Convert formatting to use ruff instead of yapf + isort (#26247)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-10-05 07:06:22 -07:00
committed by GitHub
parent 17edd8a807
commit d6953beb91
1508 changed files with 115205 additions and 94107 deletions
+65 -52
View File
@@ -28,8 +28,12 @@ from .video import VideoMediaIO
_M = TypeVar("_M")
if TYPE_CHECKING:
from .inputs import (BatchedTensorInputs, MultiModalKwargsItem,
MultiModalKwargsItems, MultiModalPlaceholderDict)
from .inputs import (
BatchedTensorInputs,
MultiModalKwargsItem,
MultiModalKwargsItems,
MultiModalPlaceholderDict,
)
else:
BatchedTensorInputs = Any
MultiModalKwargsItem = Any
@@ -37,12 +41,12 @@ else:
MultiModalPlaceholderDict = Any
global_thread_pool = ThreadPoolExecutor(
max_workers=envs.VLLM_MEDIA_LOADING_THREAD_COUNT)
max_workers=envs.VLLM_MEDIA_LOADING_THREAD_COUNT
)
atexit.register(global_thread_pool.shutdown)
class MediaConnector:
def __init__(
self,
media_io_kwargs: Optional[dict[str, dict[str, Any]]] = None,
@@ -53,9 +57,9 @@ class MediaConnector:
) -> None:
"""
Args:
media_io_kwargs: Additional args passed to process media
inputs, keyed by modalities. For example,
to set num_frames for video, set
media_io_kwargs: Additional args passed to process media
inputs, keyed by modalities. For example,
to set num_frames for video, set
`--media-io-kwargs '{"video":{"num_frames":40}}'`
connection: HTTP connection client to download media contents.
allowed_local_media_path: A local directory to load media files
@@ -63,8 +67,9 @@ class MediaConnector:
"""
super().__init__()
self.media_io_kwargs: dict[str, dict[
str, Any]] = media_io_kwargs if media_io_kwargs else {}
self.media_io_kwargs: dict[str, dict[str, Any]] = (
media_io_kwargs if media_io_kwargs else {}
)
self.connection = connection
if allowed_local_media_path:
@@ -73,11 +78,13 @@ class MediaConnector:
if not allowed_local_media_path_.exists():
raise ValueError(
"Invalid `--allowed-local-media-path`: The path "
f"{allowed_local_media_path_} does not exist.")
f"{allowed_local_media_path_} does not exist."
)
if not allowed_local_media_path_.is_dir():
raise ValueError(
"Invalid `--allowed-local-media-path`: The path "
f"{allowed_local_media_path_} must be a directory.")
f"{allowed_local_media_path_} must be a directory."
)
else:
allowed_local_media_path_ = None
@@ -107,24 +114,29 @@ class MediaConnector:
) -> _M: # type: ignore[type-var]
allowed_local_media_path = self.allowed_local_media_path
if allowed_local_media_path is None:
raise RuntimeError("Cannot load local files without "
"`--allowed-local-media-path`.")
raise RuntimeError(
"Cannot load local files without `--allowed-local-media-path`."
)
filepath = Path(url2pathname(url_spec.path))
if allowed_local_media_path not in filepath.resolve().parents:
raise ValueError(
f"The file path {filepath} must be a subpath "
f"of `--allowed-local-media-path` {allowed_local_media_path}.")
f"of `--allowed-local-media-path` {allowed_local_media_path}."
)
return media_io.load_file(filepath)
def _assert_url_in_allowed_media_domains(self, url_spec) -> None:
if self.allowed_media_domains and url_spec.hostname not in \
self.allowed_media_domains:
if (
self.allowed_media_domains
and url_spec.hostname not in self.allowed_media_domains
):
raise ValueError(
f"The URL must be from one of the allowed domains: "
f"{self.allowed_media_domains}. Input URL domain: "
f"{url_spec.hostname}")
f"{url_spec.hostname}"
)
def load_from_url(
self,
@@ -175,20 +187,19 @@ class MediaConnector:
timeout=fetch_timeout,
allow_redirects=envs.VLLM_MEDIA_URL_ALLOW_REDIRECTS,
)
future = loop.run_in_executor(global_thread_pool,
media_io.load_bytes, data)
future = loop.run_in_executor(global_thread_pool, media_io.load_bytes, data)
return await future
if url_spec.scheme == "data":
future = loop.run_in_executor(global_thread_pool,
self._load_data_url, url_spec,
media_io)
future = loop.run_in_executor(
global_thread_pool, self._load_data_url, url_spec, media_io
)
return await future
if url_spec.scheme == "file":
future = loop.run_in_executor(global_thread_pool,
self._load_file_url, url_spec,
media_io)
future = loop.run_in_executor(
global_thread_pool, self._load_file_url, url_spec, media_io
)
return await future
msg = "The URL must be either a HTTP, data or file URL."
raise ValueError(msg)
@@ -234,8 +245,9 @@ class MediaConnector:
By default, the image is converted into RGB format.
