Allow custom hook to return nil to skip a record

README.md

@@ -117,7 +117,8 @@ logic. They currently simply return the argument they take as input.
 ```ruby
 class AlgoliaSearchRecordExtractor
   # Hook to modify a record after extracting
-  def custom_hook_each(item)
+  # `node` refers to the Nokogiri HTML node of the element
+  def custom_hook_each(item, node)
     item
   end
 

lib/push.rb

@@ -145,7 +145,7 @@ class AlgoliaSearchJekyllPush < Jekyll::Command
           css_selector
           css_selector_parent
         ),
-        customRanking: ['desc(posted_at)', 'desc(title_weight)'],
+        customRanking: ['desc(posted_at)', 'desc(weight)'],
         highlightPreTag: '<span class="algolia__result-highlight">',
         highlightPostTag: '</span>'
       }

lib/record_extractor.rb

@@ -13,7 +13,7 @@ class AlgoliaSearchRecordExtractor
   end
 
   # Hook to modify a record after extracting
-  def custom_hook_each(item)
+  def custom_hook_each(item, node)
     item
   end
 
@@ -71,7 +71,7 @@ class AlgoliaSearchRecordExtractor
 
     # If initially called on a heading, we must not accept it but only accept
     # strong headings
-    level = node.name if headings.include?(node.name)
+    level = node.name if level == 'h7' && headings.include?(node.name)
 
     previous = node.previous_element
 
@@ -203,7 +203,9 @@ class AlgoliaSearchRecordExtractor
       item[:css_selector_parent] = node_css_selector(node_heading_parent(node))
       item[:weight] = weight(item)
 
-      item = custom_hook_each(item)
+      # We pass item through the user defined custom hook
+      item = custom_hook_each(item, node)
+      next if item.nil?
 
       items << item
     end

spec/fixtures/hierarchy.md

@@ -20,9 +20,13 @@ TEXT4-H2B-H1
 
 TEXT5-H3-H2B-H1
 
-
 <div>
   <h4>H4</h4>
-  <p>TEXT6-H4-H3-H2B-H1</p>
+  <p>TEXT7-H4-H3-H2B-H1</p>
 </div>
 
+## H2C
+
+TEXT8-H2C-H1
+
+

spec/record_extractor_spec.rb

@@ -111,6 +111,21 @@ describe(AlgoliaSearchRecordExtractor) do
       expect(actual.name).to eq 'h1'
       expect(actual.text).to eq 'H1'
     end
+
+    it 'should find the correct parent when indexing deep headings' do
+      # Given
+      site = get_site(algolia: { 'record_css_selector' => 'h2' })
+      test_hierarchy = extractor.new(site.file_by_name('hierarchy.md'))
+      nodes = test_hierarchy.html_nodes
+      h2 = nodes[2]
+
+      # When
+      actual = test_hierarchy.node_heading_parent(h2)
+
+      # Then
+      expect(actual.name).to eq 'h1'
+      expect(actual.text).to eq 'H1'
+    end
   end
 
   describe 'node_hierarchy' do
@@ -234,7 +249,7 @@ describe(AlgoliaSearchRecordExtractor) do
       expect(actual).to eq '#text4'
     end
 
-    it 'uses p:nth-of-type' do
+    it 'uses p:nth-of-type if no #id found' do
       # Given
       nodes = test_page.html_nodes
       p = nodes[2]
@@ -339,7 +354,7 @@ describe(AlgoliaSearchRecordExtractor) do
   describe 'custom_hook_each' do
     it 'let the user call a custom hook to modify a record' do
       # Given
-      def test_page.custom_hook_each(item)
+      def test_page.custom_hook_each(item, _)
         item[:custom_attribute] = 'foo'
         item
       end
@@ -350,6 +365,19 @@ describe(AlgoliaSearchRecordExtractor) do
       # Then
       expect(actual[0]).to include(custom_attribute: 'foo')
     end
+
+    it 'let the user discard a record by returning nil' do
+      # Given
+      def test_page.custom_hook_each(_, _)
+        nil
+      end
+
+      # When
+      actual = test_page.extract
+
+      # Then
+      expect(actual.size).to eq 0
+    end
   end
 
   describe 'custom_hook_all' do