Cast mime type to str always in SDK tests; Fix failing tests and vision logic

Author: hexbabe

docs/examples/example.ipynb

@@ -151,7 +151,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": null,
    "metadata": {
     "tags": [
      "hide-output"
@@ -165,7 +165,7 @@
     "\n",
     "robot = await connect_with_channel()\n",
     "camera = Camera.from_robot(robot, \"camera0\")\n",
-    "image = await camera.get_image(CameraMimeType.JPEG)\n",
+    "image = await camera.get_image(CameraMimeType.JPEG.value)\n",
     "pil = viam_to_pil_image(image)\n",
     "pil.save(\"foo.png\")\n",
     "\n",

examples/server/v1/client.py

@@ -34,7 +34,7 @@ async def client():
 
         print("\n#### CAMERA ####")
         camera = Camera.from_robot(robot, "camera0")
-        img = await camera.get_image(mime_type=CameraMimeType.PNG)
+        img = await camera.get_image(mime_type=CameraMimeType.PNG.value)
         assert isinstance(img, Image)
         img.show()
         await asyncio.sleep(1)

examples/server/v1/components.py

@@ -322,7 +322,7 @@ def __init__(self, name: str):
         img = Image.open(p.parent.absolute().joinpath("viam.jpeg"))
         buf = BytesIO()
         img.copy().save(buf, format="JPEG")
-        self.image = ViamImage(buf.getvalue(), CameraMimeType.JPEG)
+        self.image = ViamImage(buf.getvalue(), CameraMimeType.JPEG.value)
         img.close()
         super().__init__(name)
 

src/viam/components/camera/camera.py

@@ -74,6 +74,9 @@ async def get_images(
         """Get simultaneous images from different imagers, along with associated metadata.
         This should not be used for getting a time series of images from the same imager.
 
+        The extra parameter can be used to pass additional options to the camera resource. The filter_source_names parameter can be used to filter
+        only the images from the specified source names. When unspecified, all images are returned.
+
         ::
 
             my_camera = Camera.from_robot(robot=machine, name="my_camera")

src/viam/components/camera/client.py

@@ -60,7 +60,7 @@ async def get_images(
                 mime_type = img_data.mime_type
             else:
                 # TODO(RSDK-11728): remove this once we deleted the format field
-                mime_type = str(CameraMimeType.from_proto(img_data.format))
+                mime_type = CameraMimeType.from_proto(img_data.format).value
             img = NamedImage(img_data.source_name, img_data.image, mime_type)
             imgs.append(img)
         resp_metadata: ResponseMetadata = response.response_metadata
@@ -74,7 +74,7 @@ async def get_point_cloud(
         **kwargs,
     ) -> Tuple[bytes, str]:
         md = kwargs.get("metadata", self.Metadata()).proto
-        request = GetPointCloudRequest(name=self.name, mime_type=CameraMimeType.PCD, extra=dict_to_struct(extra))
+        request = GetPointCloudRequest(name=self.name, mime_type=CameraMimeType.PCD.value, extra=dict_to_struct(extra))
         response: GetPointCloudResponse = await self.client.GetPointCloud(request, timeout=timeout, metadata=md)
         return (response.point_cloud, response.mime_type)
 

src/viam/components/camera/service.py

@@ -22,6 +22,8 @@
 from viam.resource.rpc_service_base import ResourceRPCServiceBase
 from viam.utils import dict_to_struct, struct_to_dict
 
+from viam.media.video import CameraMimeType
+
 from . import Camera
 
 
@@ -58,15 +60,16 @@ async def GetImages(self, stream: Stream[GetImagesRequest, GetImagesResponse]) -
         )
         img_bytes_lst = []
         for img in images:
+            # TODO(RSDK-11728): remove this try except logic once we deleted the format field
             try:
                 mime_type = CameraMimeType.from_string(img.mime_type)  # this can ValueError if the mime_type is not a CameraMimeType
                 fmt = mime_type.to_proto()
             except ValueError:
-                # TODO(RSDK-11728): remove this once we deleted the format field
+                mime_type = img.mime_type
                 fmt = Format.FORMAT_UNSPECIFIED
 
             img_bytes = img.data
-            img_bytes_lst.append(Image(source_name=name, mime_type=img.mime_type, format=fmt, image=img_bytes))
+            img_bytes_lst.append(Image(source_name=name, format=fmt, mime_type=img.mime_type, image=img_bytes))
         response = GetImagesResponse(images=img_bytes_lst, response_metadata=metadata)
         await stream.send_message(response)
 

src/viam/media/utils/pil/__init__.py

@@ -25,7 +25,7 @@ def viam_to_pil_image(image: ViamImage) -> Image.Image:
     return Image.open(BytesIO(image.data), formats=LIBRARY_SUPPORTED_FORMATS)
 
 
-def pil_to_viam_image(image: Image.Image, mime_type: CameraMimeType) -> ViamImage:
+def pil_to_viam_image(image: Image.Image, mime_type: str) -> ViamImage:
     """
     Convert a PIL.Image to a ViamImage.
 
