From: Tom Clegg <tom@curoverse.com>
Date: Tue, 18 Nov 2014 16:24:58 +0000 (-0500)
Subject: 4363: Accept manifest_text in a non-UTF-8 encoding that is equal to its UTF-8 encoding.
X-Git-Tag: 1.1.0~1987^2~2
X-Git-Url: https://git.arvados.org/arvados.git/commitdiff_plain/69fae522bf6fe6dbaa64ce7d75ef4ce5616ced72

4363: Accept manifest_text in a non-UTF-8 encoding that is equal to its UTF-8 encoding.
---

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index 5dd760feee..2f81dc070b 100644
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -111,6 +111,19 @@ class Collection < ArvadosModel
     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