def encryptionkey(password, OwnerKey, Permissions, FileId1, revision=2):
# FileId1 is first string of the fileid array
# add padding string
#AR force same as iText example
#Permissions = -1836 #int(Permissions - 2**31)
password = asBytes(password) + PadString
# truncate to 32 bytes
password = password[:32]
# translate permissions to string, low order byte first
p = Permissions # + 2**32L
permissionsString = b""
for i in range(4):
byte = (p & 0xff) # seems to match what iText does
p = p >> 8
permissionsString += int2Byte(byte % 256)
hash = md5(asBytes(password))
hash.update(asBytes(OwnerKey))
hash.update(asBytes(permissionsString))
hash.update(asBytes(FileId1))
md5output = hash.digest()
if revision == 2:
key = md5output[:5]
elif revision == 3: #revision 3 algorithm - loop 50 times
for x in range(50):
md5output = md5(md5output).digest()
key = md5output[:16]
if DEBUG:
print('encryptionkey(%s,%s,%s,%s,%s)==>%s' % tuple([
hexText(str(x))
for x in (password, OwnerKey, Permissions, FileId1, revision, key)
]))
return key
def computeO(userPassword, ownerPassword, revision):
from reportlab.lib.arciv import ArcIV
#print 'digest of hello is %s' % md5('hello').digest()
assert revision in (2,3), 'Unknown algorithm revision %s' % revision
if not ownerPassword:
ownerPassword = userPassword
ownerPad = asBytes(ownerPassword) + PadString
ownerPad = ownerPad[0:32]
password = asBytes(userPassword) + PadString
userPad = password[:32]
digest = md5(ownerPad).digest()
if DEBUG: print('PadString=%s\nownerPad=%s\npassword=%s\nuserPad=%s\ndigest=%s\nrevision=%s' % (ascii(PadString),ascii(ownerPad),ascii(password),ascii(userPad),ascii(digest),revision))
if revision == 2:
O = ArcIV(digest[:5]).encode(userPad)
elif revision == 3:
for i in range(50):
digest = md5(digest).digest()
digest = digest[:16]
O = userPad
for i in range(20):
thisKey = xorKey(i, digest)
O = ArcIV(thisKey).encode(O)
if DEBUG: print('computeO(%s,%s,%s)==>%s' % tuple([hexText(str(x)) for x in (userPassword, ownerPassword, revision,O)]))
return O
def computeO(userPassword, ownerPassword, revision):
from reportlab.lib.arciv import ArcIV
#print 'digest of hello is %s' % md5('hello').digest()
assert revision in (2, 3), 'Unknown algorithm revision %s' % revision
if not ownerPassword:
ownerPassword = userPassword
ownerPad = asBytes(ownerPassword) + PadString
ownerPad = ownerPad[0:32]
password = asBytes(userPassword) + PadString
userPad = password[:32]
digest = md5(ownerPad).digest()
if DEBUG:
print(
'PadString=%s\nownerPad=%s\npassword=%s\nuserPad=%s\ndigest=%s\nrevision=%s'
% (ascii(PadString), ascii(ownerPad), ascii(password),
ascii(userPad), ascii(digest), revision))
if revision == 2:
O = ArcIV(digest[:5]).encode(userPad)
elif revision == 3:
for i in range(50):
digest = md5(digest).digest()
digest = digest[:16]
O = userPad
for i in range(20):
thisKey = xorKey(i, digest)
O = ArcIV(thisKey).encode(O)
if DEBUG:
print('computeO(%s,%s,%s)==>%s' % tuple([
hexText(str(x)) for x in (userPassword, ownerPassword, revision, O)
]))
return O
def encryptionkey(password, OwnerKey, Permissions, FileId1, revision=2):
# FileId1 is first string of the fileid array
# add padding string
#AR force same as iText example
#Permissions = -1836 #int(Permissions - 2**31)
password = asBytes(password) + PadString
# truncate to 32 bytes
password = password[:32]
# translate permissions to string, low order byte first
p = Permissions# + 2**32L
