C++ 使用 OpenSSL 进行 Base64 编码和解码

Base64 encoding and decoding with OpenSSL

提问by Bernard

I've been trying to figure out the openssl documentation for base64 decoding and encoding. I found some code snippets below

我一直在试图找出 base64 解码和编码的 openssl 文档。我在下面找到了一些代码片段

#include <openssl/sha.h>
#include <openssl/hmac.h>
#include <openssl/evp.h>
#include <openssl/bio.h>
#include <openssl/buffer.h>

char *base64(const unsigned char *input, int length)
{
  BIO *bmem, *b64;
  BUF_MEM *bptr;

  b64 = BIO_new(BIO_f_base64());
  bmem = BIO_new(BIO_s_mem());
  b64 = BIO_push(b64, bmem);
  BIO_write(b64, input, length);
  BIO_flush(b64);
  BIO_get_mem_ptr(b64, &bptr);

  char *buff = (char *)malloc(bptr->length);
  memcpy(buff, bptr->data, bptr->length-1);
  buff[bptr->length-1] = 0;

  BIO_free_all(b64);

  return buff;
}

char *decode64(unsigned char *input, int length)
{
  BIO *b64, *bmem;

  char *buffer = (char *)malloc(length);
  memset(buffer, 0, length);

  b64 = BIO_new(BIO_f_base64());
  bmem = BIO_new_mem_buf(input, length);
  bmem = BIO_push(b64, bmem);

  BIO_read(bmem, buffer, length);

  BIO_free_all(bmem);

  return buffer;
}

This only seems to work for single line strings such as "Start", the moment I introduce complex strings with newlines and spaces etc it fails horribly.

这似乎只适用于单行字符串，例如“开始”，当我引入带有换行符和空格等的复杂字符串时，它会失败得可怕。

It doesn't even have to be openssl, a simple class or set of functions that do the same thing would be fine, theres a very complicated build process for the solution and I am trying to avoid having to go in there and make multiple changes. The only reason I went for openssl is because the solution is already compiled with the libraries.

它甚至不必是 openssl，一个简单的类或一组做同样事情的函数就可以了，解决方案的构建过程非常复杂，我试图避免进入那里并进行多次更改. 我选择 openssl 的唯一原因是因为该解决方案已经使用库进行了编译。

回答by Omnifarious

Personally, I find the OpenSSL API to be so incredibly painful to use, I avoid it unless the cost of avoiding it is extremely high. I find it quite upsetting that it has become the standard API in the crypto world.

就我个人而言，我发现 OpenSSL API 使用起来非常痛苦，我避免使用它，除非避免它的成本非常高。我发现它已经成为加密世界的标准 API 非常令人沮丧。

I was feeling bored, and I wrote you one in C++. This one should even handle the edge cases that can cause security problems, like, for example, encoding a string that results in integer overflow because it's too large.

我感到无聊，我用 C++ 给你写了一个。这个甚至应该处理可能导致安全问题的边缘情况，例如，编码一个导致整数溢出的字符串，因为它太大了。

I have done some unit testing on it, so it should work.

我已经对它进行了一些单元测试，所以它应该可以工作。

#include <string>
#include <cassert>
#include <limits>
#include <stdexcept>
#include <cctype>

static const char b64_table[65] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";

static const char reverse_table[128] = {
   64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
   64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
   64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 62, 64, 64, 64, 63,
   52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 64, 64, 64, 64, 64, 64,
   64,  0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14,
   15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 64, 64, 64, 64, 64,
   64, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
   41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 64, 64, 64, 64, 64
};

