Recursing Through Multiple Text Files with References - file

Question

I have hundreds of text files in a folder which can often reference each other, and go serveral levels deep. Not sure if I am explaining this well, so I will explain with an example.
Let's say folder "A" contains 500 .txt files. The first one could be called A.txt and somewhere in there it mentions B.txt, which in turn mentions C.txt and so on. I believe the number of levels down is no more than 10.
Now, I want to find a certain text strings which relate to A.txt by programmitically going through that file, then if it sees references to other .txt files go through them as well and so on. The resulting output would be something like A_out.txt which contains everything it found based on a regex.
I started out with this using Powershell but am now a little stuck:
$files = Get-ChildItem "C:\TEST\" -Filter *.txt
$regex = ‘PCB.*;’
for ($i=0; $i -lt $files.Count; $i++) {
$infile = $files[$i].FullName
$outfile = $files[$i].BaseName + "_out.txt"
select-string $infile -Pattern $regex -AllMatches | % { $_.Matches } | % { $_.Value } > $outfile
}
It goes through every .txt file and outputs everything that matches the PCB.*; expression to its corresponding _out.txt file.
I have absolutely no idea how to now expand this to include references to the other files. I'm not even sure if this is possible in PowerShell or whether I need to use another language to achieve what I want.
I could get some office monkey's to do all this manually but if this is relatively simple to code then it would save us a lot of time. Any help would be greatly appreciated :)
/Edit
Whilst running through this in my head, I thought I could build up an array for every time another one of the files is mentioned, and then repeat the process for those as well. However, back to my original problem, I have no idea how I would go about this.
/Edit 2:
Sorry, had been away for a few days and am only just picking this up. I have been using what I've learnt from this question and a few others to come up with the following:
function Get-FileReference
{
Param($FileName, $OutputFileName='')
if ($OutputFileName -eq '')
{
Get-FileReference $FileName ($FileName -replace '.xml$', '_out.xml')
}
else
{
Select-String $FileName -Pattern 'BusinessObject.[^"rns][w.]*' -AllMatches | % { $_.Matches } | % { $_.Value } | Add-Content $OutputFileName
Set-Location C:\TEST
$References = (Select-String -Pattern '(?<=resid=")d+' -AllMatches -path $FileName | % { $_.Matches } | % { $_.Value })
Write "SC References: $References" | Out-File OUTPUT.txt -Append
foreach ($Ref in $References)
{
$count
Write "$count" | Out-File OUTPUT.txt -Append
$count++
Write "SC Reference: $Ref" | Out-File OUTPUT.txt -Append
$xml = [xml](Get-Content 'C:\TEST\package.xml')
$res = $xml.SelectSingleNode('//res[#id = $Ref]/child::resver[last()]')
$resource = $res.id + ".xml"
Write "File to Check $resource" | Out-File OUTPUT.txt -Append
Get-FileReference $resource $OutputFileName
}
}
}
$files = gci "C:\TEST" *.xml
ForEach ($file in $files) {
Get-FileReference $file.FullName
}
Following my original question, I realised that this was a little bit more extensive than I originally thought and therefore had to tinker.
These are the noteable points:
All the parent files are .xml and code that matches on
"BusinessObject" etc works as expected.
The references to other
files are not simply .txt but require a pattern match of
'(?<=resid=")d+'.
This pattern match needs to be cross referenced with another file package.xml and based on the value
it returns, the file it next needs to look into is [newname].xml
As before, those child .xml files could reference some of the
other .xml files
The code I have pasted above seems to be getting stuck in endless loops (hence why I have debugging in there at the moment) and it is not liking the use of $Ref in:
$res = $xml.SelectSingleNode('//res[#id = $Ref]/child::resver[last()]')
That results in the following error:
Exception calling "SelectSingleNode" with "1" argument(s): "Namespace Manager or XsltContext needed. This query has a prefix, variable, or user-defined function."
Since there could be hundreds of files it dies when it gets over 1000+.

Answer 1

A recursive function which tries to do what you want.
function Get-FileReference
{
Param($FileName, $OutputFileName='')
if ($OutputFileName -eq '')
{
Get-FileReference $FileName ($FileName -replace '\.txt$', '_out.txt')
}
else
{
Select-String -Pattern 'PCB.*;' -Path $FileName -AllMatches | Add-Content $OutputFileName
$References = (Select-String -Pattern '^.*\.txt' -AllMatches -path $FileName).Matches.Value
foreach ($Ref in $References)
{
Get-FileReference $Ref $OutputFileName
}
}
}
$files = gci *.txt
ForEach ($file in $files) { Get-FileReference $file.FullName }
It takes two parameters - a filename and an output filename. If called without an output filename, it assumes it's at the top of a new recursion tree and generates an output filename to append to.
If called with an output filename (i.e. by itself) it searches for PCB patterns, appends to the output, then calls itself on any file references, with the same output filename.
Assuming that file references are lines on their own with no spaces xyz.txt.