permissionsString = b""
for i in range(4):
byte = (p & 0xff) # seems to match what iText does
p = p>>8
permissionsString += int2Byte(byte % 256)
hash = md5(asBytes(password))
hash.update(asBytes(OwnerKey))
hash.update(asBytes(permissionsString))
hash.update(asBytes(FileId1))
md5output = hash.digest()
if revision==2:
key = md5output[:5]
elif revision==3: #revision 3 algorithm - loop 50 times
for x in range(50):
md5output = md5(md5output).digest()
key = md5output[:16]
if DEBUG: print('encryptionkey(%s,%s,%s,%s,%s)==>%s' % tuple([hexText(str(x)) for x in (password, OwnerKey, Permissions, FileId1, revision, key)]))
return key
def test17(self):
self.assertEqual(asUnicode(u'abc'),u'abc')
self.assertEqual(asUnicode(b'abc'),u'abc')
self.assertRaises(AttributeError,asUnicode,['abc'])
self.myAssertRaisesRegex(AttributeError,r"asUnicode\(.*'list' object has no attribute 'decode'", asUnicode,['abc'])
self.assertEqual(asUnicodeEx(u'abc'),u'abc')
self.assertEqual(asUnicodeEx(b'abc'),u'abc')
self.assertEqual(asUnicodeEx(123),u'123')
self.assertEqual(asBytes(u'abc'),b'abc')
self.assertEqual(asBytes(b'abc'),b'abc')
self.assertRaises(AttributeError,asBytes,['abc'])
self.myAssertRaisesRegex(AttributeError,"asBytes\(.*'list' object has no attribute 'encode'", asBytes,['abc'])
def create_pdf(catalog, template):
"""Creates PDF as a binary stream in memory, and returns it
This can then be used to write to disk from management commands or crons,
or returned to caller via Django views.
"""
RML_DIR = 'rml'
templateName = os.path.join(RML_DIR, template)
template = preppy.getModule(templateName)
namespace = {'products': catalog, 'RML_DIR': RML_DIR, 'IMG_DIR': 'img'}
rml = template.getOutput(namespace, quoteFunc=preppy.stdQuote)
open(os.path.join(DATA_DIR, 'latest.rml'), 'wb').write(asBytes(rml))
buf = getBytesIO()
rml2pdf.go(asBytes(rml), outputFileName=buf)
return buf.getvalue()
def pygments2xpre(s, language="python"):
"Return markup suitable for XPreformatted"
try:
from pygments import highlight
from pygments.formatters import HtmlFormatter
except ImportError:
return s
from pygments.lexers import get_lexer_by_name
rconv = lambda x: x
if isPy3:
out = getStringIO()
else:
if isUnicode(s):
s = asBytes(s)
rconv = asUnicode
out = getBytesIO()
l = get_lexer_by_name(language)
h = HtmlFormatter()
highlight(s,l,h,out)
styles = [(cls, style.split(';')[0].split(':')[1].strip())
for cls, (style, ttype, level) in h.class2style.items()
if cls and style and style.startswith('color:')]
return rconv(_2xpre(out.getvalue(),styles))
def _getCacheFileName(self):
"""Base this on the directories...same set of directories
should give same cache"""
fsEncoding = self._fsEncoding
hash = md5(b''.join(asBytes(_,enc=fsEncoding) for _ in sorted(self._dirs))).hexdigest()
from reportlab.lib.utils import get_rl_tempfile
fn = get_rl_tempfile('fonts_%s.dat' % hash)
return fn
def _getCacheFileName(self):
"""Base this on the directories...same set of directories
should give same cache"""
fsEncoding = self._fsEncoding
hash = md5(b''.join(
asBytes(_, enc=fsEncoding)
for _ in sorted(self._dirs))).hexdigest()
from reportlab.lib.utils import get_rl_tempfile
fn = get_rl_tempfile('fonts_%s.dat' % hash)
return fn
def testAsciiBase85(self):
"Test if the obvious test for whether ASCII-Base85 encoding works."
msg = "Round-trip AsciiBase85 encoding failed."
plain = 'What is the average velocity of a sparrow?'
