canon 0.1.7 → 0.1.9

data/lib/canon/diff_formatter.rb

@@ -287,7 +287,7 @@ module Canon
     # @return [String] Formatted diff output
     def format_comparison_result(comparison_result, expected, actual)
       # Detect format from expected content
-      format = Canon::Comparison.send(:detect_format, expected)
+      format = Canon::Comparison::FormatDetector.detect(expected)
 
       formatter_options = {
         use_color: @use_color,

data/lib/canon/rspec_matchers.rb

@@ -63,6 +63,15 @@ module Canon
         self
       end
 
+      # Chain method for setting match options
+      # @param match_opts [Hash] match options
+      # @return [SerializationMatcher] self for chaining
+      def with_match(**match_opts)
+        @match ||= {}
+        @match = @match.merge(match_opts)
+        self
+      end
+
       def matches?(target)
         @target = target
 
@@ -130,7 +139,7 @@ module Canon
         else
           # Fall back to detection only if format not provided
           begin
-            detected_format = Canon::Comparison.send(:detect_format, @expected)
+            detected_format = Canon::Comparison::FormatDetector.detect(@expected)
             detected_format.to_s.upcase
           rescue StandardError
             "CONTENT"
@@ -252,12 +261,22 @@ module Canon
                                diff_algorithm: diff_algorithm)
     end
 
-    def be_yaml_equivalent_to(expected)
-      SerializationMatcher.new(expected, :yaml)
+    def be_yaml_equivalent_to(expected, match_profile: nil, match: nil,
+                              preprocessing: nil, diff_algorithm: nil)
+      SerializationMatcher.new(expected, :yaml,
+                               match_profile: match_profile,
+                               match: match,
+                               preprocessing: preprocessing,
+                               diff_algorithm: diff_algorithm)
     end
 
-    def be_json_equivalent_to(expected)
-      SerializationMatcher.new(expected, :json)
+    def be_json_equivalent_to(expected, match_profile: nil, match: nil,
+                              preprocessing: nil, diff_algorithm: nil)
+      SerializationMatcher.new(expected, :json,
+                               match_profile: match_profile,
+                               match: match,
+                               preprocessing: preprocessing,
+                               diff_algorithm: diff_algorithm)
     end
 
     def be_html_equivalent_to(expected, match_profile: nil, match: nil,
@@ -287,12 +306,22 @@ module Canon
                                diff_algorithm: diff_algorithm)
     end
 
-    def be_equivalent_to(expected)
-      SerializationMatcher.new(expected, nil)
+    def be_equivalent_to(expected, match_profile: nil, match: nil,
+                         preprocessing: nil, diff_algorithm: nil)
+      SerializationMatcher.new(expected, nil,
+                               match_profile: match_profile,
+                               match: match,
+                               preprocessing: preprocessing,
+                               diff_algorithm: diff_algorithm)
     end
 
-    def be_string_equivalent_to(expected)
-      SerializationMatcher.new(expected, :string)
+    def be_string_equivalent_to(expected, match_profile: nil, match: nil,
+                                 preprocessing: nil, diff_algorithm: nil)
+      SerializationMatcher.new(expected, :string,
+                               match_profile: match_profile,
+                               match: match,
+                               preprocessing: preprocessing,
+                               diff_algorithm: diff_algorithm)
     end
 
     if defined?(::RSpec) && ::RSpec.respond_to?(:configure)

data/lib/canon/tree_diff/operation_converter.rb

@@ -2,6 +2,11 @@
 
 require_relative "../diff/diff_node"
 require_relative "../comparison/match_options"
+# OperationConverter helper modules
+require_relative "operation_converter_helpers/metadata_enricher"
+require_relative "operation_converter_helpers/reason_builder"
+require_relative "operation_converter_helpers/post_processor"
+require_relative "operation_converter_helpers/update_change_handler"
 
