Fix ruff lint and format issues

Author: akshan-main

src/diffusers/modular_pipelines/hunyuan_video1_5/before_denoise.py

@@ -245,9 +245,11 @@ def description(self) -> str:
 
     @property
     def expected_components(self) -> list[ComponentSpec]:
+        from transformers import SiglipImageProcessor, SiglipVisionModel
+
         from ...models import AutoencoderKLHunyuanVideo15
         from ...pipelines.hunyuan_video1_5.image_processor import HunyuanVideo15ImageProcessor
-        from transformers import SiglipVisionModel, SiglipImageProcessor
+
         return [
             ComponentSpec("vae", AutoencoderKLHunyuanVideo15),
             ComponentSpec(

src/diffusers/modular_pipelines/hunyuan_video1_5/decoders.py

@@ -75,7 +75,9 @@ def __call__(self, components, state: PipelineState) -> PipelineState:
         else:
             latents = block_state.latents.to(components.vae.dtype) / components.vae.config.scaling_factor
             video = components.vae.decode(latents, return_dict=False)[0]
-            block_state.videos = components.video_processor.postprocess_video(video, output_type=block_state.output_type)
+            block_state.videos = components.video_processor.postprocess_video(
+                video, output_type=block_state.output_type
+            )
 
         self.set_block_state(state, block_state)
         return components, state

src/diffusers/modular_pipelines/hunyuan_video1_5/denoise.py

@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from typing import Any
 
 import torch
 
@@ -115,7 +114,9 @@ def __call__(
 
         # Step 1: Collect model inputs
         guider_inputs = {
-            input_name: tuple(getattr(block_state, v) for v in value) if isinstance(value, tuple) else getattr(block_state, value)
+            input_name: tuple(getattr(block_state, v) for v in value)
+            if isinstance(value, tuple)
+            else getattr(block_state, value)
             for input_name, value in self._guider_input_fields.items()
         }
 
@@ -129,9 +130,7 @@ def __call__(
         for guider_state_batch in guider_state:
             components.guider.prepare_models(components.transformer)
 
-            cond_kwargs = {
-                input_name: getattr(guider_state_batch, input_name) for input_name in guider_inputs.keys()
-            }
+            cond_kwargs = {input_name: getattr(guider_state_batch, input_name) for input_name in guider_inputs.keys()}
 
             context_name = getattr(guider_state_batch, components.guider._identifier_key)
             with components.transformer.cache_context(context_name):
@@ -306,7 +305,9 @@ def __call__(
             timestep_r = None
 
         guider_inputs = {
-            input_name: tuple(getattr(block_state, v) for v in value) if isinstance(value, tuple) else getattr(block_state, value)
+            input_name: tuple(getattr(block_state, v) for v in value)
+            if isinstance(value, tuple)
+            else getattr(block_state, value)
             for input_name, value in self._guider_input_fields.items()
         }
 
@@ -316,9 +317,7 @@ def __call__(
         for guider_state_batch in guider_state:
             components.guider.prepare_models(components.transformer)
 
-            cond_kwargs = {
-                input_name: getattr(guider_state_batch, input_name) for input_name in guider_inputs.keys()
-            }
+            cond_kwargs = {input_name: getattr(guider_state_batch, input_name) for input_name in guider_inputs.keys()}
 
             context_name = getattr(guider_state_batch, components.guider._identifier_key)
             with components.transformer.cache_context(context_name):

src/diffusers/modular_pipelines/hunyuan_video1_5/encoders.py

@@ -226,12 +226,20 @@ def encode_prompt(
             )
 
         _, seq_len, _ = prompt_embeds.shape
-        prompt_embeds = prompt_embeds.repeat(1, num_videos_per_prompt, 1).view(batch_size * num_videos_per_prompt, seq_len, -1)
-        prompt_embeds_mask = prompt_embeds_mask.repeat(1, num_videos_per_prompt, 1).view(batch_size * num_videos_per_prompt, seq_len)
+        prompt_embeds = prompt_embeds.repeat(1, num_videos_per_prompt, 1).view(
+            batch_size * num_videos_per_prompt, seq_len, -1
+        )
+        prompt_embeds_mask = prompt_embeds_mask.repeat(1, num_videos_per_prompt, 1).view(
+            batch_size * num_videos_per_prompt, seq_len
+        )
 
         _, seq_len_2, _ = prompt_embeds_2.shape
-        prompt_embeds_2 = prompt_embeds_2.repeat(1, num_videos_per_prompt, 1).view(batch_size * num_videos_per_prompt, seq_len_2, -1)
-        prompt_embeds_mask_2 = prompt_embeds_mask_2.repeat(1, num_videos_per_prompt, 1).view(batch_size * num_videos_per_prompt, seq_len_2)
+        prompt_embeds_2 = prompt_embeds_2.repeat(1, num_videos_per_prompt, 1).view(
+            batch_size * num_videos_per_prompt, seq_len_2, -1
+        )
+        prompt_embeds_mask_2 = prompt_embeds_mask_2.repeat(1, num_videos_per_prompt, 1).view(
+            batch_size * num_videos_per_prompt, seq_len_2
+        )
 
         prompt_embeds = prompt_embeds.to(dtype=dtype, device=device)
         prompt_embeds_mask = prompt_embeds_mask.to(dtype=dtype, device=device)