BUG: Avoid accessing invalid page when inserting blank page under some conditions

pypdf/_writer.py

@@ -699,10 +699,8 @@ def insert_blank_page(
         If no page size is specified for a dimension, use the size of the last page.
 
         Args:
-            width: The width of the new page expressed in default user
-                space units.
-            height: The height of the new page expressed in default
-                user space units.
+            width: The width of the new page in default user space units.
+            height: The height of the new page in default user space units.
             index: Position to add the page.
 
         Returns:
@@ -711,12 +709,20 @@ def insert_blank_page(
         Raises:
             PageSizeNotDefinedError: if width and height are not defined
                 and previous page does not exist.
+            IndexError: Index is outside of [-self.get_num_pages(), self.get_num_pages()]
+        """
+        num_pages = self.get_num_pages()
+        if abs(index) <= num_pages:
+            # Use the chosen index, but do not exceed the available pages
+            fixed_index = min(index, num_pages - 1)
+            mediabox = self.pages[fixed_index].mediabox
+            if width is None or width <= 0:
+                width = mediabox.width
+            if height is None or height <= 0:
+                height = mediabox.height
+        else:
+            raise IndexError(f"Index should be in range [-{num_pages}, {num_pages}]")
 
-        """
-        if width is None or (height is None and index < self.get_num_pages()):
-            oldpage = self.pages[index]
-            width = oldpage.mediabox.width
-            height = oldpage.mediabox.height
         page = PageObject.create_blank_page(self, width, height)
         self.insert_page(page, index)
         return page

tests/test_writer.py

@@ -234,6 +234,10 @@ def writer_operate(writer: PdfWriter) -> None:
     writer.page_mode = NameObject("/UseOC")
     assert writer._get_page_mode() == "/UseOC"
     writer.insert_blank_page(width=100, height=100)
+    page = writer.insert_blank_page(width=100)
+    assert page.mediabox.height == 100
+    page = writer.insert_blank_page(height=100)
+    assert page.mediabox.width == 100
     writer.insert_blank_page()  # without parameters
 
     writer.remove_images()
@@ -251,6 +255,88 @@ def writer_operate(writer: PdfWriter) -> None:
         assert k in objects_hash, f"Missing {v}"
 
 
+def test_insert_blank_page():
+    writer = PdfWriter(clone_from=RESOURCE_ROOT / "crazyones.pdf")
+
+    old_page_count = len(writer.pages)
+
+    old_page = writer.pages[0]
+    page = writer.insert_blank_page(index=0)
+    assert len(writer.pages) == old_page_count + 1
+    assert page.mediabox.width == old_page.mediabox.width
+    assert page.mediabox.height == old_page.mediabox.height
+
+    old_page = writer.pages[0]
+    page = writer.insert_blank_page(width=10, index=0)
+    assert len(writer.pages) == old_page_count + 2
+    assert page.mediabox.width == 10
+    assert page.mediabox.height == old_page.mediabox.height
+
+    old_page = writer.pages[0]
+    page = writer.insert_blank_page(width=-10, index=0)
+    assert len(writer.pages) == old_page_count + 3
+    assert page.mediabox.width == old_page.mediabox.width
+    assert page.mediabox.height == old_page.mediabox.height
+
+    old_page = writer.pages[0]
+    page = writer.insert_blank_page(height=20, index=0)
+    assert len(writer.pages) == old_page_count + 4
+    assert page.mediabox.width == old_page.mediabox.width
+    assert page.mediabox.height == 20
+
+    old_page = writer.pages[0]
+    page = writer.insert_blank_page(height=-20, index=0)
+    assert len(writer.pages) == old_page_count + 5
+    assert page.mediabox.width == old_page.mediabox.width
+    assert page.mediabox.height == old_page.mediabox.height
+
+    page = writer.insert_blank_page(width=30, height=40, index=0)
+    assert len(writer.pages) == old_page_count + 6
+    assert page.mediabox.width == 30
+    assert page.mediabox.height == 40
+
+    old_page = writer.pages[0]
+    page = writer.insert_blank_page(width=-30, height=-40, index=0)
+    assert len(writer.pages) == old_page_count + 7
+    assert page.mediabox.width == old_page.mediabox.width
+    assert page.mediabox.height == old_page.mediabox.height
+
+    page = writer.insert_blank_page(width=50, height=60, index=len(writer.pages))
+    assert len(writer.pages) == old_page_count + 8
+    assert page.mediabox.width == 50
+    assert page.mediabox.height == 60
+
+    old_page = writer.pages[0]
+    page = writer.insert_blank_page(width=-50, height=-60, index=-len(writer.pages))
+    assert len(writer.pages) == old_page_count + 9
+    assert page.mediabox.width == old_page.mediabox.width
+    assert page.mediabox.height == old_page.mediabox.height
+
+    page = writer.insert_blank_page(width=70, height=80, index=len(writer.pages) // 2)
+    assert len(writer.pages) == old_page_count + 10
+    assert page.mediabox.width == 70
+    assert page.mediabox.height == 80
+
+    page = writer.insert_blank_page(width=70, height=80, index=-len(writer.pages) // 2)
+    assert len(writer.pages) == old_page_count + 11
+    assert page.mediabox.width == 70
+    assert page.mediabox.height == 80
+
+    num_pages = len(writer.pages)
+
+    with pytest.raises(
+        IndexError,
+        match=re.escape(f"Index should be in range [-{num_pages}, {num_pages}]"),
+    ):
+        page = writer.insert_blank_page(width=90, height=100, index=len(writer.pages) + 1)
+
+    with pytest.raises(
+        IndexError,
+        match=re.escape(f"Index should be in range [-{num_pages}, {num_pages}]"),
+    ):
+        page = writer.insert_blank_page(width=-90, height=-100, index=-len(writer.pages) - 1)
+
+
 @pytest.mark.parametrize(
     ("convert", "needs_cleanup"),
     [