calibre-web/vendor/PyPDF2/generic.py

1227 lines
44 KiB
Python
Raw Normal View History

2016-08-07 20:12:32 +00:00
# vim: sw=4:expandtab:foldmethod=marker
#
# Copyright (c) 2006, Mathieu Fenniak
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met:
#
# * Redistributions of source code must retain the above copyright notice,
# this list of conditions and the following disclaimer.
# * Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
# * The name of the author may not be used to endorse or promote products
# derived from this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
# ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
# LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
# CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
# SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
# INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
# CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
# ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
# POSSIBILITY OF SUCH DAMAGE.
"""
Implementation of generic PDF objects (dictionary, number, string, and so on)
"""
__author__ = "Mathieu Fenniak"
__author_email__ = "biziqe@mathieu.fenniak.net"
import re
from .utils import readNonWhitespace, RC4_encrypt, skipOverComment
from .utils import b_, u_, chr_, ord_
from .utils import PdfStreamError
import warnings
from . import filters
from . import utils
import decimal
import codecs
import sys
#import debugging
ObjectPrefix = b_('/<[tf(n%')
NumberSigns = b_('+-')
IndirectPattern = re.compile(b_(r"(\d+)\s+(\d+)\s+R[^a-zA-Z]"))
def readObject(stream, pdf):
tok = stream.read(1)
stream.seek(-1, 1) # reset to start
idx = ObjectPrefix.find(tok)
if idx == 0:
# name object
return NameObject.readFromStream(stream, pdf)
elif idx == 1:
# hexadecimal string OR dictionary
peek = stream.read(2)
stream.seek(-2, 1) # reset to start
if peek == b_('<<'):
return DictionaryObject.readFromStream(stream, pdf)
else:
return readHexStringFromStream(stream)
elif idx == 2:
# array object
return ArrayObject.readFromStream(stream, pdf)
elif idx == 3 or idx == 4:
# boolean object
return BooleanObject.readFromStream(stream)
elif idx == 5:
# string object
return readStringFromStream(stream)
elif idx == 6:
# null object
return NullObject.readFromStream(stream)
elif idx == 7:
# comment
while tok not in (b_('\r'), b_('\n')):
tok = stream.read(1)
tok = readNonWhitespace(stream)
stream.seek(-1, 1)
return readObject(stream, pdf)
else:
# number object OR indirect reference
if tok in NumberSigns:
# number
return NumberObject.readFromStream(stream)
peek = stream.read(20)
stream.seek(-len(peek), 1) # reset to start
if IndirectPattern.match(peek) != None:
return IndirectObject.readFromStream(stream, pdf)
else:
return NumberObject.readFromStream(stream)
class PdfObject(object):
def getObject(self):
"""Resolves indirect references."""
return self
class NullObject(PdfObject):
def writeToStream(self, stream, encryption_key):
stream.write(b_("null"))
def readFromStream(stream):
nulltxt = stream.read(4)
if nulltxt != b_("null"):
raise utils.PdfReadError("Could not read Null object")
return NullObject()
readFromStream = staticmethod(readFromStream)
class BooleanObject(PdfObject):
def __init__(self, value):
self.value = value
def writeToStream(self, stream, encryption_key):
if self.value:
stream.write(b_("true"))
else:
stream.write(b_("false"))
def readFromStream(stream):
word = stream.read(4)
if word == b_("true"):
return BooleanObject(True)
elif word == b_("fals"):
stream.read(1)
return BooleanObject(False)
else:
raise utils.PdfReadError('Could not read Boolean object')
readFromStream = staticmethod(readFromStream)
class ArrayObject(list, PdfObject):
def writeToStream(self, stream, encryption_key):
stream.write(b_("["))
for data in self:
stream.write(b_(" "))
data.writeToStream(stream, encryption_key)
stream.write(b_(" ]"))
def readFromStream(stream, pdf):
arr = ArrayObject()
tmp = stream.read(1)
if tmp != b_("["):
raise utils.PdfReadError("Could not read array")
while True:
# skip leading whitespace
tok = stream.read(1)
while tok.isspace():
tok = stream.read(1)
stream.seek(-1, 1)
# check for array ending
peekahead = stream.read(1)
if peekahead == b_("]"):
break
stream.seek(-1, 1)
# read and append obj
arr.append(readObject(stream, pdf))
return arr
readFromStream = staticmethod(readFromStream)
class IndirectObject(PdfObject):
def __init__(self, idnum, generation, pdf):
self.idnum = idnum
self.generation = generation
self.pdf = pdf
def getObject(self):
return self.pdf.getObject(self).getObject()
def __repr__(self):
return "IndirectObject(%r, %r)" % (self.idnum, self.generation)
def __eq__(self, other):
return (
other != None and
isinstance(other, IndirectObject) and
self.idnum == other.idnum and
self.generation == other.generation and
self.pdf is other.pdf
)
def __ne__(self, other):
return not self.__eq__(other)
def writeToStream(self, stream, encryption_key):
stream.write(b_("%s %s R" % (self.idnum, self.generation)))
def readFromStream(stream, pdf):
idnum = b_("")
while True:
tok = stream.read(1)
if not tok:
# stream has truncated prematurely
raise PdfStreamError("Stream has ended unexpectedly")
if tok.isspace():
break
idnum += tok
generation = b_("")
while True:
tok = stream.read(1)
if not tok:
# stream has truncated prematurely
raise PdfStreamError("Stream has ended unexpectedly")
if tok.isspace():
if not generation:
continue
break
generation += tok
r = readNonWhitespace(stream)
if r != b_("R"):
raise utils.PdfReadError("Error reading indirect object reference at byte %s" % utils.hexStr(stream.tell()))
return IndirectObject(int(idnum), int(generation), pdf)
readFromStream = staticmethod(readFromStream)
class FloatObject(decimal.Decimal, PdfObject):
def __new__(cls, value="0", context=None):
try:
return decimal.Decimal.__new__(cls, utils.str_(value), context)
except:
return decimal.Decimal.__new__(cls, str(value))
def __repr__(self):
if self == self.to_integral():
return str(self.quantize(decimal.Decimal(1)))
else:
# Standard formatting adds useless extraneous zeros.
o = "%.5f" % self
# Remove the zeros.
while o and o[-1] == '0':
o = o[:-1]
return o
def as_numeric(self):
return float(b_(repr(self)))
def writeToStream(self, stream, encryption_key):
stream.write(b_(repr(self)))
class NumberObject(int, PdfObject):
NumberPattern = re.compile(b_('[^+-.0-9]'))
ByteDot = b_(".")
def __new__(cls, value):
val = int(value)
try:
return int.__new__(cls, val)
except OverflowError:
return int.__new__(cls, 0)
def as_numeric(self):
return int(b_(repr(self)))
def writeToStream(self, stream, encryption_key):
stream.write(b_(repr(self)))
def readFromStream(stream):
num = utils.readUntilRegex(stream, NumberObject.NumberPattern)
if num.find(NumberObject.ByteDot) != -1:
return FloatObject(num)
else:
return NumberObject(num)
readFromStream = staticmethod(readFromStream)
##
# Given a string (either a "str" or "unicode"), create a ByteStringObject or a
# TextStringObject to represent the string.
def createStringObject(string):
if isinstance(string, utils.string_type):
return TextStringObject(string)
elif isinstance(string, utils.bytes_type):
try:
if string.startswith(codecs.BOM_UTF16_BE):
retval = TextStringObject(string.decode("utf-16"))
retval.autodetect_utf16 = True
return retval
else:
# This is probably a big performance hit here, but we need to
# convert string objects into the text/unicode-aware version if
# possible... and the only way to check if that's possible is
# to try. Some strings are strings, some are just byte arrays.
retval = TextStringObject(decode_pdfdocencoding(string))
retval.autodetect_pdfdocencoding = True
return retval
except UnicodeDecodeError:
return ByteStringObject(string)
else:
raise TypeError("createStringObject should have str or unicode arg")
def readHexStringFromStream(stream):
stream.read(1)
txt = ""
x = b_("")
while True:
tok = readNonWhitespace(stream)
if not tok:
# stream has truncated prematurely
raise PdfStreamError("Stream has ended unexpectedly")
if tok == b_(">"):
break
x += tok
if len(x) == 2:
txt += chr(int(x, base=16))
x = b_("")
if len(x) == 1:
x += b_("0")
if len(x) == 2:
txt += chr(int(x, base=16))
return createStringObject(b_(txt))
def readStringFromStream(stream):
tok = stream.read(1)
parens = 1
txt = b_("")
while True:
tok = stream.read(1)
if not tok:
# stream has truncated prematurely
raise PdfStreamError("Stream has ended unexpectedly")
if tok == b_("("):
parens += 1
elif tok == b_(")"):
parens -= 1
if parens == 0:
break
elif tok == b_("\\"):
tok = stream.read(1)
if tok == b_("n"):
tok = b_("\n")
elif tok == b_("r"):
tok = b_("\r")
elif tok == b_("t"):
tok = b_("\t")
elif tok == b_("b"):
tok = b_("\b")
elif tok == b_("f"):
tok = b_("\f")
elif tok == b_("c"):
tok = b_("\c")
elif tok == b_("("):
tok = b_("(")
elif tok == b_(")"):
tok = b_(")")
elif tok == b_("/"):
tok = b_("/")
elif tok == b_("\\"):
tok = b_("\\")
elif tok in (b_(" "), b_("/"), b_("%"), b_("<"), b_(">"), b_("["),
b_("]"), b_("#"), b_("_"), b_("&"), b_('$')):
# odd/unnessecary escape sequences we have encountered
tok = b_(tok)
elif tok.isdigit():