#the remainder block can be absent or from 1 to 4 bytes
for i in xrange(256):
encoded = asciiBase85Encode(plain)
decoded = asciiBase85Decode(encoded)
assert decoded == asBytes(plain, 'latin1'), msg
plain += chr(i)
def testAsciiBase85(self):
"Test if the obvious test for whether ASCII-Base85 encoding works."
msg = "Round-trip AsciiBase85 encoding failed."
plain = 'What is the average velocity of a sparrow?'
#the remainder block can be absent or from 1 to 4 bytes
for i in xrange(256):
encoded = asciiBase85Encode(plain)
decoded = asciiBase85Decode(encoded)
assert decoded == asBytes(plain,'latin1'), msg
plain += chr(i)
def _escape_and_limit(s):
s = asBytes(s)
R = []
aR = R.append
n = 0
for c in s:
c = _ESCAPEDICT[char2int(c)]
aR(c)
n += len(c)
if n>=200:
n = 0
aR('\\\n')
return ''.join(R)
def _escape_and_limit(s):
s = asBytes(s)
R = []
aR = R.append
n = 0
for c in s:
c = _ESCAPEDICT[char2int(c)]
aR(c)
n += len(c)
if n >= 200:
n = 0
aR('\\\n')
return ''.join(R)
def encodePDF(key, objectNumber, generationNumber, string, revision=None):
"Encodes a string or stream"
revision = checkRevision(revision)
#print 'encodePDF (%s, %d, %d, %s)' % (hexText(key), objectNumber, generationNumber, string)
# extend 3 bytes of the object Number, low byte first
if revision in (2, 3):
newkey = key
n = objectNumber
for i in range(3):
newkey += int2Byte(n & 0xff)
n = n >> 8
# extend 2 bytes of the generationNumber
n = generationNumber
for i in range(2):
newkey += int2Byte(n & 0xff)
n = n >> 8
md5output = md5(newkey).digest()
if revision == 2:
key = md5output[:10]
elif revision == 3:
key = md5output #all 16 bytes
from reportlab.lib.arciv import ArcIV
encrypted = ArcIV(key).encode(string)
#print 'encrypted=', hexText(encrypted)
elif revision == 5:
iv = os_urandom(16)
encrypter = pyaes.Encrypter(pyaes.AESModeOfOperationCBC(key, iv=iv))
# pkcs7 style padding so that the size of the encrypted block is multiple of 16
string_len = len(string)
padding = ""
padding_len = (16 -
(string_len % 16)) if string_len > 16 else (16 -
string_len)
if padding_len > 0:
padding = chr(padding_len) * padding_len
if isinstance(string, str):
string = (string + padding).encode("utf-8")
else:
string += asBytes(padding)
encrypted = iv + encrypter.feed(string)
encrypted += encrypter.feed()
if DEBUG:
print('encodePDF(%s,%s,%s,%s,%s)==>%s' % tuple([
hexText(str(x)) for x in (key, objectNumber, generationNumber,
string, revision, encrypted)
]))
return encrypted
def getFamilyNames(self):
"Returns a list of the distinct font families found"
if not self._fontsByFamily:
fonts = self._fonts
for font in fonts:
fam = font.familyName
if fam is None: continue
if fam in self._fontsByFamily:
self._fontsByFamily[fam].append(font)
else:
self._fontsByFamily[fam] = [font]
fsEncoding = self._fsEncoding
names = list(asBytes(_,enc=fsEncoding) for _ in self._fontsByFamily.keys())
names.sort()
return names
def getFamilyNames(self):
"Returns a list of the distinct font families found"
if not self._fontsByFamily:
fonts = self._fonts
for font in fonts:
fam = font.familyName
if fam is None: continue
if fam in self._fontsByFamily:
self._fontsByFamily[fam].append(font)
else:
self._fontsByFamily[fam] = [font]
fsEncoding = self._fsEncoding
names = list(
asBytes(_, enc=fsEncoding) for _ in self._fontsByFamily.keys())
names.sort()
return names
def testAsciiBase85RoundTrip(self):
plain = 'What is the average velocity of a sparrow?'
