This was a genius way to solve this problem. Kudos to you.

Same here. Nice to learn something new everyday:)

That's clever!

I don't know why i didn't think in replace method

Thank you for the tip!

Why I am getting this?:

"55">"100";
true

but it can be faster

It can't. It's a pure C solution, requiring the minimal possible operations to get the expected result, and is optimized to do the mapping after the filtering but before the expansion step. Your solution, on the other hand (idk whether you were referring to it as faster but still), is 10 times slower in the worst case (no spaces), and hundreds of times slower in the best case (only spaces).

It's hard to judge actual speed of python based on theoretical time complexity, it varies a lot depending on user implementation and the internal C implementation of that code. For example, a, b = min(list), max(list) is usually a bit faster than iterating once with a for loop. Your solution is about 4-5x slower, from my tests at least.