# "The number ddd may consist of one, two, or three
# octal digits; high-order overflow shall be ignored.
# Three octal digits shall be used, with leading zeros
# as needed, if the next character of the string is also
# a digit." (PDF reference 7.3.4.2, p 16)
for i in range(2):
ntok = stream.read(1)
if ntok.isdigit():
tok += ntok
else:
break
tok = b_(chr(int(tok, base=8)))
elif tok in b_("\n\r"):
# This case is hit when a backslash followed by a line
# break occurs. If it's a multi-char EOL, consume the
# second character:
tok = stream.read(1)
if not tok in b_("\n\r"):
stream.seek(-1, 1)
# Then don't add anything to the actual string, since this
# line break was escaped:
tok = b_('')
else:
raise utils.PdfReadError(r"Unexpected escaped string: %s" % tok)
txt += tok
return createStringObject(txt)
##
# Represents a string object where the text encoding could not be determined.
# This occurs quite often, as the PDF spec doesn't provide an alternate way to
# represent strings -- for example, the encryption data stored in files (like
# /O) is clearly not text, but is still stored in a "String" object.
class ByteStringObject(utils.bytes_type, PdfObject):
##
# For compatibility with TextStringObject.original_bytes. This method
# returns self.
original_bytes = property(lambda self: self)
def writeToStream(self, stream, encryption_key):
bytearr = self
if encryption_key:
bytearr = RC4_encrypt(encryption_key, bytearr)
stream.write(b_("<"))
stream.write(utils.hexencode(bytearr))
stream.write(b_(">"))
##
# Represents a string object that has been decoded into a real unicode string.
# If read from a PDF document, this string appeared to match the
# PDFDocEncoding, or contained a UTF-16BE BOM mark to cause UTF-16 decoding to
# occur.
class TextStringObject(utils.string_type, PdfObject):
autodetect_pdfdocencoding = False
autodetect_utf16 = False
##
# It is occasionally possible that a text string object gets created where
# a byte string object was expected due to the autodetection mechanism --
# if that occurs, this "original_bytes" property can be used to
# back-calculate what the original encoded bytes were.
original_bytes = property(lambda self: self.get_original_bytes())
def get_original_bytes(self):
# We're a text string object, but the library is trying to get our raw
# bytes. This can happen if we auto-detected this string as text, but
# we were wrong. It's pretty common. Return the original bytes that
# would have been used to create this object, based upon the autodetect
# method.
if self.autodetect_utf16:
return codecs.BOM_UTF16_BE + self.encode("utf-16be")
elif self.autodetect_pdfdocencoding:
return encode_pdfdocencoding(self)
else:
raise Exception("no information about original bytes")
def writeToStream(self, stream, encryption_key):
# Try to write the string out as a PDFDocEncoding encoded string. It's
# nicer to look at in the PDF file. Sadly, we take a performance hit
# here for trying...
try:
bytearr = encode_pdfdocencoding(self)
except UnicodeEncodeError:
bytearr = codecs.BOM_UTF16_BE + self.encode("utf-16be")
if encryption_key:
bytearr = RC4_encrypt(encryption_key, bytearr)
obj = ByteStringObject(bytearr)
obj.writeToStream(stream, None)
else:
stream.write(b_("("))
for c in bytearr:
if not chr_(c).isalnum() and c != b_(' '):
stream.write(b_("\\%03o" % ord_(c)))
else:
stream.write(b_(chr_(c)))
stream.write(b_(")"))
class NameObject(str, PdfObject):
delimiterPattern = re.compile(b_(r"\s+|[\(\)<>\[\]{}/%]"))
surfix = b_("/")
def writeToStream(self, stream, encryption_key):
stream.write(b_(self))
def readFromStream(stream, pdf):
debug = False
if debug: print((stream.tell()))
name = stream.read(1)
if name != NameObject.surfix:
raise utils.PdfReadError("name read error")
name += utils.readUntilRegex(stream, NameObject.delimiterPattern,
ignore_eof=True)
if debug: print(name)
try:
return NameObject(name.decode('utf-8'))
except (UnicodeEncodeError, UnicodeDecodeError) as e:
# Name objects should represent irregular characters
# with a '#' followed by the symbol's hex number
if not pdf.strict:
warnings.warn("Illegal character in Name Object", utils.PdfReadWarning)
return NameObject(name)
else:
raise utils.PdfReadError("Illegal character in Name Object")
readFromStream = staticmethod(readFromStream)
class DictionaryObject(dict, PdfObject):
def raw_get(self, key):
return dict.__getitem__(self, key)
def __setitem__(self, key, value):
if not isinstance(key, PdfObject):
raise ValueError("key must be PdfObject")
if not isinstance(value, PdfObject):
raise ValueError("value must be PdfObject")
return dict.__setitem__(self, key, value)
def setdefault(self, key, value=None):
if not isinstance(key, PdfObject):
raise ValueError("key must be PdfObject")
if not isinstance(value, PdfObject):
raise ValueError("value must be PdfObject")
return dict.setdefault(self, key, value)
def __getitem__(self, key):
return dict.__getitem__(self, key).getObject()
