C++中有免费的Base64解码代码sn-p吗？
【问题讨论】：
这是我对the implementation that was originally written by René Nyffenegger 的修改。为什么我要修改它？好吧，因为我似乎不适合使用存储在 std::string 对象中的二进制数据；）
base64.h：
#ifndef _BASE64_H_
#define _BASE64_H_
#include <vector>
#include <string>
typedef unsigned char BYTE;
std::string base64_encode(BYTE const* buf, unsigned int bufLen);
std::vector<BYTE> base64_decode(std::string const&);
#endif
base64.cpp：
#include "base64.h"
#include <iostream>
static const std::string base64_chars =
"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
"abcdefghijklmnopqrstuvwxyz"
"0123456789+/";
static inline bool is_base64(BYTE c) {
return (isalnum(c) || (c == '+') || (c == '/'));
}
std::string base64_encode(BYTE const* buf, unsigned int bufLen) {
std::string ret;
int i = 0;
int j = 0;
BYTE char_array_3[3];
BYTE char_array_4[4];
while (bufLen--) {
char_array_3[i++] = *(buf++);
if (i == 3) {
char_array_4[0] = (char_array_3[0] & 0xfc) >> 2;
char_array_4[1] = ((char_array_3[0] & 0x03) << 4) + ((char_array_3[1] & 0xf0) >> 4);
char_array_4[2] = ((char_array_3[1] & 0x0f) << 2) + ((char_array_3[2] & 0xc0) >> 6);
char_array_4[3] = char_array_3[2] & 0x3f;
for(i = 0; (i <4) ; i++)
ret += base64_chars[char_array_4[i]];
i = 0;
}
}
if (i)
{
for(j = i; j < 3; j++)
char_array_3[j] = '\0';
char_array_4[0] = (char_array_3[0] & 0xfc) >> 2;
char_array_4[1] = ((char_array_3[0] & 0x03) << 4) + ((char_array_3[1] & 0xf0) >> 4);
char_array_4[2] = ((char_array_3[1] & 0x0f) << 2) + ((char_array_3[2] & 0xc0) >> 6);
char_array_4[3] = char_array_3[2] & 0x3f;
for (j = 0; (j < i + 1); j++)
ret += base64_chars[char_array_4[j]];
while((i++ < 3))
ret += '=';
}
return ret;
}
std::vector<BYTE> base64_decode(std::string const& encoded_string) {
int in_len = encoded_string.size();
int i = 0;
int j = 0;
int in_ = 0;
BYTE char_array_4[4], char_array_3[3];
std::vector<BYTE> ret;
while (in_len-- && ( encoded_string[in_] != '=') && is_base64(encoded_string[in_])) {
char_array_4[i++] = encoded_string[in_]; in_++;
if (i ==4) {
for (i = 0; i <4; i++)
char_array_4[i] = base64_chars.find(char_array_4[i]);
char_array_3[0] = (char_array_4[0] << 2) + ((char_array_4[1] & 0x30) >> 4);
char_array_3[1] = ((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2);
char_array_3[2] = ((char_array_4[2] & 0x3) << 6) + char_array_4[3];
for (i = 0; (i < 3); i++)
ret.push_back(char_array_3[i]);
i = 0;
}
}
if (i) {
for (j = i; j <4; j++)
char_array_4[j] = 0;
for (j = 0; j <4; j++)
char_array_4[j] = base64_chars.find(char_array_4[j]);
char_array_3[0] = (char_array_4[0] << 2) + ((char_array_4[1] & 0x30) >> 4);
char_array_3[1] = ((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2);
char_array_3[2] = ((char_array_4[2] & 0x3) << 6) + char_array_4[3];
for (j = 0; (j < i - 1); j++) ret.push_back(char_array_3[j]);
}
return ret;
}
用法如下：
std::vector<BYTE> myData;
...
std::string encodedData = base64_encode(&myData[0], myData.size());
std::vector<BYTE> decodedData = base64_decode(encodedData);
【讨论】：
见Encoding and decoding base 64 with C++。
这是该页面的实现：
/*
base64.cpp and base64.h
Copyright (C) 2004-2008 René Nyffenegger
This source code is provided 'as-is', without any express or implied
warranty. In no event will the author be held liable for any damages
arising from the use of this software.
Permission is granted to anyone to use this software for any purpose,
including commercial applications, and to alter it and redistribute it
freely, subject to the following restrictions:
1. The origin of this source code must not be misrepresented; you must not
claim that you wrote the original source code. If you use this source code
in a product, an acknowledgment in the product documentation would be
appreciated but is not required.
2. Altered source versions must be plainly marked as such, and must not be
misrepresented as being the original source code.
