Reading and writing bit by bit in C++ for Huffman Encoding

问题

I'm trying to encode and decode for a Huffman coding in C++. I'm not sure where my problem is I'm able to read and write but when I decompress the file its scrambled so I'm either not encoding or decoding correctly. I think its when I'm writing and reading the file where things go wrong. So this is what I have to write the encoded file. First I store all the bitcodes from my unordered map called uMap into one string:

int i = 0, j = 0;
string fullStr = "";
for (i = 0; i < buffsize; i++) //put all codes in one string of 1's and 0's
    fullStr += uMap[buffer[i]];
unsigned char byte = 0;
i = 0;
for (j = 0; j < fullStr.length(); j++)
{

    if (i != 8)
    {
        byte |= (fullStr[j] == '1') << i; // make up one byte
        i++;
    }
    else
    {
        outf.put(byte); // write one byte at a time
        byte = 0;
        i = 0;
    }
}
if (i != 0 && i < 8)
{
    while (i<8)
    {
        byte |= 0 << i; // finish up last byte if not finished
        i++;
    }
    outf.put(byte);
}

Then on the decompress side:

int i = 0;
unsigned char byte = 0;
bitset<8> setByte;
ofstream outf(filename, ofstream::binary);
string concat = "";
string bitStr = "";
for (i = 0; i < buffLength; i++)
{
    setByte = buffer[i];
    bitStr = setByte.to_string();
    for (int j = 0; j < 8; j++)
    {
        concat += bitStr[j];
        if (uMap[concat])
        {
            //cout << "found code " << concat << " " << uMap[concat] << endl;
            outf.put(uMap[concat]);
            concat = "";
        }
    }
}
outf.close();

回答1:

The bits are being unpacked in reverse order to the packing, perhaps because you are using a different method for each. The first bit packed is going into bit 0 of the bitset (bitvalue << 0). The first bit unpacked is coming from bit 7, because setByte.to_string() creates a string with bit 7 at index 0. The question was tagged as Huffman coding but has nothing to do with it, it's about bitstream manipulation.

回答2:

Thank you Weather Vane you were correct about the read they were being reversed so I took care of that but it turns out it was also how i was writing it too.

New compress:

int i = 0, j = 0;
string fullStr = "";
for (i = 0; i < buffsize; i++) //put all codes in one string
    fullStr += uMap[buffer[i]];
for (i = 0; i < fullStr.length(); i+=8)
{
    unsigned char byte = 0;
    string str8 = "";
    if (i + 8 < fullStr.length())
        str8 = fullStr.substr(i, i + 8);
    else
        str8 = fullStr.substr(i, fullStr.length());
    for (unsigned b = 0; b != 8; ++b)
    {
        if (b < str8.length())
            byte |= (str8[b] & 1) << b; // this line was wrong before
        else
            byte |= 1 << b;
    }
    outf.put(byte);
}
int filelen = outf.tellp();
outf.close();

New decompress:

int i = 0,j=0,k=0;
unsigned char byte = 0;
bitset<8> setByte, reverseByte;
ofstream outf(filename, ofstream::binary);
string concat = "";
string bitStr = "";
string reverse = "";
int charCount = 0;
for (i = 0; i < buffLength; i++)
{
    setByte = buffer[i];
    bitStr = setByte.to_string();
    reverse = "";
    for (k = 7; k>=0; k--)
        reverse += bitStr[k];
    for (j = 0; j < 8; j++)
    {
        concat += reverse[j];
        if (uMap[concat])
        {
            outf << uMap[concat];
            charCount++;
            concat = "";
            if (charCount == origLength) // if we have written original amount stop
            {
                outf.close();
                return 1;
            }

        }
    }
}
outf.close();
return 1;

来源：https://stackoverflow.com/questions/26442093/reading-and-writing-bit-by-bit-in-c-for-huffman-encoding