 module Canon
   module TreeDiff
@@ -84,7 +89,10 @@ module Canon
         end
 
         # Post-process to detect attribute-order-only differences
-        detect_attribute_order_diffs(diff_nodes)
+        OperationConverterHelpers::PostProcessor.detect_attribute_order_diffs(
+          diff_nodes,
+          ->(dimension) { determine_normative(dimension) },
+        )
       end
 
       private
@@ -121,13 +129,19 @@ module Canon
       # @param operation [Operation] Insert operation
       # @return [DiffNode] Diff node representing insertion
       def convert_insert(operation)
-        node2 = extract_source_node(operation[:node])
+        tree_node2 = operation[:node] # TreeNode from adapter
+        node2 = extract_source_node(tree_node2)
+
+        # Enrich with path and serialized content
+        metadata = OperationConverterHelpers::MetadataEnricher.enrich(nil,
+                                                                      tree_node2, @format)
 
         diff_node = Canon::Diff::DiffNode.new(
           node1: nil,
           node2: node2,
           dimension: :element_structure,
-          reason: build_insert_reason(operation),
+          reason: OperationConverterHelpers::ReasonBuilder.build_insert_reason(operation),
+          **metadata,
         )
         # Metadata elements are informative (don't affect equivalence)
         diff_node.normative = metadata_element?(node2) ? false : determine_normative(:element_structure)
@@ -139,13 +153,20 @@ module Canon
       # @param operation [Operation] Delete operation
       # @return [DiffNode] Diff node representing deletion
       def convert_delete(operation)
-        node1 = extract_source_node(operation[:node])
+        tree_node1 = operation[:node] # TreeNode from adapter
+        node1 = extract_source_node(tree_node1)
+
+        # Enrich with path and serialized content
+        metadata = OperationConverterHelpers::MetadataEnricher.enrich(
+          tree_node1, nil, @format
+        )
 
         diff_node = Canon::Diff::DiffNode.new(
           node1: node1,
           node2: nil,
           dimension: :element_structure,
-          reason: build_delete_reason(operation),
+          reason: OperationConverterHelpers::ReasonBuilder.build_delete_reason(operation),
+          **metadata,
         )
         # Metadata elements are informative (don't affect equivalence)
         diff_node.normative = metadata_element?(node1) ? false : determine_normative(:element_structure)
@@ -159,117 +180,26 @@ module Canon
       # @param operation [Operation] Update operation
       # @return [Array<DiffNode>] Diff nodes representing updates
       def convert_update(operation)
-        node1 = extract_source_node(operation[:node1])
-        node2 = extract_source_node(operation[:node2])
-        changes = operation[:changes]
+        tree_node1 = operation[:node1] # TreeNode from adapter
+        tree_node2 = operation[:node2] # TreeNode from adapter
 
-        # Handle case where changes is a boolean or non-hash value
-        changes = {} unless changes.is_a?(Hash)
+        # Enrich with path and serialized content (shared by all DiffNodes from this operation)
+        metadata = OperationConverterHelpers::MetadataEnricher.enrich(
+          tree_node1, tree_node2, @format
+        )
 
         # Check if nodes are metadata elements
+        node1 = extract_source_node(tree_node1)
+        node2 = extract_source_node(tree_node2)
         is_metadata = metadata_element?(node1) || metadata_element?(node2)
 
