parse the cross-reference stream in pdf?

See more:

C++

C

PDF

<</DecodeParms<</Columns 4/Predictor 12>>/Filter/FlateDecode/ID[<d0d1f903c5032042884e61b2933652da><f929944ec68a3543a18a40f2e7630a9d>]/Info 23 0 R/Length 59/Root 25 0 R/Size 24/Type/XRef/W[1 2 1]>>

I use zlib to inflate the data,the use png filter arithmetic to reverse the decompressed data:
First,i decompress the stream data,i get the result with 115 bytes:
columns=4
bpc=8；
a line is 4+1,
so I convert the data to 5f*23 metrix.
hex:
02 01 00 10 00
02 00 02 6B 00
02 00 00 82 00
02 00 02 0A 00
02 00 01 6F 00
02 00 04 9F 00
02 00 01 22 00
02 00 00 66 00
02 00 08 8F 00
02 00 EE 18 00
02 01 00 A8 00
02 00 00 00 01
02 00 00 00 01
02 00 00 00 01
02 00 00 00 01
02 00 00 00 01
02 00 00 00 01
02 00 00 00 01
02 00 00 00 01
02 00 00 00 01
02 00 00 00 01
02 00 00 00 01
02 FF 01 B4 F5
Then I use PNG filter 2 algorithm to decode the decompressed data,i get the reserverd data:
01 00 10 00
01 02 7B 00
01 02 ED 00
01 04 F8 00
..................
..................
accroding to w array:
i get the obj offset:16
but i can't find obj in the offset,
Who can tell me the in the pdf parse?