The 2011 Scripting Games Advanced Event 7: Map User's Names and Twitter Names with PowerShell

The 2011 Scripting Games Advanced Event 7: Map User's Names and Twitter Names with PowerShell

My personal script:
http://2011sg.poshcode.org/1528
Average Rating: 3.50 by 2 users.
(Download it)

#
#
#  2011 Scripting Games Advanced Event 7: Map User's Names and Twitter Names with PowerShell
#
#  by F.Richard 2011-04
#  
#

#Requires -Version 2.0

[CmdletBinding()]
Param( 
  [Parameter(Mandatory = $false, Position = 0, ValueFromPipeLine = $true, ValueFromPipelineByPropertyName = $true)]   
  [ValidateNotNullOrEmpty()]   
  [String] $url = "http://www.sqlsaturday.com/70/networking.aspx",

  [Parameter(Mandatory = $false, Position = 1, ValueFromPipeLine = $False, ValueFromPipelineByPropertyName = $False)]   
  [ValidateNotNullOrEmpty()]   
  [String] $output = "TwitterName.csv"
)




# Get-WebPage
# Windows PowerShell, Invalid Certificates, and Automated Downloading
#http://blogs.technet.com/b/heyscriptingguy/archive/2010/07/25/windows-powershell-invalid-certificates-and-automated-downloading.aspx 
# + some personal modifications
Function Get-WebPage {
 <# 
.Synopsis
  Gets the content at a specified http url
.Parameter url
  Url that returns content
.Parameter file
  Optional parameter that redirects download of content to a file. If left out
  content is returned as a string
.Parameter useragent
  to define user agent
.Parameter user
  user to permit connection
.Parameter password
  password to permit connection
.Parameter domain
  user's domain to permit connection
.Parameter proxyurl
  proxy Url to permit connection
.Parameter proxyurl
  proxy port to permit connection
.Parameter force
  Forces the acceptance of content from an untrusted source (eg. invalid certificate)
.EXAMPLE
 Get-WebPage -url "http://www.mysite.com"
 return web page mysite.com
.EXAMPLE
 Get-WebPage -url "http://www.mysite.com" -user "usr" -password "pass" -domain "mydom"
 return web page mysite.com using particular user
.EXAMPLE
 Get-WebPage -url "http://www.mysite.com" -file "contentsite.htm" -proxyurl "proxy.comp.net" -proxyport "8080"
 return web page mysite.com to file "contentsite.htm" using proxy proxy.comp.net:8080
.EXAMPLE
 Get-WebPage -url "http://www.mysite.com" -useragent "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.2; .NET CLR 1.0.3705;)"
 return web page mysite.com using specific user agent
#>
Param(
[string] $url, 
[string] $file = "", 
[string] $useragent = "", 
[string] $user = "", 
[string] $password = "", 
[string] $domain = "", 
[string] $proxyurl = "", 
[string] $proxyport = "", 
[switch] $force
)
 if($force) { 
  [Net.ServicePointManager]::ServerCertificateValidationCallback = {$true} 
 }

 $webclient = New-Object system.net.webclient
    #Proxy required
 if ($proxyurl -ne "" -or !$webClient.Proxy.IsBypassed($url)) {
  $proxy = New-Object System.Net.WebProxy($proxyurl, $proxyport)
  #$proxy.Credentials = (Get-Credential).GetNetworkCredential()
  $webclient.Proxy = $proxy
 }
 if ($useragent -ne "") {
  $webclient.Headers.Add("user-agent", $useragent) # :ex "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.2; .NET CLR 1.0.3705;)"
 }

 if ($user -ne "") {
        $webclient.Credentials = New-Object System.Net.NetworkCredential($user, $password, $domain)
    }
 $webClient.UseDefaultCredentials = $true
    
 if ($file -eq "") {
  return $webclient.DownloadString($url)
 } else {
  $webclient.DownloadFile($url, $file)
    }
}



