Revamp execution provider overrides/adjustments (#1206)

* Split provider hooks into override/adjust with cached CoreML base

Replace the single resolve_inference_providers processor hook with two:
override_inference_providers (full replacement) and adjust_inference_providers
(merge options onto the base providers built by create_inference_providers).
This lets CoreML processors inherit ModelCacheDirectory + SpecializationStrategy
from the base while layering ModelFormat/MLComputeUnits on top.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HTQCZiYjJyUX11bDpbRSiB

* fix caching for execution provider by having override and adjust ways

* fix caching for execution provider by having override and adjust ways

* fix lint

* use proper pytest fixtures

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

facefusion/inference_manager.py

@@ -93,10 +93,23 @@ def resolve_static_inference_providers(module_name : str, execution_device_id :
 	module = importlib.import_module(module_name)
 	execution_providers = state_manager.get_item('execution_providers')
 
-	if hasattr(module, 'resolve_inference_providers'):
-		inference_providers = getattr(module, 'resolve_inference_providers')()
+	if hasattr(module, 'override_inference_providers'):
+		override_inference_providers = getattr(module, 'override_inference_providers')()
+
+		if override_inference_providers:
+			return override_inference_providers
+
+	if hasattr(module, 'adjust_inference_providers'):
+		adjust_inference_providers = getattr(module, 'adjust_inference_providers')()
+
+		if adjust_inference_providers:
+			inference_providers = create_inference_providers(execution_device_id, execution_providers)
+
+			for adjust_inference_provider in adjust_inference_providers:
+				for inference_provider in inference_providers:
+					if inference_provider[0] == adjust_inference_provider[0] and inference_provider[1]:
+						inference_provider[1].update(adjust_inference_provider[1])
 
-		if inference_providers:
 			return inference_providers
 
 	return create_inference_providers(execution_device_id, execution_providers)

facefusion/processors/modules/background_remover/core.py

@@ -479,7 +479,7 @@ def clear_inference_pool() -> None:
 	inference_manager.clear_inference_pool(__name__, model_names)
 
 
-def resolve_inference_providers() -> List[InferenceProvider]:
+def override_inference_providers() -> List[InferenceProvider]:
 	model_type = get_model_options().get('type')
 
 	if is_macos() and has_execution_provider('coreml') or is_windows() and has_execution_provider('directml') and model_type == 'corridor_key':

facefusion/processors/modules/face_swapper/core.py

@@ -502,7 +502,7 @@ def clear_inference_pool() -> None:
 	inference_manager.clear_inference_pool(__name__, model_names)
 
 
-def resolve_inference_providers() -> List[InferenceProvider]:
+def adjust_inference_providers() -> List[InferenceProvider]:
 	model_precision = get_model_options().get('precision')
 	model_type = get_model_options().get('type')
 
@@ -512,8 +512,7 @@ def resolve_inference_providers() -> List[InferenceProvider]:
 			[
 				(facefusion.choices.execution_provider_set.get('coreml'),
 				{
-					'ModelFormat': 'MLProgram',
-					'SpecializationStrategy': 'FastPrediction'
+					'ModelFormat': 'MLProgram'
 				})
 			]
 

facefusion/processors/modules/frame_colorizer/core.py

@@ -172,7 +172,7 @@ def clear_inference_pool() -> None:
 	inference_manager.clear_inference_pool(__name__, model_names)
 
 
-def resolve_inference_providers() -> List[InferenceProvider]:
+def override_inference_providers() -> List[InferenceProvider]:
 	if is_macos() and has_execution_provider('coreml'):
 		return [ facefusion.choices.execution_provider_set.get('cpu') ]
 

facefusion/processors/modules/frame_enhancer/core.py

@@ -558,7 +558,7 @@ def clear_inference_pool() -> None:
 	inference_manager.clear_inference_pool(__name__, model_names)
 
