ROB: cope with some images extractions issues

closes #2343:
1st case : image with images in 1 byte encoding with Separation colorspace

2nd case: similar + \n to be ignored at the end of the image data

pypdf/_xobj_image_helpers.py

@@ -125,6 +125,20 @@ def _get_imagemode(
     return mode, mode == "CMYK"
 
 
+def _extended_image_frombytes(
+    mode: str, size: Tuple[int, int], data: bytes
+) -> Any:  # Any = ImageType
+    try:
+        img = Image.frombytes(mode, size, data)
+    except ValueError as exc:
+        if len(data) % (size[0] * size[1]) != 0:
+            raise exc
+        k = size[0] * size[1] * len(mode) / len(data)
+        data = b"".join([bytes((x,) * int(k)) for x in data])
+        img = Image.frombytes(mode, size, data)
+    return img
+
+
 def _handle_flate(
     size: Tuple[int, int],
     data: bytes,
@@ -168,7 +182,7 @@ def bits2byte(data: bytes, size: Tuple[int, int], bits: int) -> bytes:
     elif mode == "4bits":
         mode = "P"
         data = bits2byte(data, size, 4)
-    img = Image.frombytes(mode, size, data)
+    img = _extended_image_frombytes(mode, size, data)
     if color_space == "/Indexed":
         from .generic import TextStringObject
 

pypdf/filters.py

@@ -285,7 +285,9 @@ def decode(
         index = 0
         while True:
             if index >= len(data):
-                logger_warning("missing EOD in ASCIIHexDecode, check if output is OK", __name__)
+                logger_warning(
+                    "missing EOD in ASCIIHexDecode, check if output is OK", __name__
+                )
                 break  # reach End Of String even if no EOD
             char = data[index : index + 1]
             if char == b">":
@@ -341,7 +343,9 @@ def decode(
         index = 0
         while True:
             if index >= len(data):
-                logger_warning("missing EOD in RunLengthDecode, check if output is OK", __name__)
+                logger_warning(
+                    "missing EOD in RunLengthDecode, check if output is OK", __name__
+                )
                 break  # reach End Of String even if no EOD
             length = data[index]
             index += 1
@@ -733,6 +737,7 @@ def _xobj_to_image(x_object_obj: Dict[str, Any]) -> Tuple[Optional[str], bytes,
     """
     from ._xobj_image_helpers import (
         Image,
+        _extended_image_frombytes,
         _get_imagemode,
         _handle_flate,
         _handle_jpx,
@@ -747,10 +752,12 @@ def _xobj_to_image(x_object_obj: Dict[str, Any]) -> Tuple[Optional[str], bytes,
     else:
         obj_as_text = x_object_obj.__repr__()
 
-    size = (x_object_obj[IA.WIDTH], x_object_obj[IA.HEIGHT])
+    size = (cast(int, x_object_obj[IA.WIDTH]), cast(int, x_object_obj[IA.HEIGHT]))
     data = x_object_obj.get_data()  # type: ignore
     if isinstance(data, str):  # pragma: no cover
         data = data.encode()
+    if len(data) % (size[0] * size[1]) == 1 and data[-1] == 0x0A:  # ie. '\n'
+        data = data[:-1]
     colors = x_object_obj.get("/Colors", 1)
     color_space: Any = x_object_obj.get("/ColorSpace", NullObject()).get_object()
     if isinstance(color_space, list) and len(color_space) == 1:
@@ -819,7 +826,7 @@ def _xobj_to_image(x_object_obj: Dict[str, Any]) -> Tuple[Optional[str], bytes,
         )
     elif mode == "CMYK":
         img, image_format, extension, invert_color = (
-            Image.frombytes(mode, size, data),
+            _extended_image_frombytes(mode, size, data),
             "TIFF",
             ".tif",
             False,
@@ -828,7 +835,7 @@ def _xobj_to_image(x_object_obj: Dict[str, Any]) -> Tuple[Optional[str], bytes,
         raise PdfReadError(f"ColorSpace field not found in {x_object_obj}")
     else:
         img, image_format, extension, invert_color = (
-            Image.frombytes(mode, size, data),
+            _extended_image_frombytes(mode, size, data),
             "PNG",
             ".png",
             False,
@@ -849,6 +856,11 @@ def _xobj_to_image(x_object_obj: Dict[str, Any]) -> Tuple[Optional[str], bytes,
         and color_space[0].get_object() == "/Indexed"
     ):
         decode = None  # decode is meanless of Indexed
+    if (
+        isinstance(color_space, ArrayObject)
+        and color_space[0].get_object() == "/Separation"
+    ):
+        decode = [1.0, 0.0] * len(img.getbands())
     if decode is not None and not all(decode[i] == i % 2 for i in range(len(decode))):
         lut: List[int] = []
         for i in range(0, len(decode), 2):

