literal.php

<?php

    $param1 = "哈哈1\t𤭢2\\3\"";
    $param2 = "10";
    $param11 = "哈\xd5\"哈1\t𤭢2\\3\"";
    $param22 = "10 union 1,2";
 
    $sql1 = "select * from user where username=" . literal($param1) . " and age=" . literal(intval($param2));
    $sql2 = "select * from user where username=" . literal($param1) . " and age=" . literal($param2);
    $sql3 = "select * from user where username=" . literal($param11) . " and age=" . literal(intval($param2));
    $sql4 = "select * from user where username=" . literal($param11) . " and age=" . literal($param22);
    $sql5 = "select * from user where username=" . literal($param11) . " and age=" . literal(intval($param22));

    echo $sql1, "\n";
    echo $sql2, "\n";
    echo $sql3, "\n";
    echo $sql4, "\n";
    echo $sql5, "\n";
   
    /**
     * 将php的字符串转换成UTF8字符数组
     */
    function stringToUTF8Array($str) {
        // 这里假设$str是utf8编码的，如果是gbk，需要用iconv("gbk", "utf8//IGNORE", $str)转换
        $i = 0;
        $j = 0;
        $len = strlen($str);
        $array = array();
        while($i < $len) {
            $byte = ord($str[$i]);
            // start with 0xxx xxxx, utf8 char has 1 byte
            if (($byte & 0x80) == 0x00) {
                $array[$j++] = $str[$i];
                $i = $i + 1;
                continue;
            }
            // start with 110x xxxx, utf8 char has 2 bytes
            if (($byte & 0xE0) == 0xC0) {
                if ($i + 1 >= $len) break;
                $array[$j++] = $str[$i] . $str[$i + 1];
                $i = $i + 2;
                continue;
            }
            // start with 1110 xxxx, utf8 char has 3 bytes
            if (($byte & 0xF0) == 0xE0) {
                if ($i + 2 >= $len) break;
                $array[$j++] = $str[$i] . $str[$i + 1] . $str[$i + 2];
                $i = $i + 3;
                continue;
            }
            // start with 1111 0xxx, utf8 char has 4 bytes
            if (($byte & 0xF8) == 0xF0) {
                if ($i + 3 >= $len) break;
                $array[$j++] = $str[$i] . $str[$i + 1] . $str[$i + 2] . $str[$i + 3];
                $i = $i + 4;
                continue;
            }
            // start with 1111 10xx, utf8 char has 5 bytes
            if (($byte & 0xFC) == 0xF8) {
                if ($i + 4 >= $len) break;
                $array[$j++] = $str[$i] . $str[$i + 1] . $str[$i + 2] . $str[$i + 3] . $str[$i + 4];
                $i = $i + 5;
                continue;
            }
            // start with 1111 110x, utf8 char has 6 bytes
            if (($byte & 0xFE) == 0xFC) {
                if ($i + 5 >= $len) break;
                $array[$j++] = $str[$i] . $str[$i+1] . $str[$i+2] . $str[$i+3] . $str[$i+4] . $str[$i+5];
                $i = $i + 6;
                continue;
            }

            // utf8 string should never reach here
            $i ++;
        }
        return $array;
    }

    /**
     * 返回$var值在代码中的表现形式
     * 比如对于字符串"hello"，返回"\"hello\""
     *     对于整型值10，返回"10"
     */
    function literal($var) {
        if (!is_scalar($var)) {
            print_r($var);
            throw new Exception("literal function expects a scalar");
        }
        if (is_string($var)) {
            return literalUTF8String($var);
        }
        if (is_bool($var)) {
            return ($var) ? "true" : "false";
        }

        return strval($var);
    }
    
    function literalUTF8String($str) {
        $array = stringToUTF8Array($str);
        $len = count($array);
        $res = "\"";
        for($i = 0; $i < $len; $i++) {
            $res = $res . literalUTF8Char($array[$i]);
        }
        $res = $res . "\"";
        return $res;
    }

    function literalUTF8Char($str) {
        if (strlen($str) == 1) {
            if ($str == "\\") {
                return "\\\\";
            }
            else if ($str == "\"") {
                return "\\\"";
            }
            else {
                return $str;
            }
        }
        else {
            return $str;
        }
    }
?>