vidformer 0.6.5__tar.gz → 0.8.0__tar.gz

{vidformer-0.6.5 → vidformer-0.8.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: vidformer
-Version: 0.6.5
+Version: 0.8.0
 Summary: A Python library for creating and viewing videos with vidformer.
 Author-email: Dominik Winecki <dominikwinecki@gmail.com>
 Requires-Python: >=3.8

{vidformer-0.6.5 → vidformer-0.8.0}/vidformer/__init__.py

@@ -1,5 +1,5 @@
 """A Python library for creating and viewing videos with vidformer."""
 
-__version__ = "0.6.5"
+__version__ = "0.8.0"
 
 from .vf import *

{vidformer-0.6.5 → vidformer-0.8.0}/vidformer/cv2/vf_cv2.py

@@ -45,6 +45,11 @@ LINE_4 = 4
 LINE_8 = 8
 LINE_AA = 16
 
+_inline_mat = vf.Filter("_inline_mat")
+_slice_mat = vf.Filter("_slice_mat")
+_slice_write_mat = vf.Filter("_slice_write_mat")
+
+
 _filter_scale = vf.Filter("Scale")
 _filter_rectangle = vf.Filter("cv2.rectangle")
 _filter_putText = vf.Filter("cv2.putText")
@@ -96,8 +101,9 @@ class Frame:
             return
 
         self._modified = True
-        self._f = _filter_scale(self._f, pix_fmt="rgb24")
-        self._fmt["pix_fmt"] = "rgb24"
+        if self._fmt["pix_fmt"] != "rgb24":
+            self._f = _filter_scale(self._f, pix_fmt="rgb24")
+            self._fmt["pix_fmt"] = "rgb24"
 
     def numpy(self):
         """
@@ -113,8 +119,112 @@ class Frame:
         assert len(frame_raster_rgb24) == self.shape[0] * self.shape[1] * 3
         raw_data_array = np.frombuffer(frame_raster_rgb24, dtype=np.uint8)
         frame = raw_data_array.reshape(self.shape)
+        frame = frame[:, :, ::-1]  # convert RGB to BGR
         return frame
 
+    def __getitem__(self, key):
+        if not isinstance(key, tuple):
+            raise NotImplementedError("Only 2D slicing is supported")
+
+        if len(key) != 2:
+            raise NotImplementedError("Only 2D slicing is supported")
+
+        if not all(isinstance(x, slice) for x in key):
+            raise NotImplementedError("Only 2D slicing is supported")
+
+        miny = key[0].start if key[0].start is not None else 0
+        maxy = key[0].stop if key[0].stop is not None else self.shape[0]
+        minx = key[1].start if key[1].start is not None else 0
+        maxx = key[1].stop if key[1].stop is not None else self.shape[1]
+
+        # handle negative indices
+        if miny < 0:
+            miny = self.shape[0] + miny
+        if maxy < 0:
+            maxy = self.shape[0] + maxy
+        if minx < 0:
+            minx = self.shape[1] + minx
+        if maxx < 0:
+            maxx = self.shape[1] + maxx
+
+        if (
+            maxy <= miny
+            or maxx <= minx
+            or miny < 0
+            or minx < 0
+            or maxy > self.shape[0]
+            or maxx > self.shape[1]
+        ):
+            raise NotImplementedError("Invalid slice")
+
+        f = _slice_mat(self._f, miny, maxy, minx, maxx)
+        fmt = self._fmt.copy()
+        fmt["width"] = maxx - minx
+        fmt["height"] = maxy - miny
+        return Frame(f, fmt)
+
+    def __setitem__(self, key, value):
+        value = frameify(value, "value")
+
+        if not isinstance(key, tuple):
+            raise NotImplementedError("Only 2D slicing is supported")
+
+        if len(key) != 2:
+            raise NotImplementedError("Only 2D slicing is supported")
+
+        if not all(isinstance(x, slice) for x in key):
+            raise NotImplementedError("Only 2D slicing is supported")
+
+        miny = key[0].start if key[0].start is not None else 0
+        maxy = key[0].stop if key[0].stop is not None else self.shape[0]
+        minx = key[1].start if key[1].start is not None else 0
+        maxx = key[1].stop if key[1].stop is not None else self.shape[1]
+
+        # handle negative indices
+        if miny < 0:
+            miny = self.shape[0] + miny
+        if maxy < 0:
+            maxy = self.shape[0] + maxy
+        if minx < 0:
+            minx = self.shape[1] + minx
+        if maxx < 0:
+            maxx = self.shape[1] + maxx
+
+        if (
+            maxy <= miny
+            or maxx <= minx
+            or miny < 0
+            or minx < 0
+            or maxy > self.shape[0]
+            or maxx > self.shape[1]
+        ):
+            raise NotImplementedError("Invalid slice")
+
+        if value.shape[0] != maxy - miny or value.shape[1] != maxx - minx:
+            raise NotImplementedError("Shape mismatch")
+
+        self._mut()
+        value._mut()
+
+        self._f = _slice_write_mat(self._f, value._f, miny, maxy, minx, maxx)
+
+
+def _inline_frame(arr):
+    assert arr.dtype == np.uint8
+    assert arr.ndim == 3
+    assert arr.shape[2] == 3
+
+    # convert BGR to RGB
+    arr = arr[:, :, ::-1]
+
+    width = arr.shape[1]
+    height = arr.shape[0]
+    pix_fmt = "rgb24"
+
+    f = _inline_mat(arr.tobytes(), width=width, height=height, pix_fmt=pix_fmt)
+    fmt = {"width": width, "height": height, "pix_fmt": pix_fmt}
+    return Frame(f, fmt)
+
 
 class VideoCapture:
     def __init__(self, path):
@@ -178,9 +288,9 @@ class VideoWriter:
         self._pix_fmt = "yuv420p"
 
     def write(self, frame):
-        if not isinstance(frame, Frame):
-            raise Exception("frame must be a vidformer.cv2.Frame object")
-        if frame._modified:
+        frame = frameify(frame, "frame")
+
+        if frame._fmt["pix_fmt"] != self._pix_fmt:
             f_obj = _filter_scale(frame._f, pix_fmt=self._pix_fmt)
             self._frames.append(f_obj)
         else:
@@ -210,6 +320,24 @@ class VideoWriter_fourcc:
         self._args = args
 
 
+def frameify(obj, field_name=None):
+    """
+    Turn an object (e.g., ndarray) into a Frame.
+    """
+
+    if isinstance(obj, Frame):
+        return obj
+    elif isinstance(obj, np.ndarray):
+        return _inline_frame(obj)
+    else:
+        if field_name is not None:
+            raise Exception(
+                f"Unsupported type for field {field_name}, expected Frame or np.ndarray"
+            )
+        else:
+            raise Exception("Unsupported type, expected Frame or np.ndarray")
+
+
 def imread(path, *args):
     if len(args) > 0:
         raise NotImplementedError("imread does not support additional arguments")
@@ -225,8 +353,7 @@ def imwrite(path, img, *args):
     if len(args) > 0:
         raise NotImplementedError("imwrite does not support additional arguments")
 
-    if not isinstance(img, Frame):
-        raise Exception("img must be a vidformer.cv2.Frame object")
+    img = frameify(img)
 
     fmt = img._fmt.copy()
     width = fmt["width"]
@@ -282,7 +409,7 @@ def rectangle(img, pt1, pt2, color, thickness=None, lineType=None, shift=None):
     cv.rectangle(	img, pt1, pt2, color[, thickness[, lineType[, shift]]]	)
     """
 
-    assert isinstance(img, Frame)
+    img = frameify(img)
     img._mut()
 
     assert len(pt1) == 2
@@ -326,7 +453,7 @@ def putText(
     cv.putText(	img, text, org, fontFace, fontScale, color[, thickness[, lineType[, bottomLeftOrigin]]]	)
     """
 
-    assert isinstance(img, Frame)
+    img = frameify(img)
     img._mut()
 
     assert isinstance(text, str)
@@ -365,7 +492,7 @@ def arrowedLine(
     """
     cv.arrowedLine(	img, pt1, pt2, color[, thickness[, line_type[, shift[, tipLength]]]]	)
     """
-    assert isinstance(img, Frame)
+    img = frameify(img)
     img._mut()
 
     assert len(pt1) == 2
@@ -399,7 +526,7 @@ def arrowedLine(
 
 
 def line(img, pt1, pt2, color, thickness=None, lineType=None, shift=None):
-    assert isinstance(img, Frame)
+    img = frameify(img)
     img._mut()
 
     assert len(pt1) == 2
@@ -429,7 +556,7 @@ def line(img, pt1, pt2, color, thickness=None, lineType=None, shift=None):
 
 
 def circle(img, center, radius, color, thickness=None, lineType=None, shift=None):
-    assert isinstance(img, Frame)
+    img = frameify(img)
     img._mut()
 
     assert len(center) == 2
@@ -480,15 +607,15 @@ def addWeighted(src1, alpha, src2, beta, gamma, dst=None, dtype=-1):
     """
     cv.addWeighted(	src1, alpha, src2, beta, gamma[, dst[, dtype]]	) -> 	dst
     """
-    assert isinstance(src1, Frame)
-    assert isinstance(src2, Frame)
+    src1 = frameify(src1, "src1")
+    src2 = frameify(src2, "src2")
     src1._mut()
     src2._mut()
 
     if dst is None:
         dst = Frame(src1._f, src1._fmt.copy())
     else:
-        assert isinstance(dst, Frame)
+        assert isinstance(dst, Frame), "dst must be a Frame"
     dst._mut()
 
     assert isinstance(alpha, float) or isinstance(alpha, int)

{vidformer-0.6.5 → vidformer-0.8.0}/vidformer/vf.py

@@ -638,6 +638,11 @@ def _json_arg(arg, skip_data_anot=False):
         if skip_data_anot:
             return {"String": arg}
         return {"Data": {"String": arg}}
+    elif type(arg) == bytes:
+        arg = list(arg)
+        if skip_data_anot:
+            return {"Bytes": arg}
+        return {"Data": {"Bytes": arg}}
     elif type(arg) == float:
         if skip_data_anot:
             return {"Float": arg}
@@ -837,7 +842,7 @@ class UDF:
             assert type(obj[type_key]) == bool
             return obj[type_key]
         else:
-            assert False
+            assert False, f"Unknown type: {type_key}"
 
     def _deser_filter(self, obj):
         assert type(obj) == dict
@@ -873,7 +878,7 @@ class UDF:
             assert type(obj[type_key]) == bool
             return obj[type_key]
         else:
-            assert False
+            assert False, f"Unknown type: {type_key}"
 
     def _host(self, socket_path: str):
         if os.path.exists(socket_path):