Fix PDF corruption in Calibre 4 (#104)

CHANGELOG.md

@@ -72,3 +72,5 @@ List of changes since the fork of Apprentice Harper's repository:
 - Fix a bug introduced with #48 that breaks DeDRM'ing on Calibre 4 (fixes #101).
 - 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).

DeDRM_plugin/epubfontdecrypt.py

@@ -25,6 +25,7 @@ import traceback
 import zlib
 import zipfile
 from zipfile import ZipInfo, ZipFile, ZIP_STORED, ZIP_DEFLATED
+from zeroedzipinfo import ZeroedZipInfo
 from contextlib import closing
 from lxml import etree
 import itertools
@@ -298,13 +299,21 @@ def decryptFontsBook(inpath, outpath):
                         zi.internal_attr = oldzi.internal_attr
                         # external attributes are dependent on the create system, so copy both.
                         zi.external_attr = oldzi.external_attr
+                        zi.volume = oldzi.volume
                         zi.create_system = oldzi.create_system
+                        zi.create_version = oldzi.create_version
+
                         if any(ord(c) >= 128 for c in path) or any(ord(c) >= 128 for c in zi.comment):
                             # If the file name or the comment contains any non-ASCII char, set the UTF8-flag
                             zi.flag_bits |= 0x800
                     except:
                         pass
 
+                    # Python 3 has a bug where the external_attr is reset to `0o600 << 16`
+                    # if it's NULL, so we need a workaround:
+                    if zi.external_attr == 0: 
+                        zi = ZeroedZipInfo(zi)
+
                     if path == "mimetype":
                         outf.writestr(zi, inf.read('mimetype'))
                     elif path == "META-INF/encryption.xml":

DeDRM_plugin/epubwatermark.py

@@ -16,6 +16,7 @@ Removes various watermarks from EPUB files
 
 import traceback
 from zipfile import ZipInfo, ZipFile, ZIP_STORED, ZIP_DEFLATED
+from zeroedzipinfo import ZeroedZipInfo
 from contextlib import closing
 from lxml import etree
 import re
@@ -133,13 +134,22 @@ def removeHTMLwatermarks(object, path_to_ebook):
                         zi.extra = oldzi.extra
                         zi.internal_attr = oldzi.internal_attr
                         zi.external_attr = oldzi.external_attr
+                        zi.volume = oldzi.volume
                         zi.create_system = oldzi.create_system
+                        zi.create_version = oldzi.create_version
+
                         if any(ord(c) >= 128 for c in path) or any(ord(c) >= 128 for c in zi.comment):
                             # If the file name or the comment contains any non-ASCII char, set the UTF8-flag
                             zi.flag_bits |= 0x800
                     except:
                         pass
 
+                    # Python 3 has a bug where the external_attr is reset to `0o600 << 16`
+                    # if it's NULL, so we need a workaround:
+                    if zi.external_attr == 0: 
+                        zi = ZeroedZipInfo(zi)
+
+
                     outf.writestr(zi, data)
         except:
             traceback.print_exc()
@@ -249,13 +259,21 @@ def removeOPFwatermarks(object, path_to_ebook):
                         zi.extra = oldzi.extra
                         zi.internal_attr = oldzi.internal_attr
                         zi.external_attr = oldzi.external_attr
+                        zi.volume = oldzi.volume
                         zi.create_system = oldzi.create_system
+                        zi.create_version = oldzi.create_version
+
                         if any(ord(c) >= 128 for c in path) or any(ord(c) >= 128 for c in zi.comment):
                             # If the file name or the comment contains any non-ASCII char, set the UTF8-flag
                             zi.flag_bits |= 0x800
                     except:
                         pass
 
+                    # Python 3 has a bug where the external_attr is reset to `0o600 << 16`
+                    # if it's NULL, so we need a workaround:
+                    if zi.external_attr == 0: 
+                        zi = ZeroedZipInfo(zi)
+
                     outf.writestr(zi, data)
         except:
             traceback.print_exc()
@@ -301,13 +319,21 @@ def removeCDPwatermark(object, path_to_ebook):
                     zi.extra = oldzi.extra
                     zi.internal_attr = oldzi.internal_attr
                     zi.external_attr = oldzi.external_attr
+                    zi.volume = oldzi.volume
                     zi.create_system = oldzi.create_system
+                    zi.create_version = oldzi.create_version
+
                     if any(ord(c) >= 128 for c in path) or any(ord(c) >= 128 for c in zi.comment):
                         # If the file name or the comment contains any non-ASCII char, set the UTF8-flag
                         zi.flag_bits |= 0x800
                 except:
                     pass
 
+                # Python 3 has a bug where the external_attr is reset to `0o600 << 16`
+                # if it's NULL, so we need a workaround:
+                if zi.external_attr == 0: 
+                    zi = ZeroedZipInfo(zi)
+
                 outf.writestr(zi, data)
         
         print("Watermark: Successfully removed cdp.info watermark")

