fix(visualization): scale top y-edge by scale_y in draw_clusters

docling/utils/visualization.py

@@ -28,7 +28,7 @@ def draw_clusters(
                 cx0, cy0, cx1, cy1 = tc.rect.to_bounding_box().as_tuple()
                 cx0 *= scale_x
                 cx1 *= scale_x
-                cy0 *= scale_x
+                cy0 *= scale_y
                 cy1 *= scale_y
 
                 draw.rectangle(
@@ -40,7 +40,7 @@ def draw_clusters(
             x0, y0, x1, y1 = c.bbox.as_tuple()
             x0 *= scale_x
             x1 *= scale_x
-            y0 *= scale_x
+            y0 *= scale_y
             y1 *= scale_y
 
             if y1 <= y0:

tests/test_visualization_utils.py

@@ -0,0 +1,37 @@
+from docling_core.types.doc import BoundingBox, CoordOrigin, DocItemLabel
+from PIL import Image, ImageDraw
+
+from docling.datamodel.base_models import Cluster
+from docling.utils.visualization import draw_clusters
+
+
+def test_draw_clusters_scales_y_edges_with_scale_y(monkeypatch):
+    """The top and bottom cluster edges must both scale by scale_y.
+
+    Under non-uniform scaling (scale_x != scale_y) the top y-coordinate was
+    scaled by scale_x, misplacing the top edge of every drawn box. Capture the
+    coordinates handed to ImageDraw.rectangle and assert both y edges use
+    scale_y.
+    """
+    captured: list = []
+    original_rectangle = ImageDraw.ImageDraw.rectangle
+
+    def capturing_rectangle(self, xy, *args, **kwargs):
+        captured.append(xy)
+        return original_rectangle(self, xy, *args, **kwargs)
+
+    monkeypatch.setattr(ImageDraw.ImageDraw, "rectangle", capturing_rectangle)
+
+    bbox = BoundingBox(l=10, t=20, r=30, b=40, coord_origin=CoordOrigin.TOPLEFT)
+    cluster = Cluster(id=0, label=DocItemLabel.TEXT, bbox=bbox)
+    image = Image.new("RGB", (200, 300))
+
+    scale_x, scale_y = 2.0, 5.0
+    draw_clusters(image, [cluster], scale_x, scale_y)
+
+    # The first rectangle is the cluster box: [(x0, y0), (x1, y1)].
+    (x0, y0), (x1, y1) = captured[0]
+    assert x0 == bbox.l * scale_x
+    assert x1 == bbox.r * scale_x
+    assert y0 == bbox.t * scale_y
+    assert y1 == bbox.b * scale_y