3. This notice may not be removed or altered from any source distribution.
René Nyffenegger rene.nyffenegger@adp-gmbh.ch
*/
static const std::string base64_chars =
"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
"abcdefghijklmnopqrstuvwxyz"
"0123456789+/";
static inline bool is_base64(unsigned char c) {
return (isalnum(c) || (c == '+') || (c == '/'));
}
std::string base64_encode(unsigned char const* bytes_to_encode, unsigned int in_len) {
std::string ret;
int i = 0;
int j = 0;
unsigned char char_array_3[3];
unsigned char char_array_4[4];
while (in_len--) {
char_array_3[i++] = *(bytes_to_encode++);
if (i == 3) {
char_array_4[0] = (char_array_3[0] & 0xfc) >> 2;
char_array_4[1] = ((char_array_3[0] & 0x03) << 4) + ((char_array_3[1] & 0xf0) >> 4);
char_array_4[2] = ((char_array_3[1] & 0x0f) << 2) + ((char_array_3[2] & 0xc0) >> 6);
char_array_4[3] = char_array_3[2] & 0x3f;
for(i = 0; (i <4) ; i++)
ret += base64_chars[char_array_4[i]];
i = 0;
}
}
if (i)
{
for(j = i; j < 3; j++)
char_array_3[j] = '\0';
char_array_4[0] = (char_array_3[0] & 0xfc) >> 2;
char_array_4[1] = ((char_array_3[0] & 0x03) << 4) + ((char_array_3[1] & 0xf0) >> 4);
char_array_4[2] = ((char_array_3[1] & 0x0f) << 2) + ((char_array_3[2] & 0xc0) >> 6);
char_array_4[3] = char_array_3[2] & 0x3f;
for (j = 0; (j < i + 1); j++)
ret += base64_chars[char_array_4[j]];
while((i++ < 3))
ret += '=';
}
return ret;
}
std::string base64_decode(std::string const& encoded_string) {
int in_len = encoded_string.size();
int i = 0;
int j = 0;
int in_ = 0;
unsigned char char_array_4[4], char_array_3[3];
std::string ret;
while (in_len-- && ( encoded_string[in_] != '=') && is_base64(encoded_string[in_])) {
char_array_4[i++] = encoded_string[in_]; in_++;
if (i ==4) {
for (i = 0; i <4; i++)
char_array_4[i] = base64_chars.find(char_array_4[i]);
char_array_3[0] = (char_array_4[0] << 2) + ((char_array_4[1] & 0x30) >> 4);
char_array_3[1] = ((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2);
char_array_3[2] = ((char_array_4[2] & 0x3) << 6) + char_array_4[3];
for (i = 0; (i < 3); i++)
ret += char_array_3[i];
i = 0;
}
}
if (i) {
for (j = i; j <4; j++)
char_array_4[j] = 0;
for (j = 0; j <4; j++)
char_array_4[j] = base64_chars.find(char_array_4[j]);
char_array_3[0] = (char_array_4[0] << 2) + ((char_array_4[1] & 0x30) >> 4);
char_array_3[1] = ((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2);
char_array_3[2] = ((char_array_4[2] & 0x3) << 6) + char_array_4[3];
for (j = 0; (j < i - 1); j++) ret += char_array_3[j];
}
return ret;
}
【讨论】：
in_len，我们知道ret的长度，为什么不在初始化时给它一个固定的长度呢？
if (i)之后，第一个for循环不是必须的。如果 char_array_4 中的字节用 0 填充，find() 返回 -1，但根本没有使用。因此，第二个 for 循环可以是：for (j = 0; j < i; j++)。另外，char_array_3[2] 的第三行完全没用，可以去掉。为什么？因为第二个块只处理其余部分（当 i 不为零时），而这只能是原始文本的一两个字节（如果有三个，那么就没有其余部分了，因为三个字节可以平滑地编码成4 个字符。）
这里有几个sn-ps。然而，这个是紧凑、高效且对 C++11 友好的：
static std::string base64_encode(const std::string &in) {
std::string out;
int val = 0, valb = -6;
for (uchar c : in) {
val = (val << 8) + c;
valb += 8;
while (valb >= 0) {
out.push_back("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"[(val>>valb)&0x3F]);
valb -= 6;
}
}
if (valb>-6) out.push_back("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"[((val<<8)>>(valb+8))&0x3F]);
while (out.size()%4) out.push_back('=');
return out;
}
static std::string base64_decode(const std::string &in) {
std::string out;
std::vector<int> T(256,-1);
for (int i=0; i<64; i++) T["ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"[i]] = i;
int val=0, valb=-8;
for (uchar c : in) {
if (T[c] == -1) break;
val = (val << 6) + T[c];
valb += 6;
if (valb >= 0) {
out.push_back(char((val>>valb)&0xFF));
valb -= 8;
}
}
return out;
