Detect binary data even when content-type doesn't contain 'charset=binary'

lib/remote_storage/riak.rb

@@ -91,7 +91,7 @@ module RemoteStorage
       timestamp = (Time.now.to_f * 1000).to_i
       object.meta["timestamp"] = timestamp
 
-      if binary_data?(content_type)
+      if binary_data?(object.content_type, data)
         save_binary_data(object, data) or halt 422
       else
         set_object_data(object, data) or halt 422
@@ -309,8 +309,15 @@ module RemoteStorage
       object.raw_data = ""
     end
 
-    def binary_data?(content_type)
+    def binary_data?(content_type, data)
-      content_type[/[^;\s]+$/] == "charset=binary"
+      return true if content_type[/[^;\s]+$/] == "charset=binary"
+
+      original_encoding = data.encoding
+      data.force_encoding("UTF-8")
+      is_binary = !data.valid_encoding?
+      data.force_encoding(original_encoding)
+
+      is_binary
     end
 
     def parent_directories_for(directory)

spec/directories_spec.rb

@@ -125,22 +125,44 @@ describe "Directories" do
       end
 
       context "with binary data" do
-        before do
+        context "charset given in content-type header" do
-          header "Content-Type", "image/jpeg; charset=binary"
+          before do
-          filename = File.join(File.expand_path(File.dirname(__FILE__)), "fixtures", "rockrule.jpeg")
+            header "Content-Type", "image/jpeg; charset=binary"
-          @image = File.open(filename, "r").read
+            filename = File.join(File.expand_path(File.dirname(__FILE__)), "fixtures", "rockrule.jpeg")
-          put "/jimmy/tasks/jaypeg.jpg", @image
+            @image = File.open(filename, "r").read
+            put "/jimmy/tasks/jaypeg.jpg", @image
+          end
+
+          it "lists the binary files" do
+            get "/jimmy/tasks/"
+
+            last_response.status.must_equal 200
+
+            content = JSON.parse(last_response.body)
+            content.must_include "jaypeg.jpg"
+            content["jaypeg.jpg"].must_be_kind_of Integer
+            content["jaypeg.jpg"].to_s.length.must_equal 13
+          end
         end
 
-        it "lists the binary files" do
+        context "no charset in content-type header" do
-          get "/jimmy/tasks/"
+          before do
+            header "Content-Type", "image/jpeg"
+            filename = File.join(File.expand_path(File.dirname(__FILE__)), "fixtures", "rockrule.jpeg")
+            @image = File.open(filename, "r").read
+            put "/jimmy/tasks/jaypeg.jpg", @image
+          end
 
-          last_response.status.must_equal 200
+          it "lists the binary files" do
+            get "/jimmy/tasks/"
 
-          content = JSON.parse(last_response.body)
+            last_response.status.must_equal 200
-          content.must_include "jaypeg.jpg"
+
-          content["jaypeg.jpg"].must_be_kind_of Integer
+            content = JSON.parse(last_response.body)
-          content["jaypeg.jpg"].to_s.length.must_equal 13
+            content.must_include "jaypeg.jpg"
+            content["jaypeg.jpg"].must_be_kind_of Integer
+            content["jaypeg.jpg"].to_s.length.must_equal 13
+          end
         end
       end
     end

spec/riak_spec.rb

@@ -185,33 +185,66 @@ describe "App with Riak backend" do
       end
 
       context "with binary data" do
-        before do
+        context "binary charset in content-type header" do
-          header "Content-Type", "image/jpeg; charset=binary"
+          before do
-          filename = File.join(File.expand_path(File.dirname(__FILE__)), "fixtures", "rockrule.jpeg")
+            header "Content-Type", "image/jpeg; charset=binary"
-          @image = File.open(filename, "r").read
+            filename = File.join(File.expand_path(File.dirname(__FILE__)), "fixtures", "rockrule.jpeg")
-          put "/jimmy/documents/jaypeg", @image
+            @image = File.open(filename, "r").read
+            put "/jimmy/documents/jaypeg", @image
+          end
+
+          it "uses the requested content type" do
+            get "/jimmy/documents/jaypeg"
+
+            last_response.status.must_equal 200
+            last_response.content_type.must_equal "image/jpeg; charset=binary"
+          end
+
+          it "delivers the data correctly" do
+            get "/jimmy/documents/jaypeg"
+
+            last_response.status.must_equal 200
+            last_response.body.must_equal @image
+          end
+
+          it "indexes the binary set" do
+            indexes = binary_bucket.get("jimmy:documents:jaypeg").indexes
+            indexes["user_id_bin"].must_be_kind_of Set
+            indexes["user_id_bin"].must_include "jimmy"
+
+            indexes["directory_bin"].must_include "documents"
+          end
         end
 
-        it "uses the requested content type" do
+        context "no binary charset in content-type header" do
-          get "/jimmy/documents/jaypeg"
+          before do
+            header "Content-Type", "image/jpeg"
+            filename = File.join(File.expand_path(File.dirname(__FILE__)), "fixtures", "rockrule.jpeg")
+            @image = File.open(filename, "r").read
+            put "/jimmy/documents/jaypeg", @image
+          end
 
-          last_response.status.must_equal 200
+          it "uses the requested content type" do
-          last_response.content_type.must_equal "image/jpeg; charset=binary"
+            get "/jimmy/documents/jaypeg"
-        end
 
-        it "delivers the data correctly" do
+            last_response.status.must_equal 200
-          get "/jimmy/documents/jaypeg"
+            last_response.content_type.must_equal "image/jpeg"
+          end
 
-          last_response.status.must_equal 200
+          it "delivers the data correctly" do
-          last_response.body.must_equal @image
+            get "/jimmy/documents/jaypeg"
-        end
 
-        it "indexes the binary set" do
+            last_response.status.must_equal 200
-          indexes = binary_bucket.get("jimmy:documents:jaypeg").indexes
+            last_response.body.must_equal @image
-          indexes["user_id_bin"].must_be_kind_of Set
+          end
-          indexes["user_id_bin"].must_include "jimmy"
 
-          indexes["directory_bin"].must_include "documents"
+          it "indexes the binary set" do
+            indexes = binary_bucket.get("jimmy:documents:jaypeg").indexes
+            indexes["user_id_bin"].must_be_kind_of Set
+            indexes["user_id_bin"].must_include "jimmy"
+
+            indexes["directory_bin"].must_include "documents"
+          end
         end
       end
 

spec/spec_helper.rb

@@ -43,7 +43,7 @@ def binary_bucket
 end
 
 def purge_all_buckets
-  [data_bucket, directory_bucket, auth_bucket].each do |bucket|
+  [data_bucket, directory_bucket, auth_bucket, binary_bucket].each do |bucket|
     bucket.keys.each {|key| bucket.delete key}
   end
 end