base32 conversion in C++

does anybody know any commonly used library for C++ 开发者_Python百科that provides methods for encoding and decoding numbers from base 10 to base 32 and viceversa?

Thanks, Stefano

[Updated] Apparently, the C++ std::setbase() IO manipulator and normal << and >> IO operators only handle bases 8, 10, and 16, and is therefore useless for handling base 32.

So to solve your issue of converting

• strings with base 10/32 representation of numbers read from some input to integers in the program
• integers in the program to strings with base 10/32 representations to be output

you will need to resort to other functions.

For converting C style strings containing base 2..36 representations to integers, you can use #include <cstdlib> and use the strtol(3) & Co. set of functions.

As for converting integers to strings with arbitrary base... I cannot find an easy answer. printf(3) style format strings only handle bases 8,10,16 AFAICS, just like std::setbase. Anyone?

Did you mean "base 10 to base 32", rather than integer to base32? The latter seems more likely and more useful; by default standard formatted I/O functions generate base 10 string format when dealing with integers.

For the base 32 to integer conversion the standard library strtol() function will do that. For the reciprocal, you don't need a library for something you can easily implement yourself (not everything is a lego brick).

Here's an example, not necessarily the most efficient, but simple;

#include <cstring>
#include <string>

long b32tol( std::string b32 )
{
    return strtol( b32.c_str(), 0, 32 ) ;
}

std::string itob32( long i )
{
    unsigned long u = *(reinterpret_cast<unsigned long*>)( &i ) ;
    std::string b32 ;

    do
    {
        int d = u % 32 ;
        if( d < 10 )
        {
            b32.insert( 0, 1, '0' + d ) ;
        }
        else
        {
            b32.insert( 0, 1, 'a' + d - 10 ) ;
        }

        u /= 32 ;

    } while( u > 0 );

    return b32 ;
}


#include <iostream>

int main() 
{ 
    long i = 32*32*11 + 32*20 + 5 ; // BK5 in base 32
    std::string b32 = itob32( i ) ;
    long ii = b32tol( b32 ) ;

    std::cout << i << std::endl ;    // Original
    std::cout << b32 << std::endl ;  // Converted to b32
    std::cout << ii << std::endl ;   // Converted back

    return 0 ;
}

In direct answer to the original (and now old) question, I don't know of any common library for encoding byte arrays in base32, or for decoding them again afterward. However, I was presented last week with a need to decode SHA1 hash values represented in base32 into their original byte arrays. Here's some C++ code (with some notable Windows/little endian artifacts) that I wrote to do just that, and to verify the results.

Note that in contrast with Clifford's code above, which, if I'm not mistaken, assumes the "base32hex" alphabet mentioned on RFC 4648, my code assumes the "base32" alphabet ("A-Z" and "2-7").

// This program illustrates how SHA1 hash values in base32 encoded form can be decoded
// and then re-encoded in base16.

#include "stdafx.h"
#include <string>
#include <vector>
#include <iostream>
#include <cassert>

using namespace std;

unsigned char Base16EncodeNibble( unsigned char value )
{
    if( value >= 0 && value <= 9 )
        return value + 48;
    else if( value >= 10 && value <= 15 )
        return (value-10) + 65;
    else //assert(false);
    {
        cout << "Error: trying to convert value: " << value << endl;
    }

    return 42; // sentinal for error condition
}

void Base32DecodeBase16Encode(const string & input, string & output)
{
    // Here's the base32 decoding:

        // The "Base 32 Encoding" section of http://tools.ietf.org/html/rfc4648#page-8
        // shows that every 8 bytes of base32 encoded data must be translated back into 5 bytes
        // of original data during a decoding process. The following code does this.

    int input_len = input.length();
    assert( input_len == 32 );
    const char * input_str = input.c_str();
    int output_len = (input_len*5)/8;
    assert( output_len == 20 );
        // Because input strings are assumed to be SHA1 hash values in base32, it is also assumed
        // that they will be 32 characters (and bytes in this case) in length, and so the output
        // string should be 20 bytes in length.
    unsigned char *output_str = new unsigned char[output_len];

    char curr_char, temp_char;
    long long temp_buffer = 0; //formerly: __int64 temp_buffer = 0;
    for( int i=0; i<input_len; i++ )
    {
        curr_char = input_str[i];
        if( curr_char >= 'A' && curr_char <= 'Z' )
            temp_char = curr_char - 'A';
        if( curr_char >= '2' && curr_char <= '7' )
            temp_char = curr_char - '2' + 26;

        if( temp_buffer )
            temp_buffer <<= 5; //temp_buffer = (temp_buffer << 5);
        temp_buffer |= temp_char;

