feat: Get-OpenGraph improvement ( Fixes #15, Fixes #16, Fixes #17 )

Author: StartAutomating

Commands/Get-OpenGraph.ps1

@@ -18,15 +18,22 @@ function Get-OpenGraph
             'https://msnbc.com/',
                 'https://fox.com/' |
                     Get-OpenGraph
+    .LINK
+        https://ogp.me/        
     #>
     [Alias('openGraph','ogp')]
     [CmdletBinding(PositionalBinding=$false)]
     param(        
     # The URL that may contain Open Graph metadata 
-    [Parameter(ValueFromPipelineByPropertyName)]
+    [Parameter(Position=0,ValueFromPipeline,ValueFromPipelineByPropertyName)]
     [Uri]
     $Url,
 
+    # Any HTML that may contain open graph metadata.
+    [Parameter(ValueFromPipelineByPropertyName)]
+    [string]
+    $Html,
+
     # A dictionary of additional Open Graph metadata to include in the result
     [Parameter(ValueFromPipelineByPropertyName)]
     [Collections.IDictionary]
@@ -40,36 +47,70 @@ function Get-OpenGraph
 
     begin {
         # Make a regex to match meta tags
-        $metaRegex = [Regex]::new('<meta.+?/>','IgnoreCase','00:00:00.1')
-        if (-not $script:OpenGraphCache) {
-            $script:OpenGraphCache = [Ordered]@{}
-        }
+        # We will match both open and closed tags.
+        $metaRegex = [Regex]::new('<meta.+?/?>','IgnoreCase','00:00:00.1')
+        # If we do not have a cache
+        if (-not $script:Cache) {
+            # create one.
+            $script:Cache = [Ordered]@{}
+        }        
     }
 
     process {
         # Declare an empty object to hold the Open Graph metadata
         $openGraphMetadata = [Ordered]@{PSTypeName='OpenGraph'}
-        if ($Url) {
-            if ($script:OpenGraphCache[$url] -and -not $Force) {
-                return $script:OpenGraphCache[$url]
+        if ($Url -and -not $PSBoundParameters['html']) {
+            # If the url is an absolute url with a scheme or http or https.
+            if ($url.Scheme -in 'http', 'https') {
+                # Get the url (if it is not cached).
+                if (-not $script:Cache[$url] -or $Force) {
+                    $script:Cache[$url] =try { 
+                            Invoke-RestMethod -Uri $Url
+                        } catch { $_ }                
+                }
+                $html = $script:Cache[$url]
+            } 
+            # Otherwise, see if the path exists
+            elseif (Test-Path $url)
+            {
+                # and get content from that path.
+                $html = Get-Content "$url" -Raw
             }
-            $restResponse = Invoke-RestMethod -Uri $Url
-            foreach ($match in $metaRegex.Matches("$restResponse")) {
+        }
+
+        # If we had any html,
+        if ($html) {
+            # find all of the `<meta>` tags.
+            foreach ($match in $metaRegex.Matches($html)) {
+                # Try to make them XML
                 $matchXml = "$match" -as [xml]
+                # If that fails,
+                if (-not $matchXml) {
+                    # try once more after explicitly closing the end tag.
+                    $matchXml = $match -replace '>$', '/>' -as [xml]
+                }
+                # If the meta tag contained a property and content,
                 if ($matchXml.meta.property -and $matchXml.meta.content) {
-                    $openGraphMetadata[$matchXml.meta.property] = $matchXml.meta.content
+                    # we will add it to our openGraph metadata.
+                    $openGraphMetadata[
+                        $matchXml.meta.property
+                    ] = $matchXml.meta.content
                 }
             }
-            $script:OpenGraphCache[$url] = $openGraphMetadata
         }
+
+        # If any data was provided
         if ($Data) {
+            # copy it into open graph metadata
             foreach ($key in $Data.Keys) {
                 $openGraphMetadata[$key] = $Data[$key]
             }
         }
 
+        # If there was no open graph metadata, return nothing.
         if (-not $openGraphMetadata.Count) { return }
 
+        # Otherwise, return our OpenGraph metadata
         [PSCustomObject]$openGraphMetadata                
     }
 }