How can i delete duplicate letter, but adding some exception letter to it

You can first capture a and g and replace with 2 times group 1.

([ag])\1+

The pattern matches:

• ([ag]) Capture group 1, match either a or g
• \1+ Repeat 1+ times the same char matched in group 1

Then replace all chars other than a g or a whitespace char, and replace with a single group 1 to remove the duplicates.

([^\sag])\1+

The pattern matches:

• ( Capture group 1
  • [^\sag] Match a non whitespace char except for a or g
• ) Close group 1
• \1+ Repeat 1+ times the same char matched in group 1

For example

import re

s = "mengganti, maaf, ppuuutttiiiihhh, mmmmeeeeerrrraaaah, maaagggz"

print(re.sub(
        r"([^\sag])\1+",
        r"\1",
        re.sub(r"([ag])\1+", r"\1\1", s))
)

Output

mengganti, maaf, putih, meraah, maaggz

See a Python demo.

Or using a single pattern with an alternation | combining the 2 patterns having 2 capture groups, and using re.sub with a lambda:

import re

pattern = r"([ag])\1+|([^\sag])\2+"
s = "mengganti, maaf, ppuuutttiiiihhh, mmmmeeeeerrrraaaah, maaagggz"
result = re.sub(
        pattern,
        lambda x: x.group(1) * 2 if x.group(1) else x.group(2),
        s
)

if result:
        print(result)

Output

mengganti, maaf, putih, meraah, maaggz

See another Python demo or a regex demo

Alternative to regex is itertools.groupby:

from itertools import groupby

def remove_dups(seq, exclude):
    # gather the intermediate results here
    result = []
    # for each letter and consecutive group it leads...
    for letter, group in groupby(seq):
        # get the length of the group
        group_len = len(list(group))
        if letter not in exclude or group_len < 2:
            # e.g., either not "a" or "g", or appears once in a row
            result.append(letter)
        else:
            # "a" or "g"; repeat by 2
            result.append(letter * 2)

    # new string with join
    return "".join(result)

or equivalently in 1-line:

def remove_dups(seq, exclude):
    return "".join(letter
                   if letter not in exclude or len(list(gr)) < 2
                   else 2 * letter
                   for letter, gr in groupby(seq))

groupby gives back consecutive groups with the key being the letter. If the letter is not to be excluded, it is put there as is; if number of consecutive letters is less than 2, it is still put there as is. If we need to exclude it, we put it with the repetition amount 2.

Sample run:

>>> remove_dups("mengganti", exclude={"a", "g"})
"mengganti"

>>> remove_dups("maaf", exclude={"a", "g"})
"maaf"

>>> remove_dups("ppuuutttiiiihhh", exclude={"a", "g"})
"putih"

>>> remove_dups("mmmmeeeeerrrraaaah", exclude={"a", "g"})
"meraah"