Find files using wild card in C#

Question

I am trying to find files from a directory:

String[] search1 = Directory.GetFiles(voiceSource, "85267-*.wav")
                                 .Select(path => Path.GetFileName(path))
                                 .ToArray();

String[] search2 = Directory.GetFiles(voiceSource, "85267 *.wav")
                                 .Select(path => Path.GetFileName(path))
                                 .ToArray();

But in search1, it selects both 85267-s.wav and 85267 -s.wav. But I want only 85267-s.wav to be selected.

search2 is doing well.

How can I do that?

Answer 1

The behaviour you are experiencing is because of short file name. Since you will get 85267-~1.WAV for 85267 -s.wav and since that matches your wild card "85267-*.wav" you get both files back.

The is explained in Directory.GetFiles Method (String, String)

Because this method checks against file names with both the 8.3 file name format and the long file name format, a search pattern similar to "1.txt" may return unexpected file names. For example, using a search pattern of "1.txt" will return "longfilename.txt" because the equivalent 8.3 file name format would be "longf~1.txt".

For workaround you can use Directory.EnumerateFiles to first select both files matching your criteria and then compare the actual(long) file name part using StartsWith. Remember EnumerateFiles does lazy evaluation.

String[] search1 = Directory.EnumerateFiles(@"C:\test", "85267-*.wav")
                         .Where(file => Path.GetFileName(file).StartsWith("85267-"))
                         .Select(path => Path.GetFileName(path))
                         .ToArray();

Answer 2

Yes, this is a side-effect of the MS-Dos 8.3 short name support that's still turned on today on most file systems. Something you can see with the DIR /X command, it displays those short names. On my machine:

C:\temp>dir /x *.wav

01/21/2015  09:11 AM                 6 85267-~1.WAV 85267 -s.wav
01/21/2015  09:11 AM                 6              85267-s.wav
               2 File(s)             12 bytes
               0 Dir(s)  235,121,160,192 bytes free

Note how the short name for "85267 -s" is missing the space. It is not a valid character in a short name. What's left over now also matches your wildcard.

That's not where the trouble ends with those short names, A wildcard like *.wav will also match a file like foobar.wavx, a completely different file type.

Short-name generation is, frankly, a relic from the previous century that ought to be turned off today. But that is not typically anything you can control yourself. You have to deal with these accidental matches and double-check what you get back. With a Regex for example.