Add random()

src/image_utils/file_utils.py

@@ -59,15 +59,15 @@ def get_images(path: Path, recursive: bool = False, allowed_extensions=[".png",
         yield Im.open(file)
 
 
-def load_cached_from_url(url: str) -> BytesIO:
+def load_cached_from_url(url: str, cache: bool = True) -> BytesIO:
     import hashlib
 
     cache_dir = Path.home() / ".cache" / "image_utils"
     cache_dir.mkdir(parents=True, exist_ok=True)
     filename = hashlib.md5(url.encode()).hexdigest()
     local_path = cache_dir / filename
 
-    if local_path.exists():
+    if cache and local_path.exists():
         return BytesIO(local_path.read_bytes())
     else:
         image_bytesio = download_file_bytes(url)

src/image_utils/im.py

@@ -18,6 +18,7 @@
 from jaxtyping import Bool, Float, Integer
 from numpy import ndarray
 from PIL import Image
+from torch import rand
 
 if importlib.util.find_spec("torch") is not None:
     import torch
@@ -111,10 +112,10 @@ class Im:
     default_normalize_mean = [0.4265, 0.4489, 0.4769]
     default_normalize_std = [0.2053, 0.2206, 0.2578]
 
-    def __init__(self, arr: Union["Im", Tensor, Image.Image, ndarray], channel_range: Optional[ChannelRange] = None, **kwargs):
+    def __init__(self, arr: Union["Im", Tensor, Image.Image, ndarray, str, Path], channel_range: Optional[ChannelRange] = None, **kwargs):
         # TODO: Add real URL checking here
         if isinstance(arr, (str, Path)) and Path(arr).exists():
-            arr = Im.open(arr)  # type: ignore
+            arr = Im.open(arr)
         elif isinstance(arr, str):
             arr = Image.open(load_cached_from_url(arr))
 
@@ -365,6 +366,13 @@ def open(filepath: Path, use_imageio=False) -> Im:
     @callable_staticmethod
     def new(h: int, w: int, color=(255, 255, 255)):
         return Im(Image.new("RGB", (w, h), color))
+
+    @callable_staticmethod
+    def random(h: int = 1080, w: int = 1920) -> Im:
+        try:
+            return Im(Image.open(load_cached_from_url(f"https://unsplash.it/{w}/{h}?random", cache=False)))
+        except:
+            return Im(Image.open(load_cached_from_url(f"https://picsum.photos/{w}/{h}?random", cache=False)))
 
     @_convert_to_datatype(desired_datatype=Tensor, desired_order=ChannelOrder.CHW, desired_range=ChannelRange.FLOAT)
     def resize(self, height: int, width: int, resampling_mode: str = "bilinear"):