}
【讨论】：
我认为这个效果更好：
#include <string>
static const char* B64chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
static const int B64index[256] =
{
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 62, 63, 62, 62, 63,
52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 0, 0, 0, 0, 0, 0,
0, 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, 0, 0, 0, 0, 63,
0, 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
};
const std::string b64encode(const void* data, const size_t &len)
{
std::string result((len + 2) / 3 * 4, '=');
unsigned char *p = (unsigned char*) data;
char *str = &result[0];
size_t j = 0, pad = len % 3;
const size_t last = len - pad;
for (size_t i = 0; i < last; i += 3)
{
int n = int(p[i]) << 16 | int(p[i + 1]) << 8 | p[i + 2];
str[j++] = B64chars[n >> 18];
str[j++] = B64chars[n >> 12 & 0x3F];
str[j++] = B64chars[n >> 6 & 0x3F];
str[j++] = B64chars[n & 0x3F];
}
if (pad) /// Set padding
{
int n = --pad ? int(p[last]) << 8 | p[last + 1] : p[last];
str[j++] = B64chars[pad ? n >> 10 & 0x3F : n >> 2];
str[j++] = B64chars[pad ? n >> 4 & 0x03F : n << 4 & 0x3F];
str[j++] = pad ? B64chars[n << 2 & 0x3F] : '=';
}
return result;
}
const std::string b64decode(const void* data, const size_t &len)
{
if (len == 0) return "";
unsigned char *p = (unsigned char*) data;
size_t j = 0,
pad1 = len % 4 || p[len - 1] == '=',
pad2 = pad1 && (len % 4 > 2 || p[len - 2] != '=');
const size_t last = (len - pad1) / 4 << 2;
std::string result(last / 4 * 3 + pad1 + pad2, '\0');
unsigned char *str = (unsigned char*) &result[0];
for (size_t i = 0; i < last; i += 4)
{
int n = B64index[p[i]] << 18 | B64index[p[i + 1]] << 12 | B64index[p[i + 2]] << 6 | B64index[p[i + 3]];
str[j++] = n >> 16;
str[j++] = n >> 8 & 0xFF;
str[j++] = n & 0xFF;
}
if (pad1)
{
int n = B64index[p[last]] << 18 | B64index[p[last + 1]] << 12;
str[j++] = n >> 16;
if (pad2)
{
n |= B64index[p[last + 2]] << 6;
str[j++] = n >> 8 & 0xFF;
}
}
return result;
}
std::string b64encode(const std::string& str)
{
return b64encode(str.c_str(), str.size());
}
std::string b64decode(const std::string& str64)
{
return b64decode(str64.c_str(), str64.size());
}
感谢Jens Alfke for pointing out a performance issue，我对这篇旧帖子做了一些修改。这个工作方式比以前更快。它的另一个优点是可以顺利处理损坏的数据。
上一版：虽然在这类问题中，速度似乎有点过头了，但只是为了好玩，我做了一些其他修改，以使这个算法成为目前最快的算法据我所知。特别感谢 GaspardP for his valuable suggestions 和不错的基准测试。
【讨论】：
strchr 的调用都会减慢解码器的速度——每个解码的字节平均循环32 次。大多数解决方案使用 256 项查找表来避免这种情况，这样会快很多。
char* 而不是std::string（只需执行char* out = &str[0]，然后使用out[j++] 而不是 str[j++]) 通过这样做，您将跳过 std::string::operator[] 所做的不必要检查。另外，避免最后一个push_back，通过分配一个多字节（std::string str;str.resize(3*((len+3)/4));，然后在最后使用out[j++]everywere 和str.resize(j);，这可能会导致非常昂贵。
delete 的情况下使用new。实际上，根本不要使用new。 @Gaspard，我不知道有任何“std::string::operator[] 进行了不必要的检查”（事实上，我有理由确定没有，至少在发布时），但如果真的绝望，你可以使用 vector<char> -无论如何，您的替换代码不是 Gaspard 建议的，除了元素访问之外，它仍然使用字符串进行所有操作，并且无论我认为这是不必要的事实，它都是安全/快速的；）
unsigned char *p，而不是char *p。如果我的输入包含字节 >= 0x80，我的 base64 字符串已损坏。添加unsigned后，似乎一切正常。
使用base-n mini lib，你可以做到以下几点：
some_data_t in[] { ... };
constexpr int len = sizeof(in)/sizeof(in[0]);
std::string encoded;
bn::encode_b64(in, in + len, std::back_inserter(encoded));
some_data_t out[len];
bn::decode_b64(encoded.begin(), encoded.end(), out);
API 是通用的，基于迭代器的。
披露：我是作者。
【讨论】：
std::size 会比sizeof hax 更好
根据 GaspardP 提出的this excellent comparison，我不会选择这个解决方案。这不是最差的，但也不是最好的。唯一的原因是它可能更容易理解。
我发现其他两个答案很难理解。它们还会在我的编译器中产生一些警告，并且在解码部分使用 find 函数会导致效率非常低。所以我决定自己动手。
标题：
#ifndef _BASE64_H_
#define _BASE64_H_
#include <vector>
#include <string>
typedef unsigned char BYTE;
class Base64
{
public:
static std::string encode(const std::vector<BYTE>& buf);
