String.Replace doesn't replace all matches

Question

Why does line2 replaces only alternating half of occurrences?

    Dim line1 As String = "AAA|BBB|CCC|CCC|CCC|CCC|EEE|FFF"
    Dim line2 As String = "AAA|BBB|CCC|CCC|CCC|CCC|EEE|FFF"
    Dim line3 As String = "AAA|BBB|CCC|CCC|CCC|CCC|EEE|FFF"

    line1 = line1.Replace("CCC", "")
    line2 = line2.Replace("|CCC|", "||")
    line3 = line3.Replace("CCC|", "|")

Result:

line1 = "AAA|BBB|||||EEE|FFF" -- OK, but fails when element is "..|ZZZCCCZZZ|.."
line2 = "AAA|BBB||CCC||CCC|EEE|FFF" -- Not OK
line3 = "AAA|BBB|||||EEE|FFF" -- OK, but fails similar to Line1 edge-case for "..|ZZZCCC|.."

I have tried using RegEx, but get similar results.

Is there really no better way than this, below?

Do While line1.Contains("|CCC|")
    line1 = line1.Replace("|CCC|", "||")
Loop

Answer 1

Once it finds the first token, it starts looking for the next one after that token. So it finds |CCC|, replaces it, then continues on and the first thing it sees is CCC| which doesn't match. It doesn't pre-scan the string looking for tokens to replace.

Consider it like this:

Given AAA|BBB|CCC|CCC|CCC|CCC|EEE|FFF

It runs to AAA|BBB|CCC| HOLD IT |CCC| was found, let's start building our string:

AAA|BBB + || (our replacement)

Now let's move on, we now have CCC|CCC|CCC|EEE|FFF left to work with.

It runs to CCC|CCC| HOLD IT |CCC| was found, let's continue adding to our string:

AAA|BBB||CCC + || (our replacement)

Now let's move on, we now have CCC|CCC|EEE|FFF and so on and so on.

EDIT: Considering the entry on MSDN describing the return value:

A string that is equivalent to the current string except that all instances of oldValue are replaced with newValue.

One could read that as what you expect that it pre-scans the string and finds all matches. I don't see anything in the MSDN doc that describes this corner case. Perhaps this is something that should be added to the MSDN doc.

Answer 2

Instead of using regular expressions or string.Replace you could parse the values, filter the ones you don't want and join them back together.

line1 = string.Join("|", line1.Split("|").Select(s => s == "CCC" ? "" : s).ToArray());

Sorry I don't know the VB equivalent.

Answer 3

For anyone in the future, I've added an extension method to overcome this limitation in the framework:

<System.Runtime.CompilerServices.Extension()>
Public Function ReplaceAll(ByVal original As String, ByVal oldValue As String, ByVal newValue As String) As String

    If newValue.Contains(oldValue) Then
        Throw New ArgumentException("New value can't be a subset of OldValue as infinite replacements can occur.", newValue)
    End If

    Dim maxIterations As Integer = original.Length \ oldValue.Length

    While maxIterations > 0 AndAlso original.Contains(oldValue)
        original = original.Replace(oldValue, newValue)
        maxIterations -= 1
    End While

    Return original

End Function