Fix tons of PDF-related issues

This commit is contained in:
NoDRM 2022-08-07 15:58:01 +02:00
parent 52cf3faa59
commit 88b0966961
2 changed files with 62 additions and 19 deletions

View File

@ -73,8 +73,9 @@ List of changes since the fork of Apprentice Harper's repository:
- Fix some more Calibre-6 bugs in the Obok plugin (should fix #114).
- Fix a bug where invalid Adobe keys could cause the plugin to stop trying subsequent keys (partially fixes #109).
- Fix DRM removal sometimes resetting the ZIP's internal "external_attr" value on Calibre 5 and newer.
- Fix PDF decryption issues on Calibre 4 (hopefully fixes #104).
- Fix tons of PDF decryption issues (hopefully fixes #104 and other PDF-related issues).
- Small Python 2 / Calibre 4 bugfix for Obok.
- Removing ancient AlfCrypto machine code libraries, moving all encryption / decryption to Python code.
- General cleanup and removal of dead code.
- Fix a bug where ADE account keys weren't automatically imported from the DeACSM plugin when importing a PDF file.
- Re-enable Xrefs in exported PDF files since the file corruption bug is hopefully fixed. Please open bug reports if you encounter new issues with PDF files.

View File

@ -117,7 +117,7 @@ def SHA256(message):
# 1 = only if present in input
# 2 = always
GEN_XREF_STM = 0
GEN_XREF_STM = 1
# This is the value for the current document
gen_xref_stm = False # will be set in PDFSerializer
@ -565,7 +565,11 @@ class PSBaseParser(object):
while 1:
self.fillbuf()
if eol:
c = bytes([self.buf[self.charpos]])
if sys.version_info[0] == 2:
c = self.buf[self.charpos]
else:
c = bytes([self.buf[self.charpos]])
# handle '\r\n'
if c == b'\n':
linebuf += c
@ -575,10 +579,17 @@ class PSBaseParser(object):
if m:
linebuf += self.buf[self.charpos:m.end(0)]
self.charpos = m.end(0)
if bytes([linebuf[-1]]) == b'\r':
eol = True
else:
break
if sys.version_info[0] == 2:
if linebuf[-1] == b'\r':
eol = True
else:
break
else:
if bytes([linebuf[-1]]) == b'\r':
eol = True
else:
break
else:
linebuf += self.buf[self.charpos:]
self.charpos = len(self.buf)
@ -954,9 +965,14 @@ class PDFStream(PDFObject):
for i in range(0, len(data), columns+1):
pred = data[i]
ent1 = data[i+1:i+1+columns]
if pred == 2:
ent1 = b''.join(bytes([(a+b) & 255]) \
for (a,b) in zip(ent0,ent1))
if sys.version_info[0] == 2:
if pred == '\x02':
ent1 = ''.join(chr((ord(a)+ord(b)) & 255) \
for (a,b) in zip(ent0,ent1))
else:
if pred == 2:
ent1 = b''.join(bytes([(a+b) & 255]) \
for (a,b) in zip(ent0,ent1))
buf += ent1
ent0 = ent1
data = buf
@ -1070,8 +1086,6 @@ class PDFXRef(object):
return (None, pos)
## PDFXRefStream
##
class PDFXRefStream(object):
def __init__(self):
@ -1404,7 +1418,10 @@ class PDFDocument(object):
x = ARC4.new(hash).decrypt(Odata) # 4
if R >= 3:
for i in range(1,19+1):
k = b''.join(bytes([c ^ i]) for c in hash )
if sys.version_info[0] == 2:
k = b''.join(chr(ord(c) ^ i) for c in hash )
else:
k = b''.join(bytes([c ^ i]) for c in hash )
x = ARC4.new(k).decrypt(x)
@ -1462,7 +1479,10 @@ class PDFDocument(object):
hash.update(docid[0]) # 3
x = ARC4.new(key).decrypt(hash.digest()[:16]) # 4
for i in range(1,19+1):
k = b''.join(bytes([c ^ i]) for c in key )
if sys.version_info[0] == 2:
k = b''.join(chr(ord(c) ^ i) for c in key )
else:
k = b''.join(bytes([c ^ i]) for c in key )
x = ARC4.new(k).decrypt(x)
u1 = x+x # 32bytes total
if R == 2:
@ -1490,8 +1510,8 @@ class PDFDocument(object):
# check owner pass:
retval = self.check_owner_password(password, docid, param)
if retval is True or retval is not None:
#print("Owner pass is valid - " + str(retval))
if retval is True or (retval is not False and retval is not None):
#print("Owner pass is valid")
if retval is True:
self.decrypt_key = self.recover_encryption_key_with_password(password, docid, param)
else:
@ -1500,7 +1520,7 @@ class PDFDocument(object):
if self.decrypt_key is None or self.decrypt_key is True or self.decrypt_key is False:
# That's not the owner password. Check if it's the user password.
retval = self.check_user_password(password, docid, param)
if retval is True or retval is not None:
if retval is True or (retval is not False and retval is not None):
#print("User pass is valid")
if retval is True:
self.decrypt_key = self.recover_encryption_key_with_password(password, docid, param)
@ -1723,7 +1743,11 @@ class PDFDocument(object):
data = data[16:]
plaintext = AES.new(key,AES.MODE_CBC,ivector).decrypt(data)
# remove pkcs#5 aes padding
cutter = -1 * plaintext[-1]
if sys.version_info[0] == 2:
cutter = -1 * ord(plaintext[-1])
else:
cutter = -1 * plaintext[-1]
plaintext = plaintext[:cutter]
return plaintext
@ -2199,7 +2223,11 @@ class PDFSerializer(object):
elif isinstance(obj, bytearray):
self.write(b'(%s)' % self.escape_string(obj))
elif isinstance(obj, bytes):
self.write(b'(%s)' % self.escape_string(obj))
# I'm not 100% sure if this is correct, but it seems to fix some PDFs ...
# If needed, revert that change.
self.write(b'<%s>' % binascii.hexlify(obj).upper())
print("ineptpdf.py: Unknown bytes element found - guessing.")
print("If this PDF is corrupted and/or doesn't work, please open a bug report.")
elif isinstance(obj, str):
self.write(b'(%s)' % self.escape_string(obj.encode('utf-8')))
elif isinstance(obj, bool):
@ -2226,6 +2254,20 @@ class PDFSerializer(object):
self.write(b'(deleted)')
else:
data = obj.get_decdata()
# Fix length:
# We've decompressed and then recompressed the PDF stream.
# Depending on the algorithm, the implementation, and the compression level,
# the resulting recompressed stream is unlikely to have the same length as the original.
# So we need to update the PDF object to contain the new proper length.
# Without this change, all PDFs exported by this plugin are slightly corrupted -
# even though most if not all PDF readers can correct that on-the-fly.
if 'Length' in obj.dic:
obj.dic['Length'] = len(data)
self.serialize_object(obj.dic)
self.write(b'stream\n')
self.write(data)