Merge pull request #2560 from archivesspace/ANW-1201

ANW-1201: bugfix for UTF-8 chars in slugs breaking the PUI
archivesspace · Jan 18, 2022 · 28185c5 · 28185c5
2 parents e057615 + a4885f1
commit 28185c5
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 7 deletions.
diff --git a/backend/app/lib/slugs/slug_helpers_generate.rb b/backend/app/lib/slugs/slug_helpers_generate.rb
@@ -43,6 +43,15 @@ def self.reset_autogenerated_slugs
   # eg. migration 129
   def self.clean_slug(slug)
     if slug
+      encoding_options = {
+        :invalid           => :replace,
+        :undef             => :replace,
+        :replace           => '',
+        :universal_newline => true
+      }
+
+      slug = slug.encode(Encoding.find('ASCII'), encoding_options)
+
       # if the slug contains two slashes (forward or backward) next to each other, completely zero it out.
       # this is intended to revert an entity to use the URI if the ID or name the slug was generated from is a URI.
       slug = "" if slug =~ /\/\// || slug =~ /\\/
@@ -59,9 +68,6 @@ def self.clean_slug(slug)
       # remove double hypens
       slug = slug.gsub("--", "")
 
-      # remove en and em dashes
-      slug = slug.gsub(/[\u2013-\u2014]/, "")
-
       # remove single quotes
       slug = slug.gsub("'", "")
 

diff --git a/backend/spec/model_resource_spec.rb b/backend/spec/model_resource_spec.rb
@@ -422,7 +422,7 @@
           expect(resource[:slug]).to eq(expected_slug)
         end
         it "cleans slug" do
-          resource = Resource.create_from_json(build(:json_resource, :is_slug_auto => true, :id_0 => "Foo Bar Baz&&&&", :id_1 => "", :id_2 => "", :id_3 => ""))
+          resource = Resource.create_from_json(build(:json_resource, :is_slug_auto => true, :id_0 => "Foo“ Bar‟ Baz—&&&&", :id_1 => "", :id_2 => "", :id_3 => ""))
           expect(resource[:slug]).to eq("foo_bar_baz")
         end
         it "dedupes slug" do

diff --git a/backend/spec/spec_slugs_helper.rb b/backend/spec/spec_slugs_helper.rb
@@ -3,6 +3,16 @@
 
 def clean_slug(slug)
   if slug
+    # remove all non-ASCII chars
+    encoding_options = {
+      :invalid           => :replace,
+      :undef             => :replace,
+      :replace           => '',
+      :universal_newline => true
+    }
+
+    slug = slug.encode(Encoding.find('ASCII'), encoding_options)
+
     # if the slug contains a slash, completely zero it out.
     # this is intended to revert an entity to use the URI if the ID or name the slug was generated from is a URL.
     slug = "" if slug =~ /\//
@@ -16,9 +26,6 @@ def clean_slug(slug)
     # remove double hypens
     slug = slug.gsub("--", "")
 
-    # remove en and em dashes
-    slug = slug.gsub(/[\u2013-\u2014]/, "")
-
     # remove single quotes
     slug = slug.gsub("'", "")