static std::string encode(const BYTE* buf, unsigned int bufLen);
static std::vector<BYTE> decode(std::string encoded_string);
};
#endif
主体：
static const BYTE from_base64[] = { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 62, 255, 62, 255, 63,
52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 255, 255, 255, 255, 255, 255,
255, 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, 255, 255, 255, 255, 63,
255, 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, 255, 255, 255, 255, 255};
static const char to_base64[] =
"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
"abcdefghijklmnopqrstuvwxyz"
"0123456789+/";
std::string Base64::encode(const std::vector<BYTE>& buf)
{
if (buf.empty())
return ""; // Avoid dereferencing buf if it's empty
return encode(&buf[0], (unsigned int)buf.size());
}
std::string Base64::encode(const BYTE* buf, unsigned int bufLen)
{
// Calculate how many bytes that needs to be added to get a multiple of 3
size_t missing = 0;
size_t ret_size = bufLen;
while ((ret_size % 3) != 0)
{
++ret_size;
++missing;
}
// Expand the return string size to a multiple of 4
ret_size = 4*ret_size/3;
std::string ret;
ret.reserve(ret_size);
for (unsigned int i=0; i<ret_size/4; ++i)
{
// Read a group of three bytes (avoid buffer overrun by replacing with 0)
size_t index = i*3;
BYTE b3[3];
b3[0] = (index+0 < bufLen) ? buf[index+0] : 0;
b3[1] = (index+1 < bufLen) ? buf[index+1] : 0;
b3[2] = (index+2 < bufLen) ? buf[index+2] : 0;
// Transform into four base 64 characters
BYTE b4[4];
b4[0] = ((b3[0] & 0xfc) >> 2);
b4[1] = ((b3[0] & 0x03) << 4) + ((b3[1] & 0xf0) >> 4);
b4[2] = ((b3[1] & 0x0f) << 2) + ((b3[2] & 0xc0) >> 6);
b4[3] = ((b3[2] & 0x3f) << 0);
// Add the base 64 characters to the return value
ret.push_back(to_base64[b4[0]]);
ret.push_back(to_base64[b4[1]]);
ret.push_back(to_base64[b4[2]]);
ret.push_back(to_base64[b4[3]]);
}
// Replace data that is invalid (always as many as there are missing bytes)
for (size_t i=0; i<missing; ++i)
ret[ret_size - i - 1] = '=';
return ret;
}
std::vector<BYTE> Base64::decode(std::string encoded_string)
{
// Make sure string length is a multiple of 4
while ((encoded_string.size() % 4) != 0)
encoded_string.push_back('=');
size_t encoded_size = encoded_string.size();
std::vector<BYTE> ret;
ret.reserve(3*encoded_size/4);
for (size_t i=0; i<encoded_size; i += 4)
{
// Get values for each group of four base 64 characters
BYTE b4[4];
b4[0] = (encoded_string[i+0] <= 'z') ? from_base64[encoded_string[i+0]] : 0xff;
b4[1] = (encoded_string[i+1] <= 'z') ? from_base64[encoded_string[i+1]] : 0xff;
b4[2] = (encoded_string[i+2] <= 'z') ? from_base64[encoded_string[i+2]] : 0xff;
b4[3] = (encoded_string[i+3] <= 'z') ? from_base64[encoded_string[i+3]] : 0xff;
// Transform into a group of three bytes
BYTE b3[3];
b3[0] = ((b4[0] & 0x3f) << 2) + ((b4[1] & 0x30) >> 4);
b3[1] = ((b4[1] & 0x0f) << 4) + ((b4[2] & 0x3c) >> 2);
b3[2] = ((b4[2] & 0x03) << 6) + ((b4[3] & 0x3f) >> 0);
// Add the byte to the return value if it isn't part of an '=' character (indicated by 0xff)
if (b4[1] != 0xff) ret.push_back(b3[0]);
if (b4[2] != 0xff) ret.push_back(b3[1]);
if (b4[3] != 0xff) ret.push_back(b3[2]);
}
return ret;
}
用法：
BYTE buf[] = "ABCD";
std::string encoded = Base64::encode(buf, 4);
// encoded = "QUJDRA=="
std::vector<BYTE> decoded = Base64::decode(encoded);
这里的好处是解码函数还可以解码 Base64 编码的 URL 变体。
【讨论】：
使用更紧凑的查找表和使用 C++17 功能的一些变化：
std::string base64_decode(const std::string_view in) {
// table from '+' to 'z'
const uint8_t lookup[] = {
62, 255, 62, 255, 63, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 255,
255, 0, 255, 255, 255, 255, 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,