        // if 8 encoded characters have been decoded into the temp location,
            // then copy them to the appropriate section of the final decoded location
        if( (i>0) && !((i+1) % 8) )
        {
            unsigned char * source = reinterpret_cast<unsigned char*>(&temp_buffer);
            //strncpy(output_str+(5*(((i+1)/8)-1)), source, 5);
            int start_index = 5*(((i+1)/8)-1);
            int copy_index = 4;
            for( int x=start_index; x<(start_index+5); x++, copy_index-- )
                output_str[x] = source[copy_index];
            temp_buffer = 0;

            // I could be mistaken, but I'm guessing that the necessity of copying
            // in "reverse" order results from temp_buffer's little endian byte order.
        }
    }

    // Here's the base16 encoding (for human-readable output and the chosen validation tests):

        // The "Base 16 Encoding" section of http://tools.ietf.org/html/rfc4648#page-10
        // shows that every byte original data must be encoded as two characters from the
        // base16 alphabet - one charactor for the original byte's high nibble, and one for
        // its low nibble.

    unsigned char out_temp, chr_temp;
    for( int y=0; y<output_len; y++ )
    {
        out_temp = Base16EncodeNibble( output_str[y] >> 4 ); //encode the high nibble
        output.append( 1, static_cast<char>(out_temp) );
        out_temp = Base16EncodeNibble( output_str[y] & 0xF ); //encode the low nibble
        output.append( 1, static_cast<char>(out_temp) );
    }

    delete [] output_str;
}

int _tmain(int argc, _TCHAR* argv[])
{
    //string input = "J3WEDSJDRMJHE2FUHERUR6YWLGE3USRH";
    vector<string> input_b32_strings, output_b16_strings, expected_b16_strings;

    input_b32_strings.push_back("J3WEDSJDRMJHE2FUHERUR6YWLGE3USRH");
    expected_b16_strings.push_back("4EEC41C9238B127268B4392348FB165989BA4A27");
    input_b32_strings.push_back("2HPUCIVW2EVBANIWCXOIQZX6N5NDIUSX");
    expected_b16_strings.push_back("D1DF4122B6D12A10351615DC8866FE6F5A345257");
    input_b32_strings.push_back("U4BDNCBAQFCPVDBL4FBG3AANGWVESI5J");
    expected_b16_strings.push_back("A7023688208144FA8C2BE1426D800D35AA4923A9");

    // Use the base conversion tool at http://darkfader.net/toolbox/convert/
    // to verify that the above base32/base16 pairs are equivalent.

    int num_input_strs = input_b32_strings.size();
    for(int i=0; i<num_input_strs; i++)
    {
        string temp;
        Base32DecodeBase16Encode(input_b32_strings[i], temp);
        output_b16_strings.push_back(temp);
    }

    for(int j=0; j<num_input_strs; j++)
    {
        cout << input_b32_strings[j] << endl;
        cout << output_b16_strings[j] << endl;
        cout << expected_b16_strings[j] << endl;

        if( output_b16_strings[j] != expected_b16_strings[j] )
        {
            cout << "Error in conversion for string " << j << endl;
        }
    }

    return 0;
}

I'm not aware of any commonly-used library devoted to base32 encoding but Crypto++ includes a public domain base32 encoder and decoder.

I don't use cpp, so correct me if I'm wrong. I wrote this code for the sake of translating it from C# to save my acquaintance the trouble. The original source, that which I used to create these methods, is on a different post, here, on stackoverflow: https://stackoverflow.com/a/10981113/13766753

That being said, here's my solution:

#include <iostream>
#include <math.h>

class Base32 {
    public:
        static std::string dict;
        static std::string encode(int number) {
            std::string result = "";
            bool negative = false;
            if (number < 0) {
                negative = true;
            }

            number = abs(number);
            do {
                result = Base32::dict[fmod(floor(number), 32)] + result;
                number /= 32;
            } while(number > 0);

            if (negative) {
                result = "-" + result;
            }
    
            return result;
        }

        static int decode(std::string str) {
            int result = 0;
            int negative = 1;
            if (str.rfind("-", 0) == 0) {
                negative = -1;
                str = str.substr(1);
            }

            for(char& letter : str) {
                result += Base32::dict.find(letter);
                result *= 32;
            }

            return result / 32 * negative;
        }
};

std::string Base32::dict = "0123456789abcdefghijklmnopqrstuvwxyz";

int main() {
    std::cout << Base32::encode(0) + "\n" << Base32::decode(Base32::encode(0)) << "\n";

    return 0;
}