-        diff_nodes = []
-
-        # Create separate DiffNode for each change dimension
-        # This ensures each dimension can be classified independently
-
-        if changes.key?(:attributes)
-          # Attribute value differences
-          # Changes can be either true (flag) or { old: ..., new: ... } (detailed)
-          if changes[:attributes].is_a?(Hash) && changes[:attributes].key?(:old)
-            old_attrs = changes[:attributes][:old]
-            new_attrs = changes[:attributes][:new]
-            diff_details = build_attribute_diff_details(old_attrs, new_attrs)
-          else
-            diff_details = "attribute values differ"
-          end
-
-          diff_node = Canon::Diff::DiffNode.new(
-            node1: node1,
-            node2: node2,
-            dimension: :attribute_values,
-            reason: diff_details,
-          )
-          diff_node.normative = is_metadata ? false : determine_normative(:attribute_values)
-          diff_nodes << diff_node
-        end
-
-        if changes.key?(:attribute_order)
-          # Attribute order differences
-          if changes[:attribute_order].is_a?(Hash) && changes[:attribute_order].key?(:old)
-            old_order = changes[:attribute_order][:old]
-            new_order = changes[:attribute_order][:new]
-            reason = "Attribute order changed: [#{old_order.join(', ')}] → [#{new_order.join(', ')}]"
-          else
-            reason = "attribute order differs"
-          end
-
-          diff_node = Canon::Diff::DiffNode.new(
-            node1: node1,
-            node2: node2,
-            dimension: :attribute_order,
-            reason: reason,
-          )
-          diff_node.normative = is_metadata ? false : determine_normative(:attribute_order)
-          diff_nodes << diff_node
-        end
-
-        if changes.key?(:value)
-          # Text content differences
-          if changes[:value].is_a?(Hash) && changes[:value].key?(:old)
-            old_val = changes[:value][:old] || ""
-            new_val = changes[:value][:new] || ""
-            preview_old = truncate_for_reason(old_val.to_s, 40)
-            preview_new = truncate_for_reason(new_val.to_s, 40)
-            reason = "Text content changed: \"#{preview_old}\" → \"#{preview_new}\""
-          else
-            reason = "text content differs"
-          end
-
-          diff_node = Canon::Diff::DiffNode.new(
-            node1: node1,
-            node2: node2,
-            dimension: :text_content,
-            reason: reason,
-          )
-          diff_node.normative = is_metadata ? false : determine_normative(:text_content)
-          diff_nodes << diff_node
-        end
-
-        if changes.key?(:label)
-          # Element name differences
-          if changes[:label].is_a?(Hash) && changes[:label].key?(:old)
-            old_label = changes[:label][:old]
-            new_label = changes[:label][:new]
-            reason = "Element name changed: <#{old_label}> → <#{new_label}>"
-          else
-            reason = "element name differs"
-          end
-
-          diff_node = Canon::Diff::DiffNode.new(
-            node1: node1,
-            node2: node2,
-            dimension: :element_structure,
-            reason: reason,
-          )
-          diff_node.normative = is_metadata ? false : determine_normative(:element_structure)
-          diff_nodes << diff_node
-        end
-
-        # If no specific changes detected, create a generic update
-        if diff_nodes.empty?
-          diff_node = Canon::Diff::DiffNode.new(
-            node1: node1,
-            node2: node2,
-            dimension: :text_content,
-            reason: "content differs",
-          )
-          diff_node.normative = is_metadata ? false : determine_normative(:text_content)
-          diff_nodes << diff_node
-        end
-
-        diff_nodes
+        # Use UpdateChangeHandler to process different change types
+        OperationConverterHelpers::UpdateChangeHandler.convert(
+          operation,
+          metadata,
+          is_metadata,
+          ->(dimension) { determine_normative(dimension) },
+        )
       end
 
       # Convert MOVE operation to DiffNode
@@ -277,14 +207,22 @@ module Canon
       # @param operation [Operation] Move operation
       # @return [DiffNode] Diff node representing move
       def convert_move(operation)
-        node1 = extract_source_node(operation[:node1])
-        node2 = extract_source_node(operation[:node2])
+        tree_node1 = operation[:node1]
+        tree_node2 = operation[:node2]
+        node1 = extract_source_node(tree_node1)
+        node2 = extract_source_node(tree_node2)
+
+        # Enrich with path and serialized content
+        metadata = OperationConverterHelpers::MetadataEnricher.enrich(
+          tree_node1, tree_node2, @format
+        )
 