"""
image_io = ImageMediaIO(image_mode=image_mode,
**self.media_io_kwargs.get("image", {}))
image_io = ImageMediaIO(
image_mode=image_mode, **self.media_io_kwargs.get("image", {})
)
try:
return self.load_from_url(
@@ -258,8 +270,9 @@ class MediaConnector:
By default, the image is converted into RGB format.
"""
image_io = ImageMediaIO(image_mode=image_mode,
**self.media_io_kwargs.get("image", {}))
image_io = ImageMediaIO(
image_mode=image_mode, **self.media_io_kwargs.get("image", {})
)
try:
return await self.load_from_url_async(
@@ -280,10 +293,10 @@ class MediaConnector:
"""
Load video from an HTTP or base64 data URL.
"""
image_io = ImageMediaIO(image_mode=image_mode,
**self.media_io_kwargs.get("image", {}))
video_io = VideoMediaIO(image_io,
**self.media_io_kwargs.get("video", {}))
image_io = ImageMediaIO(
image_mode=image_mode, **self.media_io_kwargs.get("image", {})
)
video_io = VideoMediaIO(image_io, **self.media_io_kwargs.get("video", {}))
return self.load_from_url(
video_url,
@@ -302,10 +315,10 @@ class MediaConnector:
By default, the image is converted into RGB format.
"""
image_io = ImageMediaIO(image_mode=image_mode,
**self.media_io_kwargs.get("image", {}))
video_io = VideoMediaIO(image_io,
**self.media_io_kwargs.get("video", {}))
image_io = ImageMediaIO(
image_mode=image_mode, **self.media_io_kwargs.get("image", {})
)
video_io = VideoMediaIO(image_io, **self.media_io_kwargs.get("video", {}))
return await self.load_from_url_async(
video_url,
@@ -356,7 +369,8 @@ def encode_video_base64(frames: npt.NDArray) -> str:
def argsort_mm_positions(
mm_positions: MultiModalPlaceholderDict) -> list[tuple[str, int]]:
mm_positions: MultiModalPlaceholderDict,
) -> list[tuple[str, int]]:
"""
Given a `MultiModalPlaceholderDict`, output a sequence of keys to
sort the dictionary by `offset` (starting index in the input sequence)
@@ -366,9 +380,11 @@ def argsort_mm_positions(
A list of `(modality, idx)`, which can be used to access an item
by `mm_positions[modality][idx]`.
"""
flat_items = ((modality, idx, item)
for modality, items in mm_positions.items()
for idx, item in enumerate(items))
flat_items = (
(modality, idx, item)
for modality, items in mm_positions.items()
for idx, item in enumerate(items)
)
sorted_flat_items = sorted(flat_items, key=lambda x: x[2].offset)
@@ -397,7 +413,8 @@ def group_mm_kwargs_by_modality(
raise RuntimeError(
"`group_mm_kwargs_by_modality` now requires "
"`merge_by_field_config` arg, please update your model runner "
"according to https://github.com/vllm-project/vllm/pull/25676.")
"according to https://github.com/vllm-project/vllm/pull/25676."
)
from vllm.multimodal.inputs import MultiModalKwargs, MultiModalKwargsItems
@@ -409,7 +426,9 @@ def group_mm_kwargs_by_modality(
if merge_by_field_config:
mm_kwargs_group: BatchedTensorInputs = dict(
MultiModalKwargsItems.from_seq(items_lst).get_data(
pin_memory=pin_memory))
pin_memory=pin_memory
)
)
if device is not None:
mm_kwargs_group = json_map_leaves(
@@ -440,9 +459,7 @@ def fetch_audio(
audio_url: URL of the audio file to fetch.
audio_io_kwargs: Additional kwargs passed to handle audio IO.
"""
media_io_kwargs = None if not audio_io_kwargs else {
"audio": audio_io_kwargs
}
media_io_kwargs = None if not audio_io_kwargs else {"audio": audio_io_kwargs}
media_connector = MediaConnector(media_io_kwargs=media_io_kwargs)
return media_connector.fetch_audio(audio_url)
@@ -456,9 +473,7 @@ def fetch_image(
image_url: URL of the image file to fetch.
image_io_kwargs: Additional kwargs passed to handle image IO.
"""
media_io_kwargs = None if not image_io_kwargs else {
"image": image_io_kwargs
}
media_io_kwargs = None if not image_io_kwargs else {"image": image_io_kwargs}
media_connector = MediaConnector(media_io_kwargs=media_io_kwargs)
return media_connector.fetch_image(image_url)
@@ -472,8 +487,6 @@ def fetch_video(
video_url: URL of the video file to fetch.
video_io_kwargs: Additional kwargs passed to handle video IO.
"""
media_io_kwargs = None if not video_io_kwargs else {
"video": video_io_kwargs
}
media_io_kwargs = None if not video_io_kwargs else {"video": video_io_kwargs}
media_connector = MediaConnector(media_io_kwargs=media_io_kwargs)
return media_connector.fetch_video(video_url)