Skip to content

Commit

Permalink
BUG : fix RGB FlateEncode Images(PNG) and transparency
Browse files Browse the repository at this point in the history 
Number of colors were not taken into account to process PNG Images

also properly process mask to transparency

closes py-pdf#1787
  • Loading branch information
@pubpub-zz
pubpub-zz committed May 6, 2023
1 parent 6fe1c30 commit ca44aec
Show file tree
Hide file tree
Showing 2 changed files with 61 additions and 22 deletions.
1 change: 1 addition & 0 deletions pypdf/constants.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -451,6 +451,7 @@ class GraphicsStateParameters:
SM = "/SM"
SA = "/SA"
BM = "/BM"
MASK = "/Mask" # 1-bit image mask stream
S_MASK = "/SMask" # dictionary or name, optional
CA = "/CA"
ca = "/ca"
Expand Down
82 changes: 60 additions & 22 deletions pypdf/filters.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -147,6 +147,7 @@ def decode(
columns = (
1 if decode_parms is None else decode_parms.get(LZW.COLUMNS, 1)
)
colors = 1 if decode_parms is None else decode_parms.get(LZW.COLORS, 1)
bits_per_component = (
decode_parms.get(LZW.BITS_PER_COMPONENT, DEFAULT_BITS_PER_COMPONENT)
if decode_parms
Expand All @@ -155,7 +156,7 @@ def decode(

# PNG predictor can vary by row and so is the lead byte on each row
rowlength = (
math.ceil(columns * bits_per_component / 8) + 1
math.ceil(columns * colors * bits_per_component / 8) + 1
) # number of bytes

# PNG prediction:
Expand All @@ -173,6 +174,7 @@ def _decode_png_prediction(data: str, columns: int, rowlength: int) -> bytes:
if len(data) % rowlength != 0:
raise PdfReadError("Image data is not rectangular")
prev_rowdata = (0,) * rowlength
bpp = (rowlength - 1) // columns # recomputed locally to not change params
for row in range(len(data) // rowlength):
rowdata = [
ord_(x) for x in data[(row * rowlength) : ((row + 1) * rowlength)]
Expand All @@ -182,21 +184,21 @@ def _decode_png_prediction(data: str, columns: int, rowlength: int) -> bytes:
if filter_byte == 0:
pass
elif filter_byte == 1:
for i in range(2, rowlength):
rowdata[i] = (rowdata[i] + rowdata[i - 1]) % 256
for i in range(bpp + 1, rowlength):
rowdata[i] = (rowdata[i] + rowdata[i - bpp]) % 256
elif filter_byte == 2:
for i in range(1, rowlength):
rowdata[i] = (rowdata[i] + prev_rowdata[i]) % 256
elif filter_byte == 3:
for i in range(1, rowlength):
left = rowdata[i - 1] if i > 1 else 0
left = rowdata[i - bpp] if i > bpp else 0
floor = math.floor(left + prev_rowdata[i]) / 2
rowdata[i] = (rowdata[i] + int(floor)) % 256
elif filter_byte == 4:
for i in range(1, rowlength):
left = rowdata[i - 1] if i > 1 else 0
left = rowdata[i - bpp] if i > bpp else 0
up = prev_rowdata[i]
up_left = prev_rowdata[i - 1] if i > 1 else 0
up_left = prev_rowdata[i - bpp] if i > bpp else 0
paeth = paeth_predictor(left, up, up_left)
rowdata[i] = (rowdata[i] + paeth) % 256
else:
Expand Down Expand Up @@ -647,31 +649,36 @@ def _xobj_to_image(x_object_obj: Dict[str, Any]) -> Tuple[Optional[str], bytes]:

size = (x_object_obj[IA.WIDTH], x_object_obj[IA.HEIGHT])
data = x_object_obj.get_data() # type: ignore
colors = x_object_obj.get("/Colors", 1)
color_space: Any = x_object_obj.get("/ColorSpace", NullObject()).get_object()
if (
IA.COLOR_SPACE in x_object_obj
and x_object_obj[IA.COLOR_SPACE] == ColorSpaces.DEVICE_RGB
):
# https://pillow.readthedocs.io/en/stable/handbook/concepts.html#modes
mode: Literal["1", "RGB", "P", "L", "RGBA"] = "RGB"
mode: Literal["1", "RGB", "P", "L", "RGBA", "CMYK"] = "RGB"
elif x_object_obj.get("/BitsPerComponent", 8) == 1:
mode = "1"
elif colors == 3:
mode = "RGB"
elif colors == 4:
mode = "CMYK"
# elif isinstance(colorspace,ArrayObject):
# logger_warning("ColorSpace Array not implemented; considered as RGB.\n"+
# "Please share your sample with pypdf dev team.", __name__)
# mode = "RGB"
elif "Gray" in str(color_space):
mode = "L"
else:
mode = "P"
extension = None
if SA.FILTER in x_object_obj:
if x_object_obj[SA.FILTER] == FT.FLATE_DECODE:
extension = ".png" # mime_type = "image/png"
color_space = None
if "/ColorSpace" in x_object_obj:
color_space = x_object_obj["/ColorSpace"].get_object()
if (
isinstance(color_space, ArrayObject)
and color_space[0] == "/Indexed"
):
color_space, base, hival, lookup = (
value.get_object() for value in color_space
)