255, 255, 255, 255, 63, 255, 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};
static_assert(sizeof(lookup) == 'z' - '+' + 1);
std::string out;
int val = 0, valb = -8;
for (uint8_t c : in) {
if (c < '+' || c > 'z')
break;
c -= '+';
if (lookup[c] >= 64)
break;
val = (val << 6) + lookup[c];
valb += 6;
if (valb >= 0) {
out.push_back(char((val >> valb) & 0xFF));
valb -= 8;
}
}
return out;
}
如果您没有 std::string_view，请尝试使用 std::experimental::string_view。
【讨论】：
我对@987654321@的变化：
它避免了以多次测试为代价来复制参数。
使用 uint8_t 而不是 BYTE。
为处理字符串添加了一些方便的函数，尽管通常输入数据是二进制的并且内部可能有零字节，因此通常不应将其作为字符串进行操作（这通常意味着以空值结尾的数据）。
还添加了一些强制转换来修复编译器警告（至少在 GCC 上，我还没有通过 MSVC 运行它）。
部分文件base64.hpp：
void base64_encode(string & out, const vector<uint8_t>& buf);
void base64_encode(string & out, const uint8_t* buf, size_t bufLen);
void base64_encode(string & out, string const& buf);
void base64_decode(vector<uint8_t> & out, string const& encoded_string);
// Use this if you know the output should be a valid string
void base64_decode(string & out, string const& encoded_string);
static const uint8_t from_base64[128] = {
// 8 rows of 16 = 128
// Note: only requires 123 entries, as we only lookup for <= z , which z=122
255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 62, 255, 62, 255, 63,
52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 255, 255, 0, 255, 255, 255,
255, 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, 255, 255, 255, 255, 63,
255, 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, 255, 255, 255, 255, 255
};
static const char to_base64[65] =
"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
"abcdefghijklmnopqrstuvwxyz"
"0123456789+/";
void base64_encode(string & out, string const& buf)
{
if (buf.empty())
base64_encode(out, NULL, 0);
else
base64_encode(out, reinterpret_cast<uint8_t const*>(&buf[0]), buf.size());
}
void base64_encode(string & out, std::vector<uint8_t> const& buf)
{
if (buf.empty())
base64_encode(out, NULL, 0);
else
base64_encode(out, &buf[0], buf.size());
}
void base64_encode(string & ret, uint8_t const* buf, size_t bufLen)
{
// Calculate how many bytes that needs to be added to get a multiple of 3
size_t missing = 0;
size_t ret_size = bufLen;
while ((ret_size % 3) != 0)
{
++ret_size;
++missing;
}
// Expand the return string size to a multiple of 4
ret_size = 4*ret_size/3;
ret.clear();
ret.reserve(ret_size);
for (size_t i = 0; i < ret_size/4; ++i)
{
// Read a group of three bytes (avoid buffer overrun by replacing with 0)
const size_t index = i*3;
const uint8_t b3_0 = (index+0 < bufLen) ? buf[index+0] : 0;
const uint8_t b3_1 = (index+1 < bufLen) ? buf[index+1] : 0;
const uint8_t b3_2 = (index+2 < bufLen) ? buf[index+2] : 0;
// Transform into four base 64 characters
const uint8_t b4_0 = ((b3_0 & 0xfc) >> 2);
const uint8_t b4_1 = ((b3_0 & 0x03) << 4) + ((b3_1 & 0xf0) >> 4);
const uint8_t b4_2 = ((b3_1 & 0x0f) << 2) + ((b3_2 & 0xc0) >> 6);
const uint8_t b4_3 = ((b3_2 & 0x3f) << 0);
// Add the base 64 characters to the return value
ret.push_back(to_base64[b4_0]);
ret.push_back(to_base64[b4_1]);
ret.push_back(to_base64[b4_2]);
ret.push_back(to_base64[b4_3]);
}
// Replace data that is invalid (always as many as there are missing bytes)
for (size_t i = 0; i != missing; ++i)
ret[ret_size - i - 1] = '=';
}
template <class Out>
void base64_decode_any( Out & ret, std::string const& in)
{
typedef typename Out::value_type T;
// Make sure the *intended* string length is a multiple of 4
size_t encoded_size = in.size();
while ((encoded_size % 4) != 0)
++encoded_size;
const size_t N = in.size();
ret.clear();
ret.reserve(3*encoded_size/4);
for (size_t i = 0; i < encoded_size; i += 4)
{
