canon 0.3.57 → 0.3.59

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 71a44b5ac36733a05ee0b762377f6ccaecd4ed3ae8a11527e3b067f21ded2db2
-  data.tar.gz: 1054b2885a6cd55c479c78a0935b7d18fc790753f72bdc64ebd6854fa267aeff
+  metadata.gz: 59a766120de370a517050c389b3ec809574e8e0f876b6a9b3ae351386cfc7333
+  data.tar.gz: 6a64cdc990d0d295645ec1f893e37da8b03858c06a2b3992d0e00926d04c939b
 SHA512:
-  metadata.gz: 87815f32fa049a8118764a77092e74fe23f6cd87203d67728c21d3c8ba9596f7d78e83e8201f1e8508b0ca79fb234b9e2b1c51cdb9dac27d1a8431ca7f7a7029
-  data.tar.gz: 6d51c5547e43ff83a0e1e08ca9ab84a096a57f34c1c44c469343f9cd2ac31dfaa8b15609e7c5a09fd7496a33b62b39d5cd301fe58209831810b07c4a6a762366
+  metadata.gz: 0f659e6d2708dc675e2bc0df68eb57773ca16224083249e95a9906b1f7d967c92a917882f73a810490ca3b11c228afb9b581d1ceb80b0dc674d98d46014aba42
+  data.tar.gz: 974a7dfa4cdcacb21e76952c193160022ecd83de6887b4a5582c60d5853feb5c89a3e69be559ef36484c37a8f9228da1cad8a9168829d23b3e096d51d09fca6d

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

@@ -215,9 +215,10 @@ module Canon
 
           # Check prefix closure constraint
           # If ancestors are matched, they must be matched to each other
-          node1.ancestors.each_with_index do |anc1, idx|
+          ancestors1 = node1.ancestors
+          anc2_ancestors = node2.ancestors
+          ancestors1.each_with_index do |anc1, idx|
             # Get corresponding ancestor in tree2
-            anc2_ancestors = node2.ancestors
             return false if idx >= anc2_ancestors.size
 
             anc2 = anc2_ancestors[idx]

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

@@ -81,73 +81,7 @@ module Canon
         #
         # @return [Array<String>] Path components
         def compute_path
-          # Build path from root to node
-          ancestors = @node.ancestors.reverse
-          components = ancestors.map do |ancestor|
-            path_component(ancestor)
-          end
-
-          # Add the node itself
-          components << path_component(@node)
-
-          components
-        end
-
-        # Get path component for a node
-        #
-        # @param node [TreeNode] Node to get component for
-        # @return [String]
-        def path_component(node)
-          # For element nodes: use label with sorted attributes
-          # For text nodes: use "#text"
-          # CRITICAL: Text nodes should use "#text" not "text"
-          # Check the label - actual text nodes have no label or label == "text"
-          label_str = node.label.to_s.downcase
-          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
-          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
+          @node.signature_path(include_attributes: @include_attributes)
         end
 
         # Compute signature string from path

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

@@ -50,6 +50,36 @@ 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
+
+        # Signature path components from root to this node, built by
+        # parent chaining (see NodeSignature). Memoized per variant —
+        # signatures previously re-walked the ancestor chain per node.
+        #
+        # @param include_attributes [Boolean]
+        # @return [Array<String>]
+        def signature_path(include_attributes: true)
+          if include_attributes
+            @signature_path_attrs ||= build_signature_path(true)
+          else
+            @signature_path ||= build_signature_path(false)
+          end
+        end
+
         # Check if this is a leaf node (no children)
         #
         # @return [Boolean]
@@ -57,6 +87,44 @@ 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_path(include_attributes)
+          component = signature_component(include_attributes: include_attributes)
+          if parent
+            parent.signature_path(include_attributes: include_attributes) + [component]
+          else
+            [component]
+          end
+        end
+
+        def build_signature_component(include_attributes)
+          label_str = label.to_s.downcase
+          if label.nil? || label_str.empty? || label_str == "#text" || label_str == "text"
+            return "#text"
+          end
+
+          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, :build_signature_path
+
         # Check if this is a text node (leaf with value)
         #
         # @return [Boolean]

data/lib/canon/tree_diff/matchers/hash_matcher.rb

@@ -118,11 +118,19 @@ module Canon
 
         def subtrees_match?(node1, node2)
           return false unless nodes_match?(node1, node2)
-          return false unless node1.children.size == node2.children.size
 
-          node1.children.zip(node2.children).all? do |child1, child2|
-            subtrees_match?(child1, child2)
+          children1 = node1.children
+          children2 = node2.children
+          return false unless children1.size == children2.size
+
+          index = 0
+          while index < children1.size
+            return false unless subtrees_match?(children1[index],
+                                                children2[index])
+
+            index += 1
           end
+          true
         end
 
         def nodes_match?(node1, node2)
@@ -142,6 +150,11 @@ module Canon
           return true if (text1.nil? || text1.empty?) && (text2.nil? || text2.empty?)
           return false if (text1.nil? || text1.empty?) || (text2.nil? || text2.empty?)
 
+          # Identical raw text is equivalent under every mode — skips
+          # the gsub normalization for the overwhelmingly common
+          # matched-leaf case.
+          return true if text1 == text2
+
           norm1 = normalize_text(text1)
           norm2 = normalize_text(text2)
           return true if norm1.empty? && norm2.empty?

data/lib/canon/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Canon
-  VERSION = "0.3.57"
+  VERSION = "0.3.59"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: canon
 version: !ruby/object:Gem::Version
-  version: 0.3.57
+  version: 0.3.59
 platform: ruby
 authors:
 - Ribose Inc.