Binary file search on Windows 8?

Question

I'm in a weird situation. I need to find a file based on its contents, not the name because the name is not unique...

For example, I have a file named "cat.jpg". It is a picture of a cat. I happen to have 5 million copies of that in 5 million different folders, so, while the folder names are different, the file names are the same:

cat1/cat.jpg
cat2/cat.jpg
cat3/cat.jpg
etc...

However, one of those cat.jpgs is actually a picture of a dog (and the file is still named "cat.jpg")

How can I find the one that is a dog assuming I have an exact copy of it available. I can't do filename search because then I'll just get 5 million results of "cat.jpg"

Answer 1

Here's an example using fc.exe with the binary compare flag:

for /R "C:\Some Path" %f in (*.jpg) do @(FC /B "C:\Some Other Path Maybe\dog.jpg" "%f" | FIND "FC: no differences encountered.") && echo %f

This will recursively go through the files and folders in C:\Some Path assigning the JPEGs to %f. FC will pipe the output of the comparison with dog.jpg and %f to the string comparison utility, FIND.

If the files match the output from FC will be FC: no differences encountered.

The && means that echo %f will fire if that string is found by FIND.

So if there is a match you should see something like

FC: no differences encountered.
C:\Some Path\cat5000000\cat.jpg

Answer 2

Though not a binary search, you could use powershell to get at least close (assuming all the images are exactly the same except the image...)

https://stackoverflow.com/questions/9552367/get-the-detail-informations-from-a-png-file-in-powershell

Basically, find the details you need to check against, and then use Powershell to search each image, then compare:

http://technet.microsoft.com/en-us/library/ee156812.aspx

You can use compare-object to check the object too. Either way should work, but the second method would be better / easier.

Answer 3

If you have an exact copy of it available, once you know the file size you can search for cat.jpg size:n.nnKB, or more specifically in bytes: cat.jpg size:nB.

If you have the file size and don't know the name, searching for just size:nB should work fine too, or at least narrow it down considerably.