Function Get-RegexName {
 <# 
 .Synopsis
  return array of captured regex group 
.Parameter content
  content to capture
.Parameter regex
  regular expression with group  to capture
#>
Param(
[string] $content, 
[string] $regex,
[string] $regexaction = "match",
[string] $regexoptions = "IgnoreCase"
)
    $arrName = @( )
 $matches = [regex]::Matches($content, $regex, $regexoptions)
 foreach($match in $matches) {
        if ($match.Groups["name"].Value.Trim().length -gt 0) {
    $arrName  += $match.Groups["name"].Value.Trim()
         }
    }
    return $arrName   
}


Function Get-TwitterToUser {
 <# 
 .Synopsis
  return user name from twitter name using csv file
.Parameter file
  csv file
#>
Param(
[string] $file
)
    $hashTwitterToUser = @{ } 
    Import-Csv $output | ForEach-Object { $hashTwitterToUser[$_.twitname] = $_.username }
    
    return $hashTwitterToUser
}


Function Get-UserToTwitter {
 <# 
 .Synopsis
  return twitter name from user name using csv file
.Parameter file
  csv file
#>
Param(
[string] $file
)
    $hashUserToTwitter = @{ } 
    Import-Csv $output | ForEach-Object { $hashUserToTwitter[$_.username] = $_.twitname }
    
    return $hashUserToTwitter
}



Function Get-TwitterName {
 <# 
 .Synopsis
  write all twitter name from a web page to a .csv file
.Parameter url
  url to examine default: "http://www.sqlsaturday.com/70/networking.aspx"
.Parameter output
  outup filename default: "TwitterName.csv"
#>

[CmdletBinding()]
Param( 
  [Parameter(Mandatory = $false, Position = 0, ValueFromPipeLine = $true, ValueFromPipelineByPropertyName = $true)]   
  [ValidateNotNullOrEmpty()]   
  [String] $url = "http://www.sqlsaturday.com/70/networking.aspx",

  [Parameter(Mandatory = $false, Position = 1, ValueFromPipeLine = $False, ValueFromPipelineByPropertyName = $False)]   
  [ValidateNotNullOrEmpty()]   
  [String] $output = "TwitterName.csv"
)
    # Get web page content to file
    $strCurDir = $(if ($strCurDir) {$strCurDir} else {if ($MyInvocation.MyCommand.CommandType -eq "Function") {(Get-Location).Path} else {Split-Path -parent $MyInvocation.MyCommand.Path} })
    $content = Get-WebPage -url $url -file "$strCurDir\temp.txt" -force
    $content = Get-Content "$strCurDir\temp.txt"
    Remove-Item "$strCurDir\temp.txt"

    [Array] $arrTwitUserName = @( ) 
    # get all twitter name
    # must use (?!:) trick to do not have line like this one
    # a href="http://www.twitter.com/http://twitter.com/sqlvariant
    $regex = "]*>(?.*?)"
    $arrTD = Get-RegexName -content $content -regex $regex
    foreach ($td in $arrTD) {
        # get only user name and a href ref
        $td_simplified = [regex]::Replace($td, "<[/]?(font|span|img|[ovwxp]:\w+)[^>]*?>", "$1", "IgnoreCase")
        
        # Get Twitter name
        $regex = "twitter.com/(?[^?/ ]*)\b(?!:)"
        $twitname = Get-RegexName -content $td_simplified -regex $regex
        if ($twitname) {
            # Get username        
            $username = [regex]::Replace($td_simplified, "<[/]?(a|[ovwxp]\w+)[^>]*?>", "$1", "IgnoreCase")
            
            # create new object
            $obj = New-Object PSObject
      $obj | Add-Member -MemberType noteproperty -Name "twitname" -Value $twitname.Trim()
            $obj | Add-Member -MemberType noteproperty -Name "username" -Value $username.Trim()
         $arrTwitUserName += $obj
        }
    }
    

    # Write file
    $arrTwitUserName | Export-Csv $output -NoTypeInformation

}

# main
Get-TwitterName -url $url -output $output

# test to user name from twitter name from csv file
$hashTwitterToUser = Get-TwitterToUser -file $output
Write-host "test Get-TwitterToUser: 'gunnyek' real name is"$hashTwitterToUser["gunneyk"]

# test to return user name from twitter name from csv file
$hashUserToTwitter = Get-UserToTwitter -file $output
Write-host "test Get-TwitterToUser: 'David Taylor' twitter name is"$hashUserToTwitter["David Taylor"]