::std::string base64_encode(const ::std::string &bindata)
{
   using ::std::string;
   using ::std::numeric_limits;

   if (bindata.size() > (numeric_limits<string::size_type>::max() / 4u) * 3u) {
      throw ::std::length_error("Converting too large a string to base64.");
   }

   const ::std::size_t binlen = bindata.size();
   // Use = signs so the end is properly padded.
   string retval((((binlen + 2) / 3) * 4), '=');
   ::std::size_t outpos = 0;
   int bits_collected = 0;
   unsigned int accumulator = 0;
   const string::const_iterator binend = bindata.end();

   for (string::const_iterator i = bindata.begin(); i != binend; ++i) {
      accumulator = (accumulator << 8) | (*i & 0xffu);
      bits_collected += 8;
      while (bits_collected >= 6) {
         bits_collected -= 6;
         retval[outpos++] = b64_table[(accumulator >> bits_collected) & 0x3fu];
      }
   }
   if (bits_collected > 0) { // Any trailing bits that are missing.
      assert(bits_collected < 6);
      accumulator <<= 6 - bits_collected;
      retval[outpos++] = b64_table[accumulator & 0x3fu];
   }
   assert(outpos >= (retval.size() - 2));
   assert(outpos <= retval.size());
   return retval;
}

::std::string base64_decode(const ::std::string &ascdata)
{
   using ::std::string;
   string retval;
   const string::const_iterator last = ascdata.end();
   int bits_collected = 0;
   unsigned int accumulator = 0;

   for (string::const_iterator i = ascdata.begin(); i != last; ++i) {
      const int c = *i;
      if (::std::isspace(c) || c == '=') {
         // Skip whitespace and padding. Be liberal in what you accept.
         continue;
      }
      if ((c > 127) || (c < 0) || (reverse_table[c] > 63)) {
         throw ::std::invalid_argument("This contains characters not legal in a base64 encoded string.");
      }
      accumulator = (accumulator << 6) | reverse_table[c];
      bits_collected += 6;
      if (bits_collected >= 8) {
         bits_collected -= 8;
         retval += static_cast<char>((accumulator >> bits_collected) & 0xffu);
      }
   }
   return retval;
}

回答by TCS

Here is an example of OpenSSL base64 encode/decode I wrote:

这是我编写的 OpenSSL base64 编码/解码示例：

Notice, I have some macros/classes in the code that I wrote, but none of them is important for the example. It is simply some C++ wrappers I wrote:

请注意，我编写的代码中有一些宏/类，但对于示例而言，它们都不重要。这只是我写的一些 C++ 包装器：

buffer base64::encode( const buffer& data )
{
    // bio is simply a class that wraps BIO* and it free the BIO in the destructor.

    bio b64(BIO_f_base64()); // create BIO to perform base64
    BIO_set_flags(b64,BIO_FLAGS_BASE64_NO_NL);

    bio mem(BIO_s_mem()); // create BIO that holds the result

    // chain base64 with mem, so writing to b64 will encode base64 and write to mem.
    BIO_push(b64, mem);

    // write data
    bool done = false;
    int res = 0;
    while(!done)
    {
        res = BIO_write(b64, data.data, (int)data.size);

        if(res <= 0) // if failed
        {
            if(BIO_should_retry(b64)){
                continue;
            }
            else // encoding failed
            {
                /* Handle Error!!! */
            }
        }
        else // success!
            done = true;
    }

    BIO_flush(b64);

    // get a pointer to mem's data
    char* dt;
    long len = BIO_get_mem_data(mem, &dt);

    // assign data to output
    std::string s(dt, len);

    return buffer(s.length()+sizeof(char), (byte*)s.c_str());
}

回答by Matt

This works for me, and verified no memory leaks with valgrind.

这对我有用，并验证了 valgrind 没有内存泄漏。

#include <openssl/bio.h>
#include <openssl/evp.h>
#include <cstring>
#include <memory>
#include <string>
#include <vector>

#include <iostream>

namespace {
struct BIOFreeAll { void operator()(BIO* p) { BIO_free_all(p); } };
}

std::string Base64Encode(const std::vector<unsigned char>& binary)
{
    std::unique_ptr<BIO,BIOFreeAll> b64(BIO_new(BIO_f_base64()));
    BIO_set_flags(b64.get(), BIO_FLAGS_BASE64_NO_NL);
    BIO* sink = BIO_new(BIO_s_mem());
    BIO_push(b64.get(), sink);
    BIO_write(b64.get(), binary.data(), binary.size());
    BIO_flush(b64.get());
    const char* encoded;
    const long len = BIO_get_mem_data(sink, &encoded);
    return std::string(encoded, len);
}