@@ -34,11 +34,16 @@ def pil_to_viam_image(image: Image.Image, mime_type: CameraMimeType) -> ViamImag
 
     Args:
         image (Image.Image): The image to convert.
-        mime_type (CameraMimeType): The mime type to convert the image to.
+        mime_type (str): The mime type to convert the image to. Must be of type `CameraMimeType`.
 
     Returns:
         ViamImage: The resulting ViamImage
     """
+    try:
+        mime_type = CameraMimeType.from_string(mime_type)
+    except ValueError as e:
+        raise ValueError(f"Unsupported mimetype str: {mime_type}") from e
+
     if mime_type.name in LIBRARY_SUPPORTED_FORMATS:
         buf = BytesIO()
         if image.mode == "RGBA" and mime_type == CameraMimeType.JPEG:
@@ -48,4 +53,4 @@ def pil_to_viam_image(image: Image.Image, mime_type: CameraMimeType) -> ViamImag
     else:
         raise ValueError(f"Cannot encode image to {mime_type}")
 
-    return ViamImage(data, mime_type)
+    return ViamImage(data, mime_type.value)

src/viam/media/video.py

@@ -17,6 +17,9 @@ class CameraMimeType(str, Enum):
     PNG = "image/png"
     PCD = "pointcloud/pcd"
 
+    def __str__(self) -> str:
+        return self.value
+
     @classmethod
     def from_string(cls, value: str) -> Self:
         """Return the mimetype from a string.

src/viam/services/vision/client.py

@@ -69,7 +69,11 @@ async def capture_all_from_camera(
         result = CaptureAllResult()
         result.extra = struct_to_dict(response.extra)
         if return_image:
-            mime_type = CameraMimeType.from_proto(response.image.format)
+            # TODO(RSDK-11728): remove this branching logic once we deleted the format field
+            if response.image.mime_type:
+                mime_type = response.image.mime_type
+            else:
+                mime_type = CameraMimeType.from_proto(response.image.format).value
             img = ViamImage(response.image.image, mime_type)
             result.image = img
         if return_classifications:
@@ -102,7 +106,7 @@ async def get_detections(
         **kwargs,
     ) -> List[Detection]:
         md = kwargs.get("metadata", self.Metadata()).proto
-        mime_type = CameraMimeType.JPEG
+        mime_type = CameraMimeType.JPEG.value
 
         if image.width is None or image.height is None:
             raise ViamError(f"image {image} needs to have a specified width and height")
@@ -145,7 +149,7 @@ async def get_classifications(
     ) -> List[Classification]:
         md = kwargs.get("metadata", self.Metadata()).proto
 
-        mime_type = CameraMimeType.JPEG
+        mime_type = CameraMimeType.JPEG.value
         if image.width is None or image.height is None:
             raise ViamError(f"image {image} needs to have a specified width and height")
         request = GetClassificationsRequest(
@@ -172,7 +176,7 @@ async def get_object_point_clouds(
         request = GetObjectPointCloudsRequest(
             name=self.name,
             camera_name=camera_name,
-            mime_type=CameraMimeType.PCD,
+            mime_type=CameraMimeType.PCD.value,
             extra=dict_to_struct(extra),
         )
         response: GetObjectPointCloudsResponse = await self.client.GetObjectPointClouds(request, timeout=timeout, metadata=md)

src/viam/services/vision/service.py

@@ -2,7 +2,7 @@
 
 from viam.media.video import CameraMimeType, ViamImage
 from viam.proto.common import DoCommandRequest, DoCommandResponse
-from viam.proto.component.camera import Image
+from viam.proto.component.camera import Format, Image
 from viam.proto.service.vision import (
     CaptureAllFromCameraRequest,
     CaptureAllFromCameraResponse,
@@ -36,7 +36,7 @@ class VisionRPCService(UnimplementedVisionServiceBase, ResourceRPCServiceBase):
     async def CaptureAllFromCamera(self, stream: Stream[CaptureAllFromCameraRequest, CaptureAllFromCameraResponse]) -> None:
         request = await stream.recv_message()
         assert request is not None
-        vision = self.get_resource(request.name)
+        vision: Vision = self.get_resource(request.name)
         extra = struct_to_dict(request.extra)
         timeout = stream.deadline.time_remaining() if stream.deadline else None
         result = await vision.capture_all_from_camera(
@@ -50,9 +50,15 @@ async def CaptureAllFromCamera(self, stream: Stream[CaptureAllFromCameraRequest,
         )
         img = None
         if result.image is not None:
-            fmt = result.image.mime_type.to_proto()
+            # TODO(RSDK-11728): remove this try except logic once we deleted the format field
+            try:
+                mime_type = CameraMimeType.from_string(result.image.mime_type)  # this can ValueError if mime_type is not a CameraMimeType
+                fmt = mime_type.to_proto()
+            except ValueError:
+                mime_type = result.image.mime_type
+                fmt = Format.FORMAT_UNSPECIFIED
             img_bytes = result.image.data
-            img = Image(source_name=request.camera_name, format=fmt, image=img_bytes)
+            img = Image(source_name=request.camera_name, mime_type=result.image.mime_type, format=fmt, image=img_bytes)
         response = CaptureAllFromCameraResponse(
             image=img,
             detections=result.detections,