Problems with regex in PowerShell-CodePudding

First, I am completly new to regex and trying really hard to understand but getting very frustrated with it. Please bare with me here.

I am having problem understanding regex in PowerShell. I have tested a regex on both https://regexr.com/ and on https://regex101.com/r/zT5dL7/3. I then transfer the regex to powershell but the result is not what I am getting in powershell.

I need to get hold of the actual build number which looks like '(any number)-(any number)', where any number which over time will be bigger and bigger.

I get the string from a json from our jenkins api, convert it form json and get the string that holds the buildnumber.

$r = & $c @a | ConvertFrom-Json

$t= $r.fullDisplayName | Out-String

$t -replace '[^#0-9)][0-9-] [0-9]'

Ran regex [^#0-9)][0-9-] [0-9] on input string #1737 - 1725-74033 - x64,x86 and I expect to get a reduced string like 1725-74033x64,x86 but instead I get #1737 - - , and I do not understand why.

Feel free to expand my regex if I can get rid of both x86 and x64and the leading whitespace in the beginning. So the final result would be 1725-74033.

CodePudding user response：

Apparently i didnt understand what -replace actually did as it removes the actual matches and is left with everything else. Above works just need to use -match instead of -replace and the result is in $Matches.

CodePudding user response：

What you're doing is replacing the characters which you are trying to capture.

I have assumed that the pattern you've put here is predictable. E.g. the length of the numbers and order of characters won't change.

The code

[string] $s = '#1737 - 1725-74033 - x64,x86'

$buildNumber = [regex]::Match($s, '#[0-9]{4}\s-\s([0-9]{4}-[0-9]{5})\s-\sx64,x86').Groups[1].Value

Write-Output $buildNumber

The output

PS C:\Users\ben> [string] $s = '#1737 - 1725-74033 - x64,x86'

$buildNumber = [regex]::Match($s, '#[0-9]{4}\s-\s([0-9]{4}-[0-9]{5})\s-\sx64,x86').Groups[1].Value

Write-Output $buildNumber
1725-74033

PS C:\Users\ben>

What this does is;

match the pattern of your build output
creates a group for the build number using (capture in parens)
gets the capture using regex Match() instead of regex replace.

Regex101 does a terrific job of explaining the individual bits of regex. https://regex101.com/r/xzak5z/1

CodePudding user response：

If you want to use -replace to remove unwanted characters: (this follows your example data)

PS > $s = '#1737 - 1725-74033 - x64,x86'
PS > $s -replace '(^#\d  - )|\s'
1725-74033-x64,x86

If you want to match desired characters: (this is 'AnyNumber-AnyNumber' per your descriptive text)

PS > $s -match '\d -\d '
True
PS > $matches

Name                           Value
----                           -----
0                              1725-74033