DeDRM_plugin/ineptepub.py

@@ -48,6 +48,7 @@ import base64
 import zlib
 import zipfile
 from zipfile import ZipInfo, ZipFile, ZIP_STORED, ZIP_DEFLATED
+from zeroedzipinfo import ZeroedZipInfo
 from contextlib import closing
 from lxml import etree
 from uuid import UUID
@@ -356,12 +357,23 @@ def decryptBook(userkey, inpath, outpath):
                         zi.internal_attr = oldzi.internal_attr
                         # external attributes are dependent on the create system, so copy both.
                         zi.external_attr = oldzi.external_attr
+
+                        zi.volume = oldzi.volume
                         zi.create_system = oldzi.create_system
+                        zi.create_version = oldzi.create_version
+
                         if any(ord(c) >= 128 for c in path) or any(ord(c) >= 128 for c in zi.comment):
                             # If the file name or the comment contains any non-ASCII char, set the UTF8-flag
                             zi.flag_bits |= 0x800
                     except:
                         pass
+
+                    # Python 3 has a bug where the external_attr is reset to `0o600 << 16`
+                    # if it's NULL, so we need a workaround:
+                    if zi.external_attr == 0: 
+                        zi = ZeroedZipInfo(zi)
+
+
                     if path == "META-INF/encryption.xml":
                         outf.writestr(zi, data)
                     else:

DeDRM_plugin/ineptpdf.py

@@ -51,13 +51,14 @@
 #   9.1.1 - Only support PyCryptodome; clean up the code
 #   10.0.0 - Add support for "hardened" Adobe DRM (RMSDK >= 10)
 #   10.0.2 - Fix some Python2 stuff
+#   10.0.4 - Fix more Python2 stuff
 
 """
 Decrypts Adobe ADEPT-encrypted PDF files.
 """
 
 __license__ = 'GPL v3'
-__version__ = "10.0.2"
+__version__ = "10.0.4"
 
 import codecs
 import hashlib
@@ -200,6 +201,9 @@ def nunpack(s, default=0):
     elif l == 2:
         return struct.unpack('>H', s)[0]
     elif l == 3:
+        if sys.version_info[0] == 2:
+            return struct.unpack('>L', '\x00'+s)[0]
+        else: 
             return struct.unpack('>L', bytes([0]) + s)[0]
     elif l == 4:
         return struct.unpack('>L', s)[0]
@@ -459,6 +463,9 @@ class PSBaseParser(object):
             self.hex += c
             return (self.parse_literal_hex, i+1)
         if self.hex:
+            if sys.version_info[0] == 2: 
+                self.token += chr(int(self.hex, 16))
+            else: 
                 self.token += bytes([int(self.hex, 16)])
         return (self.parse_literal, i)
 
@@ -543,10 +550,18 @@ class PSBaseParser(object):
             self.oct += c
             return (self.parse_string_1, i+1)
         if self.oct:
+            if sys.version_info[0] == 2:
+                self.token += chr(int(self.oct, 8))
+            else: 
                 self.token += bytes([int(self.oct, 8)])  
             return (self.parse_string, i)
         if c in ESC_STRING:
+
+            if sys.version_info[0] == 2:
+                self.token += chr(ESC_STRING[c])
+            else: 
                 self.token += bytes([ESC_STRING[c]])
+                
         return (self.parse_string, i+1)
 
     def parse_wopen(self, s, i):
@@ -572,14 +587,19 @@ class PSBaseParser(object):
         return (self.parse_main, i)
 
     def parse_hexstring(self, s, i):
-        m1 = END_HEX_STRING.search(s, i)
-        if not m1:
+        m = END_HEX_STRING.search(s, i)
+        if not m:
             self.token += s[i:]
             return (self.parse_hexstring, len(s))
-        j = m1.start(0)
+        j = m.start(0)
         self.token += s[i:j]
-        token = HEX_PAIR.sub(lambda m2: bytes([int(m2.group(0), 16)]),
+        if sys.version_info[0] == 2:
+            token = HEX_PAIR.sub(lambda m: chr(int(m.group(0), 16)),
+                                                 SPC.sub('', self.token))
+        else: 
+            token = HEX_PAIR.sub(lambda m: bytes([int(m.group(0), 16)]),
                                                  SPC.sub(b'', self.token))
+
         self.add_token(token)
         return (self.parse_main, j)
 