eFuncs = getFuncs('asciiBase85Encode')
for i in xrange(256):
for j,(dfunc, kind) in enumerate(getFuncs('asciiBase85Decode')):
efunc = eFuncs[j][0]
encoded = efunc(plain)
decoded = dfunc(encoded)
assert decoded == asBytes(plain,'latin1'), "Round-trip AsciiBase85 failed for %s & %s\nplain=%s\nencoded=%s\ndecoded=%s" % (
ascii(efunc),ascii(dfunc), ascii(plain), ascii(encoded), ascii(decoded))
if not j:
enc0 = encoded
dec0 = decoded
else:
assert encoded==enc0, " Python & C encodings differ failed for %s & %s\nplain=%s\nencode0=%s\nencoded=%s\ndecode0=%sdecoded=%s" % (
ascii(efunc),ascii(dfunc), ascii(plain), ascii(enc0), ascii(encoded), ascii(dec0), ascii(decoded))
assert decoded==dec0, " Python & C decodings differ failed for %s & %s\nplain=%s\nencode0=%s\nencoded=%s\ndecode0=%sdecoded=%s" % (
ascii(efunc),ascii(dfunc), ascii(plain), ascii(enc0), ascii(encoded), ascii(dec0), ascii(decoded))
plain += chr(i)
def testAsciiBase85RoundTrip(self):
plain = 'What is the average velocity of a sparrow?'
eFuncs = getFuncs('asciiBase85Encode')
for i in xrange(256):
for j,(dfunc, kind) in enumerate(getFuncs('asciiBase85Decode')):
efunc = eFuncs[j][0]
encoded = efunc(plain)
decoded = dfunc(encoded)
assert decoded == asBytes(plain,'latin1'), "Round-trip AsciiBase85 failed for %s & %s\nplain=%s\nencoded=%s\ndecoded=%s" % (
ascii(efunc),ascii(dfunc), ascii(plain), ascii(encoded), ascii(decoded))
if not j:
enc0 = encoded
dec0 = decoded
else:
assert encoded==enc0, " Python & C encodings differ failed for %s & %s\nplain=%s\nencode0=%s\nencoded=%s\ndecode0=%sdecoded=%s" % (
ascii(efunc),ascii(dfunc), ascii(plain), ascii(enc0), ascii(encoded), ascii(dec0), ascii(decoded))
assert decoded==dec0, " Python & C decodings differ failed for %s & %s\nplain=%s\nencode0=%s\nencoded=%s\ndecode0=%sdecoded=%s" % (
ascii(efunc),ascii(dfunc), ascii(plain), ascii(enc0), ascii(encoded), ascii(dec0), ascii(decoded))
plain += chr(i)
def unicodeToUTF8(tree):
tagName, attrs, children, spare = tree
newTagName = asBytes(tagName)
if attrs is None:
newAttrs = None
else:
newAttrs = {}
for key, value in list(attrs.items()):
newAttrs[key.encode('utf8')] = value.encode('utf8')
if children is None:
newChildren = None
else:
newChildren = []
for child in children:
if isinstance(child,bytesT):
newChildren.append(child.encode('utf8'))
elif isinstance(child,unicodeT):
newChildren.append(child)
else:
newChildren.append(unicodeToUTF8(child))
return (newTagName, newAttrs, newChildren, spare)
def unicodeToUTF8(tree):
tagName, attrs, children, spare = tree
newTagName = asBytes(tagName)
if attrs is None:
newAttrs = None
else:
newAttrs = {}
for key, value in list(attrs.items()):
newAttrs[key.encode('utf8')] = value.encode('utf8')
if children is None:
newChildren = None
else:
newChildren = []
for child in children:
if isinstance(child, bytesT):
newChildren.append(child.encode('utf8'))
elif isinstance(child, unicodeT):
newChildren.append(child)
else:
newChildren.append(unicodeToUTF8(child))
return (newTagName, newAttrs, newChildren, spare)
def _processLine(line, sep=',', conv=0):
if isUnicode(line):
space = u' '
dquot = u'"'
empty = u''
speol = u' \r\n'
sep = asUnicode(sep)
else:
space = b' '
dquot = b'"'
empty = b''
speol = b' \r\n'
sep = asBytes(sep)
fields = []
p = 0
ll = len(line)
ls = len(sep)
line += space
