Run length encoding using O(1) space - algorithm

Question

Can we do the run-length encoding in place(assuming the input array is very large)
We can do for the cases such as AAAABBBBCCCCDDDD
A4B4C4D4
But how to do it for the case such as ABCDEFG?
where the output would be A1B1C1D1E1F1G1

Answer 1

My first thought was to start encoding from the end, so we will use the free space (if any), after that we can shift the encoded array to the start. A problem with this approach is that it will not work for AAAAB, because there is no free space (it's not needed for A4B1) and we will try to write AAAAB1 on the first iteration.
Below is corrected solution:
(let's assume the sequence is AAABBC)
encode all groups with two or more elements and leave the rest unchanged (this will not increase length of the array) -> A3_B2C
shift everything right eliminating empty spaces after first step -> _A3B2C
encode the array from the start (reusing the already encoded groups of course) -> A3B2C1
Every step is O(n) and as far as I can see only constant additional memory is needed.
Limitations:
Digits are not supported, but that anyway would create problems with decoding as Petar Petrov mentioned.
We need some kind of "empty" character, but this can be worked around by adding zeros: A03 instead of A3_

Answer 2

C++ solution O(n) time O(1) space
string runLengthEncode(string str)
{
int len = str.length();
int j=0,k=0,cnt=0;
for(int i=0;i<len;i++)
{
j=i;
cnt=1;
while(i<len-1 && str[i]==str[i+1])
{
i++;
cnt++;
}
str[k++]=str[j];
string temp =to_string(cnt);
for(auto m:temp)
str[k++] = m;
}
str.resize(k);
return str;
}

Answer 3

null is used to indicate which items are empty and will be ignored for encoding. Also you can't encode digits (AAA2222 => A324 => 324 times 'A', but it's A3;24). Your question opens more questions.
Here's a "solution" in C#
public static void Encode(string[] input)
{
var writeIndex = 0;
var i = 0;
while (i < input.Length)
{
var symbol = input[i];
if (symbol == null)
{
break;
}
var nextIndex = i + 1;
var offset = 0;
var count = CountSymbol(input, symbol, nextIndex) + 1;
if (count == 1)
{
ShiftRight(input, nextIndex);
offset++;
}
input[writeIndex++] = symbol;
input[writeIndex++] = count.ToString();
i += count + offset;
}
Array.Clear(input, writeIndex, input.Length - writeIndex);
}
private static void ShiftRight(string[] input, int nextIndex)
{
var count = CountSymbol(input, null, nextIndex, (a, b) => a != b);
Array.Copy(input, nextIndex, input, nextIndex + 1, count);
}
private static int CountSymbol(string[] input, string symbol, int nextIndex)
{
return CountSymbol(input, symbol, nextIndex, (a, b) => a == b);
}
private static int CountSymbol(string[] input, string symbol, int nextIndex, Func<string, string, bool> cmp)
{
var count = 0;
var i = nextIndex;
while (i < input.Length && cmp(input[i], symbol))
{
count++;
i++;
}
return count;
}

Answer 4

The 1st solution does not take care of single characters. For example - 'Hi!' will not work. I've used totally different approach, used 'insert()' functions to add inplace. This take care of everything, whether the total 'same' character is > 10 or >100 or = 1.
#include<iostream>
#include<algorithm>
using namespace std;
int main(){
string name = "Hello Buddy!!";
int start = 0;
char distinct = name[0];
for(int i=1;i<name.length()+1;){
if(distinct!=name[i]){
string s = to_string(i-start);
name.insert(start+1,s);
name.erase(name.begin() + start + 1 + s.length(),name.begin() + s.length() + i);
i=start+s.length()+1;
start=i;
distinct=name[start];
continue;
}
i++;
}
cout<<name;
}
Suggest me if you find anything incorrect.

Answer 5

O(n), in-place RLE, I couldn't think better than this. It will not place a number, if chars occurence is just 1. Will also place a9a2, if the character comes 11 times.
void RLE(char *str) {
int len = strlen(str);
int count = 1, j = 0;
for (int i = 0; i < len; i++){
if (str[i] == str[i + 1])
count++;
else {
int times = count / 9;
int rem = count % 9;
for (int k = 0; k < times; k++) {
str[j++] = str[i];
_itoa(9, &str[j++], 10);
count = count - 9;
}
if (count > 1) {
str[j++] = str[i];
_itoa(rem, &str[j++], 10);
count = 1;
}
else
str[j++] = str[i];
}
}
cout << str;
}
I/P => aaabcdeeeefghijklaaaaa
O/P => a3bcde4fghijkla5

Answer 6

Inplace solution using c++ ( assumes length of encoding string is not more than actual string length):
#include <bits/stdc++.h>
#include<stdlib.h>
using namespace std;
void replacePattern(char *str)
{
int len = strlen(str);
if (len == 0)
return;
int i = 1, j = 1;
int count;
// for each character
while (str[j])
{
count = 1;
while (str[j] == str[j-1])
{
j = j + 1;
count++;
}
while(count > 0) {
int rem = count%10;
str[i++] = to_string(rem)[0];
count = count/10;
}
// copy character at current position j
// to position i and increment i and j
if (str[j])
str[i++] = str[j++];
}
// add a null character to terminate string
if(str[len-1] != str[len-2]) {
str[i] = '1';
i++;
}
str[i] = '\0';
}
// Driver code
int main()
{
char str[] = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaabccccc";
replacePattern(str);
cout << str;
return 0;
}