C语言url decode/encode (Linux Ubuntu环境)

u_char LOW_BYTE_MASK = 15;
char* HEXS = (char*)"0123456789ABCDEF";
char* URL_ALLOW = (char*)"abcdefghijklmnopqrstuvwxyz/:#?&=0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ+-,._*~@!$^()[]{}';|";

#define BudaHex2Dec(c) (c<='9'? (c-'0'):(c<='Z'? (c-55):(c-87)))
#define BudaPreWrite(max_len, dec, fail) if((max_len-=dec)<0) goto fail;
#define BudaPreWrite1(max_len, fail) if((--max_len)<0) goto fail;

int url_encode(u_char *s, u_char *d, int max_len)
{
  u_char c, c_high, c_low, *ps = s, *pd=d; 
  while(c=*(ps++))
  {
    if(strchr(URL_ALLOW, c)) { BudaPreWrite1(max_len, fail); *(pd++) = c; }
    else { BudaPreWrite(max_len, 3, fail); *(pd++) = '%'; *(pd++) = HEXS[c >> 4]; *(pd++) = HEXS[c & BYTE_LOW4_MASK]; }
  }
  *pd=0;

  succeed: return 0;
  fail: log("url_encode failed"); return -1;
}

int url_decode(u_char *s, u_char *d, int max_len)
{
  u_char c, *ps=s, *pd=d, c_high, c_low;
  while(c=*(ps++))
  {
    BudaPreWrite1(max_len, fail); 
    if (c != '%') *(pd++) = c;
    else
    {
      c = *(ps++); c_high=BudaHex2Dec(c); c = *(ps++); c_low=BudaHex2Dec(c); 
      c = *(pd++) = ((c_high<<4) | c_low); //printf("decoded ");
    }
  }
  *pd=0;

  succeed: return 0;
  fail: log("url_decode failed"); return -1;
}

参考资料

RFC 3986 - Uniform Resource Identifier (URI): Generic Syntax

gemini