I just wondered what's the easiest way to replace a string characters that must be replaced subsequently.
For example:
var str = "[Hello World]";
//enclose all occurences of [ and ] with brackets[]
str = str.Replace("[","[[]").Replace("]","[]]");
- The desired result:
[[]Hello World[]]
- The actual result:
[[[]]Hello World[]]
The reason is obviously the second replace on the already modified string.
So how to replace all occurences of "bad" characters with characters that contain "bad" characters?
A quick measurement of all approaches revealed that the StringBuilder is the most efficient way.
190kb file (all in milliseconds)
regexTime 40.5065 replaceTime 20.8891 stringBuilderTime 6.9776
7MB file
regexTime 1209.3529 replaceTime 403.3985 stringBuilderTime 175.2583
By the way, the direct StringBuilder
approach from John was twice as fast as the Aggregate approach from Sehe.
I've made an extension out of it:
public static String EncloseChars(this string input, char[] charsToEnclose, String leftSide, String rightSide) {
if (charsToEnclose == null || leftSide == null || rightSide == null)
throw new ArgumentException("Invalid arguments for EncloseChars", charsToEnclose == null ? "charsToEnclose" : leftSide == null ? "leftSide" : "rightSide");
Array.Sort(charsToEnclose);
StringBuilder sb = new StringBuilder();
foreach (char c in input) {
if (Array.BinarySearch(charsToEnclose, c) > -1)
sb.Append(leftSide).Append(c).Append(rightSide);
else
sb.Append(c);
}
return sb.ToString();
}
"[Hello World]".EncloseChars(new char[]{'[', ']'},"[","]");
How about:
Note that this avoids the multiple loops issue but creates at least as many arrays as there are characters in the input string so it might not be optimal in terms of performance.
I had the exact same problem, so I made a helper function to do just that
What about this elegant regular expression approach:
Unit test it?
Test passed
Edit @JohnMcGrant
Here is a slightly less inefficient version of your code, which has, by the way, exactly the same behaviour as the above regex:
Here's a very uncool way to do it. But it has the advantage of being pretty close to foolproof, I think, and not using regex (in case you'd rather not use regex).
What about: