string_equivilence.py

BREAK = "|"
break_charactors = ["_", "-", " "]

def eq(string1, string2):
    # I think we can look through the different cases that we have here
    # So cleary string1===string2 => True
    if string1==string2:
        return True
    # But then we want to use more inteligence to work out what we are dealing with
    # I think the next thing we should do is to beak the text appart into numeric, symbolic and alphabetical, keeping the order

    if break_down(string1) == break_down(string2):
        return True
    if break_down(string1).replace(BREAK, "") == break_down(string2).replace(BREAK, ""):
        return True
    print(break_down(string1))
    print(break_down(string2))
    return False

def xor(b1, b2):
    return bool(b1) != bool(b2)

def issymbol(char):
    return not (char.isalnum() or char in break_charactors)

def break_down(string):
    # I would say we have a few classes here
    #divider type \s _ \l\u  a break from alphabetical
    #alphabetical
    #symbolic
    #numeric
    acc = ""
    for last_char, this_char in zip(BREAK + string[:-1],string):
        # what we want here is if the previous char was upper and this is lower to create a break
        if last_char.islower() and this_char.isupper():
            #insert a break before this char
            acc += BREAK + this_char.lower()
            continue
        if last_char in break_charactors:
            acc += BREAK + this_char
            continue
        # might want to condition this on is alphanumeric
        if last_char.isalnum() and this_char.isalnum() and xor(last_char.isnumeric(), this_char.isnumeric()):
            acc += BREAK + this_char
            continue
        if this_char.isalnum():
            acc += this_char
    return acc.lower()