// Note: 'z' == 122
// Get values for each group of four base 64 characters
const uint8_t b4_0 = ( in[i+0] <= 'z') ? from_base64[static_cast<uint8_t>(in[i+0])] : 0xff;
const uint8_t b4_1 = (i+1 < N and in[i+1] <= 'z') ? from_base64[static_cast<uint8_t>(in[i+1])] : 0xff;
const uint8_t b4_2 = (i+2 < N and in[i+2] <= 'z') ? from_base64[static_cast<uint8_t>(in[i+2])] : 0xff;
const uint8_t b4_3 = (i+3 < N and in[i+3] <= 'z') ? from_base64[static_cast<uint8_t>(in[i+3])] : 0xff;
// Transform into a group of three bytes
const uint8_t b3_0 = ((b4_0 & 0x3f) << 2) + ((b4_1 & 0x30) >> 4);
const uint8_t b3_1 = ((b4_1 & 0x0f) << 4) + ((b4_2 & 0x3c) >> 2);
const uint8_t b3_2 = ((b4_2 & 0x03) << 6) + ((b4_3 & 0x3f) >> 0);
// Add the byte to the return value if it isn't part of an '=' character (indicated by 0xff)
if (b4_1 != 0xff) ret.push_back( static_cast<T>(b3_0) );
if (b4_2 != 0xff) ret.push_back( static_cast<T>(b3_1) );
if (b4_3 != 0xff) ret.push_back( static_cast<T>(b3_2) );
}
}
void base64_decode(vector<uint8_t> & out, string const& encoded_string)
{
base64_decode_any(out, encoded_string);
}
void base64_decode(string & out, string const& encoded_string)
{
base64_decode_any(out, encoded_string);
}
【讨论】：
我的版本是用于 C++Builder 的 Base64 的简单快速编码器（解码器）。
// ---------------------------------------------------------------------------
UnicodeString __fastcall TExample::Base64Encode(void *data, int length)
{
if (length <= 0)
return L"";
static const char set[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
unsigned char *in = (unsigned char*)data;
char *pos, *out = pos = new char[((length - 1) / 3 + 1) << 2];
while ((length -= 3) >= 0)
{
pos[0] = set[in[0] >> 2];
pos[1] = set[((in[0] & 0x03) << 4) | (in[1] >> 4)];
pos[2] = set[((in[1] & 0x0F) << 2) | (in[2] >> 6)];
pos[3] = set[in[2] & 0x3F];
pos += 4;
in += 3;
};
if ((length & 2) != 0)
{
pos[0] = set[in[0] >> 2];
if ((length & 1) != 0)
{
pos[1] = set[((in[0] & 0x03) << 4) | (in[1] >> 4)];
pos[2] = set[(in[1] & 0x0F) << 2];
}
else
{
pos[1] = set[(in[0] & 0x03) << 4];
pos[2] = '=';
};
pos[3] = '=';
pos += 4;
};
UnicodeString code = UnicodeString(out, pos - out);
delete[] out;
return code;
};
// ---------------------------------------------------------------------------
int __fastcall TExample::Base64Decode(const UnicodeString &code, unsigned char **data)
{
int length;
if (((length = code.Length()) == 0) || ((length & 3) != 0))
return 0;
wchar_t *str = code.c_str();
unsigned char *pos, *out = pos = new unsigned char[(length >> 2) * 3];
while (*str != 0)
{
length = -1;
int shift = 18, bits = 0;
do
{
wchar_t s = str[++length];
if ((s >= L'A') && (s <= L'Z'))
bits |= (s - L'A') << shift;
else if ((s >= L'a') && (s <= L'z'))
bits |= (s - (L'a' - 26)) << shift;
else if (((s >= L'0') && (s <= L'9')))
bits |= (s - (L'0' - 52)) << shift;
else if (s == L'+')
bits |= 62 << shift;
else if (s == L'/')
bits |= 63 << shift;
else if (s == L'=')
{
length--;
break;
}
else
{
delete[] out;
return 0;
};
}
while ((shift -= 6) >= 0);
pos[0] = bits >> 16;
pos[1] = bits >> 8;
pos[2] = bits;
pos += length;
str += 4;
};
*data = out;
return pos - out;
};
//---------------------------------------------------------------------------
【讨论】：
我用这个：
class BinaryVector {
public:
std::vector<char> bytes;
uint64_t bit_count = 0;
public:
/* Add a bit to the end */
void push_back(bool bit);
/* Return false if character is unrecognized */
bool pushBase64Char(char b64_c);
};
void BinaryVector::push_back(bool bit)
{
if (!bit_count || bit_count % 8 == 0) {
bytes.push_back(bit << 7);
}
else {
uint8_t next_bit = 8 - (bit_count % 8) - 1;
bytes[bit_count / 8] |= bit << next_bit;
}
bit_count++;
}