while (ll > 0 and (line[ll-1] in speol)): ll -= 1
while p < ll:
#Skip unquoted space at the start of a field
while p<ll and line[p]==space: p += 1
field = empty
ql = 0
while p < ll:
#Skip unquoted space at the end of a field
if ql == 0 and line[p] == space:
q = p
while q < ll and line[q] == space:
q += 1
if q >= ll:
break
elif line[q:q+ls] == sep:
p = q
if ql == 0 and line[p:p+ls] == sep:
break
elif line[p:p+1] == dquot:
if ql == 0:
ql = 1
elif line[p+1:p+2]==dquot:
field += dquot
p += 1
else:
ql = 0
else:
field += line[p:p+1]
p += 1
p += ls
if conv:
try:
fields.append(int(field))
except ValueError:
try:
fields.append(float(field))
except ValueError:
fields.append(field)
else:
fields.append(field)
if line[ll-ls:ll]==sep:
fields.append(empty) #extra field when there's a separator at the end
return fields
def bytes3(x):
if isinstance(x, basestring):
return asBytes(x)
else:
return b''.join([chr(k) for k in x])
def prepare(self, document, overrideID=None):
# get ready to do encryption
if DEBUG:
print('StandardEncryption.prepare(...) - revision %d' %
self.revision)
if self.prepared:
raise ValueError("encryption already prepared!")
# get the unescaped string value of the document id (first array element).
# we allow one to be passed in instead to permit reproducible tests
# of our algorithm, but in real life overrideID will always be None
if overrideID:
internalID = overrideID
else:
externalID = document.ID() # initialize it...
internalID = document.signature.digest()
#AR debugging
if CLOBBERID:
internalID = "xxxxxxxxxxxxxxxx"
if DEBUG:
print('userPassword = %r' % self.userPassword)
print('ownerPassword = %r' % self.ownerPassword)
print('internalID = %r' % internalID)
self.P = int(self.permissionBits() - 2**31)
if CLOBBERPERMISSIONS: self.P = -44 # AR hack
if DEBUG:
print("self.P = %s" % repr(self.P))
if self.revision == 5:
# Init vectro for AES cipher (should be 16 bytes null array)
iv = b'\x00' * 16
# Random User salts
uvs = os_urandom(8)
uks = os_urandom(8)
# the main encryption key
self.key = asBytes(os_urandom(32))
if DEBUG:
print("uvs (hex) = %s" % hexText(uvs))
print("uks (hex) = %s" % hexText(uks))
print("self.key (hex) = %s" % hexText(self.key))
# Calculate the sha-256 hash of the User password (U)
md = sha256(asBytes(self.userPassword[:127]) + uvs)
self.U = md.digest() + uvs + uks
if DEBUG:
print("self.U (hex) = %s" % hexText(self.U))
# Calculate the User encryption key (UE)
md = sha256(asBytes(self.userPassword[:127]) + uks)
encrypter = pyaes.Encrypter(
pyaes.AESModeOfOperationCBC(md.digest(), iv=iv))
self.UE = encrypter.feed(self.key)
self.UE += encrypter.feed()
if DEBUG:
print("self.UE (hex) = %s" % hexText(self.UE))
# Random Owner salts
ovs = os_urandom(8)
oks = os_urandom(8)
# Calculate the hash of the Owner password (U)
md = sha256(asBytes(self.ownerPassword[:127]) + ovs + self.U)
self.O = md.digest() + ovs + oks
if DEBUG:
print("self.O (hex) = %s" % hexText(self.O))
# Calculate the User encryption key (OE)
md = sha256(asBytes(self.ownerPassword[:127]) + oks + self.U)
encrypter = pyaes.Encrypter(
pyaes.AESModeOfOperationCBC(md.digest(), iv=iv))
self.OE = encrypter.feed(self.key)
self.OE += encrypter.feed()
if DEBUG:
print("self.OE (hex) = %s" % hexText(self.OE))
# Compute permissions array
permsarr = [
self.P
& 0xFF, # store the permission value in the first 32-bits
