Introduce DeprecatedPipelineMixin to simplify pipeline deprecation process (#11596)

* update

* update

* update

* update

* update

* update

* update

src/diffusers/pipelines/stable_diffusion_ldm3d/pipeline_stable_diffusion_ldm3d.py

@@ -37,7 +37,7 @@ from ...utils import (
     unscale_lora_layers,
 )
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin
 from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
 
 
@@ -178,6 +178,7 @@ class LDM3DPipelineOutput(BaseOutput):
 
 
 class StableDiffusionLDM3DPipeline(
+    DeprecatedPipelineMixin,
     DiffusionPipeline,
     StableDiffusionMixin,
     TextualInversionLoaderMixin,
@@ -185,6 +186,8 @@ class StableDiffusionLDM3DPipeline(
     StableDiffusionLoraLoaderMixin,
     FromSingleFileMixin,
 ):
+    _last_supported_version = "0.33.1"
+
     r"""
     Pipeline for text-to-image and 3D generation using LDM3D.
 

src/diffusers/pipelines/stable_diffusion_panorama/pipeline_stable_diffusion_panorama.py

@@ -33,7 +33,7 @@ from ...utils import (
     unscale_lora_layers,
 )
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin
 from ..stable_diffusion import StableDiffusionPipelineOutput
 from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
 
@@ -156,12 +156,15 @@ def retrieve_timesteps(
 
 
 class StableDiffusionPanoramaPipeline(
+    DeprecatedPipelineMixin,
     DiffusionPipeline,
     StableDiffusionMixin,
     TextualInversionLoaderMixin,
     StableDiffusionLoraLoaderMixin,
     IPAdapterMixin,
 ):
+    _last_supported_version = "0.33.1"
+
     r"""
     Pipeline for text-to-image generation using MultiDiffusion.
 

src/diffusers/pipelines/stable_diffusion_safe/pipeline_stable_diffusion_safe.py

@@ -14,7 +14,7 @@ from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel
 from ...schedulers import KarrasDiffusionSchedulers
 from ...utils import deprecate, is_torch_xla_available, logging
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin
 from . import StableDiffusionSafePipelineOutput
 from .safety_checker import SafeStableDiffusionSafetyChecker
 
@@ -29,7 +29,9 @@ else:
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
-class StableDiffusionPipelineSafe(DiffusionPipeline, StableDiffusionMixin, IPAdapterMixin):
+class StableDiffusionPipelineSafe(DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin, IPAdapterMixin):
+    _last_supported_version = "0.33.1"
+
     r"""
     Pipeline based on the [`StableDiffusionPipeline`] for text-to-image generation using Safe Latent Diffusion.
 

src/diffusers/pipelines/stable_diffusion_sag/pipeline_stable_diffusion_sag.py

@@ -34,7 +34,7 @@ from ...utils import (
     unscale_lora_layers,
 )
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin
 from ..stable_diffusion import StableDiffusionPipelineOutput
 from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
 
@@ -107,7 +107,11 @@ class CrossAttnStoreProcessor:
 
 
 # Modified to get self-attention guidance scale in this paper (https://huggingface.co/papers/2210.00939) as an input
-class StableDiffusionSAGPipeline(DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, IPAdapterMixin):
+class StableDiffusionSAGPipeline(
+    DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, IPAdapterMixin
+):
+    _last_supported_version = "0.33.1"
+
     r"""
     Pipeline for text-to-image generation using Stable Diffusion.
 

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth.py

@@ -33,7 +33,7 @@ from ...utils import (
 )
 from ...utils.torch_utils import randn_tensor
 from ...video_processor import VideoProcessor
-from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin
 from . import TextToVideoSDPipelineOutput
 
 
@@ -68,8 +68,13 @@ EXAMPLE_DOC_STRING = """
 
 
 class TextToVideoSDPipeline(
-    DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, StableDiffusionLoraLoaderMixin
+    DeprecatedPipelineMixin,
+    DiffusionPipeline,
+    StableDiffusionMixin,
+    TextualInversionLoaderMixin,
+    StableDiffusionLoraLoaderMixin,
 ):
+    _last_supported_version = "0.33.1"
     r"""
     Pipeline for text-to-video generation.
 

