PHP 的 base_convert 函数能在任意进制之间转换数字，这是常识。那么请你不要实际运行，用常识判断一下，这句代码运行的结果：

echo base_convert('http://demon.tw', 16, 10);

如果你的答案是 222，那么恭喜你答对了，其实上面那句代码跟这句是一样的：

echo base_convert('de', 16, 10);

也就是说，base_convert 函数会忽略掉该进制以外的其他字符。下面通过 base_convert 函数的 C 源码来分析原因，base_convert 函数定义在 PHP 源码的 ext/standard/math.c 中：

/* {
   {
   { proto string base_convert(string number, int frombase, int tobase)   Converts a number in a string from any base <= 36 to any base <= 36 */PHP_FUNCTION(base_convert){    zval **number, **frombase, **tobase, temp;    char *result;    if (ZEND_NUM_ARGS() != 3 || zend_get_parameters_ex(3, &number, &frombase, &tobase) == FAILURE) {        WRONG_PARAM_COUNT;    }    convert_to_string_ex(number);    convert_to_long_ex(frombase);    convert_to_long_ex(tobase);    if (Z_LVAL_PP(frombase) < 2 || Z_LVAL_PP(frombase) > 36) {        php_error_docref(NULL TSRMLS_CC, E_WARNING, "Invalid `from base' (%ld)", Z_LVAL_PP(frombase));        RETURN_FALSE;    }    if (Z_LVAL_PP(tobase) < 2 || Z_LVAL_PP(tobase) > 36) {        php_error_docref(NULL TSRMLS_CC, E_WARNING, "Invalid `to base' (%ld)", Z_LVAL_PP(tobase));        RETURN_FALSE;    }    if(_php_math_basetozval(*number, Z_LVAL_PP(frombase), &temp) != SUCCESS) {        RETURN_FALSE;    }    result = _php_math_zvaltobase(&temp, Z_LVAL_PP(tobase) TSRMLS_CC);    RETVAL_STRING(result, 0);}

前面几行都是解析和校验参数是否正确，关键代码是 _php_math_basetozval 和 _php_math_zvaltobase 函数，_php_math_basetozval 定义如下：

/* {
   {
   { _php_math_basetozval *//* * Convert a string representation of a base(2-36) number to a zval. */PHPAPI int _php_math_basetozval(zval *arg, int base, zval *ret){    long num = 0;    double fnum = 0;    int i;    int mode = 0;    char c, *s;    long cutoff;    int cutlim;    if (Z_TYPE_P(arg) != IS_STRING || base < 2 || base > 36) {        return FAILURE;    }    s = Z_STRVAL_P(arg);    cutoff = LONG_MAX / base;    cutlim = LONG_MAX % base;        for (i = Z_STRLEN_P(arg); i > 0; i--) {        c = *s++;        /* might not work for EBCDIC */        if (c >= '0' && c <= '9')             c -= '0';        else if (c >= 'A' && c <= 'Z')             c -= 'A' - 10;        else if (c >= 'a' && c <= 'z')             c -= 'a' - 10;        else            continue;        if (c >= base)            continue;                switch (mode) {        case 0: /* Integer */            if (num < cutoff || (num == cutoff && c <= cutlim)) {                num = num * base + c;                break;            } else {                fnum = num;                mode = 1;            }            /* fall-through */        case 1: /* Float */            fnum = fnum * base + c;        }       }    if (mode == 1) {        ZVAL_DOUBLE(ret, fnum);    } else {        ZVAL_LONG(ret, num);    }    return SUCCESS;}/* }}} */

代码太长看起来很烦，关键是这一段：

for (i = Z_STRLEN_P(arg); i > 0; i--) {        c = *s++;        /* might not work for EBCDIC */        if (c >= '0' && c <= '9')             c -= '0';        else if (c >= 'A' && c <= 'Z')             c -= 'A' - 10;        else if (c >= 'a' && c <= 'z')             c -= 'a' - 10;        else            continue;        if (c >= base)            continue;

遍历字符串，碰到除了 [0-9a-zA-Z] 以外的字符只是用 continue 直接跳到下一次循环，所以其他字符并不影响进制的转换。而且当 c 大于 base 时也是直接跳到下一次循环，所以该进制以外的其他字母亦不会影响进制的转换。这是 base_convert 函数的一个 BUG 呢，还是设计者有意为之？