feat(weight): Add more weight values in default weight

lib/record_extractor.rb

@@ -135,10 +135,9 @@ def node_css_selector(node)
     node.css_path.gsub('html > body > ', '')
   end
 
-  # Returns a custom numeric value representing how relevant to its hierarchy
-  # this record is. This value can be used in the custom ranking to display more
-  # relevant records first.
-  def weight(data)
+  # The more words are in common between this node and its parent heading, the
+  # higher the score
+  def weight_heading_relevance(data)
     # Get list of unique words in headings
     title_words = %i(title h1 h2 h3 h4 h5 h6)
                   .select { |title| data.key?(title) }
@@ -152,21 +151,38 @@ def weight(data)
     (title_words & text_words).size
   end
 
+  # Returns a weight based on the tag_name
+  def weight_tag_name(item)
+    tag_name = item[:tag_name]
+    # No a heading, no weight
+    return 0 unless  %w(h1 h2 h3 h4 h5 h6).include?(tag_name)
+    # h1: 100, h2: 90, ..., h6: 50
+    100 - (tag_name.gsub('h', '').to_i - 1) * 10
+  end
+
+  # Returns an object of all weights
+  def weight(item, index)
+    {
+      tag_name: weight_tag_name(item),
+      heading_relevance: weight_heading_relevance(item),
+      position: index
+    }
+  end
+
   def extract
     items = []
     html_nodes.each_with_index do |node, index|
       next unless node.text.size > 0
 
       item = metadata.clone
-      item[:objectID] = "#{item[:slug]}_#{index}"
       item.merge!(node_hierarchy(node))
       item[:tag_name] = node.name
       item[:raw_html] = node_raw_html(node)
       item[:text] = node_text(node)
       item[:unique_hierarchy] = unique_hierarchy(item)
       item[:css_selector] = node_css_selector(node)
       item[:css_selector_parent] = node_css_selector(node_heading_parent(node))
-      item[:weight] = weight(item)
+      item[:weight] = weight(item, index)
 
       # We pass item through the user defined custom hook
       item = custom_hook_each(item, node)

spec/record_extractor_spec.rb

@@ -316,7 +316,7 @@
     end
   end
 
-  describe 'weight' do
+  describe 'weight_heading_relevance' do
     it 'gets the number of words in text also in the title' do
       # Given
       data = {
@@ -325,7 +325,7 @@
       }
 
       # When
-      actual = page_file.weight(data)
+      actual = page_file.weight_heading_relevance(data)
 
       # Then
       expect(actual).to eq 2
@@ -341,7 +341,7 @@
       }
 
       # When
-      actual = page_file.weight(data)
+      actual = page_file.weight_heading_relevance(data)
 
       # Then
       expect(actual).to eq 3
@@ -357,7 +357,7 @@
       }
 
       # When
-      actual = page_file.weight(data)
+      actual = page_file.weight_heading_relevance(data)
 
       # Then
       expect(actual).to eq 2
@@ -372,7 +372,7 @@
       }
 
       # When
-      actual = page_file.weight(data)
+      actual = page_file.weight_heading_relevance(data)
 
       # Then
       expect(actual).to eq 2
@@ -386,7 +386,7 @@
       }
 
       # When
-      actual = page_file.weight(data)
+      actual = page_file.weight_heading_relevance(data)
 
       # Then
       expect(actual).to eq 1
@@ -407,13 +407,71 @@
       }
 
       # When
-      actual = page_file.weight(data)
+      actual = page_file.weight_heading_relevance(data)
 
       # Then
       expect(actual).to eq 2
     end
   end
 
+  describe 'weight_tag_name' do
+    it 'gives a score of 0 to non-headings' do
+      # Given
+      data = {
+        tag_name: 'p'
+      }
+
+      # When
+      actual = page_file.weight_tag_name(data)
+
+      # Then
+      expect(actual).to eq 0
+    end
+    it 'gives a score of 100 to h1' do
+      # Given
+      data = {
+        tag_name: 'h1'
+      }
+
+      # When
+      actual = page_file.weight_tag_name(data)
+
+      # Then
+      expect(actual).to eq 100
+    end
+    it 'gives a score of 40 to h6' do
+      # Given
+      data = {
+        tag_name: 'h6'
+      }
+
+      # When
+      actual = page_file.weight_tag_name(data)
+
+      # Then
+      expect(actual).to eq 50
+    end
+  end
+
+  describe 'weight' do
+    it 'returns an object with all weights' do
+      # Given
+      item = {
+        tag_name: 'p'
+      }
+      allow(page_file).to receive(:weight_tag_name) { 10 }
+      allow(page_file).to receive(:weight_heading_relevance) { 20 }
+
+      # When
+      actual = page_file.weight(item, 42)
+
+      # Then
+      expect(actual).to include(tag_name: 10)
+      expect(actual).to include(heading_relevance: 20)
+      expect(actual).to include(position: 42)
+    end
+  end
+
   describe 'custom_hook_each' do
     it 'let the user call a custom hook to modify a record' do
       # Given