imaging-server-kit 0.0.10__tar.gz → 0.0.12__tar.gz

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: imaging-server-kit
-Version: 0.0.10
+Version: 0.0.12
 Summary: Deploy image processing algorithms in FastAPI servers. 
 Author-email: Mallory Wittwer <mallory.wittwer@epfl.ch>
 License: Copyright (c) 2024, EPFL

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/docs/sections/algorithm_server.md

@@ -23,7 +23,7 @@ To turn this function into to an algorithm server, we first need to modify the f
 
 ## Make the function return a list of data tuples
 
-The function should return a list of tuples ("data tuples"). Each tuple represents one distinct output of the algorithm.
+The function should return a list of tuples. Each tuple has three elements and represents a distinct output of the algorithm.
 
 Our threshold algorithm only has one output: the segmentation mask resulting from thresholding.
 
@@ -36,22 +36,22 @@ def threshold_algo_server(image: np.ndarray, threshold: float):
 The data tuples are inspired from Napari's [LayerDataTuple](https://napari.org/0.4.15/guides/magicgui.html?highlight=layerdatatuple) model. They include three elements:
 
 - The *first element* is the **data**, usually in the form of a Numpy array. The shape and interpretation of the axes depends on what the output represents (cf. table below).
-- The *second element* is optional **metadata** associated with the output, represented as a Python dictionary. It can be empty (`{}`). These metadata are used to affect how the output is displayed in client apps.
+- The *second element* is a Python dictionary representing **metadata** associated with the output. It can be empty (`{}`). These metadata are used to affect how the output is displayed in client apps (e.g. Napari).
 - The *third element* is the output type: a string identifying what the output represents.
 
 | Output type       | Description                                                                                           |
 | ----------------- | ----------------------------------------------------------------------------------------------------- |
-| `"image"`         | An image or image-like data (incl. 3D and RGB) as a nD array.                                         |
-| `"mask"`          | A segmentation mask (2D, 3D) as integer nD array. Integers represent the **object class**.            |
+| `"image"`         | An image or image-like data (incl. 3D and RGB) as a numpy array.                                         |
+| `"mask"`          | A segmentation mask (2D, 3D) as integer numpy array. Integers represent the **object class**.            |
 | `"instance_mask"` | A segmentation mask (2D, 3D) as integer nD array. Integers represent **object instances**.            |
 | `"points"`        | A collection of point coordinates (array of shape (N, 2) or (N, 3)).                                  |
-| `"boxes"`         | A collection of boxes (array of shape (N, 4) representing the top-left and bottom-right box corners). |
+| `"boxes"`         | A collection of boxes (array of shape (N, 4, 2) representing the box corners). |
 | `"vectors"`       | Array of vectors in the Napari [Vectors](https://napari.org/stable/howtos/layers/vectors.html) format.|
 | `"tracks"`        | Array of tracks in the Napari [Tracks](https://napari.org/stable/howtos/layers/tracks.html) format.   |
 | `"class"`         | A class label (for image classification).                                                             |
 | `"text"`          | A string of text (for example, for image captioning).                                                 |
 
-The function can return an arbitrary number of outputs, following the pattern:
+The function can return multiple outputs, following the pattern:
 
 ```
 # Body of the function
@@ -68,10 +68,11 @@ The function parameters can only be **numpy arrays**, **numeric values** (`int`,
 
 ## Decorate the function with `@algorithm_server`
 
-This serves several purposes:
+This has several purposes:
 
 - It converts the python function to a FastAPI server with predefined routes (cf. [API Endpoints](api_endpoints)).
 - It enables the server to validate algorithm parameters when receiving requests.
+- It tells client apps (Napari, QuPath) how to render the parameters in the user interface.
 - Optional info about the algorithm server can be added to populate its `info` page.
 
 Below is our decorated threshold algorithm function:

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/docs/sections/getting_started.md

@@ -10,7 +10,7 @@ serverkit demo
 
 This will start a demo web server for a simple intensity threshold algorithm.
 
-The server will be running on http://localhost:8000. Open your web browser and navigate to this page to check that the server is running. You should see a web page displaying information about the running threshold algorithm server.
+The server will be running on http://localhost:8000. Open your web browser and navigate to this page to check that the server is running. You should see a web page displaying information about the threshold algorithm server.
 
 ![screenshot](../assets/screenshot_running_server.png)
 
@@ -42,7 +42,7 @@ You can also interact with the algorithm server via QuPath. To try it out, you n
 
 ## Usage from Python
 
-It is also possible to connect to an algorithm server and run computation from Python directly:
+It is also possible to connect to an algorithm server and run computations from Python directly:
 
 ```python
 from imaging_server_kit import Client

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/examples/basic/auto_threshold.py

@@ -14,7 +14,7 @@ from imaging_server_kit import algorithm_server, ImageUI, DropDownUI
     parameters={
         "image": ImageUI(),
         "method": DropDownUI(
-            default="otsu",
+            default="Otsu",
             title="Method",
             description="Auto-threshold method.",
             items=["Otsu", "Li"],

