pad truncated data in bits2byte instead of reading out of bounds

Author: metsw24-max

pypdf/generic/_image_xobject.py

@@ -131,6 +131,12 @@ def bits2byte(data: bytes, size: tuple[int, int], bits: int) -> bytes:
 
     byte_buffer = bytearray(buffer_size)
     mask = (1 << bits) - 1
+
+    required = size[1] * ((size[0] * bits + 7) // 8)
+    if len(data) < required:
+        logger_warning("Image data is not rectangular. Adding padding.", source=__name__)
+        data += b"\x00" * (required - len(data))
+
     data_index = 0
     bit = 8 - bits
     for y in range(size[1]):

tests/generic/test_image_xobject.py

@@ -276,3 +276,10 @@ def test_bits2byte__limit() -> None:
             match=r"^Requested buffer size 76500000 exceeds limit of 75000000\.$"
     ):
         bits2byte(data=b"TEST", size=(9000, 8500), bits=8)
+
+
+def test_bits2byte__truncated_data(caplog) -> None:
+    # 4x4 image at 2 bits per sample needs 4 bytes; provide only 1.
+    result = bits2byte(data=b"\x00", size=(4, 4), bits=2)
+    assert result == bytes(16)
+    assert "Image data is not rectangular. Adding padding." in caplog.text