def get_obj(self, obj_num, gen_num):
if not self.ready:
raise Exception(
'get_obj can only be called after the document is scanned completely.'
)
startxref_found = False
startxref = -1
for increment in range(len(self.increments)):
increment = -(increment + 1) # increment from -1 to -len
xref_section = self.increments[increment]['xref_section']
offset = xref_section.get_obj_offset(obj_num, gen_num)
if offset is None:
# offset is None <=> obj_num not found
continue
elif isinstance(offset, tuple):
# TODO: handle the case where obj is not already cached
return self.compressed_obj[offset]
elif offset > 0:
if self.offset_obj.get(offset) is None:
temp_f = open(self.__f.name, 'rb')
temp_f.seek(offset, io.SEEK_SET)
self.offset_obj[offset] = PdfObject.create_from_file(
temp_f, self)
return self.offset_obj[offset]
elif offset == 0:
# offset = 0 <=> obj_num is free at gen_num
return None
raise Exception('Object not found')
def decode_objstm(objstmobj, doc):
result = {}
streamObj = objstmobj.value
objstmobj_no = objstmobj.obj_no
if not isinstance(streamObj, PdfStreamObject):
raise ValueError(
'objstmobj is not a PdfIndirectObject containing a PdfStreamObject'
)
objbytestream = io.BufferedReader(io.BytesIO(streamObj.decode()))
# N pairs of integers
# 1st int is obj no of the compressed object
# 2nd int is byte offset of that object, relative to the first obj
objbytestream.seek(0, io.SEEK_SET)
utils.seek_until(objbytestream,
syntax.NON_WHITESPACES,
ignore_comment=True)
numbers = []
N = 0
First = 0
try:
# TODO: assuming both N and First have direct obj values
N = int(str(streamObj.dict['N'].value))
First = int(str(streamObj.dict['First'].value))
if N < 0 or First < 0:
raise Exception(
f'Invalid N or First field in ObjStm at offset {org_pos}.')
except Exception as ex:
raise Exception(
f'Invalid N or First field in ObjStm at offset {org_pos}.') from ex
for _ in range(2 * N):
utils.seek_until(objbytestream,
syntax.NON_WHITESPACES,
ignore_comment=True)
numobj = PdfNumericObject.create_from_file(objbytestream)
try:
temp = int(str(numobj.value))
if temp < 0:
raise Exception(
f'Invalid obj no./offset in ObjStm at offset {org_pos}.')
numbers += [temp]
except Exception as ex:
raise Exception(f'Invalid ObjStm at offset {org_pos}.') from ex
for idx, p in enumerate(utils.chunks(numbers, 2)):
# gen no, of object stream and of any compressed object is implicitly 0
objbytestream.seek(First + p[1], io.SEEK_SET)
result[objstmobj_no, idx] = PdfIndirectObject(
PdfObject.create_from_file(objbytestream, doc), p[0], 0)
# TODO: check for orphaned bytes between compressed objectes?
return result
def parse_linear(self, f, progress_cb=None):
'''Initialize a PdfDocument from a opened PDF file f from the beginning'''
def print_progress():
print('', end="\r")
print(f'{f.tell() / filesize * 100:5.2f}% processed',
end='',
flush=True)
if progress_cb is not None:
progress_cb(f'{f.tell() / filesize * 100:5.2f}% processed',
read=f.tell(),
total=filesize)
f.seek(0, io.SEEK_SET)
filesize = os.fstat(f.fileno()).st_size
print_progress()
# First line is header
s, eol_marker = utils.read_until(f, syntax.EOL)
header = re.match(rb'%PDF-(\d+\.\d+)', s)
if header:
self.version = Decimal(header.group(1).decode('iso-8859-1'))
f.seek(len(eol_marker), io.SEEK_CUR)
else:
raise Exception('Not a PDF file')
while True:
utils.seek_until(f, syntax.NON_WHITESPACES, ignore_comment=False)
if f.tell() >= filesize:
break
org_pos = f.tell()
s, eol_marker = utils.read_until(f, syntax.EOL)
if s == b'startxref': # the last startxref always override the ones before
utils.seek_until(f,
syntax.NON_WHITESPACES,
ignore_comment=True)
t, _ = utils.read_until(f, syntax.EOL)
self.startxref = int(t)
self.increments[-1]['startxref'] = self.startxref
continue
elif s == b'xref':
f.seek(-4, io.SEEK_CUR)
self.increments[-1]['xref_section'] = PdfXRefSection(f)
self.offset_xref[org_pos] = self.increments[-1]['xref_section']
continue
elif s == b'trailer':
utils.seek_until(f,
syntax.NON_WHITESPACES,
ignore_comment=True)
self.increments[-1][
'trailer'] = PdfDictionaryObject.create_from_file(f, self)
continue
elif s == b'%%EOF':
# TODO: check if trailer dict immediately precedes %%EOF
# since we are seeking until non-ws, the only case EOF marker
# does not appear by itself it when it is preceded by some
# whitespaces, which should be ignored
self.increments[-1]['eof'] = True
f.seek(5 + len(eol_marker), io.SEEK_CUR)
continue
elif s[0:1] == b'%':
# otherwise, it is a comment, ignore the whole remaining line
utils.seek_until(f, syntax.EOL)
continue
#else:
f.seek(org_pos, io.SEEK_SET)
if self.increments[-1]['eof']:
self.increments += [{
'body': [],
'xref_section': None,
'trailer': None,
'startxref': None,
'eof': False
}]
# TODO: how to handle object parse error?
new_obj = PdfObject.create_from_file(f, self)
self.increments[-1]['body'] += [new_obj]
self.offset_obj[org_pos] = new_obj
if isinstance(new_obj.value, PdfStreamObject
) and new_obj.value.dict.get('Type') == 'ObjStm':
self.offset_obj_streams[org_pos] = new_obj
print_progress()
print('', end="\r")
print('100% processed ')
if progress_cb is not None:
progress_cb('100% processed', read=f.tell(), total=filesize)
self.ready = True
print('Decoding object streams...')