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/examples/servers/serverkit-skimage-log/main.py

@@ -127,9 +127,14 @@ def skimage_log_server(
         },  # sigmas = numpy array representing the point size
     }
 
-    return [
-        (points, points_params, "points"),
-    ]
+    if len(points):
+        return [
+            (points, points_params, "points"),
+        ]
+    else:
+        return [
+            ("No points were detected.", {}, "text")
+        ]
 
 
 if __name__ == "__main__":

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/src/imaging_server_kit/_version.py

@@ -17,5 +17,5 @@ __version__: str
 __version_tuple__: VERSION_TUPLE
 version_tuple: VERSION_TUPLE
 
-__version__ = version = '0.0.10'
-__version_tuple__ = version_tuple = (0, 0, 10)
+__version__ = version = '0.0.12'
+__version_tuple__ = version_tuple = (0, 0, 12)

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/src/imaging_server_kit/core/decorator.py

@@ -11,7 +11,6 @@ from .encoding import decode_contents
 class BaseParamsModel(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
-
 def decode_image_array(cls, v, dimensionality) -> "np.ndarray":
     image_array = decode_contents(v)
 
@@ -58,7 +57,6 @@ def parse_params(parameters: dict) -> BaseModel:
                 "widget_type"
             ] = param_details.widget_type
 
-
             if param_details.widget_type in ["image", "mask"]:
                 validated_func = partial(
                     decode_image_array, dimensionality=param_details.dimensionality

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/src/imaging_server_kit/core/geometry.py

@@ -27,7 +27,11 @@ def mask2features(segmentation_mask: np.ndarray) -> List[Feature]:
         for detection_id, contour in enumerate(polygons.points, start=1):
             coords = np.array(contour)
             coords = np.vstack([coords, coords[0]])  # Close the polygon for QuPath
-            geom = Polygon(coordinates=[coords.tolist()])
+            try:
+                geom = Polygon(coordinates=[coords.tolist()], validate=True)
+            except ValueError:
+                # TODO: not sure what to do when this happens (very occasionally)
+                print("Found an invalid polygon...")
             feature = Feature(
                 geometry=geom, 
                 properties={
@@ -75,7 +79,11 @@ def instance_mask2features(segmentation_mask: np.ndarray) -> List[Feature]:
         for contour in polygons.points:
             coords = np.array(contour)
             coords = np.vstack([coords, coords[0]])  # Close the polygon for QuPath
-            geom = Polygon(coordinates=[coords.tolist()])
+            try:
+                geom = Polygon(coordinates=[coords.tolist()], validate=True)
+            except ValueError:
+                # TODO: not sure what to do when this happens (very occasionally)
+                print("Found an invalid polygon...")
             feature = Feature(
                 geometry=geom, 
                 properties={
@@ -170,14 +178,18 @@ def boxes2features(boxes: np.ndarray) -> List[Feature]:
         coords.append(
             coords[0]
         )  # Add the first element at the end to close the Polygon
-        geom = Polygon(coordinates=[coords])
+        try:
+            geom = Polygon(coordinates=[coords], validate=True)
+        except ValueError:
+            print("Invalid box polygon found. Expected an array of shape (N, 4, D) representing the corners of the box.")
+            geom = None  # TODO: should we handle this better?
         features.append(Feature(geometry=geom, properties={"Detection ID": i}))
     return features
 
 
 def features2boxes(features):
     boxes = np.array([feature["geometry"]["coordinates"] for feature in features])
-    boxes = np.squeeze(boxes)
+    boxes = np.array([box[0] for box in boxes])  # We get back a shape of (N, 1, 5, 2) - so we remove dim. 1
     boxes = boxes[:, :-1] # Remove the last element
     boxes = boxes[:, :, ::-1]  # Invert XY
     return boxes

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/src/imaging_server_kit/hub/multialgo_server.py

@@ -13,9 +13,15 @@ from contextlib import asynccontextmanager
 import asyncio
 import imaging_server_kit as serverkit
 from imaging_server_kit import AlgorithmServer
-from imaging_server_kit.core import parse_algo_params_schema, encode_contents, decode_contents
-
-templates_dir = importlib.resources.files("imaging_server_kit.core").joinpath("templates")
+from imaging_server_kit.core import (
+    parse_algo_params_schema,
+    encode_contents,
+    decode_contents,
+)
+
+templates_dir = importlib.resources.files("imaging_server_kit.core").joinpath(
+    "templates"
+)
 static_dir = importlib.resources.files("imaging_server_kit.core").joinpath("static")
 
 templates = Jinja2Templates(directory=str(templates_dir))
@@ -31,7 +37,9 @@ class MultiAlgorithmServer:  # TODO: could this inherit from algoserver somehow?
 