##
# Retrieves XMP (Extensible Metadata Platform) data relevant to the
# this object, if available.
# <p>
# Stability: Added in v1.12, will exist for all future v1.x releases.
# @return Returns a {@link #xmp.XmpInformation XmlInformation} instance
# that can be used to access XMP metadata from the document. Can also
# return None if no metadata was found on the document root.
def getXmpMetadata(self):
metadata = self.get("/Metadata", None)
if metadata == None:
return None
metadata = metadata.getObject()
from . import xmp
if not isinstance(metadata, xmp.XmpInformation):
metadata = xmp.XmpInformation(metadata)
self[NameObject("/Metadata")] = metadata
return metadata
##
# Read-only property that accesses the {@link
# #DictionaryObject.getXmpData getXmpData} function.
# <p>
# Stability: Added in v1.12, will exist for all future v1.x releases.
xmpMetadata = property(lambda self: self.getXmpMetadata(), None, None)
def writeToStream(self, stream, encryption_key):
stream.write(b_("<<\n"))
for key, value in list(self.items()):
key.writeToStream(stream, encryption_key)
stream.write(b_(" "))
value.writeToStream(stream, encryption_key)
stream.write(b_("\n"))
stream.write(b_(">>"))
def readFromStream(stream, pdf):
debug = False
tmp = stream.read(2)
if tmp != b_("<<"):
raise utils.PdfReadError("Dictionary read error at byte %s: stream must begin with '<<'" % utils.hexStr(stream.tell()))
data = {}
while True:
tok = readNonWhitespace(stream)
if tok == b_('\x00'):
continue
elif tok == b_('%'):
stream.seek(-1, 1)
skipOverComment(stream)
continue
if not tok:
# stream has truncated prematurely
raise PdfStreamError("Stream has ended unexpectedly")
if debug: print(("Tok:", tok))
if tok == b_(">"):
stream.read(1)
break
stream.seek(-1, 1)
key = readObject(stream, pdf)
tok = readNonWhitespace(stream)
stream.seek(-1, 1)
value = readObject(stream, pdf)
if not data.get(key):
data[key] = value
elif pdf.strict:
# multiple definitions of key not permitted
raise utils.PdfReadError("Multiple definitions in dictionary at byte %s for key %s" \
% (utils.hexStr(stream.tell()), key))
else:
warnings.warn("Multiple definitions in dictionary at byte %s for key %s" \
% (utils.hexStr(stream.tell()), key), utils.PdfReadWarning)
pos = stream.tell()
s = readNonWhitespace(stream)
if s == b_('s') and stream.read(5) == b_('tream'):
eol = stream.read(1)
# odd PDF file output has spaces after 'stream' keyword but before EOL.
# patch provided by Danial Sandler
while eol == b_(' '):
eol = stream.read(1)
assert eol in (b_("\n"), b_("\r"))
if eol == b_("\r"):
# read \n after
if stream.read(1) != b_('\n'):
stream.seek(-1, 1)
# this is a stream object, not a dictionary
assert "/Length" in data
length = data["/Length"]
if debug: print(data)
if isinstance(length, IndirectObject):
t = stream.tell()
length = pdf.getObject(length)
stream.seek(t, 0)
data["__streamdata__"] = stream.read(length)
if debug: print("here")
#if debug: print(binascii.hexlify(data["__streamdata__"]))
e = readNonWhitespace(stream)
ndstream = stream.read(8)
if (e + ndstream) != b_("endstream"):
# (sigh) - the odd PDF file has a length that is too long, so
# we need to read backwards to find the "endstream" ending.
# ReportLab (unknown version) generates files with this bug,
# and Python users into PDF files tend to be our audience.
# we need to do this to correct the streamdata and chop off
# an extra character.
pos = stream.tell()
stream.seek(-10, 1)
end = stream.read(9)
if end == b_("endstream"):
# we found it by looking back one character further.
data["__streamdata__"] = data["__streamdata__"][:-1]
else:
if debug: print(("E", e, ndstream, debugging.toHex(end)))
stream.seek(pos, 0)
raise utils.PdfReadError("Unable to find 'endstream' marker after stream at byte %s." % utils.hexStr(stream.tell()))
else:
stream.seek(pos, 0)
if "__streamdata__" in data:
return StreamObject.initializeFromDictionary(data)
else:
retval = DictionaryObject()
retval.update(data)
return retval
readFromStream = staticmethod(readFromStream)
class TreeObject(DictionaryObject):
def __init__(self):
DictionaryObject.__init__(self)
def hasChildren(self):
return '/First' in self
def __iter__(self):
return self.children()
def children(self):
if not self.hasChildren():
raise StopIteration
child = self['/First']
while True:
yield child
if child == self['/Last']:
raise StopIteration
child = child['/Next']
def addChild(self, child, pdf):
childObj = child.getObject()
child = pdf.getReference(childObj)
assert isinstance(child, IndirectObject)
if '/First' not in self:
self[NameObject('/First')] = child
self[NameObject('/Count')] = NumberObject(0)
prev = None
else:
prev = self['/Last']
self[NameObject('/Last')] = child