         diff_node = Canon::Diff::DiffNode.new(
           node1: node1,
           node2: node2,
           dimension: :element_position,
-          reason: build_move_reason(operation),
+          reason: OperationConverterHelpers::ReasonBuilder.build_move_reason(operation),
+          **metadata,
         )
         # Metadata elements are informative (don't affect equivalence)
         is_metadata = metadata_element?(node1) || metadata_element?(node2)
@@ -299,14 +237,22 @@ module Canon
       def convert_merge(operation)
         # Merge combines multiple nodes into one
         # node1 represents the source nodes, node2 is the merged result
-        node1 = extract_source_node(operation[:nodes]&.first)
-        node2 = extract_source_node(operation[:result])
+        tree_node1 = operation[:nodes]&.first
+        tree_node2 = operation[:result]
+        node1 = extract_source_node(tree_node1)
+        node2 = extract_source_node(tree_node2)
+
+        # Enrich with path and serialized content
+        metadata = OperationConverterHelpers::MetadataEnricher.enrich(
+          tree_node1, tree_node2, @format
+        )
 
         diff_node = Canon::Diff::DiffNode.new(
           node1: node1,
           node2: node2,
           dimension: :element_structure,
           reason: "merged #{operation[:nodes]&.length || 0} nodes",
+          **metadata,
         )
         diff_node.normative = true # Merges are structural changes, always normative
         diff_node
@@ -319,14 +265,22 @@ module Canon
       def convert_split(operation)
         # Split divides one node into multiple
         # node1 is the original, node2 represents the split results
-        node1 = extract_source_node(operation[:node])
-        node2 = extract_source_node(operation[:results]&.first)
+        tree_node1 = operation[:node]
+        tree_node2 = operation[:results]&.first
+        node1 = extract_source_node(tree_node1)
+        node2 = extract_source_node(tree_node2)
+
+        # Enrich with path and serialized content
+        metadata = OperationConverterHelpers::MetadataEnricher.enrich(
+          tree_node1, tree_node2, @format
+        )
 
         diff_node = Canon::Diff::DiffNode.new(
           node1: node1,
           node2: node2,
           dimension: :element_structure,
           reason: "split into #{operation[:results]&.length || 0} nodes",
+          **metadata,
         )
         diff_node.normative = true # Splits are structural changes, always normative
         diff_node
@@ -337,14 +291,22 @@ module Canon
       # @param operation [Operation] Upgrade operation
       # @return [DiffNode] Diff node representing upgrade
       def convert_upgrade(operation)
-        node1 = extract_source_node(operation[:node1])
-        node2 = extract_source_node(operation[:node2])
+        tree_node1 = operation[:node1]
+        tree_node2 = operation[:node2]
+        node1 = extract_source_node(tree_node1)
+        node2 = extract_source_node(tree_node2)
+
+        # Enrich with path and serialized content
+        metadata = OperationConverterHelpers::MetadataEnricher.enrich(
+          tree_node1, tree_node2, @format
+        )
 
         diff_node = Canon::Diff::DiffNode.new(
           node1: node1,
           node2: node2,
           dimension: :element_hierarchy,
           reason: "promoted to higher level",
+          **metadata,
         )
         diff_node.normative = determine_normative(:element_hierarchy)
         diff_node
@@ -355,14 +317,22 @@ module Canon
       # @param operation [Operation] Downgrade operation
       # @return [DiffNode] Diff node representing downgrade
       def convert_downgrade(operation)
-        node1 = extract_source_node(operation[:node1])
-        node2 = extract_source_node(operation[:node2])
+        tree_node1 = operation[:node1]
+        tree_node2 = operation[:node2]
+        node1 = extract_source_node(tree_node1)
+        node2 = extract_source_node(tree_node2)
+
+        # Enrich with path and serialized content
+        metadata = OperationConverterHelpers::MetadataEnricher.enrich(
+          tree_node1, tree_node2, @format
+        )
 
         diff_node = Canon::Diff::DiffNode.new(
           node1: node1,
           node2: node2,
           dimension: :element_hierarchy,
           reason: "demoted to lower level",
+          **metadata,
         )
         diff_node.normative = determine_normative(:element_hierarchy)
         diff_node
@@ -378,32 +348,6 @@ module Canon
         tree_node.respond_to?(:source_node) ? tree_node.source_node : tree_node
       end
 