tests/test_images.py

@@ -255,3 +255,35 @@ def test_cmyk_no_filter():
     name = "iss2522.pdf"
     reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
     reader.pages[0].images[0].image
+
+
+@pytest.mark.enable_socket()
+def test_separation_1byte_to_rgb_inverted():
+    """Cf #2343"""
+    url = "https://github.com/py-pdf/pypdf/files/13679585/test2_P038-038.pdf"
+    name = "iss2343.pdf"
+    reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
+    url = (
+        "https://private-user-images.githubusercontent.com/4083478/"
+        "320313822-b7f41897-96ef-4ea6-b165-5ef307a92b87.png?"
+        "jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.jlaFEusLfEElwynVjtUk5VtMB1bZAjKfPJdYX3L_gBY"
+    )
+    name = "iss2343.png"
+    img = Image.open(BytesIO(get_data_from_url(url, name=name)))
+    assert image_similarity(reader.pages[0].images[0].image, img) >= 0.99
+
+
+@pytest.mark.enable_socket()
+def test_data_with_lf():
+    """Cf #2343"""
+    url = "https://github.com/py-pdf/pypdf/files/13946477/panda.pdf"
+    name = "iss2343b.pdf"
+    reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
+    url = (
+        "https://private-user-images.githubusercontent.com/4083478/"
+        "320594409-1120b0cf-a67a-403f-aa1a-9a191cbc087f.png?"
+        "jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MTI1OTc5MTAsIm5iZiI6MTcxMjU5NzYxMCwicGF0aCI6Ii80MDgzNDc4LzMyMDU5NDQwOS0xMTIwYjBjZi1hNjdhLTQwM2YtYWExYS05YTE5MWNiYzA4N2YucG5nP1gtQW16LUFsZ29yaXRobT1BV1M0LUhNQUMtU0hBMjU2JlgtQW16LUNyZWRlbnRpYWw9QUtJQVZDT0RZTFNBNTNQUUs0WkElMkYyMDI0MDQwOCUyRnVzLWVhc3QtMSUyRnMzJTJGYXdzNF9yZXF1ZXN0JlgtQW16LURhdGU9MjAyNDA0MDhUMTczMzMwWiZYLUFtei1FeHBpcmVzPTMwMCZYLUFtei1TaWduYXR1cmU9YWMxY2U0NzU1NWVkYzQwNWViNmYyOTFjOWQ0ZTkyN2Q3NmM1NTg4MGM4ZGZiYzRhYjg4MmVkNjRiZDcxZGRkMSZYLUFtei1TaWduZWRIZWFkZXJzPWhvc3QmYWN0b3JfaWQ9MCZrZXlfaWQ9MCZyZXBvX2lkPTAifQ.mDhPeMtRdV2ut1683_5JTJfxfb-wYSFGbWITR5RnRNw"
+    )
+    name = "iss2343b0.png"
+    img = Image.open(BytesIO(get_data_from_url(url, name=name)))
+    assert image_similarity(reader.pages[8].images[9].image, img) >= 0.99