在Ruby的json中压缩单个字段

Base-64 encode gzip输出，以使其可读字符，从而使其有效的UTF-8.这将使数据扩展约三分之一，抵消了您的一些压缩.您还可以使用更有效的字符进行编码，例如Base-85，以减少影响，在这种情况下扩展约四分之一.通过一些工作，你应该能够把它降低到接近七分之一的涨幅.

下面是用C语言编写的示例代码，它将字节编码为1..127中的符号，这些符号都是有效的UTF-8.(JSON不允许字符串中有空字节.)由此产生的扩展约为1.145倍.

#include <stddef.h>

// Encode, reading six or seven bits from bin[0..len-1] to encode each symbol
// to *enc, where a symbol is one byte in the range 1..127. enc must have room
// for at least ceil((len * 4) / 3) symbols. The average number of encoded
// symbols for random input is 1.1454 * len. The number of encoded symbols is
// returned.
size_t enc127(char *enc, unsigned char const *bin, size_t len) {
    unsigned buf = 0;
    int bits = 0;
    size_t i = 0, k = 0;
    for (;;) {
        if (bits < 7) {
            if (i == len)
                break;
            buf = (buf << 8) | bin[i++];
            bits += 8;
        }
        unsigned sym = ((buf >> (bits - 7)) & 0x7f) + 1;
        if (sym > 0x7e) {
            enc[k++] = 0x7f;
            bits -= 6;
        }
        else {
            enc[k++] = sym;
            bits -= 7;
        }
    }
    if (bits)
        enc[k++] = ((buf << (7 - bits)) & 0x7f) + 1;
    return k;
}

// Decode, converting each symbol from enc, which must be in the range 1..127,
// into 6 or 7 bits in the output, from which 8 bits at a time is written to
// bin. bin must have room for at least floor((len * 7) / 8) bytes. The number
// of decoded bytes is returned.
size_t dec127(unsigned char *bin, char const *enc, size_t len) {
    unsigned buf = 0;
    int bits = 0;
    size_t k = 0;
    for (size_t i = 0; i < len; i++) {
        unsigned sym = enc[i];
        if (sym == 0x7f) {
            buf = (buf << 6) | 0x3f;
            bits += 6;
        }
        else {
            buf = (buf << 7) | (sym - 1);
            bits += 7;
        }
        if (bits >= 8) {
            bin[k++] = buf >> (bits - 8);
            bits -= 8;
        }
    }
    return k;
}