ENH: add decode_as_image() to ContentStreams

closes py-pdf#2613

pypdf/generic/_data_structures.py

@@ -947,6 +947,26 @@ def flate_encode(self, level: int = -1) -> "EncodedStreamObject":
  retval._data = FlateDecode.encode(b_(self._data), level)
  return retval
 
+ def decode_as_image(self) -> Any:
+ """
+ Try to decode the stream object as an image
+
+ Returns:
+ a PIL image if proper decoding has been found
+ """
+ from ..filters import _xobj_to_image
+
+ if self.get("/Subtype", "") != "/Image":
+ try:
+ msg = f"{self.indirect_reference} does not seems to be an Image" # pragma: no cover
+ except AttributeError:
+ msg = f"{self.__repr__()} object does not seems to be an Image" # pragma: no cover
+ logger_warning(msg, __name__)
+ extension, byte_stream, img = _xobj_to_image(self)
+ if extension is None:
+ return None # pragma: no cover
+ return img
+
 
 class DecodedStreamObject(StreamObject):
  pass

tests/test_images.py

@@ -346,3 +346,24 @@ def test_corrupted_jpeg_iss2266(pdf, pdf_name, images, images_name, filtr):
  print(fn) # noqa: T201
  img = Image.open(BytesIO(zf.read(fn)))
  assert image_similarity(reader.pages[p].images[i].image, img) >= 0.99
+
+
+@pytest.mark.enable_socket()
+def test_extract_image_from_object(caplog):
+ url = "https://github.com/py-pdf/pypdf/files/15176076/B2.pdf"
+ name = "iss2613.pdf"
+ reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
+ image = reader.pages[0]["/Resources"]["/Pattern"]["/P1"]["/Resources"]["/XObject"][
+ "/X1"
+ ].decode_as_image()
+ assert isinstance(image, Image.Image)
+ with pytest.raises(Exception):
+ co = reader.pages[0].get_contents()
+ co.decode_as_image()
+ assert "does not seems to be an Image" in caplog.text
+ caplog.clear()
+ co.indirect_reference = "for_test"
+ with pytest.raises(Exception):
+ co = reader.pages[0].get_contents()
+ co.decode_as_image()
+ assert "does not seems to be an Image" in caplog.text