self.P >> 8 & 0xFF,
self.P >> 16 & 0xFF,
self.P >> 24 & 0xFF,
0xFF,
0xFF,
0xFF,
0xFF,
ord(
'T'
), # 'T' if EncryptMetaData is True (default), 'F' otherwise
ord('a'), # a, d, b are magic values
ord('d'),
ord('b'),
0x01, # trailing zeros will be ignored
0x01,
0x01,
0x01
]
# the permission array should be enrypted in the Perms field
encrypter = pyaes.Encrypter(
pyaes.AESModeOfOperationCBC(self.key, iv=iv))
self.Perms = encrypter.feed(bytes3(permsarr))
self.Perms += encrypter.feed()
if DEBUG:
print("self.Perms (hex) = %s" % hexText(self.Perms))
elif self.revision in (2, 3):
self.O = computeO(self.userPassword, self.ownerPassword,
self.revision)
if DEBUG:
print("self.O (as hex) = %s" % hexText(self.O))
#print "\nself.O", self.O, repr(self.O)
self.key = encryptionkey(self.userPassword,
self.O,
self.P,
internalID,
revision=self.revision)
if DEBUG:
print("self.key (hex) = %s" % hexText(self.key))
self.U = computeU(self.key,
revision=self.revision,
documentId=internalID)
if DEBUG:
print("self.U (as hex) = %s" % hexText(self.U))
self.objnum = self.version = None
self.prepared = 1
def _AsciiHexEncode(self, input): # also based on piddlePDF
"Helper function used by images"
output = getStringIO()
for char in asBytes(input):
output.write('%02x' % char2int(char))
return output.getvalue()
def _AsciiHexEncode(self, input): # also based on piddlePDF
"Helper function used by images"
output = getStringIO()
for char in asBytes(input):
output.write('%02x' % char2int(char))
return output.getvalue()
def value(self,v):
self._value = asBytes(v)
self._recalc = True
def asciiBase85Decode(input):
"""Decodes input using ASCII-Base85 coding.
This is not normally used - Acrobat Reader decodes for you
- but a round trip is essential for testing."""
#strip all whitespace
stripped = ''.join(asNative(input).split())
#check end
assert stripped[-2:] == '~>', 'Invalid terminator for Ascii Base 85 Stream'
stripped = stripped[:-2] #chop off terminator
#may have 'z' in it which complicates matters - expand them
stripped = stripped.replace('z','!!!!!')
# special rules apply if not a multiple of five bytes.
whole_word_count, remainder_size = divmod(len(stripped), 5)
#print '%d words, %d leftover' % (whole_word_count, remainder_size)
#assert remainder_size != 1, 'invalid Ascii 85 stream!'
cut = 5 * whole_word_count
body, lastbit = stripped[0:cut], stripped[cut:]
out = [].append
for i in range(whole_word_count):
offset = i*5
c1 = ord(body[offset]) - 33
c2 = ord(body[offset+1]) - 33
c3 = ord(body[offset+2]) - 33
c4 = ord(body[offset+3]) - 33
c5 = ord(body[offset+4]) - 33
num = ((85**4) * c1) + ((85**3) * c2) + ((85**2) * c3) + (85*c4) + c5
temp, b4 = divmod(num,256)
temp, b3 = divmod(temp,256)
b1, b2 = divmod(temp, 256)
assert num == 16777216 * b1 + 65536 * b2 + 256 * b3 + b4, 'dodgy code!'
out(chr(b1))
out(chr(b2))
out(chr(b3))
out(chr(b4))
#decode however many bytes we have as usual
if remainder_size > 0:
while len(lastbit) < 5:
lastbit = lastbit + '!'
c1 = ord(lastbit[0]) - 33
c2 = ord(lastbit[1]) - 33
c3 = ord(lastbit[2]) - 33
c4 = ord(lastbit[3]) - 33
c5 = ord(lastbit[4]) - 33
num = (((85*c1+c2)*85+c3)*85+c4)*85 + (c5
+(0,0,0xFFFFFF,0xFFFF,0xFF)[remainder_size])
temp, b4 = divmod(num,256)
temp, b3 = divmod(temp,256)
b1, b2 = divmod(temp, 256)
assert num == 16777216 * b1 + 65536 * b2 + 256 * b3 + b4, 'dodgy code!'
