<#
.SYNOPSIS
    Bulk-converts Microsoft Publisher (.pub) files to PDF, saving each PDF in
    the SAME folder as the original .pub file. Includes a timeout per file so
    a single problem file (e.g. corrupted) can't hang the whole batch.

.DESCRIPTION
    Microsoft Publisher is being retired (support ends October 1, 2026).
    This script automates Microsoft's own recommended archival step: export
    every .pub file to PDF, at printing-quality resolution.

    Each file is converted in its own background job with its own Publisher
    instance. If a file doesn't finish within the timeout, that job (and any
    stray Publisher process) is force-killed, the file is logged as "Timed
    Out", and the script moves on to the next file automatically.

.PARAMETER Path
    Folder to search for .pub files. Defaults to C:\ (your whole C: drive).

.PARAMETER SkipExisting
    If set, skips a .pub file when a same-named .pdf already exists next to
    it. Useful for resuming a run that was interrupted partway through.

.PARAMETER TimeoutSeconds
    How long to wait for a single file to convert before giving up on it and
    moving on. Defaults to 120 seconds (2 minutes), which is generous for
    all but unusually large or problematic files.

.EXAMPLE
    .\Convert-PubFiles.ps1 -Path "E:\" -SkipExisting
#>

param(
    [string]$Path = "C:\",
    [switch]$SkipExisting,
    [int]$TimeoutSeconds = 120
)

$ErrorActionPreference = "Continue"
$logPath = Join-Path $env:USERPROFILE "Desktop\Pub_Conversion_Log.csv"
$results = New-Object System.Collections.Generic.List[Object]

Write-Host "Searching for .pub files under $Path ..." -ForegroundColor Cyan
$pubFiles = Get-ChildItem -Path $Path -Filter *.pub -Recurse -File -ErrorAction SilentlyContinue

if ($pubFiles.Count -eq 0) {
    Write-Host "No .pub files found under $Path" -ForegroundColor Yellow
    return
}

Write-Host "Found $($pubFiles.Count) .pub file(s). Starting conversion...`n" -ForegroundColor Cyan
Write-Host "Per-file timeout: $TimeoutSeconds seconds`n" -ForegroundColor DarkGray

# Clear out any stray Publisher processes before starting
Get-Process MSPUB -ErrorAction SilentlyContinue | Stop-Process -Force -ErrorAction SilentlyContinue
Start-Sleep -Seconds 1

# The actual conversion work, run inside a background job per file so we
# can enforce a timeout on it (a hung COM call can't otherwise be interrupted).
$conversionScript = {
    param($pubPath, $pdfPath)

    $PDF_FORMAT = 2       # pbFixedFormatTypePDF
    $PRINTING_INTENT = 3  # pbIntentPrinting

    $pub = $null
    $doc = $null
    try {
        $pub = New-Object -ComObject Publisher.Application
        $doc = $pub.Open($pubPath)
        $doc.ExportAsFixedFormat($PDF_FORMAT, $pdfPath, $PRINTING_INTENT)
        $doc.Close()
        $pub.Quit()
        return "OK"
    } catch {
        return "ERROR: $($_.Exception.Message)"
    } finally {
        if ($doc) { try { $doc.Close() } catch {} }
        if ($pub) { try { $pub.Quit() } catch {} }
    }
}

$count = 0
foreach ($file in $pubFiles) {
    $count++
    $baseName = [System.IO.Path]::GetFileNameWithoutExtension($file.Name)
    $folder = $file.DirectoryName
    $pdfPath = Join-Path $folder ($baseName + ".pdf")

    Write-Host "[$count/$($pubFiles.Count)] $($file.FullName)"

    $status = "OK"
    $note = ""

    # --- Skip if output already exists and -SkipExisting was passed ---
    if ($SkipExisting -and (Test-Path $pdfPath)) {
        Write-Host "   Skipped (PDF already exists)" -ForegroundColor DarkGray
        $results.Add([PSCustomObject]@{ File = $file.FullName; Status = "Skipped"; Note = "PDF already exists" })
        continue
    }

    if (Test-Path $pdfPath) {
        Remove-Item $pdfPath -Force -ErrorAction SilentlyContinue
    }

    # --- Run the conversion in a background job with a timeout ---
    $job = Start-Job -ScriptBlock $conversionScript -ArgumentList $file.FullName, $pdfPath
    $completed = Wait-Job -Job $job -Timeout $TimeoutSeconds

    if ($null -eq $completed) {
        # Timed out - kill the job and any stray Publisher process it spawned
        Stop-Job -Job $job -ErrorAction SilentlyContinue
        Remove-Job -Job $job -Force -ErrorAction SilentlyContinue
        Get-Process MSPUB -ErrorAction SilentlyContinue | Stop-Process -Force -ErrorAction SilentlyContinue
        Start-Sleep -Seconds 1

        $status = "Timed Out"
        $note = "Exceeded $TimeoutSeconds second timeout - possibly corrupted or unusually complex file"
        Write-Host "   TIMED OUT - skipping to next file" -ForegroundColor Red
    } else {
        $jobResult = Receive-Job -Job $job
        Remove-Job -Job $job -Force -ErrorAction SilentlyContinue

        if ($jobResult -eq "OK" -and (Test-Path $pdfPath)) {
            Write-Host "   -> PDF created" -ForegroundColor Green
        } else {
            $status = "ERROR"
            $note = if ($jobResult) { $jobResult } else { "Unknown failure - no PDF produced" }
            Write-Host "   $note" -ForegroundColor Red
        }
    }

    $results.Add([PSCustomObject]@{ File = $file.FullName; Status = $status; Note = $note })

    # Periodically save the log as we go, in case the batch is interrupted
    if ($count % 10 -eq 0) {
        $results | Export-Csv -Path $logPath -NoTypeInformation
    }
}

# Final cleanup, just in case
Get-Process MSPUB -ErrorAction SilentlyContinue | Stop-Process -Force -ErrorAction SilentlyContinue

# --- Write final log ---
$results | Export-Csv -Path $logPath -NoTypeInformation
Write-Host "`nDone. $($results.Count) file(s) processed." -ForegroundColor Cyan
$timedOut = ($results | Where-Object { $_.Status -eq "Timed Out" }).Count
$errored = ($results | Where-Object { $_.Status -eq "ERROR" }).Count
if ($timedOut -gt 0 -or $errored -gt 0) {
    Write-Host "$timedOut file(s) timed out, $errored file(s) errored - check the log to review them." -ForegroundColor Yellow
}
Write-Host "Log saved to: $logPath" -ForegroundColor Cyan