self[NameObject('/Count')] = NumberObject(self[NameObject('/Count')] + 1)
if prev:
prevRef = pdf.getReference(prev)
assert isinstance(prevRef, IndirectObject)
childObj[NameObject('/Prev')] = prevRef
prev[NameObject('/Next')] = child
parentRef = pdf.getReference(self)
assert isinstance(parentRef, IndirectObject)
childObj[NameObject('/Parent')] = parentRef
def removeChild(self, child):
childObj = child.getObject()
if NameObject('/Parent') not in childObj:
raise ValueError("Removed child does not appear to be a tree item")
elif childObj[NameObject('/Parent')] != self:
raise ValueError("Removed child is not a member of this tree")
found = False
prevRef = None
prev = None
curRef = self[NameObject('/First')]
cur = curRef.getObject()
lastRef = self[NameObject('/Last')]
last = lastRef.getObject()
while cur != None:
if cur == childObj:
if prev == None:
if NameObject('/Next') in cur:
# Removing first tree node
nextRef = cur[NameObject('/Next')]
next = nextRef.getObject()
del next[NameObject('/Prev')]
self[NameObject('/First')] = nextRef
self[NameObject('/Count')] = self[NameObject('/Count')] - 1
else:
# Removing only tree node
assert self[NameObject('/Count')] == 1
del self[NameObject('/Count')]
del self[NameObject('/First')]
if NameObject('/Last') in self:
del self[NameObject('/Last')]
else:
if NameObject('/Next') in cur:
# Removing middle tree node
nextRef = cur[NameObject('/Next')]
next = nextRef.getObject()
next[NameObject('/Prev')] = prevRef
prev[NameObject('/Next')] = nextRef
self[NameObject('/Count')] = self[NameObject('/Count')] - 1
else:
# Removing last tree node
assert cur == last
del prev[NameObject('/Next')]
self[NameObject('/Last')] = prevRef
self[NameObject('/Count')] = self[NameObject('/Count')] - 1
found = True
break
prevRef = curRef
prev = cur
if NameObject('/Next') in cur:
curRef = cur[NameObject('/Next')]
cur = curRef.getObject()
else:
curRef = None
cur = None
if not found:
raise ValueError("Removal couldn't find item in tree")
del childObj[NameObject('/Parent')]
if NameObject('/Next') in childObj:
del childObj[NameObject('/Next')]
if NameObject('/Prev') in childObj:
del childObj[NameObject('/Prev')]
def emptyTree(self):
for child in self:
childObj = child.getObject()
del childObj[NameObject('/Parent')]
if NameObject('/Next') in childObj:
del childObj[NameObject('/Next')]
if NameObject('/Prev') in childObj:
del childObj[NameObject('/Prev')]
if NameObject('/Count') in self:
del self[NameObject('/Count')]
if NameObject('/First') in self:
del self[NameObject('/First')]
if NameObject('/Last') in self:
del self[NameObject('/Last')]
class StreamObject(DictionaryObject):
def __init__(self):
self._data = None
self.decodedSelf = None
def writeToStream(self, stream, encryption_key):
self[NameObject("/Length")] = NumberObject(len(self._data))
DictionaryObject.writeToStream(self, stream, encryption_key)
del self["/Length"]
stream.write(b_("\nstream\n"))
data = self._data
if encryption_key:
data = RC4_encrypt(encryption_key, data)
stream.write(data)
stream.write(b_("\nendstream"))
def initializeFromDictionary(data):
if "/Filter" in data:
retval = EncodedStreamObject()
else:
retval = DecodedStreamObject()
retval._data = data["__streamdata__"]
del data["__streamdata__"]
del data["/Length"]
retval.update(data)
return retval
initializeFromDictionary = staticmethod(initializeFromDictionary)
def flateEncode(self):
if "/Filter" in self:
f = self["/Filter"]
if isinstance(f, ArrayObject):
f.insert(0, NameObject("/FlateDecode"))
else:
newf = ArrayObject()
newf.append(NameObject("/FlateDecode"))
newf.append(f)
f = newf
else:
f = NameObject("/FlateDecode")
retval = EncodedStreamObject()
retval[NameObject("/Filter")] = f
retval._data = filters.FlateDecode.encode(self._data)
return retval
class DecodedStreamObject(StreamObject):
def getData(self):
return self._data
def setData(self, data):
self._data = data
class EncodedStreamObject(StreamObject):
def __init__(self):
self.decodedSelf = None
def getData(self):
if self.decodedSelf:
# cached version of decoded object
return self.decodedSelf.getData()
else:
# create decoded object
decoded = DecodedStreamObject()
decoded._data = filters.decodeStreamData(self)
for key, value in list(self.items()):
if not key in ("/Length", "/Filter", "/DecodeParms"):
decoded[key] = value
self.decodedSelf = decoded
return decoded._data
def setData(self, data):
raise utils.PdfReadError("Creating EncodedStreamObject is not currently supported")
class RectangleObject(ArrayObject):
"""
This class is used to represent *page boxes* in PyPDF2. These boxes include:
* :attr:`artBox <PyPDF2.pdf.PageObject.artBox>`
* :attr:`bleedBox <PyPDF2.pdf.PageObject.bleedBox>`
