AVR Freaks

Subtract 1 from each digit in the number sequence.

Sketchy:

char buffer[10];
uint8_t output = 0;

sprintf(buffer, "%08d", invalue);

for (int n = 0; n < strlen(buffer)) {
    if (buffer[n] == '0') {
        buffer[n] = '1';
    }
    output <<= 1;
    output |= (buffer[n] - '1');
}

say the invalue was 1122211 then after the sprintf() buffer should hold "01122211". The for() loop will iterate 8 times (it always will because of the %08 padding). If there are some leading 0's the first if() converts '0's to '1's. Each time round the loop the output is shifted one place left. As it starts with 0 the first iteration this does nothing. The first |= with n=0 will encounter the leading '1' that started as 0. it subtracts '1' from this to leave binary 0 and this is OR'd. Nothing changes, output is still 0. Next time round (and from now on) the if() does no character replacements as it's all 1's and 2's from now on. For the next two 1's the - '1' turns them into 0's so nothing yet is added to output. On the next time around it hits the first '2'. '2' - '1' is 1 so this time a 1 is |'d into the output. As always the result is <<= 1 each time round so this should now OR 11100 into the result. So you end with 11100 which is the right value for the 1122211 input.

This encoding (using 1 or 2 out of 0..9 in decimal positions) seems hugely inefficient! Wonder why the original encoding did not just put 0b0011100 into a single byte in the first place ??

that's still not clear whether it's a string of ASCII coded characters: - a '0', followed by a 'd', followed by a '1', followed by a '1', followed by a '2', etc  

or an actual numeric value.

"0d" isn't a standard notation for anything.

Likewise!

really?

and why did it need to be decimal?