-      # Determine update dimension based on what changed
-      #
-      # @param operation [Operation] Update operation
-      # @return [Symbol] Match dimension
-      def determine_update_dimension(operation)
-        changes = operation[:changes] || {}
-
-        # Check what actually changed
-        if changes.key?(:attribute_order)
-          # Only attribute order changed
-          :attribute_order
-        elsif changes.key?(:attributes)
-          # Attribute values changed
-          :attribute_values
-        elsif changes.key?(:value)
-          # Text content changed
-          :text_content
-        elsif changes.key?(:label)
-          # Element name changed (rare)
-          :element_structure
-        else
-          # Default to text_content for generic updates
-          :text_content
-        end
-      end
-
       # Determine if a diff is normative based on match options
       #
       # @param dimension [Symbol] Match dimension
@@ -417,153 +361,6 @@ module Canon
         behavior != :ignore
       end
 
-      # Build reason string for INSERT operation
-      #
-      # @param operation [Operation] Operation
-      # @return [String] Reason description
-      def build_insert_reason(operation)
-        node = operation[:node]
-        content = operation[:content]
-
-        if node.respond_to?(:label)
-          # Include content preview for clarity
-          "Element inserted: #{content || "<#{node.label}>"}"
-        else
-          "Element inserted"
-        end
-      end
-
-      # Build reason string for DELETE operation
-      #
-      # @param operation [Operation] Operation
-      # @return [String] Reason description
-      def build_delete_reason(operation)
-        node = operation[:node]
-        content = operation[:content]
-
-        if node.respond_to?(:label)
-          # Include content preview for clarity
-          "Element deleted: #{content || "<#{node.label}>"}"
-        else
-          "Element deleted"
-        end
-      end
-
-      # Build reason string for UPDATE operation
-      #
-      # @param operation [Operation] Operation
-      # @return [String] Reason description
-      def build_update_reason(operation)
-        change_type = operation[:change_type] || "content"
-        "updated #{change_type}"
-      end
-
-      # Build reason string for MOVE operation
-      #
-      # @param operation [Operation] Operation
-      # @return [String] Reason description
-      def build_move_reason(operation)
-        from_pos = operation[:from_position]
-        to_pos = operation[:to_position]
-
-        if from_pos && to_pos
-          "moved from position #{from_pos} to #{to_pos}"
-        else
-          "moved to different position"
-        end
-      end
-
-      # Detect INSERT/DELETE pairs that differ only in attribute order
-      # and reclassify them to use the attribute_order dimension
-      #
-      # @param diff_nodes [Array<DiffNode>] Diff nodes to process
-      # @return [Array<DiffNode>] Processed diff nodes
-      def detect_attribute_order_diffs(diff_nodes)
-        # Group nodes by parent and element type
-        deletes = diff_nodes.select { |dn| dn.node1 && !dn.node2 }
-        inserts = diff_nodes.select { |dn| !dn.node1 && dn.node2 }
-
-        # For each DELETE, try to find a matching INSERT
-        deletes.each do |delete_node|
-          node1 = delete_node.node1
-          next unless node1.respond_to?(:name) && node1.respond_to?(:attributes)
-
-          # Skip if node has no attributes (can't be attribute order diff)
-          next if node1.attributes.nil? || node1.attributes.empty?
-
-          # Find inserts with same element name at same position
-          matching_insert = inserts.find do |insert_node|
-            node2 = insert_node.node2
-            next false unless node2.respond_to?(:name) && node2.respond_to?(:attributes)
-            next false unless node1.name == node2.name
-
-            # Must have attributes to differ in order
-            next false if node2.attributes.nil? || node2.attributes.empty?
-
-            # Check if they differ only in attribute order
