I need to be able to identify some large binary files which have been copied and renamed between secure servers. To do this, I would like to be able to hash the first X bytes and the last X bytes of all the files. I need to do this with only what is available on a standard Windows 10 system with no additional software installed, so PowerShell seems like the right choice.
Some things that don't work:
- I cannot read the entire file in, then extract the parts of the file I want to hash. The objective I'm trying to achieve is to minimize the amount of the file I need to read, and reading the entire file defeats that purpose.
- Reading moderately large portions of a file into a PowerShell variable appears to be pretty slow, so
$hash.ComputeHash($moderatelyLargeVariable)doesn't seem like a viable solution.
I'm pretty sure I need to do $hash.ComputeHash($stream) where $stream only streams part of the file.
Thus far I've tried:
function Get-FileStreamHash {
param (
$FilePath,
$Algorithm
)
$hash = [Security.Cryptography.HashAlgorithm]::Create($Algorithm)
## METHOD 0: See description below
$stream = ([IO.StreamReader]"${FilePath}").BaseStream
$hashValue = $hash.ComputeHash($stream)
## END of part I need help with
# Convert to a hexadecimal string
$hexHashValue = -join ($hashValue | ForEach-Object { "{0:x2}" -f $_ })
$stream.Close()
# return
$hexHashValue
}
Method 0: This works, but it's streaming the whole file and thus doesn't solve my problem. For a 3GB file this takes about 7 seconds on my machine.
Method 1: $hashValue = $hash.ComputeHash((Get-Content -Path $FilePath -Stream "")). This also is streaming the whole file, and it also takes forever. For the same 3GB file it takes something longer than 5 minutes (I cancelled at that point, and don't know what the total duration would be).
Method 2: $hashValue = $hash.ComputeHash((Get-Content -Path $FilePath -Encoding byte -TotalCount $qtyBytes -Stream "")). This is the same as Method 1, except that it limits the content to $qtyBytes. At 1000000 (1MB) it takes 18 seconds. I think that means Method 1 would have taken ~15 hours, 7700x slower than Method 0.
Is there a way to do something like Method 2 (limit what is read) but without the slow down? And if so, is there a good way to do it on just the end of the file?
Thanks!