 
-def resolve_inference_providers() -> List[InferenceProvider]:
+def adjust_inference_providers() -> List[InferenceProvider]:
 	model_precision = get_model_options().get('precision')
 
 	if is_macos() and has_execution_provider('coreml') and model_precision == 'fp16':
@@ -566,8 +566,7 @@ def resolve_inference_providers() -> List[InferenceProvider]:
 		[
 			(facefusion.choices.execution_provider_set.get('coreml'),
 			{
-				'ModelFormat': 'MLProgram',
-				'SpecializationStrategy': 'FastPrediction'
+				'ModelFormat': 'MLProgram'
 			})
 		]
 

tests/test_inference_manager.py

@@ -1,10 +1,12 @@
-from unittest.mock import patch
+from types import SimpleNamespace
+from unittest.mock import Mock, patch
 
 import pytest
 from onnxruntime import InferenceSession
 
 from facefusion import content_analyser, state_manager
-from facefusion.inference_manager import INFERENCE_POOL_SET, get_inference_pool
+from facefusion.execution import resolve_cache_path
+from facefusion.inference_manager import INFERENCE_POOL_SET, get_inference_pool, resolve_static_inference_providers
 
 
 @pytest.fixture(scope = 'module', autouse = True)
@@ -12,7 +14,6 @@ def before_all() -> None:
 	state_manager.init_item('execution_device_ids', [ 0 ])
 	state_manager.init_item('execution_providers', [ 'cpu' ])
 	state_manager.init_item('download_providers', [ 'github' ])
-	content_analyser.pre_check()
 
 
 def test_get_inference_pool() -> None:
@@ -30,3 +31,26 @@ def test_get_inference_pool() -> None:
 		assert isinstance(INFERENCE_POOL_SET.get('cli').get('facefusion.content_analyser.nsfw_1.nsfw_2.nsfw_3.0.cpu').get('nsfw_1'), InferenceSession)
 
 	assert INFERENCE_POOL_SET.get('cli').get('facefusion.content_analyser.nsfw_1.nsfw_2.nsfw_3.0.cpu').get('nsfw_1') == INFERENCE_POOL_SET.get('ui').get('facefusion.content_analyser.nsfw_1.nsfw_2.nsfw_3.0.cpu').get('nsfw_1')
+
+
+@pytest.fixture
+def override_module() -> SimpleNamespace:
+	return SimpleNamespace(override_inference_providers = Mock(return_value = [ ('CoreMLExecutionProvider', { 'ModelFormat': 'MLProgram' }) ]))
+
+
+@pytest.fixture
+def adjust_module() -> SimpleNamespace:
+	return SimpleNamespace(adjust_inference_providers = Mock(return_value = [ ('CoreMLExecutionProvider', { 'ModelFormat': 'MLProgram' }) ]))
+
+
+def test_resolve_static_inference_providers(override_module : SimpleNamespace, adjust_module : SimpleNamespace) -> None:
+	state_manager.init_item('execution_providers', ['coreml'])
+	resolve_static_inference_providers.cache_clear()
+
+	with patch('facefusion.inference_manager.importlib', Mock(import_module = Mock(return_value = override_module))):
+		assert resolve_static_inference_providers('override_module', 0) == [ ('CoreMLExecutionProvider', { 'ModelFormat': 'MLProgram' }) ]
+
+	with patch('facefusion.inference_manager.importlib', Mock(import_module = Mock(return_value = adjust_module))):
+		assert resolve_static_inference_providers('adjust_module', 0) == [ ('CoreMLExecutionProvider', { 'SpecializationStrategy': 'FastPrediction', 'ModelCacheDirectory': resolve_cache_path(), 'ModelFormat': 'MLProgram' }) ]
+
+	assert resolve_static_inference_providers('test', 0) == [ ('CoreMLExecutionProvider', { 'SpecializationStrategy': 'FastPrediction', 'ModelCacheDirectory': resolve_cache_path() }) ]