fix(error): Correctly calculate record size in bytes

Author: pixelastic

lib/jekyll/algolia/error_handler.rb

@@ -183,7 +183,7 @@ def self.invalid_credentials?(error, _context = {})
       # should not happen as we proactively check for record size before pushing
       # them. If it still happens it means that the value set in max_record_size
       # is not matching the value in the plan.
-      def self.record_too_big_api?(error, _contex = {})
+      def self.record_too_big_api?(error, _context = {})
         details = error_hash(error.message)
         return false if details == false
 

lib/jekyll/algolia/extractor.rb

@@ -54,13 +54,20 @@ def self.add_unique_object_id(record)
       #
       # content - The HTML content to parse
       def self.extract_raw_records(content)
-        AlgoliaHTMLExtractor.run(
+        records = AlgoliaHTMLExtractor.run(
           content,
           options: {
             css_selector: Configurator.algolia('nodes_to_index'),
             tags_to_exclude: 'script,style,iframe'
           }
         )
+        # We remove objectIDs, as the will be added at the very end, after all
+        # the hooks and shrinkage
+        records.each do |record|
+          record.delete(:objectID)
+        end
+
+        records
       end
     end
   end

lib/jekyll/algolia/indexer.rb

@@ -346,7 +346,7 @@ def self.set_settings(settings)
       # their dashboard
       #
       # When users change some settings in their dashboard, those settings might
-      # get overwritten by the pluging. We can't prevent that, but we can warn
+      # get overwritten by the plugin. We can't prevent that, but we can warn
       # them when we detect they changed something.
       def self.warn_of_manual_dashboard_editing(changed_keys)
         # Transform the hash into readable YAML

lib/jekyll/algolia/shrinker.rb

@@ -10,7 +10,7 @@ module Shrinker
       # Public: Get the byte size of the object once converted to JSON
       # - record: The record to estimate
       def self.size(record)
-        record.to_json.length
+        record.to_json.bytesize
       end
 
       # Public: Attempt to reduce the size of the record by reducing the size of
@@ -33,7 +33,7 @@ def self.fit_to_size(raw_record, max_size)
         record[:excerpt_html] = record[:excerpt_text]
         return record if size(record) <= max_size
 
-        # We halve the excerpts
+        # We half the excerpts
         excerpt_words = record[:excerpt_text].split(/\s+/)
         shortened_excerpt = excerpt_words[0...excerpt_words.size / 2].join(' ')
         record[:excerpt_text] = shortened_excerpt

spec/jekyll/algolia/shrinker_spec.rb

@@ -16,10 +16,15 @@
       it { should eq 2 }
     end
 
-    describe 'should return a size in bytes' do
+    describe 'should return the size if no special chars' do
       let(:input) { { foo: 'bar' } }
       it { should eq 13 }
     end
+
+    describe 'should return a size in bytes, even with UTF-8 chars' do
+      let(:input) { { foo: '“æé' } }
+      it { should eq 17 }
+    end
   end
 
   describe '.fit_to_size' do