I cannot do the translation before converting to utf8 because the string can be in different character sets.

You can't work with strings if you don't know how they were encoded. Without that info, all you have is a bunch of meaningless bytes.

Once you get your encoding problem straightened out, you might be interested in Text::Unidecode. I'm not sure how it handles the euro symbol specifically, but it's great at ASCIIfying data.