def _readCode(self):
toread = self.bitspercode
value = 0
while toread > 0:
if self.bytepos >= len(self.data):
return -1
nextbits = pypdfOrd(self.data[self.bytepos])
bitsfromhere = 8 - self.bitpos
if bitsfromhere > toread:
bitsfromhere = toread
value |= ((nextbits >> (8 - self.bitpos - bitsfromhere)) &
(0xFF >>
(8 - bitsfromhere))) << (toread - bitsfromhere)
toread -= bitsfromhere
self.bitpos += bitsfromhere
if self.bitpos >= 8:
self.bitpos = 0
self.bytepos = self.bytepos + 1
return value
def encode(data, decodeParms=None):
"""
Encodes chunks of 4-byte sequences of textual or bytes data according
to the base-85 ASCII encoding algorithm.
:param data: a str or byte sequence of values.
:return: ASCII85-encoded data in bytes format (equal to str in Python
2).
"""
if sys.version_info[0] < 3:
result = str()
filler = "\x00" if type(data) is str else b"\x00"
if type(data) not in (str, bytes):
raise TypeError(
"Expected str or bytes type for data, got %s instead" %
type(data))
for group in range(int(math.ceil(len(data) / 4.0))):
decimalRepr = 0
ascii85 = str()
groupWidth = min(4, len(data) - 4 * group)
if groupWidth < 4:
data = data + (4 - groupWidth) * filler
for byte in range(4):
decimalRepr += pypdfOrd(
data[4 * group + byte]) << 8 * (4 - byte - 1)
# If all bytes are 0, we turn them into a single 'z' character
if decimalRepr == 0 and groupWidth == 4:
ascii85 = "z"
else:
for i in range(5):
ascii85 = chr(decimalRepr % 85 + 33) + ascii85
decimalRepr = int(decimalRepr / 85.0)
# In case of a partial group of four bytes, the standard says:
# «Finally, it shall write only the first n + 1 characters of the
# resulting group of 5.» - ISO 32000 (2008), sec. 7.4.3
result += ascii85[:min(5, groupWidth + 1)]
return ("<~" + result + "~>").encode("LATIN1")
else:
return base64.a85encode(data, adobe=True)
def decode(data, decodeParms=None):
"""
:param data: flate-encoded data.
:param decodeParms: a dictionary of parameter values.
:return: the flate-decoded data.
:rtype: bytes
"""
data = decompress(data)
predictor = 1
if decodeParms:
try:
predictor = decodeParms.get("/Predictor", 1)
except AttributeError:
pass # Usually an array with a null object was read
# predictor 1 == no predictor
if predictor != 1:
# The /Columns param. has 1 as the default value; see ISO 32000,
# §7.4.4.3 LZWDecode and FlateDecode Parameters, Table 8
columns = decodeParms.get("/Columns", 1)
# PNG prediction:
if 10 <= predictor <= 15:
output = BytesIO()
# PNG prediction can vary from row to row
row_length = columns + 1
assert len(data) % row_length == 0
prev_rowdata = (0, ) * row_length
for row in range(len(data) // row_length):
rowdata = [
pypdfOrd(x)
for x in data[(row * row_length):((row + 1) *
row_length)]
]
filterByte = rowdata[0]
if filterByte == 0:
pass
elif filterByte == 1:
for i in range(2, row_length):
rowdata[i] = (rowdata[i] + rowdata[i - 1]) % 256
elif filterByte == 2:
for i in range(1, row_length):
rowdata[i] = (rowdata[i] + prev_rowdata[i]) % 256
elif filterByte == 3:
for i in range(1, row_length):
left = rowdata[i - 1] if i > 1 else 0
floor = math.floor(left + prev_rowdata[i]) / 2
rowdata[i] = (rowdata[i] + int(floor)) % 256
elif filterByte == 4:
for i in range(1, row_length):
left = rowdata[i - 1] if i > 1 else 0
up = prev_rowdata[i]
up_left = prev_rowdata[i - 1] if i > 1 else 0
paeth = paethPredictor(left, up, up_left)
rowdata[i] = (rowdata[i] + paeth) % 256
else:
# Unsupported PNG filter
raise PdfReadError("Unsupported PNG filter %r" %
filterByte)
prev_rowdata = rowdata
for d in rowdata:
if version_info < (3, 0):
output.write(chr(d))
else:
output.write(bytes([d]))
data = output.getvalue()
else:
# unsupported predictor
raise PdfReadError("Unsupported flatedecode predictor %r" %
predictor)
return data
def decode(data, decodeParms=None):
"""
:param data: flate-encoded data.
:param decodeParms: a dictionary of parameter values.
:return: the flate-decoded data.
:rtype: bytes
"""
data = decompress(data)
predictor = 1
if decodeParms:
try:
predictor = decodeParms.get("/Predictor", 1)
except AttributeError:
pass # Usually an array with a null object was read
# predictor 1 == no predictor
if predictor != 1:
# The /Columns param. has 1 as the default value; see ISO 32000,
# §7.4.4.3 LZWDecode and FlateDecode Parameters, Table 8
columns = decodeParms.get("/Columns", 1)
colors = decodeParms.get("/Colors", 8)
# PNG prediction:
if 10 <= predictor <= 15:
output = BytesIO(
) ### BytesIO au lieu de StringIO, car PIL attend des bytes (Pas de mon fait)
# PNG prediction can vary from row to row
row_length = columns * colors + 1
assert len(data) % row_length == 0
prev_rowdata = (0, ) * row_length
for row in range(len(data) // row_length):
rowdata = [
pypdfOrd(x)
for x in data[(row * row_length):((row + 1) *
row_length)]
]
filterByte = rowdata[0]
if filterByte == 0:
pass
elif filterByte == 1:
for i in range(1, row_length):
if i <= 3: ### On vise hors de la limite gauche, donc 0 (as per les specs)
rowdata[i] = (rowdata[i] + 0) % 256
else:
rowdata[i] = (
rowdata[i] + rowdata[i - colors]
) % 256 ### Pas le byte d'avant ! Son homologue dans le pixel d'avant
### (TODO : vérifier si le bitdepth < 8, auquel cas il faut
### bel et bien prendre simplement le byte d'avant)
elif filterByte == 2:
for i in range(1, row_length):
rowdata[i] = (rowdata[i] + prev_rowdata[i]) % 256
elif filterByte == 3:
for i in range(1, row_length):
left = rowdata[i - 1] if i > 1 else 0
floor = math.floor(left + prev_rowdata[i]) / 2
rowdata[i] = (rowdata[i] + int(floor)) % 256
elif filterByte == 4:
for i in range(1, row_length):
left = rowdata[i - 1] if i > 1 else 0
up = prev_rowdata[i]
up_left = prev_rowdata[i - 1] if i > 1 else 0
paeth = paethPredictor(left, up, up_left)
rowdata[i] = (rowdata[i] + paeth) % 256
else:
# Unsupported PNG filter
raise PdfReadError("Unsupported PNG filter %r" %
filterByte)
prev_rowdata = rowdata
for d in rowdata[
1:]: ### Ignorer le premier bit de chaque scanline
if version_info < (3, 0):
output.write(chr(d))
else:
output.write(bytes([d]))
data = output.getvalue()
else:
# unsupported predictor
raise PdfReadError("Unsupported flatedecode predictor %r" %
predictor)
return data