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth_img2img.py

@@ -34,7 +34,7 @@ from ...utils import (
 )
 from ...utils.torch_utils import randn_tensor
 from ...video_processor import VideoProcessor
-from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin
 from . import TextToVideoSDPipelineOutput
 
 
@@ -103,8 +103,13 @@ def retrieve_latents(
 
 
 class VideoToVideoSDPipeline(
-    DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, StableDiffusionLoraLoaderMixin
+    DeprecatedPipelineMixin,
+    DiffusionPipeline,
+    StableDiffusionMixin,
+    TextualInversionLoaderMixin,
+    StableDiffusionLoraLoaderMixin,
 ):
+    _last_supported_version = "0.33.1"
     r"""
     Pipeline for text-guided video-to-video generation.
 

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_zero.py

@@ -24,7 +24,7 @@ from ...utils import (
     unscale_lora_layers,
 )
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin
 from ..stable_diffusion import StableDiffusionSafetyChecker
 
 
@@ -296,12 +296,14 @@ def create_motion_field_and_warp_latents(motion_field_strength_x, motion_field_s
 
 
 class TextToVideoZeroPipeline(
+    DeprecatedPipelineMixin,
     DiffusionPipeline,
     StableDiffusionMixin,
     TextualInversionLoaderMixin,
     StableDiffusionLoraLoaderMixin,
     FromSingleFileMixin,
 ):
+    _last_supported_version = "0.33.1"
     r"""
     Pipeline for zero-shot text-to-video generation using Stable Diffusion.
 

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_zero_sdxl.py

@@ -35,7 +35,7 @@ from ...utils import (
     unscale_lora_layers,
 )
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin
 
 
 if is_invisible_watermark_available():
@@ -346,11 +346,13 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
 
 
 class TextToVideoZeroSDXLPipeline(
+    DeprecatedPipelineMixin,
     DiffusionPipeline,
     StableDiffusionMixin,
     StableDiffusionXLLoraLoaderMixin,
     TextualInversionLoaderMixin,
 ):
+    _last_supported_version = "0.33.1"
     r"""
     Pipeline for zero-shot text-to-video generation using Stable Diffusion XL.
 

src/diffusers/pipelines/unclip/pipeline_unclip.py

@@ -24,7 +24,7 @@ from ...models import PriorTransformer, UNet2DConditionModel, UNet2DModel
 from ...schedulers import UnCLIPScheduler
 from ...utils import is_torch_xla_available, logging
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, ImagePipelineOutput
 from .text_proj import UnCLIPTextProjModel
 
 
@@ -38,7 +38,7 @@ else:
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
-class UnCLIPPipeline(DiffusionPipeline):
+class UnCLIPPipeline(DeprecatedPipelineMixin, DiffusionPipeline):
     """
     Pipeline for text-to-image generation using unCLIP.
 
@@ -69,6 +69,7 @@ class UnCLIPPipeline(DiffusionPipeline):
 
     """
 
+    _last_supported_version = "0.33.1"
     _exclude_from_cpu_offload = ["prior"]
 
     prior: PriorTransformer

src/diffusers/pipelines/unclip/pipeline_unclip_image_variation.py

@@ -29,7 +29,7 @@ from ...models import UNet2DConditionModel, UNet2DModel
 from ...schedulers import UnCLIPScheduler
 from ...utils import is_torch_xla_available, logging
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, ImagePipelineOutput
 from .text_proj import UnCLIPTextProjModel
 
 
@@ -43,7 +43,7 @@ else:
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 
-class UnCLIPImageVariationPipeline(DiffusionPipeline):
+class UnCLIPImageVariationPipeline(DeprecatedPipelineMixin, DiffusionPipeline):
     """
     Pipeline to generate image variations from an input image using UnCLIP.
 