// Assumes no newlines or extra characters in encoded string
std::vector<unsigned char> Base64Decode(const char* encoded)
{
    std::unique_ptr<BIO,BIOFreeAll> b64(BIO_new(BIO_f_base64()));
    BIO_set_flags(b64.get(), BIO_FLAGS_BASE64_NO_NL);
    BIO* source = BIO_new_mem_buf(encoded, -1); // read-only source
    BIO_push(b64.get(), source);
    const int maxlen = strlen(encoded) / 4 * 3 + 1;
    std::vector<unsigned char> decoded(maxlen);
    const int len = BIO_read(b64.get(), decoded.data(), maxlen);
    decoded.resize(len);
    return decoded;
}

int main()
{
    const char* msg = "hello";
    const std::vector<unsigned char> binary(msg, msg+strlen(msg));
    const std::string encoded = Base64Encode(binary);
    std::cout << "encoded = " << encoded << std::endl;
    const std::vector<unsigned char> decoded = Base64Decode(encoded.c_str());
    std::cout << "decoded = ";
    for (unsigned char c : decoded) std::cout << c;
    std::cout << std::endl;
    return 0;
}

Compile:

编译：

g++ -lcrypto main.cc

Output:

输出：

encoded = aGVsbG8=
decoded = hello

回答by Satish

Improved TCS answer to remove macros/datastructures

改进了 TCS 答案以删除宏/数据结构

unsigned char *encodeb64mem( unsigned char *data, int len, int *lenoutput )
{
// bio is simply a class that wraps BIO* and it free the BIO in the destructor.

BIO *b64 = BIO_new(BIO_f_base64()); // create BIO to perform base64
BIO_set_flags(b64, BIO_FLAGS_BASE64_NO_NL);

BIO *mem = BIO_new(BIO_s_mem()); // create BIO that holds the result

// chain base64 with mem, so writing to b64 will encode base64 and write to mem.
BIO_push(b64, mem);

// write data
bool done = false;
int res = 0;
while(!done)
{
    res = BIO_write(b64, data, len);

    if(res <= 0) // if failed
    {
        if(BIO_should_retry(b64)){
            continue;
        }
        else // encoding failed
        {
            /* Handle Error!!! */
        }
    }
    else // success!
        done = true;
}

BIO_flush(b64);

// get a pointer to mem's data
unsigned char* output;
*lenoutput = BIO_get_mem_data(mem, &output);

// assign data to output
//std::string s(dt, len2);

return output;
}

To write to file

写入文件

int encodeb64(unsigned char* input, const char* filenm, int leni)
{
BIO *b64 = BIO_new(BIO_f_base64());
BIO_set_flags(b64,BIO_FLAGS_BASE64_NO_NL);

BIO *file = BIO_new_file(filenm, "w");
BIO *mem = BIO_new(BIO_f_buffer());
BIO_push(b64, mem);
BIO_push(mem, file);

// write data
bool done = false;
int res = 0;
while(!done)
{
    res = BIO_write(b64, input, leni);

    if(res <= 0) // if failed
    {
        if(BIO_should_retry(b64)){
            continue;
        }
        else // encoding failed
        {
            /* Handle Error!!! */
        }
    }
    else // success!
        done = true;
}

BIO_flush(b64);
BIO_pop(b64);
BIO_free_all(b64);

    return 0;
}

Base64 encoding from file to file. Many times due to file constraint we have read in chunks of data and do encoding. Below is the code.

从文件到文件的 Base64 编码。很多时候由于文件限制，我们读取了大块数据并进行编码。下面是代码。

int encodeb64FromFile(const char* input, const char* outputfilename)
{
BIO *b64 = BIO_new(BIO_f_base64());
BIO_set_flags(b64,BIO_FLAGS_BASE64_NO_NL);
int leni = 3*64;
unsigned char *data[3*64];
BIO *file = BIO_new_file(outputfilename, "w");
BIO *mem = BIO_new(BIO_f_buffer());
BIO_push(b64, mem);
BIO_push(mem, file);

