """Create the search instruction dictionary."""
         instruction_dict = super().__call__(augmentation_instruction=augmentation_instruction)
-        instruction_dict["positive"] = augmentation_instruction.augmentation_metadata["positive"]
+        instruction_dict["positive"] = augmentation_instruction.augmentation_metadata["positive"]  # type: ignore[index]
         return instruction_dict
 
     def _create_mission_id(self, augmentation_instruction: AugmentationInstruction) -> str:
         image_name = self._flat_image_name(augmentation_instruction.image_name)
-        positive = augmentation_instruction.augmentation_metadata["positive"]
+        positive = augmentation_instruction.augmentation_metadata["positive"]  # type: ignore[index]
         return f"{self.action_type}_ispositive{positive}_{augmentation_instruction.annotation_id}_{image_name}"
 
 

src/emma_datasets/augmentations/simbot_augmentators/base_augmentator.py

@@ -201,7 +201,7 @@ def _merge_instructions(
                 action_instructions.extend(instructions)
             else:
                 action_instructions.extend(object_instructions)
-        return action_instructions
+        return [dict(instruction) for instruction in action_instructions]
 
     def _compute_distance_to_object(
         self, object_annotation: dict[str, Any], robot_position: NDArray[np.float32]
@@ -223,7 +223,7 @@ def _get_instructions_from_attributes(
     ) -> tuple[list[AugmentationInstruction], int]:
         instructions = []
         bbox_centers = [
-            compute_bbox_center_coords(instruction.bbox) for instruction in instruction_list
+            compute_bbox_center_coords(instruction.bbox) for instruction in instruction_list  # type: ignore[arg-type]
         ]
 
         left2right = np.argsort([bbox_center[0] for bbox_center in bbox_centers])

src/emma_datasets/augmentations/simbot_augmentators/break_augmentator.py

@@ -82,7 +82,7 @@ def __call__(
                     attributes=SimBotObjectAttributes(
                         readable_name=readable_name,
                         color=self._get_color(readable_name),
-                        distance=distance_to_object,
+                        distance=distance_to_object,  # type: ignore[arg-type]
                     ),
                     bbox=self._get_bbox(image_annotation),
                     image_name=image_name,
@@ -93,7 +93,7 @@ def __call__(
                 break_instructions_dict[object_class].append(instruction)
 
         break_instructions = self._merge_instructions(break_instructions_dict, annotation_id)
-        return break_instructions
+        return break_instructions  # type: ignore[return-value]
 
     @classmethod
     def from_yaml_config(  # type: ignore[override]

src/emma_datasets/augmentations/simbot_augmentators/clean_augmentator.py

@@ -89,7 +89,7 @@ def __call__(
                     attributes=SimBotObjectAttributes(
                         readable_name=cleaning_readable_name,
                         color=self._get_color(readable_name),
-                        distance=distance_to_object,
+                        distance=distance_to_object,  # type: ignore[arg-type]
                     ),
                     bbox=self._get_bbox(image_annotation),
                     image_name=image_name,
@@ -101,7 +101,7 @@ def __call__(
                 clean_instructions_dict[object_class].append(instruction)
 
         clean_instructions = self._merge_instructions(clean_instructions_dict, annotation_id)
-        return clean_instructions
+        return clean_instructions  # type: ignore[return-value]
 
     @classmethod
     def from_yaml_config(  # type: ignore[override]

src/emma_datasets/augmentations/simbot_augmentators/fill_augmentator.py

@@ -104,7 +104,7 @@ def __call__(  # noqa: WPS231
                     attributes=SimBotObjectAttributes(
                         readable_name=readable_name,
                         color=self._get_color(readable_name),
-                        distance=distance_to_object,
+                        distance=distance_to_object,  # type: ignore[arg-type]
                     ),
                     bbox=self._get_bbox(image_annotation),
                     image_name=image_name,
@@ -116,7 +116,7 @@ def __call__(  # noqa: WPS231
                 clean_instructions_dict[object_class].append(instruction)
 
         clean_instructions = self._merge_instructions(clean_instructions_dict, annotation_id)
-        return clean_instructions
+        return clean_instructions  # type: ignore[return-value]
 
     @classmethod
     def from_yaml_config(  # type: ignore[override]

src/emma_datasets/augmentations/simbot_augmentators/goto_augmentator.py

@@ -76,7 +76,7 @@ def __call__(
                     attributes=SimBotObjectAttributes(
                         readable_name=readable_name,
                         color=self._get_color(readable_name),
-                        distance=distance_to_object,
+                        distance=distance_to_object,  # type: ignore[arg-type]
                     ),
                     bbox=self._get_bbox(image_annotation),
                     image_name=image_name,
@@ -89,7 +89,7 @@ def __call__(
         navigation_instructions = self._merge_instructions(
             navigation_instructions_dict, annotation_id
         )
-        return navigation_instructions
+        return navigation_instructions  # type: ignore[return-value]
 