* :attr:`cropBox <PyPDF2.pdf.PageObject.cropBox>`
* :attr:`mediaBox <PyPDF2.pdf.PageObject.mediaBox>`
* :attr:`trimBox <PyPDF2.pdf.PageObject.trimBox>`
"""
def __init__(self, arr):
# must have four points
assert len(arr) == 4
# automatically convert arr[x] into NumberObject(arr[x]) if necessary
ArrayObject.__init__(self, [self.ensureIsNumber(x) for x in arr])
def ensureIsNumber(self, value):
if not isinstance(value, (NumberObject, FloatObject)):
value = FloatObject(value)
return value
def __repr__(self):
return "RectangleObject(%s)" % repr(list(self))
def getLowerLeft_x(self):
return self[0]
def getLowerLeft_y(self):
return self[1]
def getUpperRight_x(self):
return self[2]
def getUpperRight_y(self):
return self[3]
def getUpperLeft_x(self):
return self.getLowerLeft_x()
def getUpperLeft_y(self):
return self.getUpperRight_y()
def getLowerRight_x(self):
return self.getUpperRight_x()
def getLowerRight_y(self):
return self.getLowerLeft_y()
def getLowerLeft(self):
return self.getLowerLeft_x(), self.getLowerLeft_y()
def getLowerRight(self):
return self.getLowerRight_x(), self.getLowerRight_y()
def getUpperLeft(self):
return self.getUpperLeft_x(), self.getUpperLeft_y()
def getUpperRight(self):
return self.getUpperRight_x(), self.getUpperRight_y()
def setLowerLeft(self, value):
self[0], self[1] = [self.ensureIsNumber(x) for x in value]
def setLowerRight(self, value):
self[2], self[1] = [self.ensureIsNumber(x) for x in value]
def setUpperLeft(self, value):
self[0], self[3] = [self.ensureIsNumber(x) for x in value]
def setUpperRight(self, value):
self[2], self[3] = [self.ensureIsNumber(x) for x in value]
def getWidth(self):
return self.getUpperRight_x() - self.getLowerLeft_x()
def getHeight(self):
return self.getUpperRight_y() - self.getLowerLeft_y()
lowerLeft = property(getLowerLeft, setLowerLeft, None, None)
"""
Property to read and modify the lower left coordinate of this box
in (x,y) form.
"""
lowerRight = property(getLowerRight, setLowerRight, None, None)
"""
Property to read and modify the lower right coordinate of this box
in (x,y) form.
"""
upperLeft = property(getUpperLeft, setUpperLeft, None, None)
"""
Property to read and modify the upper left coordinate of this box
in (x,y) form.
"""
upperRight = property(getUpperRight, setUpperRight, None, None)
"""
Property to read and modify the upper right coordinate of this box
in (x,y) form.
"""
class Field(TreeObject):
"""
A class representing a field dictionary. This class is accessed through
:meth:`getFields()<PyPDF2.PdfFileReader.getFields>`
"""
def __init__(self, data):
DictionaryObject.__init__(self)
attributes = ("/FT", "/Parent", "/Kids", "/T", "/TU", "/TM", "/Ff",
"/V", "/DV", "/AA")
for attr in attributes:
try:
self[NameObject(attr)] = data[attr]
except KeyError:
pass
fieldType = property(lambda self: self.get("/FT"))
"""
Read-only property accessing the type of this field.
"""
parent = property(lambda self: self.get("/Parent"))
"""
Read-only property accessing the parent of this field.
"""
kids = property(lambda self: self.get("/Kids"))
"""
Read-only property accessing the kids of this field.
"""
name = property(lambda self: self.get("/T"))
"""
Read-only property accessing the name of this field.
"""
altName = property(lambda self: self.get("/TU"))
"""
Read-only property accessing the alternate name of this field.
"""
mappingName = property(lambda self: self.get("/TM"))
"""
Read-only property accessing the mapping name of this field. This
name is used by PyPDF2 as a key in the dictionary returned by
:meth:`getFields()<PyPDF2.PdfFileReader.getFields>`
"""
flags = property(lambda self: self.get("/Ff"))
"""
Read-only property accessing the field flags, specifying various
characteristics of the field (see Table 8.70 of the PDF 1.7 reference).
"""
value = property(lambda self: self.get("/V"))
"""
Read-only property accessing the value of this field. Format
varies based on field type.
"""
defaultValue = property(lambda self: self.get("/DV"))
"""
Read-only property accessing the default value of this field.
"""
additionalActions = property(lambda self: self.get("/AA"))
"""
Read-only property accessing the additional actions dictionary.
This dictionary defines the field's behavior in response to trigger events.
See Section 8.5.2 of the PDF 1.7 reference.
"""
class Destination(TreeObject):
"""
A class representing a destination within a PDF file.
See section 8.2.1 of the PDF 1.6 reference.
:param str title: Title of this destination.
:param int page: Page number of this destination.
:param str typ: How the destination is displayed.
:param args: Additional arguments may be necessary depending on the type.
:raises PdfReadError: If destination type is invalid.