FILE *fp = fopen(input, "rb");
while ((leni = fread(data, 1, sizeof data, fp)) > 0) {
    // write data
    bool done = false;
    int res = 0;
    while(!done)
    {
        res = BIO_write(b64, data, leni);

        if(res <= 0) // if failed
        {
            if(BIO_should_retry(b64)){
                continue;
            }
            else // encoding failed
            {
                /* Handle Error!!! */
            }
        }
        else // success!
            done = true;
    }

 }

 BIO_flush(b64);
BIO_pop(b64);
BIO_free_all(b64);
fclose(fp);

return 0;
 }

回答by fsenart

  #include <openssl/bio.h>

  typedef unsigned char byte;      

  namespace base64 {
    static void Encode(const byte* in, size_t in_len,
                       char** out, size_t* out_len) {
      BIO *buff, *b64f;
      BUF_MEM *ptr;

      b64f = BIO_new(BIO_f_base64());
      buff = BIO_new(BIO_s_mem());
      buff = BIO_push(b64f, buff);

      BIO_set_flags(buff, BIO_FLAGS_BASE64_NO_NL);
      BIO_set_close(buff, BIO_CLOSE);
      BIO_write(buff, in, in_len);
      BIO_flush(buff);

      BIO_get_mem_ptr(buff, &ptr);
      (*out_len) = ptr->length;
      (*out) = (char *) malloc(((*out_len) + 1) * sizeof(char));
      memcpy(*out, ptr->data, (*out_len));
      (*out)[(*out_len)] = '
#include <iostream>
#include <stdlib.h>
#include <openssl/evp.h>

char *base64(const unsigned char *input, int length) {
  const auto pl = 4*((length+2)/3);
  auto output = reinterpret_cast<char *>(calloc(pl+1, 1)); //+1 for the terminating null that EVP_EncodeBlock adds on
  const auto ol = EVP_EncodeBlock(reinterpret_cast<unsigned char *>(output), input, length);
  if (pl != ol) { std::cerr << "Whoops, encode predicted " << pl << " but we got " << ol << "\n"; }
  return output;
}

unsigned char *decode64(const char *input, int length) {
  const auto pl = 3*length/4;
  auto output = reinterpret_cast<unsigned char *>(calloc(pl+1, 1));
  const auto ol = EVP_DecodeBlock(output, reinterpret_cast<const unsigned char *>(input), length);
  if (pl != ol) { std::cerr << "Whoops, decode predicted " << pl << " but we got " << ol << "\n"; }
  return output;
}
';

      BIO_free_all(buff);
    }

    static void Decode(const char* in, size_t in_len,
                       byte** out, size_t* out_len) {
      BIO *buff, *b64f;

      b64f = BIO_new(BIO_f_base64());
      buff = BIO_new_mem_buf((void *)in, in_len);
      buff = BIO_push(b64f, buff);
      (*out) = (byte *) malloc(in_len * sizeof(char));

      BIO_set_flags(buff, BIO_FLAGS_BASE64_NO_NL);
      BIO_set_close(buff, BIO_CLOSE);
      (*out_len) = BIO_read(buff, (*out), in_len);
      (*out) = (byte *) realloc((void *)(*out), ((*out_len) + 1) * sizeof(byte));
      (*out)[(*out_len)] = '##代码##';

      BIO_free_all(buff);
    }
  }

回答by mtrw

Rather than using the BIO_interface it's much easier to use the EVP_interface. For instance:

而不是使用BIO_界面，它更容易使用的EVP_界面。例如：

##代码##

The EVP functions include a streaming interface too, see the man page.

EVP 功能也包括流接口，请参阅手册页。

回答by David Gelhar

Base64 is really pretty simple; you shouldn't have trouble finding any number of implementations via a quick Google. For example hereis a reference implementation in C from the Internet Software Consortium, with detailed comments explaining the process.

Base64 真的很简单；您应该可以通过快速 Google 轻松找到任何数量的实现。例如，这里是来自 Internet Software Consortium 的 C 参考实现，其中包含解释该过程的详细注释。

The openssl implementation layers a lot of complexity with the "BIO" stuff that's not (IMHO) very useful if all you're doing is decoding/encoding.

如果您所做的只是解码/编码，则openssl 实现将很多复杂性与“BIO”内容（恕我直言）非常有用。