     @classmethod
     def from_yaml_config(  # type: ignore[override]

src/emma_datasets/augmentations/simbot_augmentators/open_close_augmentator.py

@@ -78,8 +78,8 @@ def __call__(
                     object_id=object_type,
                     attributes=SimBotObjectAttributes(
                         readable_name=readable_name,
-                        color=self._object_color_map.get(object_class, None),
-                        distance=distance_to_object,
+                        color=self._object_color_map.get(object_class, None),  # type: ignore[arg-type]
+                        distance=distance_to_object,  # type: ignore[arg-type]
                     ),
                     bbox=self._get_bbox(image_annotation),
                     image_name=image_name,
@@ -90,7 +90,7 @@ def __call__(
                 open_instructions_dict[object_class].append(instruction)
 
         open_instructions = self._merge_instructions(open_instructions_dict, annotation_id)
-        return open_instructions
+        return open_instructions  # type: ignore[return-value]
 
     @classmethod
     def from_yaml_config(  # type: ignore[override]

src/emma_datasets/augmentations/simbot_augmentators/pickup_augmentator.py

@@ -77,7 +77,7 @@ def __call__(
                     attributes=SimBotObjectAttributes(
                         readable_name=readable_name,
                         color=self._get_color(readable_name),
-                        distance=distance_to_object,
+                        distance=distance_to_object,  # type: ignore[arg-type]
                     ),
                     bbox=self._get_bbox(image_annotation),
                     image_name=image_name,
@@ -88,7 +88,7 @@ def __call__(
                 pickup_instructions_dict[object_class].append(instruction)
 
         pickup_instructions = self._merge_instructions(pickup_instructions_dict, annotation_id)
-        return pickup_instructions
+        return pickup_instructions  # type: ignore[return-value]
 
     @classmethod
     def from_yaml_config(  # type: ignore[override]

src/emma_datasets/augmentations/simbot_augmentators/place_augmentator.py

@@ -89,7 +89,7 @@ def __call__(
                     attributes=SimBotObjectAttributes(
                         readable_name=readable_name,
                         color=self._get_color(readable_name),
-                        distance=distance_to_object,
+                        distance=distance_to_object,  # type: ignore[arg-type]
                     ),
                     bbox=self._get_bbox(image_annotation),
                     image_name=image_name,
@@ -100,7 +100,7 @@ def __call__(
                 place_instructions_dict[object_class].append(instruction)
 
         place_instructions = self._merge_instructions(place_instructions_dict, annotation_id)
-        return place_instructions
+        return place_instructions  # type: ignore[return-value]
 
     @classmethod
     def from_yaml_config(  # type: ignore[override]

src/emma_datasets/augmentations/simbot_augmentators/scan_augmentator.py

@@ -77,7 +77,7 @@ def __call__(
                     attributes=SimBotObjectAttributes(
                         readable_name=readable_name,
                         color=self._get_color(readable_name),
-                        distance=distance_to_object,
+                        distance=distance_to_object,  # type: ignore[arg-type]
                     ),
                     bbox=self._get_bbox(image_annotation),
                     image_name=image_name,
@@ -88,7 +88,7 @@ def __call__(
                 scan_instructions_dict[object_class].append(instruction)
 
         scan_instructions = self._merge_instructions(scan_instructions_dict, annotation_id)
-        return scan_instructions
+        return scan_instructions  # type: ignore[return-value]
 
     @classmethod
     def from_yaml_config(  # type: ignore[override]

src/emma_datasets/augmentations/simbot_augmentators/search_augmentator.py

@@ -99,7 +99,7 @@ def __call__(  # noqa: WPS231
                 SimBotObjectAttributes(
                     readable_name=readable_name,
                     color=self._get_color(readable_name),
-                    distance=distance_to_object,
+                    distance=distance_to_object,  # type: ignore[arg-type]
                 )
             )
 

src/emma_datasets/augmentations/simbot_augmentators/toggle_augmentator.py

@@ -77,7 +77,7 @@ def __call__(
                     attributes=SimBotObjectAttributes(
                         readable_name=readable_name,
                         color=self._get_color(readable_name),
-                        distance=distance_to_object,
+                        distance=distance_to_object,  # type: ignore[arg-type]
                     ),
                     bbox=self._get_bbox(image_annotation),
                     image_name=image_name,
@@ -88,7 +88,7 @@ def __call__(
                 toggle_instructions_dict[object_class].append(instruction)
 
         toggle_instructions = self._merge_instructions(toggle_instructions_dict, annotation_id)
-        return toggle_instructions
+        return toggle_instructions  # type: ignore[return-value]
 
