this implementation of md5 is working well on strings however it does not give correct results on files

Question

I am using an implementation of md5 algorithm written in C. It works on strings well. However it gives wrong results on files. I am using the code below to convert the file to char* and then send it to md5 function:

    struct stat st; 

    if (stat(fileName, &st) == 0)
    {

        // read the file into a char* in order to send it to md5 function
        std::string fdata = "";
        std::ifstream filePtr(fileName);
        std::string dummy = "";
        while((st.st_mode & 0100000) && !filePtr.eof())
        {
            filePtr >> dummy;
            fdata += dummy;

        }

        if(st.st_mode & 0100000)
        {
            const char* fileData = fdata.c_str();
            unsigned* d = md5(fileData, strlen(fileData));
            WBunion u;

            for (unsigned j = 0; j < 4; ++j)
            {
                u.w = d[j];
                for (unsigned k = 0; k < 4; ++k)
                {
                    printf("%02x",u.b[k]);
                }
            }
            printf("\n");

        }

Note: The bitwise operation is just to check whether the file is a regular file(not a folder).
What am I doing wrong? Is the overloaded >> operator to read the file the problem?

Thank you in advance,

Link to the implementation of md5 algorithm that I am using: http://rosettacode.org/wiki/MD5#C.2B.2B

what if the file is binary and there's a zero in it? `strlen(fileData)` wouldn't give you the right length... — mark, Jul 02 '14 at 16:30
Off topic but md5 is broken. You should use a modern hashing algorithm. — Brandon Yates, Jul 02 '14 at 17:42

score 2 · Accepted Answer · edited May 23 '17 at 12:25

filePtr >> fdata clears fdata on each iteration of the loop, which means the string ends up containing the very last word in the file. A correct and simple way to slurp the whole file into an std::string is by using a stringstream.

Also, you must open the file in binary mode for MD5 to work correctly, and call fdata.size() instead of strlen(fileData) in order to handle files with NUL characters.

this implementation of md5 is working well on strings however it does not give correct results on files

1 Answers1