LINQ lamba expression to replace multiple characters in a string?

Question

Hi,

Is it possible to write a lambda expression that will iterate through array of objects and replace all occurences of 'X', 'Y', ' ' and 'Z' in one of the properties?

E.g.

return query.Select(x => { x.SomePropertyName= x.SomePropertyName.Trim().Replace(' ', "_"); return x; }).ToList();

For some reason, a query above doesn't replace a single character, when I need to replace multiple characters.

Thank you

Answer 1

If you want to use LINQ, you can use Select, ToArray and the String(char[]) constructor like this:

var result = query.ToList();
foreach (var x in result)
{
    x.SomeProperty =
        new string(x.SomeProperty
                    .Select(c => (c == 'X' || ... || c == ' ') ? '_' : c)
                    .ToArray());
}

Note that LINQ is not intended to be used to cause side-effects, but to create new enumerables from existing enumerables. So a foreach loop is better here.

---

But why not simply a chain of Replace calls?

var result = query.ToList();
foreach (var x in result)
{
    x.SomeProperty = x.SomeProperty
                      .Replace('X', '_')
                      .Replace('Y', '_')
                      .Replace('Z', '_')
                      .Replace(' ', '_');
}

---

Or are you trying to replace a single character with a sequence of characters? Then use the String.Replace overload that takes two strings:

var result = query.ToList();
foreach (var x in result)
{
    x.SomeProperty = x.SomeProperty.Replace(" ", "ABC");
}

Answer 2

When I want to replace one of a number of characters with one single other character, I often use a combination of string.Split and string.Join:

char[] unwanted = new[] {'X', 'Y', 'Z'};
query = query.Select(x =>
{
    x.SomePropertyName = string.Join("_", x.SomePropertyName.Split(unwanted));
    return x;
});

This will replace any occurrence of 'X', 'Y' or 'Z' with '_'. You can combine this with the looping construct of your choice.

As discussed in the comments, using Select does not really add any value in this case. A normal foreach loop would do the job, and would even produce more compact code:

char[] unwanted = new[] {'X', 'Y', 'Z'};
foreach(var x in query)
{
    x.SomePropertyName = string.Join("_", x.SomePropertyName.Split(unwanted));
};

Answer 3

If you're in the mood to use a regular expressions, you can do something like this:

query.Select(
    x => { 
        x.SomePropertyName = Regex.Replace(x.SomePropertyName, @"[XYZ\s]", "_"); 
        return x; })

Answer 4

loop through each character in the string and replace it if necessary.

 var xyz = new HashSet<char>(new[] { 'X', 'Y', 'Z' });
 char replacement = '_';
 var results = query
  .Select(x =>
     {
      var value = x.SomePropertyName
     .Trim()
     .Select(y => xyz.Contains(y) ? replacement : y)
     .ToArray();
      x.SomePropertyName = new string(value);
      return x;
     })
  .ToList();

Answer 5

Although you requested LINQ, this (ab)use of LINQ relies on side-effects and clutters the code. An alternative to use with an array of objects (which you stated is the case in your question) is to use the Array.ForEach method:

// using Fredrik Mörk's approach
Array.ForEach(query, x =>
    x.SomePropertyName = string.Join("_", x.SomePropertyName.Split(unwanted)));

// using Regex
Array.ForEach(query, x => x.SomePropertyName =
    Regex.Replace(x.SomePropertyName, "[XYZ]", "_", RegexOptions.IgnoreCase));

Of course use of regex requires some knowledge. Be mindful of using metacharacters that hold a special meaning in regex. You can sanitize your characters by using the Regex.Escape method.