Merge branch 'master' into 4194-keep-logging

[arvados.git] / services / api / app / models / collection.rb

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index 08c50d9480aad17e247e38d049f278cd5d29e837..66d7addb45528cbab407ba6f62edb7d439f70cd6 100644 (file)
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -5,6 +5,7 @@ class Collection < ArvadosModel
   include KindAndEtag
   include CommonApiTemplate
 
+  before_validation :check_encoding
   before_validation :check_signatures
   before_validation :strip_manifest_text
   before_validation :set_portable_data_hash
@@ -125,6 +126,28 @@ class Collection < ArvadosModel
     end
   end
 
+  def check_encoding
+    if manifest_text.encoding.name == 'UTF-8' and manifest_text.valid_encoding?
+      true
+    else
+      begin
+        # If Ruby thinks the encoding is something else, like 7-bit
+        # ASCII, but its stored bytes are equal to the (valid) UTF-8
+        # encoding of the same string, we declare it to be a UTF-8
+        # string.
+        utf8 = manifest_text
+        utf8.force_encoding Encoding::UTF_8
+        if utf8.valid_encoding? and utf8 == manifest_text.encode(Encoding::UTF_8)
+          manifest_text = utf8
+          return true
+        end
+      rescue
+      end
+      errors.add :manifest_text, "must use UTF-8 encoding"
+      false
+    end
+  end
+
   def redundancy_status
     if redundancy_confirmed_as.nil?
       'unconfirmed'