Question

I am trying to remove all the vowels from a string except for the first and last character. I have tried with 2 expressions and using 2 ways but in vain. I have described them below. Does anybody has a regular expression for this?

e.g.

original string — source = apeaple

after regex — source_modified = apple (this is what is expected)

I tried the expression ([a-zA-Z])[aeiouAEIOU]([a-zA-Z]) but this expression is removing repeated character as well. So the following is happening when i apply the above expression

code used —

Regex reg = new Regex("([a-zA-Z])[aeiouAEIOU]([a-zA-Z])");
string source_modified = reg.Replace(source, "");

original string — source = apeaple

after code execution — source_modified = aple (repeating character removed)

code used — string source_modified = Regex.Replace(source, "([a-zA-Z])[aeiouAEIOU]([a-zA-Z])", "$1" + "$2");

original string — source = apeaple

after code execution — source_modified = apaple (just 1 vowel gets removed)

i also tried ([a-zA-Z])[aeiouAEIOU]*([a-zA-Z]) but this is removing just 1 vowel and not all. So the following is happening when i apply the above expression

code used —

Regex reg = new Regex("([a-zA-Z])[aeiouAEIOU]*([a-zA-Z])");
string source_modified = reg.Replace(source, "");

original string — source = apeaple

after code execution — source_modified = “” (all characters are removed)

code used — string source_modified = Regex.Replace(source, "([a-zA-Z])[aeiouAEIOU]*([a-zA-Z])", "$1" + "$2");

original string — source = apeaple

after code execution — source_modified = apeple

Answer 1

You need some lookaround like so

(?<!^)[aouieyAOUIEY](?!$)

C# supports it and it’s very powerful

string resultString = null;
try {
    resultString = Regex.Replace(subjectString, "(?<!^)[aeui](?!$)", "");
} catch (ArgumentException ex) {
    // Syntax error in the regular expression
}

Update 1

T.W.R.Cole informs me that there is a special rule in the English language (“this doesn’t work for words like “Anyanka” where an inner ‘y’ is used as a consonant”)

The following change should do this, using the technique of negative lookahead:

(?<!^)([aouie]|y(?![aouie]))(?!$)

This time enable the regex modifier that matches case insensitive, it makes the regex simpler than the original

if a y followed by another y still means that the y is a consonant (euh… is there such a word) and thus should not disappear than a y must be listed in the last character class as well :

(?<!^)([aouie]|y(?![aouiey]))(?!$)

I repeat that I used C# as my regex dialect which has good support for lookaround techniques.