Zlib uncompress returns Z_DATA_ERROR - visual-c++

Question

I am working on a client server application, where client compresses a 2MB data sends to the server, server receives the data uncompresses it and writes it to a file.
For some packets uncompression was failing and I added MD5 sum to both client side and server side code and also debugged using uncompression at the client side after compressing the data. The same parameters that passes for uncompress function in client side is failing with Z_DATA_ERROR in the server side. The data's MD5sum seems same. Am totally clueless what I could do next.
Server Side cod looks like this:
int ret = uncompress((Bytef*)unCompressedBuffer, &dwUncompressedBytes, (const Bytef*) receivedBuffer+525, dwBlockLength);
if (ret == Z_OK)
{
}
else
{
std::cout << " Uncompression failed for Block: " << iBlock << std::endl;
std::cout << " PacketType: 4" << " Block Number:" << iBlock << " Length:" << dwBlockLength << "Error:" << ret << std::endl;
PrintMD5SumResult((PBYTE)receivedBuffer+525, compressedSize-525);
std::cout << " Uncompressed MD5 Checksum:0";
PrintMD5SumResult((PBYTE)unCompressedBuffer, dwUncompressedBytes);
}
}
Client Code Looks like this:
int ret = compress2(l_pCompressData + 4, &destLen,
(const Bytef*) pBlockData, dwBlockSize, 6);
memcpy(m_pWriteBuffer+525, l_pCompressData, destLen);
m_dwWriteBytes = destLen+525;
std::cout << " \n Compressed MD5 Sum:0";
PrintMD5SumResult(m_pWriteBuffer, m_dwWriteBytes);
PrintMD5SumResult(m_pWriteBuffer+525, m_dwWriteBytes-525);
int ret = uncompress(m_pUnCompressData, &uncomLen, (const Bytef*)m_pWriteBuffer+525, destLen);
if(ret != Z_OK)
{
std::cout << " Uncompression has failed." << std::endl;
}
else
{
//std::cout << " UnCompressed MD5 Sum:0";
//PrintMD5SumResult((PBYTE)m_pUnCompressData, md5Output, dwBlockSize);
}
// Write the 2MB to the network
WriteDataOverNetwork(m_NetworkStream, m_pWriteBuffer, m_dwWriteBytes, &dwNumBytes, TRUE);
I narrowed down the problem to the following piece of code in zlib - but have a hard time understanding it. In the inflate() function, (ZSWAP32(hold)) != state->check) this statement is failing. Can someone help me out here? MD5sum used here is from Boton C++ library.
case CHECK:
if (state->wrap) {
NEEDBITS(32);
out -= left;
strm->total_out += out;
state->total += out;
if (out)
strm->adler = state->check =
UPDATE(state->check, put - out, out);
out = left;
if ((
#ifdef GUNZIP
state->flags ? hold :
#endif
ZSWAP32(hold)) != state->check) {
strm->msg = (char *)"incorrect data check";
state->mode = BAD;
break;
}

Answer 1

I also met this issue recently when I used zlib to do in-memory compression/decompression. The code is as follow:
size_t size = 1048576;
void *data;
void *comp_data;
uLong comp_data_len;
void *uncomp_data;
uLong uncomp_data_len;
void *temp;
int ret;
data = calloc(1, size); // data is filled with all zeros
comp_data_len = size * 1.01 + 12;
comp_data = calloc(1, size);
ret = compress(comp_data, &comp_data_len, data, size); //here ret is Z_OK.
uncomp_data_len = size;
uncomp_data = calloc(1, uncomp_data_len);
ret = uncompress(uncomp_data, &uncomp_data_len, comp_data, comp_data_len); //here ret is Z_OK
temp = calloc(1, 496);
for (i = 0; i < 100; i++)
{
//here fill some random data to temp
memcpy((char*)data + i * 100, temp, 496);
ret = compress(comp_data, &comp_data_len, data, size); //here ret is Z_OK.
ret = uncompress(uncomp_data, &uncomp_data_len, comp_data, comp_data_len); //here ret sometimes is Z_OK, sometimes is Z_DATA_ERROR!!!
}
I also traced the code and found that it failed at the statement "inflate() function, (ZSWAP32(hold)) != state->check)" too. So I cannot believe that the function uncompress is related to the data pattern. Am I wrong?
I also noticed that compress function calls deflate to do compression, deflate processes data every 64k, so need I split it to 64k blocks, compress each block one by one, then uncompress can work well?

Answer 2

i don't know whether it‘s the right answer ,maybe it help ！my English is so poor，hope you can understand. perhaps the parameters convert to another has bugs . when they convert the info maybe lose ! i meet the same problem , after use the source code type the problem has been solved (Bytef\uLongf\ uLong,etc). the wed is Chinese you can use Google to translate.
http://www.360doc.com/content/13/0927/18/11217914_317498849.shtml

Answer 3

This is my test.the arry[] can be larger,same time the sour[]/dest[]/destLen/Len will be changed.using the source code type the problem has been solved. Hope will be helpful.
my code as follow:
#include <stdio.h>
#include "zlib.h"
int main(){
//the buffer can be larger
Bytef arry[] = "中文测试 yesaaaaa bbbbb ccccc ddddd 中文测试 yesaaaaa bbbbb ccccc ddddd 中文测试yesaaaaa bbbbb ccccc ddddd 中文测试 yes 我是一名军人！";
//buffer length
int size = sizeof(arry);
//store the uncompressed data
Bytef sour[2500];
//store the compressed data
Bytef dest[2500];
//压缩后的数据可能比源数据要大
unsigned long destLen = 2500;
//解压数据时因为不知道源数据大小，设置时长度尽可能大一些。以免出错
unsigned long Len = 2500;
int ret = -1;
ret = compress(dest,&destLen,arry,size);
//dest[destLen] = '\0';
printf("ret = %d\ndest = %s\n", ret, dest);
ret = uncompress(sour,&Len,dest,destLen);
//sour[size-1] = '\0';
printf("ret = %d\nsour = %s\n", ret, sour);
return 0;
}