Merge remote-tracking branch 'upstream/master' into configurable-schemes-bug-fix

Author: Simeon F. Willbanks

.travis.yml

@@ -7,7 +7,6 @@ before_install:
 script: "bundle exec rake"
 
 rvm:
-  - 1.8.7
   - 1.9.2
   - 1.9.3
   - 2.0.0

CHANGELOG.md

@@ -1,6 +1,8 @@
 # CHANGELOG
 
-## 1.2.0
+## 1.3.0
+
+1.2.0 didn't actually include the following changes. Yanked that release.
 
   * CamoFilter now camos https images. #96 josh
 

lib/html/pipeline/camo_filter.rb

@@ -14,14 +14,17 @@ class Pipeline
     # Context options:
     #   :asset_proxy (required) - Base URL for constructed asset proxy URLs.
     #   :asset_proxy_secret_key (required) - The shared secret used to encode URLs.
+    #   :asset_proxy_whitelist - Array of host Strings or Regexps to skip
+    #                            src rewriting.
     #
     # This filter does not write additional information to the context.
     class CamoFilter < Filter
       # Hijacks images in the markup provided, replacing them with URLs that
       # go through the github asset proxy.
       def call
+        return unless asset_proxy_enabled?
+
         doc.search("img").each do |element|
-          next if context[:disable_asset_proxy]
           next if element['src'].nil?
 
           begin
@@ -30,10 +33,11 @@ def call
             next
           end
 
-          next if uri.hostname.nil?
-          next if uri.hostname.match(/(^|\.)github(app)?\.com$/)
+          next if uri.host.nil?
+          next if asset_host_whitelisted?(uri.host)
 
           element['src'] = asset_proxy_url(uri.to_s)
+          element['data-canonical-src'] = uri.to_s
         end
         doc
       end
@@ -55,7 +59,12 @@ def asset_url_hash(url)
         OpenSSL::HMAC.hexdigest(digest, asset_proxy_secret_key, url)
       end
 
-      # Private: the hostname to use for generated asset proxied URLs.
+      # Private: Return true if asset proxy filter should be enabled
+      def asset_proxy_enabled?
+        !context[:disable_asset_proxy]
+      end
+
+      # Private: the host to use for generated asset proxied URLs.
       def asset_proxy_host
         context[:asset_proxy]
       end
@@ -64,6 +73,16 @@ def asset_proxy_secret_key
         context[:asset_proxy_secret_key]
       end
 
+      def asset_proxy_whitelist
+        context[:asset_proxy_whitelist] || []
+      end
+
+      def asset_host_whitelisted?(host)
+        asset_proxy_whitelist.any? do |test|
+          test.is_a?(String) ? host == test : test.match(host)
+        end
+      end
+
       # Private: helper to hexencode a string. Each byte ends up encoded into
       # two characters, zero padded value in the range [0-9a-f].
       def hexencode(str)

lib/html/pipeline/version.rb

@@ -1,5 +1,5 @@
 module HTML
   class Pipeline
-    VERSION = "1.2.0"
+    VERSION = "1.3.0"
   end
 end

test/html/pipeline/camo_filter_test.rb

@@ -8,14 +8,17 @@ def setup
     @asset_proxy_secret_key = 'ssssh-secret'
     @options = {
       :asset_proxy            => @asset_proxy_url,
-      :asset_proxy_secret_key => @asset_proxy_secret_key
+      :asset_proxy_secret_key => @asset_proxy_secret_key,
+      :asset_proxy_whitelist  => [/(^|\.)github\.com$/]
     }
   end
 
   def test_camouflaging_http_image_urls
     orig = %(<p><img src="http://twitter.com/img.png"></p>)
     assert_includes 'img src="' + @asset_proxy_url,
       CamoFilter.call(orig, @options).to_s
+    assert_includes 'data-canonical-src="http://twitter.com/img.png"',
+      CamoFilter.call(orig, @options).to_s
   end
 
   def test_doesnt_rewrite_dotcom_image_urls
@@ -42,12 +45,6 @@ def test_camouflaging_github_prefixed_image_urls
       CamoFilter.call(orig, @options).to_s
   end
 
-  def test_doesnt_rewrite_dotcom_app_image_urls
-    orig = %(<p><img src="https://githubapp.com/img.png"></p>)
-    assert_equal "<p><img src=\"https://githubapp.com/img.png\"></p>",
-      CamoFilter.call(orig, @options).to_s
-  end
-
   def test_doesnt_rewrite_absolute_image_urls
     orig = %(<p><img src="/img.png"></p>)
     assert_equal "<p><img src=\"/img.png\"></p>",