DeDRM_plugin/zeroedzipinfo.py

@@ -0,0 +1,30 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+
+"""
+Python 3's "zipfile" has an annoying bug where the `external_attr` field 
+of a ZIP file cannot be set to 0. However, if the original DRMed ZIP has 
+that set to 0 then we want the DRM-free ZIP to have that as 0, too. 
+See https://github.com/python/cpython/issues/87713
+
+We cannot just set the "external_attr" to 0 as the code to save the ZIP
+resets that variable. 
+
+So, here's a class that inherits from ZipInfo and ensures that EVERY 
+read access to that variable will return a 0 ...
+
+"""
+
+import zipfile
+
+class ZeroedZipInfo(zipfile.ZipInfo):
+    def __init__(self, zinfo):
+        for k in self.__slots__:
+            if hasattr(zinfo, k):
+                setattr(self, k, getattr(zinfo, k))
+
+    def __getattribute__(self, name):
+        if name == "external_attr":
+            return 0
+        return object.__getattribute__(self, name)

DeDRM_plugin/zipfilerugged.py

@@ -394,6 +394,19 @@ class ZipInfo (object):
             extra = extra[ln+4:]
 
 
+class ZeroedZipInfo(ZipInfo):
+    def __init__(self, zinfo):
+        for k in self.__slots__:
+            if hasattr(zinfo, k):
+                setattr(self, k, getattr(zinfo, k))
+
+    def __getattribute__(self, name):
+        if name == "external_attr":
+            return 0
+        return object.__getattribute__(self, name)
+
+
+
 class _ZipDecrypter:
     """Class to handle decryption of files stored within a ZIP archive.
 

DeDRM_plugin/zipfix.py

@@ -26,6 +26,7 @@ import sys, os
 
 import zlib
 import zipfilerugged
+from zipfilerugged import ZipInfo, ZeroedZipInfo
 import getopt
 from struct import unpack
 
@@ -36,12 +37,6 @@ _FILENAME_OFFSET = 30
 _MAX_SIZE = 64 * 1024
 _MIMETYPE = 'application/epub+zip'
 
-class ZipInfo(zipfilerugged.ZipInfo):
-    def __init__(self, *args, **kwargs):
-        if 'compress_type' in kwargs:
-            compress_type = kwargs.pop('compress_type')
-        super(ZipInfo, self).__init__(*args, **kwargs)
-        self.compress_type = compress_type
 
 class fixZip:
     def __init__(self, zinput, zoutput):
@@ -117,7 +112,8 @@ class fixZip:
         # if epub write mimetype file first, with no compression
         if self.ztype == 'epub':
             # first get a ZipInfo with current time and no compression
-            mimeinfo = ZipInfo(b'mimetype',compress_type=zipfilerugged.ZIP_STORED)
+            mimeinfo = ZipInfo(b'mimetype')
+            mimeinfo.compress_type = zipfilerugged.ZIP_STORED
             mimeinfo.internal_attr = 1 # text file
             try:
                 # if the mimetype is present, get its info, including time-stamp
@@ -129,8 +125,16 @@ class fixZip:
                 mimeinfo.internal_attr = oldmimeinfo.internal_attr
                 mimeinfo.external_attr = oldmimeinfo.external_attr
                 mimeinfo.create_system = oldmimeinfo.create_system
+                mimeinfo.create_version = oldmimeinfo.create_version
+                mimeinfo.volume = oldmimeinfo.volume
             except:
                 pass
+
+            # Python 3 has a bug where the external_attr is reset to `0o600 << 16`
+            # if it's NULL, so we need a workaround:
+            if mimeinfo.external_attr == 0: 
+                mimeinfo = ZeroedZipInfo(mimeinfo)
+
             self.outzip.writestr(mimeinfo, _MIMETYPE.encode('ascii'))
 
         # write the rest of the files
@@ -145,13 +149,23 @@ class fixZip:
                     zinfo.filename = local_name
 
                 # create new ZipInfo with only the useful attributes from the old info
-                nzinfo = ZipInfo(zinfo.filename, zinfo.date_time, compress_type=zinfo.compress_type)
+                nzinfo = ZipInfo(zinfo.filename)
+                nzinfo.date_time = zinfo.date_time
+                nzinfo.compress_type = zinfo.compress_type
                 nzinfo.comment=zinfo.comment
                 nzinfo.extra=zinfo.extra
                 nzinfo.internal_attr=zinfo.internal_attr
                 nzinfo.external_attr=zinfo.external_attr
                 nzinfo.create_system=zinfo.create_system
+                nzinfo.create_version = zinfo.create_version
+                nzinfo.volume = zinfo.volume
                 nzinfo.flag_bits = zinfo.flag_bits & 0x800  # preserve UTF-8 flag
+
+                # Python 3 has a bug where the external_attr is reset to `0o600 << 16`
+                # if it's NULL, so we need a workaround:
+                if nzinfo.external_attr == 0: 
+                    nzinfo = ZeroedZipInfo(nzinfo)
+
                 self.outzip.writestr(nzinfo,data)
 
         self.bzf.close()