canon 0.3.57 → 0.3.58

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 71a44b5ac36733a05ee0b762377f6ccaecd4ed3ae8a11527e3b067f21ded2db2
-  data.tar.gz: 1054b2885a6cd55c479c78a0935b7d18fc790753f72bdc64ebd6854fa267aeff
+  metadata.gz: c1b3bba3210fd3ca95a66a4a3e439f6b9e1a0f95b30ddf4d90f4069a2b5a6d8b
+  data.tar.gz: fadf6bdf8fb892fbb3998d53754f599b17fac80b099936d77a7dae0881e4f5c5
 SHA512:
-  metadata.gz: 87815f32fa049a8118764a77092e74fe23f6cd87203d67728c21d3c8ba9596f7d78e83e8201f1e8508b0ca79fb234b9e2b1c51cdb9dac27d1a8431ca7f7a7029
-  data.tar.gz: 6d51c5547e43ff83a0e1e08ca9ab84a096a57f34c1c44c469343f9cd2ac31dfaa8b15609e7c5a09fd7496a33b62b39d5cd301fe58209831810b07c4a6a762366
+  metadata.gz: e551b02b98c802c3d2593c1f684c89d271b223892fb32dbd20f186fe6cb13e06d18bd079839298325eb1d8a798b962876af9c95177763268ae291fc503828552
+  data.tar.gz: 8485c2e7e3049386ff977fca9c4162b44d439a2c218067f3aefe4a3b596d0f351b60c0c18a7f9fd0d5bcbf77192a0525c84dac7c1b35fe7f92369737f6dc2308

data/lib/canon/tree_diff/core/node_signature.rb

@@ -106,50 +106,11 @@ module Canon
           if node.label.nil? || label_str.empty? || label_str == "#text" || label_str == "text"
             "#text"
           else
-            component = node.label.to_s
-
-            # Include sorted attributes to distinguish nodes with same label
-            # but different attributes (while ignoring attribute order)
-            # Only include attributes if requested (for hash matching)
-            if @include_attributes && !node.attributes.empty?
-              sorted_attrs = node.attributes.sort.to_h
-              attrs_str = sorted_attrs.map { |k, v| "#{k}=#{v}" }.join(",")
-              component += "{#{attrs_str}}"
-            end
-
-            # CRITICAL: For whitespace-sensitive HTML elements, include the text value
-            # in the signature to prevent incorrect matching of nodes with different whitespace
-            if @include_attributes && whitespace_sensitive?(node) && node.value
-              # Include text value in signature for whitespace-sensitive elements
-              # Use inspect to make whitespace visible and handle special characters
-              component += "[text=#{node.value.inspect}]"
-            end
-
-            component
+            # Memoized on the node — see TreeNode#signature_component.
+            node.signature_component(include_attributes: @include_attributes)
           end
         end
 
-        # Check if a node is in a whitespace-sensitive context
-        #
-        # HTML elements where whitespace is significant: <pre>, <code>, <textarea>, <script>, <style>
-        #
-        # @param node [TreeNode] Node to check
-        # @return [Boolean] True if node is whitespace-sensitive
-        def whitespace_sensitive?(node)
-          return false unless node
-
-          # List of HTML elements where whitespace is semantically significant
-          whitespace_sensitive_tags = %w[pre code textarea script style]
-
-          # Check if this node is whitespace-sensitive
-          if node.is_a?(TreeNode)
-            label = node.label.to_s.downcase
-            return true if whitespace_sensitive_tags.include?(label)
-          end
-
-          false
-        end
-
         # Compute signature string from path
         #
         # @return [String]

data/lib/canon/tree_diff/core/tree_node.rb

@@ -50,6 +50,22 @@ module Canon
           @weight = nil
         end
 
+        # Signature path component (see NodeSignature): the label plus
+        # sorted attributes (and whitespace-sensitive text value) when
+        # +include_attributes+. Memoized per variant — signatures
+        # rebuild ancestor components for every node they cover, which
+        # is quadratic over the tree without it.
+        #
+        # @param include_attributes [Boolean]
+        # @return [String]
+        def signature_component(include_attributes: true)
+          if include_attributes
+            @signature_component_attrs ||= build_signature_component(true)
+          else
+            @signature_component ||= build_signature_component(false)
+          end
+        end
+
         # Check if this is a leaf node (no children)
         #
         # @return [Boolean]
@@ -57,6 +73,30 @@ module Canon
           children.empty?
         end
 
+        # Elements where whitespace is semantically significant; their
+        # text value participates in the signature (see NodeSignature).
+        WHITESPACE_SENSITIVE_TAGS = %w[pre code textarea script style].freeze
+
+        def build_signature_component(include_attributes)
+          component = label.to_s
+
+          # Sorted attributes distinguish same-label nodes while
+          # ignoring attribute order.
+          if include_attributes && !attributes.empty?
+            component += "{#{attributes.sort.map { |k, v| "#{k}=#{v}" }.join(',')}}" # rubocop:disable Style/StringConcatenation
+          end
+
+          if include_attributes &&
+              WHITESPACE_SENSITIVE_TAGS.include?(label.to_s.downcase) && value
+            # inspect makes whitespace visible and handles special
+            # characters.
+            component += "[text=#{value.inspect}]"
+          end
+
+          component
+        end
+        private :build_signature_component
+
         # Check if this is a text node (leaf with value)
         #
         # @return [Boolean]

data/lib/canon/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Canon
-  VERSION = "0.3.57"
+  VERSION = "0.3.58"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: canon
 version: !ruby/object:Gem::Version
-  version: 0.3.57
+  version: 0.3.58
 platform: ruby
 authors:
 - Ribose Inc.