/* Converts one Base64 character to 6 bits */
bool BinaryVector::pushBase64Char(char c)
{
uint8_t d;
// A to Z
if (c > 0x40 && c < 0x5b) {
d = c - 65; // Base64 A is 0
}
// a to z
else if (c > 0x60 && c < 0x7b) {
d = c - 97 + 26; // Base64 a is 26
}
// 0 to 9
else if (c > 0x2F && c < 0x3a) {
d = c - 48 + 52; // Base64 0 is 52
}
else if (c == '+') {
d = 0b111110;
}
else if (c == '/') {
d = 0b111111;
}
else if (c == '=') {
d = 0;
}
else {
return false;
}
push_back(d & 0b100000);
push_back(d & 0b010000);
push_back(d & 0b001000);
push_back(d & 0b000100);
push_back(d & 0b000010);
push_back(d & 0b000001);
return true;
}
bool loadBase64(std::vector<char>& b64_bin, BinaryVector& vec)
{
for (char& c : b64_bin) {
if (!vec.pushBase64Char(c)) {
return false;
}
}
return true;
}
使用vec.bytes 访问转换后的数据。
【讨论】：
我先做了自己的版本，然后找到了这个话题。
为什么我的版本看起来比这里展示的其他版本更简单？难道我做错了什么？我没有测试它的速度。
inline char const* b64units = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
inline char* b64encode(void const* a, int64_t b) {
ASSERT(a != nullptr);
if (b > 0) {
uint8_t const* aa = static_cast<uint8_t const*>(a);
uint8_t v = 0;
int64_t bp = 0;
int64_t sb = 0;
int8_t off = 0;
int64_t nt = ((b + 2) / 3) * 4;
int64_t nd = (b * 8) / 6;
int64_t tl = ((b * 8) % 6) ? 1 : 0;
int64_t nf = nt - nd - tl;
int64_t ri = 0;
char* r = new char[nt + 1]();
for (int64_t i = 0; i < nd; i++) {
v = (aa[sb] << off) | (aa[sb + 1] >> (8 - off));
v >>= 2;
r[ri] = b64units[v];
ri += 1;
bp += 6;
sb = (bp / 8);
off = (bp % 8);
}
if (tl > 0) {
v = (aa[sb] << off);
v >>= 2;
r[ri] = b64units[v];
ri += 1;
}
for (int64_t i = 0; i < nf; i++) {
r[ri] = '=';
ri += 1;
}
return r;
} else return nullptr;
}
P.S.：我的方法效果很好。我用Node.js测试了它：
let data = 'stackabuse.com';
let buff = new Buffer(data);
let base64data = buff.toString('base64');
【讨论】：
这是一个单独的 hpp 文件，它使用 vector
#pragma once
#include <string>
#include <vector>
#include <stdexcept>
#include <cstdint>
namespace base64
{
inline static const char kEncodeLookup[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
inline static const char kPadCharacter = '=';
using byte = std::uint8_t;
inline std::string encode(const std::vector<byte>& input)
{
std::string encoded;
encoded.reserve(((input.size() / 3) + (input.size() % 3 > 0)) * 4);
std::uint32_t temp{};
auto it = input.begin();
for(std::size_t i = 0; i < input.size() / 3; ++i)
{
temp = (*it++) << 16;
temp += (*it++) << 8;
temp += (*it++);
encoded.append(1, kEncodeLookup[(temp & 0x00FC0000) >> 18]);
encoded.append(1, kEncodeLookup[(temp & 0x0003F000) >> 12]);
encoded.append(1, kEncodeLookup[(temp & 0x00000FC0) >> 6 ]);
encoded.append(1, kEncodeLookup[(temp & 0x0000003F) ]);
}
switch(input.size() % 3)
{
case 1:
temp = (*it++) << 16;
encoded.append(1, kEncodeLookup[(temp & 0x00FC0000) >> 18]);
encoded.append(1, kEncodeLookup[(temp & 0x0003F000) >> 12]);
encoded.append(2, kPadCharacter);
break;
case 2:
temp = (*it++) << 16;
temp += (*it++) << 8;
encoded.append(1, kEncodeLookup[(temp & 0x00FC0000) >> 18]);
encoded.append(1, kEncodeLookup[(temp & 0x0003F000) >> 12]);
encoded.append(1, kEncodeLookup[(temp & 0x00000FC0) >> 6 ]);
encoded.append(1, kPadCharacter);
break;
}
return encoded;
}
std::vector<byte> decode(const std::string& input)
{
if(input.length() % 4)
throw std::runtime_error("Invalid base64 length!");
std::size_t padding{};
if(input.length())
{
if(input[input.length() - 1] == kPadCharacter) padding++;
if(input[input.length() - 2] == kPadCharacter) padding++;
}
std::vector<byte> decoded;
decoded.reserve(((input.length() / 4) * 3) - padding);
std::uint32_t temp{};
auto it = input.begin();
while(it < input.end())
{
for(std::size_t i = 0; i < 4; ++i)
{
temp <<= 6;
