3 回答
至少,一个简单的反向扫描确实为编码部分提供了一个(看似)好的解决方案。我正在从右到左进行一次扫描,并用出现计数覆盖字符串的部分。
char * enc(char * ip)
{
int r,op;
int l=strlen(ip);
r=l-1;
char curr;
op=r;
int curr_count=1,mod_curr_count;
while(r>=0)
{
curr=ip[r];
while(ip[--r]==curr)
{
curr_count++;
}
if(curr_count!=1)
{
while(curr_count)
{
mod_curr_count=curr_count%10;
ip[op--]=(char)(mod_curr_count+48);
curr_count/=10;
}
ip[op--]=curr;
curr_count=1;
}
else
{
ip[op--]=curr;
}
}
ip=ip+op+1;
return ip;
}
输入: aaaaaaaaaaabbbfffffffffffffffqqqqqqqqqqqqqqqqqccccpoii
输出: a12b3f15q18c4poi2
To do an in-place encoding, the encoded string must never be longer than the original string. Suppose we assume the following encoding rules:
- No numeric digits in the original string (so no count delimiter characters are needed)
- A run length of 1 is never explicitly coded (so
abc
remainsabc
)
I believe that with these assumptions, a run-length encoding is not ambiguous and will never be longer than the string itself. Then the following algorithm (pseudocode) should do the job of encoding in place:
currentChar ← string[0]
nextOutputPos ← 1
nextReadPos ← 1
count ← 1
while (nextReadPos < length of string) {
nextChar ← string[nextReadPos++];
if (nextChar == currentChar) {
count++;
} else {
if (count > 1) {
write (count as a string) to string at position nextOutputPos
nextOutputPos ← nextOutputPos + (length of count as a string)
}
string[nextOutputPos++] ← currentChar ← nextChar;
}
}
At the end, the encoded string is contained in the half-open range [0, nextOutputPos
) of string
.
这是 Java 中使用正则表达式的一种可能性:
String str = "aaabbbc"; // string to be encoded
StringBuilder sb = new StringBuilder(); // to hold encoded string
for (String s : str.split("(?<=(.))(?!\\1)")) {
sb.append(s.charAt(0));
if (s.length() > 1) // only append length if it's > 1
sb.append(s.length());
}
System.out.println(sb.toString());
a3b3c