@@ -73,6 +73,7 @@ class UnCLIPImageVariationPipeline(DiffusionPipeline):
             Scheduler used in the super resolution denoising process (a modified [`DDPMScheduler`]).
     """
 
+    _last_supported_version = "0.33.1"
     decoder: UNet2DConditionModel
     text_proj: UnCLIPTextProjModel
     text_encoder: CLIPTextModelWithProjection

src/diffusers/pipelines/unidiffuser/pipeline_unidiffuser.py

@@ -28,7 +28,7 @@ from ...utils import (
 )
 from ...utils.outputs import BaseOutput
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline
 from .modeling_text_decoder import UniDiffuserTextDecoder
 from .modeling_uvit import UniDiffuserModel
 
@@ -62,7 +62,7 @@ class ImageTextPipelineOutput(BaseOutput):
     text: Optional[Union[List[str], List[List[str]]]]
 
 
-class UniDiffuserPipeline(DiffusionPipeline):
+class UniDiffuserPipeline(DeprecatedPipelineMixin, DiffusionPipeline):
     r"""
     Pipeline for a bimodal image-text model which supports unconditional text and image generation, text-conditioned
     image generation, image-conditioned text generation, and joint image-text generation.
@@ -96,6 +96,7 @@ class UniDiffuserPipeline(DiffusionPipeline):
             original UniDiffuser paper uses the [`DPMSolverMultistepScheduler`] scheduler.
     """
 
+    _last_supported_version = "0.33.1"
     # TODO: support for moving submodules for components with enable_model_cpu_offload
     model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae->text_decoder"
 

src/diffusers/pipelines/wuerstchen/pipeline_wuerstchen.py

@@ -21,7 +21,7 @@ from transformers import CLIPTextModel, CLIPTokenizer
 from ...schedulers import DDPMWuerstchenScheduler
 from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring
 from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, ImagePipelineOutput
 from .modeling_paella_vq_model import PaellaVQModel
 from .modeling_wuerstchen_diffnext import WuerstchenDiffNeXt
 
@@ -56,7 +56,7 @@ EXAMPLE_DOC_STRING = """
 """
 
 
-class WuerstchenDecoderPipeline(DiffusionPipeline):
+class WuerstchenDecoderPipeline(DeprecatedPipelineMixin, DiffusionPipeline):
     """
     Pipeline for generating images from the Wuerstchen model.
 

src/diffusers/pipelines/wuerstchen/pipeline_wuerstchen_combined.py

@@ -18,7 +18,7 @@ from transformers import CLIPTextModel, CLIPTokenizer
 
 from ...schedulers import DDPMWuerstchenScheduler
 from ...utils import deprecate, replace_example_docstring
-from ..pipeline_utils import DiffusionPipeline
+from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline
 from .modeling_paella_vq_model import PaellaVQModel
 from .modeling_wuerstchen_diffnext import WuerstchenDiffNeXt
 from .modeling_wuerstchen_prior import WuerstchenPrior
@@ -40,7 +40,7 @@ TEXT2IMAGE_EXAMPLE_DOC_STRING = """
 """
 
 
-class WuerstchenCombinedPipeline(DiffusionPipeline):
+class WuerstchenCombinedPipeline(DeprecatedPipelineMixin, DiffusionPipeline):
     """
     Combined Pipeline for text-to-image generation using Wuerstchen
 
@@ -68,6 +68,7 @@ class WuerstchenCombinedPipeline(DiffusionPipeline):
             The scheduler to be used for prior pipeline.
     """
 
+    _last_supported_version = "0.33.1"
     _load_connected_pipes = True
 
     def __init__(

tests/pipelines/test_pipelines_common.py

@@ -1115,6 +1115,14 @@ class PipelineTesterMixin:
         gc.collect()
         backend_empty_cache(torch_device)
 
+        # Skip tests for pipelines that inherit from DeprecatedPipelineMixin
+        from diffusers.pipelines.pipeline_utils import DeprecatedPipelineMixin
+
+        if hasattr(self, "pipeline_class") and issubclass(self.pipeline_class, DeprecatedPipelineMixin):
+            import pytest
+
+            pytest.skip(reason=f"Deprecated Pipeline: {self.pipeline_class.__name__}")
+
     def tearDown(self):
         # clean up the VRAM after each test in case of CUDA runtime errors
         super().tearDown()