Directory.GetFiles of certain extension

Question

Is there a way to simplify this linq expression, or is there a better way of doing this?

Directory.GetFiles(dir, "*.*", SearchOption.AllDirectories)
         .Where(s => s.EndsWith(".jpg", StringComparison.OrdinalIgnoreCase) ||
                     s.EndsWith(".gif", StringComparison.OrdinalIgnoreCase) ||
                     s.EndsWith(".png", StringComparison.OrdinalIgnoreCase) ||
                     ...);

Basically I want to return all files of a certain extension. Unfortunately, this method isn't very flexible. I'd rather be able to add extensions to a list and have Directory.GetFiles return those extensions. Is that possible?

Communard

Duplicate: #163662 – Powel

Possible duplicate of Can you call Directory.GetFiles() with multiple filters? – Theran

Answer 1

If you would like to do your filtering in LINQ, you can do it like this:

var ext = new List<string> { "jpg", "gif", "png" };
var myFiles = Directory
    .EnumerateFiles(dir, "*.*", SearchOption.AllDirectories)
    .Where(s => ext.Contains(Path.GetExtension(s).TrimStart(".").ToLowerInvariant()));

Now ext contains a list of allowed extensions; you can add or remove items from it as necessary for flexible filtering.

Edmondson

I can perform tests, but do you know which one would be more efficient? – Communard

Interesting that it works... From the documentation on GetFiles, it doesn't seem like it would accept the | like you would expect from a regex. The docs say it only recognizes * and ?, but hey, if it works, it works =) – Amaral

Note: This doesn't seem to work on Win7 x64 running .NET 4.5. Throws Illegal characters in path due to the pipe. – Erb

@SimonWhitehead Never mind, I misinterpreted a piece of code in a post on Microsoft's web site, where a similarly-named method has been interpreting the pipes internally. I removed the non-working part of the solution (and SSL checked only the first one). – Edmondson

Note for the passerby: you might get a performance boost out of the newer .NET 4.0 Directory.EnumerateFiles as mentioned in answer by @nekizalb – Segmental

Answer 2

Doesn't the Directory.GetFiles(String, String) overload already do that? You would just do Directory.GetFiles(dir, "*.jpg", SearchOption.AllDirectories)

If you want to put them in a list, then just replace the "*.jpg" with a variable that iterates over a list and aggregate the results into an overall result set. Much clearer than individually specifying them. =)

Something like...

foreach(String fileExtension in extensionList){
    foreach(String file in Directory.GetFiles(dir, fileExtension, SearchOption.AllDirectories)){
        allFiles.Add(file);
    }
}

For .NET 4+: if your directories are large, using EnumerateFiles instead of GetFiles can potentially be more efficient

Amaral

Please be aware of the fact that while using wildcard * there is always chance that you can mess up the intended logic . For instance you have two files Test1.xls and Test2.xlsx and if you use filter *.xls to filter out xls file , Directory.GetFiles ** returns both xls as well as xlsx files **. [Read MSDN for more info]: msdn.microsoft.com/en-us/library/ms143316%28v=vs.110%29.aspx – Gaidano

effectively, xls and xlsx are the same file that modern day Excel can read? – Ford

@Ford It's not an issue for Excel, it's an issue for other systems that try to read Excel data. But yes it's an exceedingly rare edge case. – Cohen

Answer 3

I would have done using just a single line like:

List<string> imageFiles = Directory.GetFiles(dir, "*.*", SearchOption.AllDirectories).Where(file => new string[] { ".jpg", ".gif", ".png" }.Contains(Path.GetExtension(file))).ToList();

Morphine

This works, of course, but the lambda expression you pass to .Where() will create one string[] every time it's invoked; that is, for every file under dir. If dir is, say, @"C:\" then this could potentially create tens or hundreds of thousands of short-lived "garbage" arrays. I definitely think it'd be worth it to define your array outside of .Where(), even if it means having to write that dreaded second line of code. – Annabellannabella