Paragraph conditional nested regular expression (recursion)

Question

I need a regular expression that matches the paragraph: '&Start a'(first in sample text) until '&end a' (last end from sample text). The problem is that sometimes '&end a' is not explicitly specified, and is sometimes written as '&end'. The problem is even bigger when you have '&Start b' and '&end b' (which is sometimes '&end' as well, hence the confusion).

A sample block of target for this regex is (sorry for putting it as code block):

junk text

&Start a <

fulfilling text

fulfilling text

&Start b

&Start c

&end c

fulfilling text

&end

&end <

junk text

So the regex should match all paragraph starting and ending with the lines which contain the < symbol, though it's not included in the original text. (i.e. with the &Start X we want, and skipping the '&Start Y' '&end' (or '&end Y') groups until the '&end' (or '&end X') we want.

This is not a simple implementation. The expression I am working with is the following:

&start a([^&]*)(&end a|&end) 

Which maches well isolated '&start a' '&end' paragraphs, but when other '&start Y' lines come in between, the script gets confused. I might use some If statment that jumps the undesired blocks... Here is a more complicated approach of the case:

junk text

&Start a <

fulfilling text

fulfilling text

&Start b

&Start c

&end

fulfilling text

&end

&end <

junk text

Without specifying any '&end'. Note1: '&start X' is always defined, but '&end X' can be as well '&end', but always corresponds to the closest start upfront. Note2: I can't change much the structure of my regex because of stack overflow errors, but rather adapt it to this specific case.

Sorry for the weird explaination but I hope somebody can find any viable advise.

Thank you

Edit:

#@ -split "`n" | ForEach-Object { $_.trim() } |

$files = Get-ChildItem "$PSScriptRoot" # root path

for($i=0; $i -lt $files.Count; $i++){

    #iterate through files from the current folder.
    $data = Get-Content -Path $files[$i].FullName

    # parse DisabledFeatures.txt file as array of strings (1 string per line of the file)
    $feature = Get-Content DisabledFeatures.txt

    #iterate for each string entry in $feature array (read from txt file)
    for($counter=0; counter -lt $feature.Count; counter++){

        #retrieve array value to use it in the main algorythm
        $groupID = "$feature"

        $data | ForEach-Object -Begin { $ignore = $false; $levels = 0 } -Process {
            #Start ignoring text after we've found the trigger
            if($_ -match "^`#ifdef $groupID") { $ignore = $true }   
            #Track nested groups
            elseif($ignore) {
                if ($_ -match '^`#ifdef') { $levels++ }
                elseif ($_ -match '`#endif') {
                    if($levels -ge 1) { $levels-- }
                    #If no nesting, we've hit the end of our targeted group. Stop ignoring
                    else { $ignore = $false }
                }
            }
            #Write line
            else { $_ }
        }  
    }
}

Answer 1

A pure regex-solution is probably not the best solution for this problem. It can probably be done, but it would likely be very complex and unreadable. I would use a simple parser for this. Example:

function Remove-TextGroup {
    param(
        [Parameter(Mandatory=$true)]
        [string[]]$Data,
        [Parameter(Mandatory=$true)]
        [string]$GroupID
    )

    $Data | ForEach-Object -Begin { $ignore = $false; $levels = 0 } -Process {
        #Start ignoring text after we've found the trigger
        if($_ -match "^&start $GroupID") { $ignore = $true }   
        #Track nested groups
        elseif($ignore) {
            if ($_ -match '^&start') { $levels++ }
            elseif ($_ -match '^&end') {
                if($levels -ge 1) { $levels-- }
                #If no nesting, we've hit the end of our targeted group. Stop ignoring
                else { $ignore = $false }
            }
        }
        #Write line
        else { $_ }

    }
}

Usage:

$data = @"
junk text

&Start a <

fulfilling text

fulfilling text

&Start b

&Start c

&end

fulfilling text

&end

&end <

junk text
"@ -split "`n" | ForEach-Object { $_.trim() } |
#Remove empty lines
Where-Object { $_ }

Remove-TextGroup -Data $data -GroupID a    

#Or to read from file.. 
#$data = Get-Content -Path Myfile.txt
Remove-TextGroup -Data $data -GroupID a    

Output:

junk text
junk text

If the files are big, I would optimize the sample above to use a streamreader for reading the file.