     @classmethod
     def from_yaml_config(  # type: ignore[override]

src/emma_datasets/commands/create_simbot_augmentations_from_vision_data.py

@@ -136,7 +136,7 @@ def gather(self) -> None:
             for action_type, annotations_per_action_type in metadata_per_action_type.items():
                 final_metadata.update(
                     self.vision_data_augmentations[action_type].post_process_metadata(
-                        annotations_per_action_type, self._class_thresholds
+                        annotations_per_action_type, self._class_thresholds  # type: ignore[arg-type]
                     )
                 )
                 progress.advance(task_id)
@@ -202,7 +202,7 @@ def __iter__(self) -> Iterator[dict[Any, Any]]:
                         annotations=annotations,
                         robot_position=robot_position,
                         image_name=image_name,
-                        class_thresholds=self._class_thresholds,
+                        class_thresholds=self._class_thresholds,  # type: ignore[arg-type]
                         room_name=room_name,
                     )
                 )

src/emma_datasets/commands/create_simbot_trajectories_from_dynamo_db.py

@@ -655,7 +655,7 @@ def process_highl_level_key(self, session_id: str) -> HighLevelKey:
             high_level_key = self._high_level_key_processor(session_id.split("/")[1])
         except Exception:
             logger.error(f"Could not convert the session id {session_id} to a high level key")
-            return None
+            return None  # type: ignore[return-value]
 
         return high_level_key
 

src/emma_datasets/datamodels/datasets/utils/simbot_utils/ambiguous_data.py

@@ -165,7 +165,7 @@ def _filter_bboxes_based_on_area(
         if thresholds is None:
             threshold = self._default_min_area
         else:
-            threshold = min(thresholds[0] * 5, self._default_min_area)
+            threshold = min(thresholds[0] * 5, self._default_min_area)  # type: ignore[assignment]
         for bbox in candidate_bboxes:
             if compute_bbox_area(bbox) > threshold:
                 filtered_bboxes.append(bbox)
@@ -413,7 +413,7 @@ def __call__(
         """Filter the vision augmentation data."""
         action = instruction_instance.actions[0]
         if action.type == "Search":
-            search_object_metadata = action.search.get("selected_object", None)
+            search_object_metadata = action.search.get("selected_object", None)  # type: ignore[attr-defined]
             if search_object_metadata is None:
                 return False
             target_object_id = search_object_metadata["id"]

src/emma_datasets/datamodels/datasets/utils/simbot_utils/high_level_key_processor.py

@@ -238,7 +238,7 @@ def __call__(self, highlevel_key: str) -> HighLevelKey:
             if secondary_key == "YesterdayMachine_01" and decoded_key.target_object == "Carrot_01":
                 secondary_key = "YesterdayMachine_01_from_Carrot"
 
-            template_metadata = template_metadata[secondary_key]
+            template_metadata = template_metadata[secondary_key]  # type: ignore[index]
 
         for decoded_key_field in self.decoded_key_fields:
             decoded_key_value = getattr(decoded_key, decoded_key_field)
@@ -247,11 +247,11 @@ def __call__(self, highlevel_key: str) -> HighLevelKey:
                 and decoded_key_value is not None
             )
             if should_get_object_synonym:
-                template_metadata[decoded_key_field] = get_object_synonym(decoded_key_value)
+                template_metadata[decoded_key_field] = get_object_synonym(decoded_key_value)  # type: ignore[index]
             else:
-                template_metadata[decoded_key_field] = decoded_key_value
+                template_metadata[decoded_key_field] = decoded_key_value  # type: ignore[index]
 
-        formatted_paraphrases = self.get_paraphrases(template_metadata, decoded_key=decoded_key)
+        formatted_paraphrases = self.get_paraphrases(template_metadata, decoded_key=decoded_key)  # type: ignore[arg-type]
         return HighLevelKey(
             decoded_key=decoded_key,
             paraphrases=formatted_paraphrases,

src/emma_datasets/datamodels/datasets/winoground.py

@@ -21,10 +21,12 @@ class WinogroundInstance(BaseInstance):
     num_main_preds: int
     collapsed_tag: str
 
+    @property
     def modality(self) -> MediaType:
         """Returns the data modality for Winoground."""
         return MediaType.image
 
+    @property
     def features_path(self) -> Union[Path, list[Path]]:
         """Returns the features path for Winoground images."""
         return Settings().paths.winoground_features.joinpath(f"{self.id}.pt")