if (*it >= 0x41 && *it <= 0x5A) temp |= *it - 0x41;
else if(*it >= 0x61 && *it <= 0x7A) temp |= *it - 0x47;
else if(*it >= 0x30 && *it <= 0x39) temp |= *it + 0x04;
else if(*it == 0x2B) temp |= 0x3E;
else if(*it == 0x2F) temp |= 0x3F;
else if(*it == kPadCharacter)
{
switch(input.end() - it)
{
case 1:
decoded.push_back((temp >> 16) & 0x000000FF);
decoded.push_back((temp >> 8 ) & 0x000000FF);
return decoded;
case 2:
decoded.push_back((temp >> 10) & 0x000000FF);
return decoded;
default:
throw std::runtime_error("Invalid padding in base64!");
}
}
else throw std::runtime_error("Invalid character in base64!");
++it;
}
decoded.push_back((temp >> 16) & 0x000000FF);
decoded.push_back((temp >> 8 ) & 0x000000FF);
decoded.push_back((temp ) & 0x000000FF);
}
return decoded;
}
}
【讨论】：
这是我写的一篇文章，它使用联合和位域来获得最大的效率和可读性。
const char PADDING_CHAR = '=';
const char* ALPHABET = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
const uint8_t DECODED_ALPHBET[128]={0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,62,0,0,0,63,52,53,54,55,56,57,58,59,60,61,0,0,0,0,0,0,0,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,0,0,0,0,0,0,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,0,0,0,0,0};
/**
* Given a string, this function will encode it in 64b (with padding)
*/
std::string encodeBase64(const std::string& binaryText)
{
std::string encoded((binaryText.size()/3 + (binaryText.size()%3 > 0)) << 2, PADDING_CHAR);
const char* bytes = binaryText.data();
union
{
uint32_t temp = 0;
struct
{
uint32_t first : 6, second : 6, third : 6, fourth : 6;
} tempBytes;
};
std::string::iterator currEncoding = encoded.begin();
for(uint32_t i = 0, lim = binaryText.size() / 3; i < lim; ++i, bytes+=3)
{
temp = bytes[0] << 16 | bytes[1] << 8 | bytes[2];
(*currEncoding++) = ALPHABET[tempBytes.fourth];
(*currEncoding++) = ALPHABET[tempBytes.third];
(*currEncoding++) = ALPHABET[tempBytes.second];
(*currEncoding++) = ALPHABET[tempBytes.first];
}
switch(binaryText.size() % 3)
{
case 1:
temp = bytes[0] << 16;
(*currEncoding++) = ALPHABET[tempBytes.fourth];
(*currEncoding++) = ALPHABET[tempBytes.third];
break;
case 2:
temp = bytes[0] << 16 | bytes[1] << 8;
(*currEncoding++) = ALPHABET[tempBytes.fourth];
(*currEncoding++) = ALPHABET[tempBytes.third];
(*currEncoding++) = ALPHABET[tempBytes.second];
break;
}
return encoded;
}
/**
* Given a 64b padding-encoded string, this function will decode it.
*/
std::string decodeBase64(const std::string& base64Text)
{
if( base64Text.empty() )
return "";
assert((base64Text.size()&3) == 0 && "The base64 text to be decoded must have a length devisible by 4!");
uint32_t numPadding = (*std::prev(base64Text.end(),1) == PADDING_CHAR) + (*std::prev(base64Text.end(),2) == PADDING_CHAR);
std::string decoded((base64Text.size()*3>>2) - numPadding, '.');
union
{
uint32_t temp;
char tempBytes[4];
};
const uint8_t* bytes = reinterpret_cast<const uint8_t*>(base64Text.data());
std::string::iterator currDecoding = decoded.begin();
for(uint32_t i = 0, lim = (base64Text.size() >> 2) - (numPadding!=0); i < lim; ++i, bytes+=4)
{
temp = DECODED_ALPHBET[bytes[0]] << 18 | DECODED_ALPHBET[bytes[1]] << 12 | DECODED_ALPHBET[bytes[2]] << 6 | DECODED_ALPHBET[bytes[3]];
(*currDecoding++) = tempBytes[2];
(*currDecoding++) = tempBytes[1];
(*currDecoding++) = tempBytes[0];
}
switch (numPadding)
{
case 2:
temp = DECODED_ALPHBET[bytes[0]] << 18 | DECODED_ALPHBET[bytes[1]] << 12;
(*currDecoding++) = tempBytes[2];
break;
case 1:
temp = DECODED_ALPHBET[bytes[0]] << 18 | DECODED_ALPHBET[bytes[1]] << 12 | DECODED_ALPHBET[bytes[2]] << 6;
(*currDecoding++) = tempBytes[2];
(*currDecoding++) = tempBytes[1];
break;
}
return decoded;
}
【讨论】：