Show tooltip for malicious looking links

Such as those with IDN homographs or embedded
right-to-left (RTLO) characters.

Autolinked hrefs should be escaped

3 files changed, 88 insertions, 11 deletions

diff --git a/lib/banzai/filter/autolink_filter.rb b/lib/banzai/filter/autolink_filter.rb
index deda4b1872e..f3061bad4ff 100644
--- a/lib/banzai/filter/autolink_filter.rb
+++ b/lib/banzai/filter/autolink_filter.rb
@@ -8,6 +8,10 @@ module Banzai
     #
     # Based on HTML::Pipeline::AutolinkFilter
     #
+    # Note that our CommonMark parser, `commonmarker` (using the autolink extension)
+    # handles standard autolinking, like http/https. We detect additional
+    # schemes (smb, rdar, etc).
+    #
     # Context options:
     #   :autolink  - Boolean, skips all processing done by this filter when false
     #   :link_attr - Hash of attributes for the generated links
@@ -107,10 +111,13 @@ module Banzai
           end
         end
 
-        # match has come from node.to_html above, so we know it's encoded
-        # correctly.
+        # Since this came from a Text node, make sure the new href is encoded.
+        # `commonmarker` percent encodes the domains of links it handles, so
+        # do the same (instead of using `normalized_encode`).
+        href_safe = Addressable::URI.encode(match).html_safe
+
         html_safe_match = match.html_safe
-        options = link_options.merge(href: html_safe_match)
+        options         = link_options.merge(href: href_safe)
 
         content_tag(:a, html_safe_match, options) + dropped
       end
diff --git a/lib/banzai/filter/external_link_filter.rb b/lib/banzai/filter/external_link_filter.rb
index 4f60b6f84c6..61ee3eac216 100644
--- a/lib/banzai/filter/external_link_filter.rb
+++ b/lib/banzai/filter/external_link_filter.rb
@@ -4,17 +4,29 @@ module Banzai
   module Filter
     # HTML Filter to modify the attributes of external links
     class ExternalLinkFilter < HTML::Pipeline::Filter
-      SCHEMES = ['http', 'https', nil].freeze
+      SCHEMES      = ['http', 'https', nil].freeze
+      RTLO         = "\u202E".freeze
+      ENCODED_RTLO = '%E2%80%AE'.freeze
 
       def call
         links.each do |node|
-          uri = uri(node['href'].to_s)
-
-          node.set_attribute('href', uri.to_s) if uri
+          # URI.parse does stricter checking on the url than Addressable,
+          # such as on `mailto:` links. Since we've been using it, do an
+          # initial parse for validity and then use Addressable
+          # for IDN support, etc
+          uri = uri_strict(node['href'].to_s)
+          if uri
+            node.set_attribute('href', uri.to_s)
+            addressable_uri = addressable_uri(node['href'])
+          else
+            addressable_uri = nil
+          end
 
-          if SCHEMES.include?(uri&.scheme) && !internal_url?(uri)
-            node.set_attribute('rel', 'nofollow noreferrer noopener')
-            node.set_attribute('target', '_blank')
+          unless internal_url?(addressable_uri)
+            punycode_autolink_node!(addressable_uri, node)
+            sanitize_link_text!(node)
+            add_malicious_tooltip!(addressable_uri, node)
+            add_nofollow!(addressable_uri, node)
           end
         end
 
@@ -23,12 +35,18 @@ module Banzai
 
       private
 
-      def uri(href)
+      def uri_strict(href)
         URI.parse(href)
       rescue URI::Error
         nil
       end
 
+      def addressable_uri(href)
+        Addressable::URI.parse(href)
+      rescue Addressable::URI::InvalidURIError
+        nil
+      end
+
       def links
         query = 'descendant-or-self::a[@href and not(@href = "")]'
         doc.xpath(query)
@@ -45,6 +63,57 @@ module Banzai
       def internal_url
         @internal_url ||= URI.parse(Gitlab.config.gitlab.url)
       end
+
+      # Only replace an autolink with an IDN with it's punycode
+      # version if we need emailable links.  Otherwise let it
+      # be shown normally and the tooltips will show the
+      # punycode version.
+      def punycode_autolink_node!(uri, node)
+        return unless uri
+        return unless context[:emailable_links]
+
+        unencoded_uri_str = Addressable::URI.unencode(node['href'])
+
+        if unencoded_uri_str == node.content && idn?(uri)
+          node.content = uri.normalize
+        end
+      end
+
+      # escape any right-to-left (RTLO) characters in link text
+      def sanitize_link_text!(node)
+        node.inner_html = node.inner_html.gsub(RTLO, ENCODED_RTLO)
+      end
+
+      # If the domain is an international domain name (IDN),
+      # let's expose with a tooltip in case it's intended
+      # to be malicious. This is particularly useful for links
+      # where the link text is not the same as the actual link.
+      # We will continue to show the unicode version of the domain
+      # in autolinked link text, which could contain emojis, etc.
+      #
+      # Also show the tooltip if the url contains the RTLO character,
+      # as this is an indicator of a malicious link
+      def add_malicious_tooltip!(uri, node)
+        if idn?(uri) || has_encoded_rtlo?(uri)
+          node.add_class('has-tooltip')
+          node.set_attribute('title', uri.normalize)
+        end
+      end
+
+      def add_nofollow!(uri, node)
+        if SCHEMES.include?(uri&.scheme)
+          node.set_attribute('rel', 'nofollow noreferrer noopener')
+          node.set_attribute('target', '_blank')
+        end
+      end
+
+      def idn?(uri)
+        uri&.normalized_host&.start_with?('xn--')
+      end
+
+      def has_encoded_rtlo?(uri)
+        uri&.to_s&.include?(ENCODED_RTLO)
+      end
     end
   end
 end
diff --git a/lib/banzai/pipeline/email_pipeline.rb b/lib/banzai/pipeline/email_pipeline.rb
index 0f4dd9d143d..13e6a990407 100644
--- a/lib/banzai/pipeline/email_pipeline.rb
+++ b/lib/banzai/pipeline/email_pipeline.rb
@@ -12,6 +12,7 @@ module Banzai
       def self.transform_context(context)
         super(context).merge(
           only_path: false,
+          emailable_links: true,
           no_sourcepos: true
         )
       end