Merge branch 'main' into feature/sidebar-ide-style

Author: MohamedAli1937

imagelab-backend/app/operators/conversions/gray_to_binary.py

@@ -6,7 +6,7 @@
 
 class GrayToBinary(BaseOperator):
     def compute(self, image: np.ndarray) -> np.ndarray:
-        threshold_value = float(self.params.get("thresholdValue", 0))
+        threshold_value = float(self.params.get("thresholdValue", 127))
         max_value = float(self.params.get("maxValue", 255))
         _, dst = cv2.threshold(image, threshold_value, max_value, cv2.THRESH_BINARY)
         return dst

imagelab-backend/app/services/pipeline_executor.py

@@ -43,8 +43,8 @@ def execute_pipeline(request: PipelineRequest) -> PipelineResponse:
             t_fail = time.perf_counter()
             return PipelineResponse(
                 success=False,
-                error=f"Unknown operator '{step.type}' at step {i}",
-                step=i,
+                error=f"Unknown operator '{step.type}' at step {i + 1}",
+                step=i + 1,
                 timings=PipelineTimings(total_ms=(t_fail - t_start_total) * 1000, steps=step_timings),
             )
 
@@ -60,8 +60,8 @@ def execute_pipeline(request: PipelineRequest) -> PipelineResponse:
             t_fail = time.perf_counter()
             return PipelineResponse(
                 success=False,
-                error=f"Error in step {i} ({step.type}): {type(e).__name__}: {e}",
-                step=i,
+                error=f"Error in step {i + 1} ({step.type}): {type(e).__name__}: {e}",
+                step=i + 1,
                 timings=PipelineTimings(total_ms=(t_fail - t_start_total) * 1000, steps=step_timings),
             )
 

imagelab-backend/tests/operators/filtering/test_contour_detection.py

@@ -50,9 +50,11 @@ def test_contour_detection_supports_gray_bgr_bgra(channels):
 
     result = op.compute(image.copy())
 
-    if channels == 4:
-        assert result.shape == (100, 100, 3)  # BGRA -> BGR conversion path
+    if channels == 1:
+        # Grayscale input is promoted to BGR so the contour can be drawn in colour.
+        assert result.shape == (100, 100, 3)
     else:
+        # BGR (3) and BGRA (4) inputs keep their channel count.
         assert result.shape == image.shape
 
 

imagelab-backend/tests/test_blurring_operators.py

@@ -1,4 +1,5 @@
 import numpy as np
+import pytest
 
 from app.operators.blurring.blur import Blur
 from app.operators.blurring.gaussian_blur import GaussianBlur
@@ -56,11 +57,10 @@ def test_custom_params_output_shape(self, color_image):
         result = GaussianBlur({"widthSize": 5, "heightSize": 5}).compute(color_image)
         assert result.shape == color_image.shape
 
-    def test_even_kernel_corrected_to_odd(self, color_image):
-        result_even = GaussianBlur({"widthSize": 4, "heightSize": 4}).compute(color_image)
-        result_odd = GaussianBlur({"widthSize": 5, "heightSize": 5}).compute(color_image)
-        assert result_even.shape == color_image.shape
-        np.testing.assert_array_equal(result_even, result_odd)
+    def test_even_kernel_raises_value_error(self, color_image):
+        # Validator rejects even kernels with a "Did you mean 3 or 5?" message.
+        with pytest.raises(ValueError, match="odd"):
+            GaussianBlur({"widthSize": 4, "heightSize": 4}).compute(color_image)
 
     def test_grayscale_input(self, grayscale_image):
         result = GaussianBlur({}).compute(grayscale_image)
@@ -102,11 +102,10 @@ def test_custom_params_output_shape(self, color_image):
         result = MedianBlur({"kernelSize": 5}).compute(color_image)
         assert result.shape == color_image.shape
 
-    def test_even_kernel_corrected_to_odd(self, color_image):
-        result_even = MedianBlur({"kernelSize": 4}).compute(color_image)
-        result_odd = MedianBlur({"kernelSize": 5}).compute(color_image)
-        assert result_even.shape == color_image.shape
-        np.testing.assert_array_equal(result_even, result_odd)
+    def test_even_kernel_raises_value_error(self, color_image):
+        # Validator rejects even kernels with a "Did you mean 3 or 5?" message.
+        with pytest.raises(ValueError, match="odd"):
+            MedianBlur({"kernelSize": 4}).compute(color_image)
 
     def test_grayscale_input(self, grayscale_image):
         result = MedianBlur({}).compute(grayscale_image)
@@ -129,9 +128,11 @@ def test_median_blur_removes_salt_pepper_noise(self):
             np.abs(result.astype(int) - clean.astype(int)).mean() < np.abs(noisy.astype(int) - clean.astype(int)).mean()
         )
 
-    def test_kernel_size_1_is_identity(self, color_image):
-        result = MedianBlur({"kernelSize": 1}).compute(color_image)
-        np.testing.assert_array_equal(result, color_image)
+    def test_kernel_size_1_rejected(self, color_image):
+        # MedianBlur's validator deliberately forbids ksize=1 (a no-op) even though
+        # OpenCV would accept it. See app/operators/blurring/validation.py.
+        with pytest.raises(ValueError, match=">= 3"):
+            MedianBlur({"kernelSize": 1}).compute(color_image)
 
     def test_small_image_no_crash(self):
         img = np.zeros((3, 3, 3), dtype=np.uint8)

imagelab-backend/tests/test_pipeline_executor.py

@@ -45,8 +45,8 @@ def test_multi_step_pipeline(make_request, sample_image_b64):
 def test_unknown_operator_gives_clear_error(make_request):
     res = execute_pipeline(make_request([PipelineStep(type="not_a_real_op")]))
     assert res.success is False
-    assert res.step == 0
-    assert "at step 0" in res.error
+    assert res.step == 1
+    assert "at step 1" in res.error
     assert "not_a_real_op" in res.error
     assert "Unknown operator" in res.error
 
@@ -58,8 +58,8 @@ def test_error_includes_correct_step_index(make_request):
     ]
     res = execute_pipeline(make_request(steps))
     assert res.success is False
-    assert res.step == 1  # first step succeeds, second should fail
-    assert "at step 1" in res.error
+    assert res.step == 2  # first step succeeds, second (1-indexed: 2) should fail
+    assert "at step 2" in res.error
     assert "bad_operator_step_one" in res.error
 
 

imagelab-backend/tests/test_thresholding_operators.py

@@ -1,5 +1,4 @@
 import numpy as np
-import pytest
 
 from app.operators.thresholding.adaptive_threshold import AdaptiveThreshold
 from app.operators.thresholding.apply_threshold import ApplyThreshold
@@ -15,9 +14,6 @@ def test_output_is_uint8(self, color_image):
         result = ApplyThreshold({}).compute(color_image)
         assert result.dtype == np.uint8
 
-    @pytest.mark.xfail(
-        strict=True, reason="maxValue defaults to 0 — known bug, fix in fix/apply-threshold-default-max-value"
-    )
     def test_default_params_produces_non_empty_output(self, color_image):
         result = ApplyThreshold({}).compute(color_image)
         assert result.max() > 0

imagelab-frontend/src/hooks/useBlocklyWorkspace.ts

@@ -1,4 +1,3 @@
-/* eslint-disable react-hooks/preserve-manual-memoization */
 import { useRef, useEffect, useState, useCallback } from "react";
 import * as Blockly from "blockly";
 
@@ -150,6 +149,11 @@ export function useBlocklyWorkspace({ isDark = false }: UseBlocklyWorkspaceOptio
       resizeObserverRef.current = observer;
     }
     updateBlockStats(ws); // Initial stats calculation if any blocks loaded
+    // isDark is intentionally omitted from deps: initWorkspace is a one-shot
+    // initializer (guarded by workspaceRef.current). Live theme toggles are
+    // handled by the setTheme useEffect above, so adding isDark here would
+    // dispose and recreate the workspace on every toggle, losing user blocks.
+    // eslint-disable-next-line react-hooks/exhaustive-deps
   }, [setSelectedBlock, updateBlockStats]);
 
   useEffect(() => {