Extracting substring from a string - powershell

Question

Struggling to extract value within square brackets from below strings using PowerShell
in relation to any Facility C Loan [?10%?] per cent. per annum;
"Facility A Commitments" means the aggregate of the Facility A Commitments, being [????????10 million?????] at the date of this Agreement.
Output required:
10%
10 million

Answer 1

With a single, multiline string in memory (PSv4+):
$str = #'
in relation to any Facility C Loan [?10%?] per cent. per annum;
"Facility A Commitments" means the aggregate of the Facility A Commitments, being [????????10 million?????] at the date of this Agreement.
'#
[regex]::matches($str,'\[\?+([^?]+)\?+\]').ForEach({ $_.Groups[1].Value })
Using the pipeline with Get-Content and Select-String for line-by-line processing (PSv3+):
$lines = #'
in relation to any Facility C Loan [?10%?] per cent. per annum;
"Facility A Commitments" means the aggregate of the Facility A Commitments, being [????????10 million?????] at the date of this Agreement.
'# -split '\r?\n'
# Substitute your `Get-Content someFile.txt` call for $lines
$lines |
Select-String '\[\?+([^?]+)\?+\]' |
ForEach-Object { $_.Matches.Groups[1].Value }
Explanation of regex \[\?+([^?]+)\?+\]:
\[ matches a literal [
\?+ matches one or more (+) literal ?
([^?]+) is a capture group ((...)) that matches one or more (+) characters from the set of characters ([...]) that are not (^) part of the set, i.e., any character that is not the ? character - this is the value of interest to extract.
\?+ matches one or more literal ?
\] matches a literal ]
[regex]::Matches() and the .Matches property on the objects that Select-String emits is a collection of [System.Text.RegularExpressions.Match] objects, whose .Groups property contains both the full match (index 0) and what each capture group captured (1 containing the 1st capture group's value, ...).

Answer 2

This is your regex for both cases:
(?<=\[\?+)[^\?]*(?=\?+\])
You can play with it at https://regex101.com
But this does not support non-fixed width look behinds (the first plus). It should work in .NET/PowerShell though.
This will be good for you:
https://www.regular-expressions.info/lookaround.html

Answer 3

For the first one run:
$message -match '\[\?(\d*%)\?\]'
echo $Matches[1]
For the second one:
\[\?*(\d* million)\?*\]
echo $Matches[1]
In each iteration you can simple as if $message -match '...' returns $True, than check the values inside $Matches variable (this is a system variable to hold the result of the regex.