Thread: Anyone know a good, easy to implement MD5 function (for file checksums)

Even if I knew where to start, I'd probably write something half as efficient as others have already done, so why try? Does anyone have/know a fast, easy to implement function/library for calculating MD5sums on files?

There's a program in the GNU coreutils called "md5sum". I'm not sure what it uses, but it's quite fast. You could execute that program if you're using Linux.

Ah, here's the source for it, I think. http://www.koders.com/c/fid3D64E21C0...A6C6A9EB4.aspx

Not sure if that's useful or not . . . Googling for "md5sum c library" shows at least one other implementation.

Originally Posted by dwks

There's a program in the GNU coreutils called "md5sum". I'm not sure what it uses, but it's quite fast. You could execute that program if you're using Linux.

Ah, here's the source for it, I think. http://www.koders.com/c/fid3D64E21C0...A6C6A9EB4.aspx

Not sure if that's useful or not . . . Googling for "md5sum c library" shows at least one other implementation.

dammit. Why is everything I find written for linux? I need a function/library that'll actually compile in Windows.

and why the hell can't I find any CRC examples that are in C, NOT C++? This is utterly annoying.

Originally Posted by Viper187

Even if I knew where to start, I'd probably write something half as efficient as others have already done, so why try? Does anyone have/know a fast, easy to implement function/library for calculating MD5sums on files?

http://cr.yp.to/2004-494/gaim/0.81-src/md5.c

You'll need to create your own .h file to provide prototypes. There are only a few functions, so that should be simple enough.

I would have thought that there is very little in the md5sum code that is linux specific.

--
Mats

Originally Posted by matsp

I would have thought that there is very little in the md5sum code that is linux specific.

I doubt there is anything but purely portable code in the algorithm itself. Don't assume things won't compile in Windows until you've tried it.

Originally Posted by brewbuck

http://cr.yp.to/2004-494/gaim/0.81-src/md5.c

You'll need to create your own .h file to provide prototypes. There are only a few functions, so that should be simple enough.

There's a md5.h here:
http://cr.yp.to/2004-494/gaim/0.81-src/md5.h

I just built that on my Windows machine with

Code:

gcc -Wall -o md5 md5.c -DTEST

. The only two changes I had to make was to include stdio.h in the "TEST" section, and put an "int" in front of the main().

The second teststring doesn't seem to match the md5 value in the comments to the same string - not sure why that is. All the others match. (The second strings produces: 4d52535c7692376e7e7e205940a93ae9)


--
Mats

The code that brewbuck posted a link to is pretty straight forward. CRC32 is not much easier, since it also involves a table of pre-computed values.

--
Mats

Originally Posted by Viper187

The more I look at this stuff, I think I'd rather use CRC32/64 for this project, if anyone could point me to an easy to use example in C. I don't like the prospect of dealing with all these char arrays of hashes vs a simple array of 32/64 bit integers. The thing I intend to write will recurse subdirectories looking for duplicate files by comparing checksums.

I don't see what the difference in difficulty is between an array of ints versus an array of chars. CRC32 has a significant possibility of collisions. The code is right there, just use it.