BUG: Avoid exceeding recursion depth when retrieving image mode (#2251)

pypdf/_xobj_image_helpers.py

@@ -34,17 +34,24 @@
     "", "1", "RGB", "2bits", "4bits", "P", "L", "RGBA", "CMYK"
 ]
 
+MAX_IMAGE_MODE_NESTING_DEPTH: int = 10
+
 
 def _get_imagemode(
     color_space: Union[str, List[Any], Any],
     color_components: int,
     prev_mode: mode_str_type,
+    depth: int = 0,
 ) -> Tuple[mode_str_type, bool]:
     """
     Returns
         Image mode not taking into account mask(transparency)
         ColorInversion is required (like for some DeviceCMYK)
     """
+    if depth > MAX_IMAGE_MODE_NESTING_DEPTH:
+        raise PdfReadError(
+            "Color spaces nested too deep. If required, consider increasing MAX_IMAGE_MODE_NESTING_DEPTH."
+        )
     if isinstance(color_space, NullObject):
         return "", False
     if isinstance(color_space, str):
@@ -63,22 +70,22 @@ def _get_imagemode(
         color_space = color_space[1]
         if isinstance(color_space, IndirectObject):
             color_space = color_space.get_object()
-        mode2, invert_color = _get_imagemode(color_space, color_components, prev_mode)
+        mode2, invert_color = _get_imagemode(color_space, color_components, prev_mode, depth + 1)
         if mode2 in ("RGB", "CMYK"):
             mode2 = "P"
         return mode2, invert_color
     elif color_space[0] == "/Separation":
         color_space = color_space[2]
         if isinstance(color_space, IndirectObject):
             color_space = color_space.get_object()
-        mode2, invert_color = _get_imagemode(color_space, color_components, prev_mode)
+        mode2, invert_color = _get_imagemode(color_space, color_components, prev_mode, depth + 1)
         return mode2, True
     elif color_space[0] == "/DeviceN":
         color_components = len(color_space[1])
         color_space = color_space[2]
         if isinstance(color_space, IndirectObject):  # pragma: no cover
             color_space = color_space.get_object()
-        mode2, invert_color = _get_imagemode(color_space, color_components, prev_mode)
+        mode2, invert_color = _get_imagemode(color_space, color_components, prev_mode, depth + 1)
         return mode2, invert_color
 
     mode_map = {

tests/test_xobject_image_helpers.py

@@ -0,0 +1,27 @@
+"""Test the pypdf._xobj_image_helpers module."""
+from io import BytesIO
+
+import pytest
+
+from pypdf import PdfReader
+from pypdf.errors import PdfReadError
+
+from . import get_data_from_url
+
+
+@pytest.mark.enable_socket()
+def test_get_imagemode_recursion_depth():
+    """Avoid infinite recursion for nested color spaces."""
+    url = "https://github.com/py-pdf/pypdf/files/12814018/out1.pdf"
+    name = "issue2240.pdf"
+    # Simple example: Just let the color space object reference itself.
+    # The alternative would be to generate a chain of referencing objects.
+    content = get_data_from_url(url, name=name)
+    source = b"\n10 0 obj\n[ /DeviceN [ /HKS#2044#20K /Magenta /Yellow /Black ] 7 0 R 11 0 R 12 0 R ]\nendobj\n"
+    target = b"\n10 0 obj\n[ /DeviceN [ /HKS#2044#20K /Magenta /Yellow /Black ] 10 0 R 11 0 R 12 0 R ]\nendobj\n"
+    reader = PdfReader(BytesIO(content.replace(source, target)))
+    with pytest.raises(
+            PdfReadError,
+            match="Color spaces nested too deep. If required, consider increasing MAX_IMAGE_MODE_NESTING_DEPTH."
+    ):
+        reader.pages[0].images[0]