if isinstance(color_space, ArrayObject) and color_space[0] == "/Indexed":
color_space, base, hival, lookup = (
value.get_object() for value in color_space
)
img = Image.frombytes(mode, size, data)
if color_space == "/Indexed":
from .generic import ByteStringObject
Expand All @@ -685,7 +692,10 @@ def _xobj_to_image(x_object_obj: Dict[str, Any]) -> Tuple[Optional[str], bytes]:
else:
img.putpalette(lookup.get_data())
img = img.convert("L" if base == ColorSpaces.DEVICE_GRAY else "RGB")
elif color_space is not None and color_space[0] == "/ICCBased":
elif (
not isinstance(color_space, NullObject)
and color_space[0] == "/ICCBased"
):
# see Table 66 - Additional Entries Specific to an ICC Profile
# Stream Dictionary
icc_profile = color_space[1].get_object()
Expand All @@ -695,17 +705,27 @@ def _xobj_to_image(x_object_obj: Dict[str, Any]) -> Tuple[Optional[str], bytes]:
mode_map = {
"/DeviceGray": "L",
"/DeviceRGB": "RGB",
"/DeviceCMYK": "RGBA",
"/DeviceCMYK": "CMYK", # used to be "RGBA" but this is seems not in accordance withFlateEncode Spec
}
mode = (
mode_map.get(color_space) # type: ignore
or {1: "L", 3: "RGB", 4: "RGBA"}.get(color_components)
or list(mode_map.values())[color_components]
or mode
) # type: ignore
img = Image.frombytes(mode, size, data)
alpha = None
if G.S_MASK in x_object_obj: # add alpha channel
alpha = Image.frombytes("L", size, x_object_obj[G.S_MASK].get_data())
elif G.MASK in x_object_obj: # add alpha channel
alpha = Image.frombytes("1", size, x_object_obj[G.MASK].get_data())
if alpha is not None:
scale = x_object_obj[G.S_MASK].get("/Decode", [0.0, 1.0])
if (scale[1] - scale[0]) != 1.0:
alpha = alpha.point(
lambda v: 255.0 * (v / 255.0 * (scale[1] - scale[0]) + scale[0])
)
img.putalpha(alpha)

img_byte_arr = BytesIO()
img.convert("RGBA").save(img_byte_arr, format="PNG")
data = img_byte_arr.getvalue()
Expand All @@ -723,7 +743,25 @@ def _xobj_to_image(x_object_obj: Dict[str, Any]) -> Tuple[Optional[str], bytes]:
extension = ".png" # mime_type = "image/png"
data = b_(data)
elif x_object_obj[SA.FILTER] == FT.DCT_DECODE:
extension = ".jpg" # mime_type = "image/jpeg"
img = Image.open(BytesIO(data))
alpha = None
if G.S_MASK in x_object_obj: # add alpha channel
alpha = Image.frombytes("L", size, x_object_obj[G.S_MASK].get_data())
elif G.MASK in x_object_obj: # add alpha channel
alpha = Image.frombytes("1", size, x_object_obj[G.MASK].get_data())
else:
extension = ".jpg" # mime_type = "image/jpeg"
if alpha is not None:
scale = x_object_obj[G.S_MASK].get("/Decode", [0.0, 1.0])
if (scale[1] - scale[0]) != 1.0:
alpha = alpha.point(
lambda v: 255.0 * (v / 255.0 * (scale[1] - scale[0]) + scale[0])
)
img.putalpha(alpha)
extension = ".jp2" # mime_type = "image/jp2"
img_byte_arr = BytesIO()
img.save(img_byte_arr, format="JPEG2000")
data = img_byte_arr.getvalue()
elif x_object_obj[SA.FILTER] == "/JPXDecode":
extension = ".jp2" # mime_type = "image/x-jp2"
elif x_object_obj[SA.FILTER] == FT.CCITT_FAX_DECODE:
Expand Down

0 comments on commit ca44aec

Please sign in to comment.