##############################
#
# (C) Ondrej Sevecek, www.sevecek.com - 2014
#
# This is a SharePoint 2010 and SharePoint 2013 and newer warm up PowerShell script
# which automatically enumerates all web applications, for each in turns, all its site collections
# with their web sites and inside that all files in any type of folder or subfolder.
# The Get-AllSPFolders function does not use recursion, although it would be a logical step
# because the subfolders are recursively stored in their folders and document libraries.
#
# After enumeration, the script tries to download/touch all files with certain file extensions
# such as .ASPX files, HTML files and other infrastructure files. The script does not touch
# data files such as .DOCX or .XLSX etc.
#
# The script also generates some statistics about the numbers of files it found and the times
# it took it to enumerate and also download the files
#


function DBG ([string] $msg) {

  Write-Host ("{0:s} $msg`r`n" -f (Get-Date))
}


function Get-AllSPFolders([object] $folder)
{
  [System.Collections.ArrayList] $outFolders = @($folder)

  if (($folder -ne $null) -and (($folder -is [Microsoft.SharePoint.SPFolder]) -or ($folder -is [Microsoft.SharePoint.SPWeb]))) {

    if ($folder -is [Microsoft.SharePoint.SPWeb]) {

      DBG ('Get subfolders for web site: {0}' -f $folder.Url)
      $outFolders.AddRange($folder.Folders)
    }

    $i = 0;
    while ($i -lt $outFolders.Count) {

      DBG ('Get subfolders for folder: {0} | # = {1}' -f $outFolders[$i].Url, $outFolders[$i].SubFolders.Count)

      if ($outFolders[$i].SubFolders.Count -gt 0) {

        $outFolders.AddRange($outFolders[$i].SubFolders)
      }

      $i ++
    }
  }

  DBG ('Folders found: {0}' -f $outFolders.Count)

  return ,$outFolders
}


function Get-AllSPFiles (
  [bool] $includeCA,
  [string[]] $includedExts = @(
    '.aspx', '.html', '.htm',
    '.master', '.preview', '.css', '.bmp', '.jpg', '.jpeg', '.png', '.gif', '.js', '.xml', '.xsl', '.xaml', '.spcolor', '.spfont', '.webpart', '.dwp'
    '.txt' )
  )
{
  [System.Collections.ArrayList] $urls = @()
  $results = New-Object PSCustomObject

  Add-Member -Input $results -MemberType NoteProperty -Name enumStartTime -Value (Get-Date)
  Add-Member -Input $results -MemberType NoteProperty -Name statsPerSiteCol -Value (New-Object System.Collections.ArrayList)

  $webApps = Get-SPWebApplication -IncludeCentralAdministration:$includeCA

  $allFiles = 0
  foreach ($oneApp in $webApps) {

    DBG ('Web application: {0}' -f $oneApp.Url)

    $sites = Get-SPSite -WebApplication $oneApp

    foreach ($oneSite in $sites) {

      DBG ('Site collection: {0}' -f $oneSite.Url)

      $webs = $oneSite.Allwebs

      $siteColAllFiles = 0
      $siteColOutFiles = 0
   
      foreach ($oneWeb in $webs) {

        DBG ('Web site: {0}' -f $oneWeb.Url)

        $folders = Get-AllSPFolders $oneWeb

        foreach ($oneFolder in $folders) {

          foreach ($oneFile in $oneFolder.Files) {

            $allFiles ++
            $siteColAllFiles ++

            if ([System.IO.Path]::GetExtension($oneFile.Url) -in $includedExts) {

              if ($oneFile.Url -match '\Ahttp(?:s|)://.+') {

                $oneFileUrl = $oneFile.Url
 
              } else {

                $oneFileUrl = '{0}/{1}' -f $oneWeb.Url, $oneFile.Url
              }

              DBG ('File: {0}' -f $oneFileUrl)
              [void] $urls.Add($oneFileUrl)
  
              $siteColOutFiles ++
            }
          }
        }
      }

      $siteColStat = New-Object PSCustomObject
      Add-Member -Input $siteColStat -MemberType NoteProperty -Name webApp -Value $oneApp.Url
      Add-Member -Input $siteColStat -MemberType NoteProperty -Name siteCol -Value $oneSite.Url
      Add-Member -Input $siteColStat -MemberType NoteProperty -Name allFiles -Value $siteColAllFiles
      Add-Member -Input $siteColStat -MemberType NoteProperty -Name outFiles -Value $siteColOutFiles
      [void] $results.statsPerSiteCol.Add($siteColStat)
    }
  }

  Add-Member -Input $results -MemberType NoteProperty -Name enumEndTime -Value (Get-Date)
  Add-Member -Input $results -MemberType NoteProperty -Name enumMinutes -Value ([Math]::Round(($results.enumEndTime - $results.enumStartTime).TotalMinutes, 1))
  Add-Member -Input $results -MemberType NoteProperty -Name allFiles -Value $allFiles
  Add-Member -Input $results -MemberType NoteProperty -Name outFiles -Value $urls.Count
  Add-Member -Input $results -MemberType NoteProperty -Name urls -Value $urls

  return $results
}


Add-PSSnapin Microsoft.SharePoint.PowerShell

$stats = Get-AllSPFiles $true

$webClient = New-Object System.Net.WebClient
$webClient.Headers.Add('User-Agent', 'Sevecek-SharePoint-Keep-Alive')
$webClient.Credentials = [System.Net.CredentialCache]::DefaultCredentials

[DateTime] $startTime = Get-Date

foreach ($oneUrl in $stats.urls) {
  
  [string] $httpOutput = ''
  $httpOutput = $webClient.DownloadString($oneUrl)
  
  if ($httpOutput.Length -lt 1) {

    DBG ('Empty result: {0}' -f $oneUrl)
  }
}

[DateTime] $endTime = Get-Date

DBG ('Enum time taken: {0}' -f $stats.enumMinutes)
DBG ('Files: all = {0} | touched = {1}' -f $stats.allFiles, $stats.outFiles)
DBG ('Download time taken: {0:N1}' -f ($endTime - $startTime).TotalMinutes)
DBG ('Site collection stats: {0}' -f ($stats.statsPerSiteCol | ft -Auto | Out-String))