if progress_cb is not None:
progress_cb('Decoding object streams...',
read=f.tell(),
total=filesize)
for k in self.offset_obj_streams:
from objstm import decode_objstm
self.compressed_obj = {
**(self.compressed_obj),
**(decode_objstm(self.offset_obj_streams[k], self))
}
print('Done')
if progress_cb is not None:
progress_cb('Done', read=f.tell(), total=filesize)
def parse_normal(self, f, progress_cb=None):
'''Initialize a PdfDocument from a opened PDF file f by reading xref and trailers. After this is called, offset_obj, offset_obj_streams, compressed_obj, offset_xref_trailer, all xref sections are ready'''
f.seek(0, io.SEEK_SET)
filesize = os.fstat(f.fileno()).st_size
# First line is header
s, eol_marker = utils.read_until(f, syntax.EOL)
header = re.match(rb'%PDF-(\d+\.\d+)', s)
if header:
self.version = Decimal(header.group(1).decode('iso-8859-1'))
f.seek(len(eol_marker), io.SEEK_CUR)
else:
raise Exception('Not a PDF file')
# read from end of file, find xref
eof_found = -1
startxref_found = -1
temp_line = b''
temp_count = 2
temp_offset = 0
for line in utils.rlines(f):
temp_offset -= len(line)
if line.rstrip() == b'%%EOF':
eof_found = temp_offset
if eof_found != -1 and temp_count == 0:
if line.rstrip() == b'startxref':
startxref_found = temp_offset
break
else:
raise Exception(
'startxref not found at 2 lines before EOF marker')
elif eof_found != -1:
temp_count -= 1
temp_line = line
xref_offset = int(temp_line.decode('iso-8859-1'))
self.startxref = xref_offset
# The only required part for a trailer (and marks the end of an increment) is startxref and %%EOF
self.increments[-1]['startxref'] = xref_offset
self.increments[-1]['eof'] = True
inuse_count = 0
while True:
f.seek(xref_offset, io.SEEK_SET)
xref_section, trailer = self.get_xref_trailer_at_offset(
f, xref_offset)
self.offset_xref_trailer[xref_offset] = (xref_section, trailer)
for subsec in xref_section.subsections:
inuse_count += len(subsec.inuse_entry)
self.increments[0]['xref_section'] = xref_section
self.increments[0]['trailer'] = trailer
if trailer.get('Prev') is None:
break
if trailer['Prev'].value - int(trailer['Prev'].value) != 0:
raise Exception(
f'Prev must be an integer, in trailer dict at offset {xref_offset}'
)
xref_offset = int(trailer['Prev'].value) # must not be indirect
self.increments = [{
'body': [],
'xref_section': None,
'trailer': None,
'startxref': None,
'eof': False
}] + self.increments
self.increments[0]['startxref'] = xref_offset
self.ready = True
inuse_parsed_count = 0
# parse each in use obj num
for inc in self.increments:
for subsec in inc['xref_section'].subsections:
for entry in subsec.inuse_entry:
if entry.get('compressed'):
inuse_parsed_count += 1
continue
offset = entry['offset']
f.seek(offset, io.SEEK_SET)
new_obj = PdfObject.create_from_file(f, self)
if not isinstance(
new_obj,
PdfIndirectObject) or new_obj.obj_no != entry[
'obj_no'] or new_obj.gen_no != entry['gen_no']:
raise Exception(
f'Invalid obj referenced by xref at offset {offset}'
)
self.offset_obj[offset] = new_obj
if isinstance(new_obj.value,
PdfStreamObject) and new_obj.value.dict.get(
'Type') == 'ObjStm':
self.offset_obj_streams[offset] = new_obj
inuse_parsed_count += 1
print('', end="\r")
print(
f'{inuse_parsed_count / inuse_count * 100:5.2f}% processed',
end='',
flush=True)
if progress_cb is not None:
progress_cb(
f'{inuse_parsed_count / inuse_count * 100:5.2f}% processed',
read=inuse_parsed_count,
total=inuse_count)
print('Decoding object streams...')
if progress_cb is not None:
progress_cb('Decoding object streams...',
read=inuse_parsed_count,
total=inuse_count)
for k in self.offset_obj_streams:
from objstm import decode_objstm
self.compressed_obj = {
**(self.compressed_obj),
**(decode_objstm(self.offset_obj_streams[k], self))
}
print('', end="\r")
print('100% processed ')
if progress_cb is not None:
progress_cb('100% processed',
read=inuse_parsed_count,
total=inuse_count)
print('Done')
if progress_cb is not None:
progress_cb('Done', read=inuse_parsed_count, total=inuse_count)