def __init__(self, raw, ident, user_encoding, log, try_extra_data_fix=False): self.log = log self.compression_type = raw[:2] self.records, self.records_size = struct.unpack('>HH', raw[8:12]) self.encryption_type, = struct.unpack('>H', raw[12:14]) if ident == 'TEXTREAD': self.codepage = 1252 if len(raw) <= 16: self.codec = 'cp1252' self.extra_flags = 0 self.title = _('Unknown') self.language = 'ENGLISH' self.sublanguage = 'NEUTRAL' self.exth_flag, self.exth = 0, None self.ancient = True self.first_image_index = -1 self.mobi_version = 1 else: self.ancient = False self.doctype = raw[16:20] self.length, self.type, self.codepage, self.unique_id, \ self.version = struct.unpack('>LLLLL', raw[20:40]) try: self.codec = { 1252: 'cp1252', 65001: 'utf-8', }[self.codepage] except (IndexError, KeyError): self.codec = 'cp1252' if not user_encoding else user_encoding log.warn('Unknown codepage %d. Assuming %s' % (self.codepage, self.codec)) # Some KF8 files have header length == 256 (generated by kindlegen # 2.7?). See https://bugs.launchpad.net/bugs/1067310 max_header_length = 0x100 if (ident == 'TEXTREAD' or self.length < 0xE4 or self.length > max_header_length or (try_extra_data_fix and self.length == 0xE4)): self.extra_flags = 0 else: self.extra_flags, = struct.unpack('>H', raw[0xF2:0xF4]) if self.compression_type == 'DH': self.huff_offset, self.huff_number = struct.unpack('>LL', raw[0x70:0x78]) toff, tlen = struct.unpack('>II', raw[0x54:0x5c]) tend = toff + tlen self.title = raw[toff:tend] if tend < len(raw) else _('Unknown') langcode = struct.unpack('!L', raw[0x5C:0x60])[0] langid = langcode & 0xFF sublangid = (langcode >> 10) & 0xFF self.language = main_language.get(langid, 'ENGLISH') self.sublanguage = sub_language.get(sublangid, 'NEUTRAL') self.mobi_version = struct.unpack('>I', raw[0x68:0x6c])[0] self.first_image_index = struct.unpack('>L', raw[0x6c:0x6c + 4])[0] self.exth_flag, = struct.unpack('>L', raw[0x80:0x84]) self.exth = None if not isinstance(self.title, unicode): self.title = self.title.decode(self.codec, 'replace') if self.exth_flag & 0x40: try: self.exth = EXTHHeader(raw[16 + self.length:], self.codec, self.title) self.exth.mi.uid = self.unique_id if self.exth.mi.is_null('language'): try: self.exth.mi.language = mobi2iana(langid, sublangid) except: self.log.exception('Unknown language code') except: self.log.exception('Invalid EXTH header') self.exth_flag = 0 self.ncxidx = NULL_INDEX if len(raw) >= 0xF8: self.ncxidx, = struct.unpack_from(b'>L', raw, 0xF4) # Ancient PRC files from Baen can have random values for # mobi_version, so be conservative if self.mobi_version == 8 and len(raw) >= (0xF8 + 16): self.dividx, self.skelidx, self.datpidx, self.othidx = \ struct.unpack_from(b'>4L', raw, 0xF8) # need to use the FDST record to find out how to properly # unpack the raw_ml into pieces it is simply a table of start # and end locations for each flow piece self.fdstidx, self.fdstcnt = struct.unpack_from(b'>2L', raw, 0xC0) # if cnt is 1 or less, fdst section number can be garbage if self.fdstcnt <= 1: self.fdstidx = NULL_INDEX else: # Null values self.skelidx = self.dividx = self.othidx = self.fdstidx = \ NULL_INDEX
def __init__(self, raw, ident, user_encoding, log, try_extra_data_fix=False): self.log = log self.compression_type = raw[:2] self.records, self.records_size = struct.unpack('>HH', raw[8:12]) self.encryption_type, = struct.unpack('>H', raw[12:14]) if ident == 'TEXTREAD': self.codepage = 1252 if len(raw) <= 16: self.codec = 'cp1252' self.extra_flags = 0 self.title = _('Unknown') self.language = 'ENGLISH' self.sublanguage = 'NEUTRAL' self.exth_flag, self.exth = 0, None self.ancient = True self.first_image_index = -1 self.mobi_version = 1 else: self.ancient = False self.doctype = raw[16:20] self.length, self.type, self.codepage, self.unique_id, \ self.version = struct.unpack('>LLLLL', raw[20:40]) try: self.codec = { 1252: 'cp1252', 65001: 'utf-8', }[self.codepage] except (IndexError, KeyError): self.codec = 'cp1252' if not user_encoding else user_encoding log.warn('Unknown codepage %d. Assuming %s' % (self.codepage, self.codec)) # Some KF8 files have header length == 264 (generated by kindlegen # 2.9?). See https://bugs.launchpad.net/bugs/1179144 max_header_length = 500 # We choose 500 for future versions of kindlegen if (ident == 'TEXTREAD' or self.length < 0xE4 or self.length > max_header_length or (try_extra_data_fix and self.length == 0xE4)): self.extra_flags = 0 else: self.extra_flags, = struct.unpack('>H', raw[0xF2:0xF4]) if self.compression_type == 'DH': self.huff_offset, self.huff_number = struct.unpack( '>LL', raw[0x70:0x78]) toff, tlen = struct.unpack('>II', raw[0x54:0x5c]) tend = toff + tlen self.title = raw[toff:tend] if tend < len(raw) else _('Unknown') langcode = struct.unpack('!L', raw[0x5C:0x60])[0] langid = langcode & 0xFF sublangid = (langcode >> 10) & 0xFF self.language = main_language.get(langid, 'ENGLISH') self.sublanguage = sub_language.get(sublangid, 'NEUTRAL') self.mobi_version = struct.unpack('>I', raw[0x68:0x6c])[0] self.first_image_index = struct.unpack('>L', raw[0x6c:0x6c + 4])[0] self.exth_flag, = struct.unpack('>L', raw[0x80:0x84]) self.exth = None if not isinstance(self.title, unicode): self.title = self.title.decode(self.codec, 'replace') if self.exth_flag & 0x40: try: self.exth = EXTHHeader(raw[16 + self.length:], self.codec, self.title) self.exth.mi.uid = self.unique_id if self.exth.mi.is_null('language'): try: self.exth.mi.language = mobi2iana( langid, sublangid) except: self.log.exception('Unknown language code') except: self.log.exception('Invalid EXTH header') self.exth_flag = 0 self.ncxidx = NULL_INDEX if len(raw) >= 0xF8: self.ncxidx, = struct.unpack_from(b'>L', raw, 0xF4) # Ancient PRC files from Baen can have random values for # mobi_version, so be conservative if self.mobi_version == 8 and len(raw) >= (0xF8 + 16): self.dividx, self.skelidx, self.datpidx, self.othidx = \ struct.unpack_from(b'>4L', raw, 0xF8) # need to use the FDST record to find out how to properly # unpack the raw_ml into pieces it is simply a table of start # and end locations for each flow piece self.fdstidx, self.fdstcnt = struct.unpack_from( b'>2L', raw, 0xC0) # if cnt is 1 or less, fdst section number can be garbage if self.fdstcnt <= 1: self.fdstidx = NULL_INDEX else: # Null values self.skelidx = self.dividx = self.othidx = self.fdstidx = \ NULL_INDEX
def __init__(self, raw, ident, user_encoding, log, try_extra_data_fix=False): self.log = log self.compression_type = raw[:2] self.records, self.records_size = struct.unpack(">HH", raw[8:12]) self.encryption_type, = struct.unpack(">H", raw[12:14]) if ident == "TEXTREAD": self.codepage = 1252 if len(raw) <= 16: self.codec = "cp1252" self.extra_flags = 0 self.title = _("Unknown") self.language = "ENGLISH" self.sublanguage = "NEUTRAL" self.exth_flag, self.exth = 0, None self.ancient = True self.first_image_index = -1 self.mobi_version = 1 else: self.ancient = False self.doctype = raw[16:20] self.length, self.type, self.codepage, self.unique_id, self.version = struct.unpack(">LLLLL", raw[20:40]) try: self.codec = {1252: "cp1252", 65001: "utf-8"}[self.codepage] except (IndexError, KeyError): self.codec = "cp1252" if not user_encoding else user_encoding log.warn("Unknown codepage %d. Assuming %s" % (self.codepage, self.codec)) # Some KF8 files have header length == 264 (generated by kindlegen # 2.9?). See https://bugs.launchpad.net/bugs/1179144 max_header_length = 500 # We choose 500 for future versions of kindlegen if ( ident == "TEXTREAD" or self.length < 0xE4 or self.length > max_header_length or (try_extra_data_fix and self.length == 0xE4) ): self.extra_flags = 0 else: self.extra_flags, = struct.unpack(">H", raw[0xF2:0xF4]) if self.compression_type == "DH": self.huff_offset, self.huff_number = struct.unpack(">LL", raw[0x70:0x78]) toff, tlen = struct.unpack(">II", raw[0x54:0x5C]) tend = toff + tlen self.title = raw[toff:tend] if tend < len(raw) else _("Unknown") langcode = struct.unpack("!L", raw[0x5C:0x60])[0] langid = langcode & 0xFF sublangid = (langcode >> 10) & 0xFF self.language = main_language.get(langid, "ENGLISH") self.sublanguage = sub_language.get(sublangid, "NEUTRAL") self.mobi_version = struct.unpack(">I", raw[0x68:0x6C])[0] self.first_image_index = struct.unpack(">L", raw[0x6C : 0x6C + 4])[0] self.exth_flag, = struct.unpack(">L", raw[0x80:0x84]) self.exth = None if not isinstance(self.title, unicode): self.title = self.title.decode(self.codec, "replace") if self.exth_flag & 0x40: try: self.exth = EXTHHeader(raw[16 + self.length :], self.codec, self.title) self.exth.mi.uid = self.unique_id if self.exth.mi.is_null("language"): try: self.exth.mi.language = mobi2iana(langid, sublangid) except: self.log.exception("Unknown language code") except: self.log.exception("Invalid EXTH header") self.exth_flag = 0 self.ncxidx = NULL_INDEX if len(raw) >= 0xF8: self.ncxidx, = struct.unpack_from(b">L", raw, 0xF4) # Ancient PRC files from Baen can have random values for # mobi_version, so be conservative if self.mobi_version == 8 and len(raw) >= (0xF8 + 16): self.dividx, self.skelidx, self.datpidx, self.othidx = struct.unpack_from(b">4L", raw, 0xF8) # need to use the FDST record to find out how to properly # unpack the raw_ml into pieces it is simply a table of start # and end locations for each flow piece self.fdstidx, self.fdstcnt = struct.unpack_from(b">2L", raw, 0xC0) # if cnt is 1 or less, fdst section number can be garbage if self.fdstcnt <= 1: self.fdstidx = NULL_INDEX else: # Null values self.skelidx = self.dividx = self.othidx = self.fdstidx = NULL_INDEX