#print 'decoding: %d %d %d %d %d -> %d -> %d %d %d %d' % (
# c1,c2,c3,c4,c5,num,b1,b2,b3,b4)
#the last character needs 1 adding; the encoding loses
#data by rounding the number to x bytes, and when
#divided repeatedly we get one less
if remainder_size == 2:
lastword = chr(b1)
elif remainder_size == 3:
lastword = chr(b1) + chr(b2)
elif remainder_size == 4:
lastword = chr(b1) + chr(b2) + chr(b3)
else:
lastword = ''
out(lastword)
r = ''.join(out.__self__)
return asBytes(r,enc='latin1')
def asciiBase85Decode(input):
"""Decodes input using ASCII-Base85 coding.
This is not normally used - Acrobat Reader decodes for you
- but a round trip is essential for testing."""
#strip all whitespace
stripped = ''.join(asNative(input).split())
#check end
assert stripped[
-2:] == '~>', 'Invalid terminator for Ascii Base 85 Stream'
stripped = stripped[:-2] #chop off terminator
#may have 'z' in it which complicates matters - expand them
stripped = stripped.replace('z', '!!!!!')
# special rules apply if not a multiple of five bytes.
whole_word_count, remainder_size = divmod(len(stripped), 5)
#print '%d words, %d leftover' % (whole_word_count, remainder_size)
#assert remainder_size != 1, 'invalid Ascii 85 stream!'
cut = 5 * whole_word_count
body, lastbit = stripped[0:cut], stripped[cut:]
out = [].append
for i in range(whole_word_count):
offset = i * 5
c1 = ord(body[offset]) - 33
c2 = ord(body[offset + 1]) - 33
c3 = ord(body[offset + 2]) - 33
c4 = ord(body[offset + 3]) - 33
c5 = ord(body[offset + 4]) - 33
num = ((85**4) * c1) + ((85**3) * c2) + (
(85**2) * c3) + (85 * c4) + c5
temp, b4 = divmod(num, 256)
temp, b3 = divmod(temp, 256)
b1, b2 = divmod(temp, 256)
assert num == 16777216 * b1 + 65536 * b2 + 256 * b3 + b4, 'dodgy code!'
out(chr(b1))
out(chr(b2))
out(chr(b3))
out(chr(b4))
#decode however many bytes we have as usual
if remainder_size > 0:
while len(lastbit) < 5:
lastbit = lastbit + '!'
c1 = ord(lastbit[0]) - 33
c2 = ord(lastbit[1]) - 33
c3 = ord(lastbit[2]) - 33
c4 = ord(lastbit[3]) - 33
c5 = ord(lastbit[4]) - 33
num = (((85 * c1 + c2) * 85 + c3) * 85 + c4) * 85 + (
c5 + (0, 0, 0xFFFFFF, 0xFFFF, 0xFF)[remainder_size])
temp, b4 = divmod(num, 256)
temp, b3 = divmod(temp, 256)
b1, b2 = divmod(temp, 256)
assert num == 16777216 * b1 + 65536 * b2 + 256 * b3 + b4, 'dodgy code!'
#print 'decoding: %d %d %d %d %d -> %d -> %d %d %d %d' % (
# c1,c2,c3,c4,c5,num,b1,b2,b3,b4)
#the last character needs 1 adding; the encoding loses
#data by rounding the number to x bytes, and when
#divided repeatedly we get one less
if remainder_size == 2:
lastword = chr(b1)
elif remainder_size == 3:
lastword = chr(b1) + chr(b2)
elif remainder_size == 4:
lastword = chr(b1) + chr(b2) + chr(b3)
else:
lastword = ''
out(lastword)
r = ''.join(out.__self__)
return asBytes(r, enc='latin1')