automatically inferring og:description

Author: dmalan

_layouts/page.html

@@ -7,11 +7,9 @@
         <meta charset="utf-8">
         <meta name="viewport" content="initial-scale=1, width=device-width">
 
-        {%- if page.refresh -%}
+        {%- if page.refresh %}
         <meta content="{{ page.refresh }}" http-equiv="refresh">
-        {%- endif -%}
-
-        <meta property="og:description" content="{{ page.description | default: site.cs50.description }}">
+        {%- endif %}
 
         <meta property="og:image" content="{{ page.image | default: site.cs50.image }}">
 
@@ -25,7 +23,7 @@
                     {{- site.cs50.title -}}
                 {%- endunless -%}
             {%- endif -%}
-        {%- endcapture -%}
+        {%- endcapture %}
 
         <meta property="og:title" content="{{ title }}">
 

lib/jekyll-theme-cs50.rb

@@ -431,6 +431,60 @@ def relative_url(input)
   end
 end
 
+Jekyll::Hooks.register [:pages], :post_render do |page|
+
+  # Skip if not HTML
+  next if page.output_ext != ".html"
+
+  # Parse HTML
+  doc = Nokogiri::HTML(page.output)
+  main = doc.at_css("main")
+  next if main.nil?
+
+  # If page.description
+  if page.data.key?("description")
+    description = page.data["description"]
+
+  # Else infer
+  else
+
+    # Remove the page's title (i.e., first h1 tag)
+    main.css("h1").first&.remove
+        
+    # Remove any table of contents
+    main.css("ul#markdown-toc").first&.remove
+        
+    # Remove any spoilers
+    main.css("details")&.remove
+        
+    # Strip tags
+    text = main.text.strip
+
+    # Clean up whitespace
+    text = text.gsub(/\s+/, " ").strip
+      
+    # Truncate to max_length, breaking at word boundary
+    max_length = 160
+    if text.length > max_length
+      description = text[0...(max_length - 3)]
+      last_space = description .rindex(" ")
+      description = description[0...last_space] if last_space && last_space > 0
+      description += "..."
+    end
+  end
+
+  # Inject description
+  head = doc.at_css("head")
+  next if head.nil?
+  meta = Nokogiri::XML::Node.new("meta", doc)
+  meta["property"] = "og:description"
+  meta["content"] = CGI.escapeHTML(description)
+  head.add_child(meta)
+
+  # Update HTML
+  page.output = doc.to_html
+end
+
 Jekyll::Hooks.register [:site], :post_render do |site|
 
   # Paths of pages