         # Centralized exception handlers
         @self.app.exception_handler(RequestValidationError)
-        async def validation_exception_handler(request: Request, exc: RequestValidationError):
+        async def validation_exception_handler(
+            request: Request, exc: RequestValidationError
+        ):
             return JSONResponse(
                 status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
                 content={"error_type": "validation_error", "detail": exc.errors()},
@@ -40,12 +48,14 @@ class MultiAlgorithmServer:  # TODO: could this inherit from algoserver somehow?
         @self.app.exception_handler(Exception)
         async def generic_exception_handler(request: Request, exc: Exception):
             from fastapi import HTTPException
+
             if isinstance(exc, HTTPException):
                 raise exc
             return JSONResponse(
                 status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
                 content={"error_type": "internal_server_error", "message": str(exc)},
             )
+
         self.app.mount("/static", StaticFiles(directory=str(static_dir)), name="static")
         self.services = [server_name]
 
@@ -119,8 +129,10 @@ class MultiAlgorithmServer:  # TODO: could this inherit from algoserver somehow?
             request: Request = ...,
         ):
             if algorithm_name not in self.algorithms:
-                raise HTTPException(status_code=status.HTTP_404_NOT_FOUND,
-                                    detail=f"Algorithm {algorithm_name} not found")
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail=f"Algorithm {algorithm_name} not found",
+                )
             algo_info = self.algorithms[algorithm_name].get("algo_info")
             algo_params_schema = get_algo_params(algorithm_name)
             algo_params = parse_algo_params_schema(algo_params_schema)
@@ -142,8 +154,10 @@ class MultiAlgorithmServer:  # TODO: could this inherit from algoserver somehow?
             request: Request = ...,
         ):
             if algorithm_name not in self.algorithms:
-                raise HTTPException(status_code=status.HTTP_404_NOT_FOUND,
-                                    detail=f"Algorithm {algorithm_name} not found")
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail=f"Algorithm {algorithm_name} not found",
+                )
 
             parameters_model = self.algorithms.get(algorithm_name, {}).get(
                 "parameters_model"
@@ -156,34 +170,26 @@ class MultiAlgorithmServer:  # TODO: could this inherit from algoserver somehow?
             except ValidationError as e:
                 raise HTTPException(status_code=422, detail=e.errors())
 
-            # `array` fields must be decoded (TODO: is there a better way of handling this?)
-            if "image" in data.keys():
-                data["image"] = decode_contents(data["image"])
-            elif "mask" in data.keys():
-                data["mask"] = decode_contents(data["mask"])
-            elif "points" in data.keys():
-                data["points"] = decode_contents(data["points"])
-            elif "vectors" in data.keys():
-                data["vectors"] = decode_contents(data["vectors"])
-            elif "shapes" in data.keys():
-                data["shapes"] = decode_contents(data["shapes"])
-            elif "tracks" in data.keys():
-                data["tracks"] = decode_contents(data["tracks"])
-
-            return await self._run_algo_logic(algorithm_name, data)
+            decoded_params = algo_params.dict()
+
+            return await self._run_algo_logic(algorithm_name, decoded_params)
 
         @self.app.get("/{algorithm_name}/parameters", response_model=dict)
         def get_algo_params(algorithm_name):
             if algorithm_name not in self.algorithms:
-                raise HTTPException(status_code=status.HTTP_404_NOT_FOUND,
-                                    detail=f"Algorithm {algorithm_name} not found")
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail=f"Algorithm {algorithm_name} not found",
+                )
             return self.get_algorithm_params(algorithm_name)
 
         @self.app.get("/{algorithm_name}/sample_images", response_model=dict)
         def get_sample_images(algorithm_name):
             if algorithm_name not in self.algorithms:
-                raise HTTPException(status_code=status.HTTP_404_NOT_FOUND,
-                                    detail=f"Algorithm {algorithm_name} not found")
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail=f"Algorithm {algorithm_name} not found",
+                )
             return self.get_algorithm_sample_images(algorithm_name)
 
     async def _serialize_result_tuple(self, result_data_tuple):
@@ -228,7 +234,5 @@ class MultiAlgorithmServer:  # TODO: could this inherit from algoserver somehow?
         server = self.algorithms.get(algorithm_name)
         load_funct = server.get("load_funct")
         images = load_funct()
-        encoded_images = [
-            {"sample_image": encode_contents(image)} for image in images
-        ]
+        encoded_images = [{"sample_image": encode_contents(image)} for image in images]
         return {"sample_images": encoded_images}

{imaging_server_kit-0.0.10 → imaging_server_kit-0.0.12}/src/imaging_server_kit.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: imaging-server-kit
-Version: 0.0.10
+Version: 0.0.12
 Summary: Deploy image processing algorithms in FastAPI servers. 
 Author-email: Mallory Wittwer <mallory.wittwer@epfl.ch>
 License: Copyright (c) 2024, EPFL