Faster way to delete up to the 4th colon in a variable?semicolon

Question

I'm trying to query MSMQ with Powershell, and then read (in my case) XML out of the BodyStream. Stripping out the crap before the XML is slow... like 200-500ms per record.

Pulling the XML out seems okay fast (below), but that replace takes forever.

$msmq = Get-MsmqQueue -Name "myqueuelovesme"
$var = Receive-MsmqQueue -input $msmq -Peek -Count 500 -Timeout 5000 -RetrieveBody | foreach {
$sr = $_.BodyStream.ToArray()
$pr = $enc.GetString($sr)
[xml]$clean= $pr -creplace ".*:Message:"
select-xml -xml $clean -XPath "/MyXML/Defin" | foreach {$_.node}
}
$var

The contents of bodystream look like this:

F:01.02.03:1234567890:Message:<MyXML><Defin xmlns:dsi="http:// [...]

The XML always starts after that 4th semicolon, so I don't think I can split/join (as the number of semicolons in the XML can change). I see that creplace is faster than replace, but neither is great.

Any suggestions? (And heck, is the rest of the code okay?)

Answer 1

Tonnes of options I imagine here. Not sure which would be the best one. I will try and add some as we go. For sure other solutions will show up. You said you don't think you can split/join well.... BLAM!

'F:01.02.03:1234567890:Message:<MyXML><Defin xmlns:dsi="http:// [...]'.Split(":",5)[-1]

Split the string by colons but only up to 5 elements. Then return only the last element which would be <MyXML><Defin.... You would apply this to your implementation as:

[xml]$clean= $pr.Split(":",5)[-1]

You could also just split on the first < using the same logic.

[xml]$clean= "<$($pr.Split("<",2)[-1])"

Since we are splitting on the < we just add it back to the string.

I did a Measure-Command test using an indexOf and Split. Running both 100000 times took about 2 seconds each. They were both neck and neck. Same test with -creplace took about 7 seconds. Measure-Command is not perfect and its results can be questioned but it would seem either answer here (so far) would perform better.