sitecore-text-classifier.ps1

$page = Get-Item -Path '/sitecore/content/Habitat/Home/Modules/Feature/Accounts'

$links = [Sitecore.Globals]::LinkDatabase
$processed = New-Object System.Collections.ArrayList

# template names of items that will be linked from the presentation field that are not datasources
$skip = @('Device','Layout','View rendering', 'Controller rendering', 'Test Definition', 'Multivariate Test Variable')

# shared and final layout
$layouts = @('{F1A1FE9E-A60C-4DDB-A3A0-BB5B29FE732E}', '{04BF00DB-F5FB-41F7-8AB7-22408372A981}')

function GetContent($item, $layout = $False)
{
    # Prevent circular reference infinite recursion
    if ($processed -contains $item.ID.ToString())
    {
        return @()
    }
    $processed.Add($item.ID.ToString())
    
    # Read content fields stripping out markup, quotes, and white spaces
    $values = $item.Fields `
        | Where { $_.Name -notmatch '^__' } `
        | Select -ExpandProperty 'Value' `
        | % { $_ -replace "<[\w/][^<>]*[\w/]?>", ""} `
        | % { $_ -replace "[\s\t]+", " "} `
        | %{ $_ -replace '[''"]', ''} `
        | %{ $_ -replace '\n|(nbsp;)', ' '} `
    
    $content = @()
    
    # Ignore GUIDs and single word values
    $content += $values `
        | Where { $_ -notmatch '[\d\w]{8}-[\d\w]{4}-[\d\w]{4}-[\d\w]{4}-[\d\w]{12}' } `
        | Where { $_.Split(' ').Length -gt 1 }
        
    # Recursively resolve references
    $content += $values `
        | Where { $_ -match '[\d\w]{8}-[\d\w]{4}-[\d\w]{4}-[\d\w]{4}-[\d\w]{12}' } `
        | %{ $_.Split('|') | %{ $_ } } `
        | Where { [Sitecore.Data.ID]::IsID($_) } `
        | %{ Get-Item . -ID $_ } `
        | %{ GetContent $_ $False }

    # Read content from datasources
    if ($layout)
    {
        $content += $links.GetReferences($item) `
            | Where { $_.SourceItemVersion -eq $item.Version.Number } `
            | Where { $layouts -contains $_.SourceFieldID.ToString() } `
            | Where { [Sitecore.Data.ID]::IsID($_.TargetItemID) } `
            | % { Get-Item . -ID $_.TargetItemID } `
            | Where { $skip -notcontains $_.Template.Name }`
            | % { GetContent $_ $False }
    }
    
    return $content
}

$content = GetContent $page $True `
    | Where { $_ -match '\D+' } `
    | %{ $_ -replace '\.$', ''} `
    | Sort-Object `
    | Get-Unique
    
$text = [String]::Join('. ', $content)

$keywords = Invoke-WebRequest `
         -Uri 'https://westus.api.cognitive.microsoft.com/text/analytics/v2.0/keyPhrases' `
         -Body "{'documents': [ { 'language': 'en', 'id': '$($page.ID.ToString())', 'text': '$text' } ]}" `
         -ContentType "application/json" `
         -Headers @{'Ocp-Apim-Subscription-Key' = '<use-your-own-key>'} `
         -Method 'Post' `
         -UseBasicParsing | ConvertFrom-Json

$sentiment = Invoke-WebRequest `
         -Uri 'https://westus.api.cognitive.microsoft.com/text/analytics/v2.0/sentiment' `
         -Body "{'documents': [ { 'language': 'en', 'id': '$($page.ID.ToString())', 'text': '$text' } ]}" `
         -ContentType "application/json" `
         -Headers @{'Ocp-Apim-Subscription-Key' = '<use-your-own-key>'} `
         -Method 'Post' `
         -UseBasicParsing | ConvertFrom-Json
         
Write-Host "The text is $($sentiment.documents[0].score*100)% positive"

# Limit to keyw phrases where all words are capitalized
$keywords.documents[0].keyPhrases `
    | Where { $_ -cmatch '^([A-Z]\w+\s?)*$' } `
    | %{ Write-Host $_ }