-            next false unless attributes_equal_ignoring_order?(
-              node1.attributes, node2.attributes
-            )
-
-            # Ensure same content (text and children structure)
-            nodes_same_except_attr_order?(node1, node2)
-          end
-
-          next unless matching_insert
-
-          # Found an attribute-order-only difference
-          # Reclassify both nodes to use attribute_order dimension
-          delete_node.dimension = :attribute_order
-          delete_node.reason = "attribute order changed"
-          delete_node.normative = determine_normative(:attribute_order)
-
-          matching_insert.dimension = :attribute_order
-          matching_insert.reason = "attribute order changed"
-          matching_insert.normative = determine_normative(:attribute_order)
-        end
-
-        diff_nodes
-      end
-
-      # Check if two attribute hashes are equal ignoring order
-      #
-      # @param attrs1 [Hash] First attribute hash
-      # @param attrs2 [Hash] Second attribute hash
-      # @return [Boolean] True if attributes are equal (ignoring order)
-      def attributes_equal_ignoring_order?(attrs1, attrs2)
-        return true if attrs1.nil? && attrs2.nil?
-        return false if attrs1.nil? || attrs2.nil?
-
-        # Convert to hashes if needed
-        attrs1 = attrs1.to_h if attrs1.respond_to?(:to_h)
-        attrs2 = attrs2.to_h if attrs2.respond_to?(:to_h)
-
-        # Compare as sets (order-independent)
-        attrs1.sort.to_h == attrs2.sort.to_h
-      end
-
-      # Check if two nodes are the same except for attribute order
-      #
-      # @param node1 [Nokogiri::XML::Node] First node
-      # @param node2 [Nokogiri::XML::Node] Second node
-      # @return [Boolean] True if nodes are same except attribute order
-      def nodes_same_except_attr_order?(node1, node2)
-        # Same text content
-        return false if node1.text != node2.text
-
-        # Same number of children
-        return false if node1.children.length != node2.children.length
-
-        # If has children, they should have same structure
-        if node1.children.any?
-          node1.children.zip(node2.children).all? do |child1, child2|
-            child1.name == child2.name
-          end
-        else
-          true
-        end
-      end
-
       # Check if a node is a metadata/presentation element
       #
       # @param node [Object] Node to check (could be TreeNode or Nokogiri node)
@@ -583,49 +380,6 @@ module Canon
         # Check if it's in our metadata elements list
         METADATA_ELEMENTS.include?(element_name)
       end
-
-      # Build detailed reason for attribute differences
-      #
-      # @param old_attrs [Hash] Old attributes
-      # @param new_attrs [Hash] New attributes
-      # @return [String] Detailed reason
-      def build_attribute_diff_details(old_attrs, new_attrs)
-        old_keys = Set.new(old_attrs.keys)
-        new_keys = Set.new(new_attrs.keys)
-
-        missing = old_keys - new_keys
-        extra = new_keys - old_keys
-        changed = (old_keys & new_keys).reject do |k|
-          old_attrs[k] == new_attrs[k]
-        end
-
-        parts = []
-        parts << "Missing: #{missing.to_a.join(', ')}" if missing.any?
-        parts << "Extra: #{extra.to_a.join(', ')}" if extra.any?
-        if changed.any?
-          parts << "Changed: #{changed.map do |k|
-            "#{k}=\"#{truncate_for_reason(old_attrs[k],
-                                          20)}\" → \"#{truncate_for_reason(new_attrs[k],
-                                                                           20)}\""
-          end.join(', ')}"
-        end
-
-        parts.any? ? "Attributes differ (#{parts.join('; ')})" : "Attribute values differ"
-      end
-
-      # Truncate text for reason messages
-      #
-      # @param text [String] Text to truncate
-      # @param max_length [Integer] Maximum length
-      # @return [String] Truncated text
-      def truncate_for_reason(text, max_length)
-        return "" if text.nil?
-
-        text = text.to_s
-        return text if text.length <= max_length
-
-        "#{text[0...max_length - 3]}..."
-      end
     end
   end
 end