Valid ``typ`` arguments (see PDF spec for details):
/Fit No additional arguments
/XYZ [left] [top] [zoomFactor]
/FitH [top]
/FitV [left]
/FitR [left] [bottom] [right] [top]
/FitB No additional arguments
/FitBH [top]
/FitBV [left]
"""
def __init__(self, title, page, typ, *args):
DictionaryObject.__init__(self)
self[NameObject("/Title")] = title
self[NameObject("/Page")] = page
self[NameObject("/Type")] = typ
# from table 8.2 of the PDF 1.7 reference.
if typ == "/XYZ":
(self[NameObject("/Left")], self[NameObject("/Top")],
self[NameObject("/Zoom")]) = args
elif typ == "/FitR":
(self[NameObject("/Left")], self[NameObject("/Bottom")],
self[NameObject("/Right")], self[NameObject("/Top")]) = args
elif typ in ["/FitH", "/FitBH"]:
self[NameObject("/Top")], = args
elif typ in ["/FitV", "/FitBV"]:
self[NameObject("/Left")], = args
elif typ in ["/Fit", "/FitB"]:
pass
else:
raise utils.PdfReadError("Unknown Destination Type: %r" % typ)
def getDestArray(self):
return ArrayObject([self.raw_get('/Page'), self['/Type']] + [self[x] for x in ['/Left', '/Bottom', '/Right', '/Top', '/Zoom'] if x in self])
def writeToStream(self, stream, encryption_key):
stream.write(b_("<<\n"))
key = NameObject('/D')
key.writeToStream(stream, encryption_key)
stream.write(b_(" "))
value = self.getDestArray()
value.writeToStream(stream, encryption_key)
key = NameObject("/S")
key.writeToStream(stream, encryption_key)
stream.write(b_(" "))
value = NameObject("/GoTo")
value.writeToStream(stream, encryption_key)
stream.write(b_("\n"))
stream.write(b_(">>"))
title = property(lambda self: self.get("/Title"))
"""
Read-only property accessing the destination title.
:rtype: str
"""
page = property(lambda self: self.get("/Page"))
"""
Read-only property accessing the destination page number.
:rtype: int
"""
typ = property(lambda self: self.get("/Type"))
"""
Read-only property accessing the destination type.
:rtype: str
"""
zoom = property(lambda self: self.get("/Zoom", None))
"""
Read-only property accessing the zoom factor.
:rtype: int, or ``None`` if not available.
"""
left = property(lambda self: self.get("/Left", None))
"""
Read-only property accessing the left horizontal coordinate.
:rtype: int, or ``None`` if not available.
"""
right = property(lambda self: self.get("/Right", None))
"""
Read-only property accessing the right horizontal coordinate.
:rtype: int, or ``None`` if not available.
"""
top = property(lambda self: self.get("/Top", None))
"""
Read-only property accessing the top vertical coordinate.
:rtype: int, or ``None`` if not available.
"""
bottom = property(lambda self: self.get("/Bottom", None))
"""
Read-only property accessing the bottom vertical coordinate.
:rtype: int, or ``None`` if not available.
"""
class Bookmark(Destination):
def writeToStream(self, stream, encryption_key):
stream.write(b_("<<\n"))
for key in [NameObject(x) for x in ['/Title', '/Parent', '/First', '/Last', '/Next', '/Prev'] if x in self]:
key.writeToStream(stream, encryption_key)
stream.write(b_(" "))
value = self.raw_get(key)
value.writeToStream(stream, encryption_key)
stream.write(b_("\n"))
key = NameObject('/Dest')
key.writeToStream(stream, encryption_key)
stream.write(b_(" "))
value = self.getDestArray()
value.writeToStream(stream, encryption_key)
stream.write(b_("\n"))
stream.write(b_(">>"))
def encode_pdfdocencoding(unicode_string):
retval = b_('')
for c in unicode_string:
try:
retval += b_(chr(_pdfDocEncoding_rev[c]))
except KeyError:
raise UnicodeEncodeError("pdfdocencoding", c, -1, -1,
"does not exist in translation table")
return retval
def decode_pdfdocencoding(byte_array):
retval = u_('')
for b in byte_array:
c = _pdfDocEncoding[ord_(b)]
if c == u_('\u0000'):
raise UnicodeDecodeError("pdfdocencoding", utils.barray(b), -1, -1,
