Is creating a Guid out of an MD5 hash instead of String valid?

Question

I am trying to implement a method for detecting duplicate files. I have an MD5 hashing method (let's ignore the fact that MD5 is broken) as below:

using(MD5 hasher = MD5.Create())
using(FileStream fs = File.OpenRead("SomeFile"))
{
    byte[] hashBytes = hasher.ComputeHash(fs);
    string hashString = string.Join(string.Empty, hashBytes.Select(x => x.ToString("X2"))); 
}

Instead of creating a string out of the hashBytes can I simply create a Guid out of it like so?

Guid hashGuid = new Guid(hashBytes);

Would it still be valid or will I lose uniqueness?

Using MD5 is actually a version 3 GUID: https://stackoverflow.com/questions/20342058/which-uuid-version-to-use There are a couple bytes in the version 3 GUID which are hardcoded, but I'm not having any luck finding the reference to the exact details at the moment. — Bryce Wagner, Apr 05 '19 at 11:41

score 9 · Accepted Answer · answered Apr 05 '19 at 10:16

MD5 hashes and Guid essentially both express 128 bits of binary, so:

plus: you won't lose any uniqueness
plus: the fact that Guid is a value-type means that you avoids an allocation compared to string...
minus: ... but if you're going to display it anywhere, you might actually end up allocating multiple strings (i.e. rendering the same Guid multiple times)
minus: there is a semantic meaning to Guid that won't really be respected/expected here
minus: Guid default formatting isn't the same as how MD5 hashes are usually expressed
minus: Guid endianness is a mess, so if you want to get between raw bytes and any text representation: tread very carefully; it is not what you expect

Thanks Marc, avoiding allocation was the reason I chose `Guid` but after benchmarking it turned out it's 50% slower than just dealing with strings. — MaYaN, Apr 05 '19 at 10:18

score 0 · Answer 2 · answered Apr 05 '19 at 10:15

0

Not sure if it's the best idea, but since both values are 128-bits, you wouldn't be losing any data, assuming that you aren't trying to convert textual representation of the MD5.

Just convert MD5 bytes directly to GUID, without converting it to a string first.

answered Apr 05 '19 at 10:15

Can Poyrazoğlu

33,241
48
191
389

I am converting the bytes directly to `Guid` – MaYaN Apr 05 '19 at 10:16

score -3 · Answer 3 · answered Apr 05 '19 at 11:29

Guids can guarantee uniqueness only if they are generated properly by calling Guid.NewGuid(). By constructing Guids from MD5 bytes you gain zero uniqueness. You only store your bytes in a data structure named "Globally Unique IDentifier", that could potentially be not unique.

Do this experiment: create two Guids using the same byte array for both. Do you expect the Guids to be different or equal?

Is creating a Guid out of an MD5 hash instead of String valid?

3 Answers3