chore: add unit tests

Dockerfile

@@ -1,8 +1,6 @@
 FROM pytorchlightning/pytorch_lightning:1.6.4-py3.9-torch1.9
 LABEL org.opencontainers.image.source=https://github.com/Greenroom-Robotics/nuclio_lighting_flash
 
-COPY ./nuclio_lighting_flash /opt/nuclio
-
 # Remote nvidia lists as they have a borked GPG
 RUN rm /etc/apt/sources.list.d/nvidia-ml.list /etc/apt/sources.list.d/cuda.list
 
@@ -13,6 +11,10 @@ RUN apt-get install ffmpeg libsm6 libxext6 -y
 # Install lighting flash and it's deps
 RUN pip install lightning-flash icevision 'lightning-flash[image]'
 
+COPY ./nuclio_lighting_flash /opt/nuclio
+
 WORKDIR /opt/nuclio
 
-CMD python ./test_flash_model_handler.py
+# Run the tests
+# Nuclio will overwrite this CMD when it is deployed
+CMD pytest .

nuclio_lighting_flash/fixtures/fixtures_nuclio.py

@@ -0,0 +1,38 @@
+import base64
+from dataclasses import dataclass
+
+
+class Logger:
+    def info(self, message: str):
+        ...
+
+
+class UserData:
+    model = None
+
+
+@dataclass
+class Response:
+    body: str
+    headers: dict
+    content_type: str
+    status_code: int
+
+
+class Context:
+    logger = Logger()
+    user_data = UserData()
+    Response = Response
+
+
+class Event:
+    def __init__(self, image_path: str, threshold: float):
+        """
+        Create an event with the image body serialised as base64
+        """
+        with open(image_path, "rb") as image_file:
+            image_base64 = base64.b64encode(image_file.read()).decode("ascii")
+            self.body = {
+                "image": image_base64,
+                "threshold": threshold,
+            }

nuclio_lighting_flash/test_flash_model_handler.py

@@ -4,11 +4,18 @@
 
 from flash_model_handler import FlashModelHandler
 
-model = ObjectDetector(
-    head="efficientdet", backbone="d0", num_classes=91, image_size=1024
-)
-model_handler = FlashModelHandler(model=model, image_size=1024, labels={25: "giraffe"})
-image = Image.open(os.path.join(os.getcwd(), "./fixtures/giraffe.jpg"))
-
-result = model_handler.infer(image, 0)
-print(result)
+
+def test_flash_model_handler():
+    """
+    Runs the flash model handler (bypassing nuclio)
+    Should detect giraffes in the COCO image
+    """
+    model = ObjectDetector(head="efficientdet", backbone="d0", num_classes=91, image_size=1024)
+    model_handler = FlashModelHandler(model=model, image_size=1024, labels={25: "giraffe"})
+    image = Image.open(os.path.join(os.getcwd(), "./fixtures/giraffe.jpg"))
+
+    result = model_handler.infer(image, 0.0)
+
+    assert len(result) == 2
+    assert result[0]["label"] == "giraffe"
+    assert result[1]["label"] == "giraffe"

nuclio_lighting_flash/test_main.py

@@ -0,0 +1,37 @@
+import json
+import os
+
+from main import init_context, handler
+from fixtures.fixtures_nuclio import Context, Event
+
+
+def test_main():
+    """
+    Runs main nuclio function with a fake nuclio context
+    """
+    context = Context()
+    init_context(context)
+
+    event = Event(image_path=os.path.join(os.getcwd(), "./fixtures/giraffe.jpg"), threshold=0.5)
+    response = handler(context, event)
+    assert response.body == json.dumps(
+        [
+            {
+                "confidence": 0.9171872138977051,
+                "label": "giraffe",
+                "points": [596.2103271484375, 276.046875, 962.4700927734375, 759.198974609375],
+                "type": "rectangle",
+            },
+            {
+                "confidence": 0.8744097352027893,
+                "label": "giraffe",
+                "points": [
+                    84.43610382080078,
+                    734.1155395507812,
+                    297.03192138671875,
+                    840.5248413085938,
+                ],
+                "type": "rectangle",
+            },
+        ]
+    )