"does not exist in translation table")
retval += c
return retval
_pdfDocEncoding = (
u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'),
u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'),
u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'), u_('\u0000'),
u_('\u02d8'), u_('\u02c7'), u_('\u02c6'), u_('\u02d9'), u_('\u02dd'), u_('\u02db'), u_('\u02da'), u_('\u02dc'),
u_('\u0020'), u_('\u0021'), u_('\u0022'), u_('\u0023'), u_('\u0024'), u_('\u0025'), u_('\u0026'), u_('\u0027'),
u_('\u0028'), u_('\u0029'), u_('\u002a'), u_('\u002b'), u_('\u002c'), u_('\u002d'), u_('\u002e'), u_('\u002f'),
u_('\u0030'), u_('\u0031'), u_('\u0032'), u_('\u0033'), u_('\u0034'), u_('\u0035'), u_('\u0036'), u_('\u0037'),
u_('\u0038'), u_('\u0039'), u_('\u003a'), u_('\u003b'), u_('\u003c'), u_('\u003d'), u_('\u003e'), u_('\u003f'),
u_('\u0040'), u_('\u0041'), u_('\u0042'), u_('\u0043'), u_('\u0044'), u_('\u0045'), u_('\u0046'), u_('\u0047'),
u_('\u0048'), u_('\u0049'), u_('\u004a'), u_('\u004b'), u_('\u004c'), u_('\u004d'), u_('\u004e'), u_('\u004f'),
u_('\u0050'), u_('\u0051'), u_('\u0052'), u_('\u0053'), u_('\u0054'), u_('\u0055'), u_('\u0056'), u_('\u0057'),
u_('\u0058'), u_('\u0059'), u_('\u005a'), u_('\u005b'), u_('\u005c'), u_('\u005d'), u_('\u005e'), u_('\u005f'),
u_('\u0060'), u_('\u0061'), u_('\u0062'), u_('\u0063'), u_('\u0064'), u_('\u0065'), u_('\u0066'), u_('\u0067'),
u_('\u0068'), u_('\u0069'), u_('\u006a'), u_('\u006b'), u_('\u006c'), u_('\u006d'), u_('\u006e'), u_('\u006f'),
u_('\u0070'), u_('\u0071'), u_('\u0072'), u_('\u0073'), u_('\u0074'), u_('\u0075'), u_('\u0076'), u_('\u0077'),
u_('\u0078'), u_('\u0079'), u_('\u007a'), u_('\u007b'), u_('\u007c'), u_('\u007d'), u_('\u007e'), u_('\u0000'),
u_('\u2022'), u_('\u2020'), u_('\u2021'), u_('\u2026'), u_('\u2014'), u_('\u2013'), u_('\u0192'), u_('\u2044'),
u_('\u2039'), u_('\u203a'), u_('\u2212'), u_('\u2030'), u_('\u201e'), u_('\u201c'), u_('\u201d'), u_('\u2018'),
u_('\u2019'), u_('\u201a'), u_('\u2122'), u_('\ufb01'), u_('\ufb02'), u_('\u0141'), u_('\u0152'), u_('\u0160'),
u_('\u0178'), u_('\u017d'), u_('\u0131'), u_('\u0142'), u_('\u0153'), u_('\u0161'), u_('\u017e'), u_('\u0000'),
u_('\u20ac'), u_('\u00a1'), u_('\u00a2'), u_('\u00a3'), u_('\u00a4'), u_('\u00a5'), u_('\u00a6'), u_('\u00a7'),
u_('\u00a8'), u_('\u00a9'), u_('\u00aa'), u_('\u00ab'), u_('\u00ac'), u_('\u0000'), u_('\u00ae'), u_('\u00af'),
u_('\u00b0'), u_('\u00b1'), u_('\u00b2'), u_('\u00b3'), u_('\u00b4'), u_('\u00b5'), u_('\u00b6'), u_('\u00b7'),
u_('\u00b8'), u_('\u00b9'), u_('\u00ba'), u_('\u00bb'), u_('\u00bc'), u_('\u00bd'), u_('\u00be'), u_('\u00bf'),
u_('\u00c0'), u_('\u00c1'), u_('\u00c2'), u_('\u00c3'), u_('\u00c4'), u_('\u00c5'), u_('\u00c6'), u_('\u00c7'),
u_('\u00c8'), u_('\u00c9'), u_('\u00ca'), u_('\u00cb'), u_('\u00cc'), u_('\u00cd'), u_('\u00ce'), u_('\u00cf'),
u_('\u00d0'), u_('\u00d1'), u_('\u00d2'), u_('\u00d3'), u_('\u00d4'), u_('\u00d5'), u_('\u00d6'), u_('\u00d7'),
u_('\u00d8'), u_('\u00d9'), u_('\u00da'), u_('\u00db'), u_('\u00dc'), u_('\u00dd'), u_('\u00de'), u_('\u00df'),
u_('\u00e0'), u_('\u00e1'), u_('\u00e2'), u_('\u00e3'), u_('\u00e4'), u_('\u00e5'), u_('\u00e6'), u_('\u00e7'),
u_('\u00e8'), u_('\u00e9'), u_('\u00ea'), u_('\u00eb'), u_('\u00ec'), u_('\u00ed'), u_('\u00ee'), u_('\u00ef'),
u_('\u00f0'), u_('\u00f1'), u_('\u00f2'), u_('\u00f3'), u_('\u00f4'), u_('\u00f5'), u_('\u00f6'), u_('\u00f7'),
u_('\u00f8'), u_('\u00f9'), u_('\u00fa'), u_('\u00fb'), u_('\u00fc'), u_('\u00fd'), u_('\u00fe'), u_('\u00ff')
)
assert len(_pdfDocEncoding) == 256
_pdfDocEncoding_rev = {}
for i in range(256):
char = _pdfDocEncoding[i]
if char == u_("\u0000"):
continue
assert char not in _pdfDocEncoding_rev
_pdfDocEncoding_rev[char] = i