canon 0.3.30 → 0.3.31

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 308416a45e76807d58f00f40a88d5d6f53617d2a9dac468da8add04fa8175a2f
-  data.tar.gz: 2fa30986100df7b8238263546513705513635bd6c39f1552f0e0f8774f510f5f
+  metadata.gz: 7c7f349890b131b4280956b65c069369fee17b526db3ffacaa4560c3b3beb485
+  data.tar.gz: dc2179259f0a7212de6b6aaddd50e61a32fba4627d4b853d60e00172cb819c10
 SHA512:
-  metadata.gz: 0ffdb20ffc91117fdf24afb405f1684aea837316c1481ed5cc7e435bff4bf5d409876d9ca5851e187d3cd128b192ce3516b49114e6ac5d56a327dd87330f0ec9
-  data.tar.gz: aaeac239ac4707d8034eb92c8205753e3ec94ff203bf0b4023c33d0b7a92d57742e632ee29998de5a2ed041e45ea7bf2453b2479bb24ed9b9164dacba8db4ab1
+  metadata.gz: 6e3085a46f5ae85c542a419ae35b961529fc76d1f232d72f3599eb54b1f6588e0bbc9d8f1ffcdb6e31e61a5dcf4a3d8c884ddf0999f9a1f2062e22997640052a
+  data.tar.gz: 3573d0da07ad60bd4aac8a3e3ffb149d61c400a2708c10d8dcca9280abae0dc12ac7d442786fd12be62ea584345e67dd2807ca63bc598578a4013e9159cd0ad7

data/lib/canon/comparison/xml_comparator.rb

@@ -74,6 +74,25 @@ module Canon
           # Store resolved match options hash for use in comparison logic
           opts[:match_opts] = match_opts_hash
 
+          # FAST PATH: leptris Merkle digest — equal root-subtree
+          # digests plus identical document-level skeletons prove
+          # content identity (modulo whitespace-only text nodes and
+          # attribute order), which implies equivalence wherever
+          # canon's whitespace handling is the same symmetric
+          # parse-time strip. Callers whose whitespace semantics
+          # differ from that skip the gate: strict attribute order
+          # (invisible to the digest), user-configured whitespace
+          # element lists, xml:space documents (checked inside the
+          # gate), and verbose callers who need the report.
+          if !(opts[:verbose] ||
+                 match_opts_hash[:attribute_order] == :strict ||
+                 match_opts_hash[:preserve_whitespace_elements] ||
+                 match_opts_hash[:collapse_whitespace_elements] ||
+                 match_opts_hash[:strip_whitespace_elements]) &&
+              Xml::DigestGate.equal?(n1, n2)
+            return true
+          end
+
           # Create child_opts with resolved options
           child_opts = opts.merge(child_opts)
 

data/lib/canon/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Canon
-  VERSION = "0.3.30"
+  VERSION = "0.3.31"
 end

data/lib/canon/xml/digest_gate.rb

@@ -0,0 +1,94 @@
+# frozen_string_literal: true
+
+module Canon
+  module Xml
+    # Document-level Merkle-digest equivalence gate over leptris'
+    # content-defined subtree digests (libleptris #869, leptris-ruby
+    # Node#digest since 1.9.144).
+    #
+    # Equal root digests prove the root subtrees carry identical
+    # content (whitespace-only nodes dropped), and every canon match
+    # behavior is a relaxation of content identity — so equal digests
+    # imply equivalence under any option combination EXCEPT the two
+    # restrictions that compare things the digest deliberately
+    # ignores (attribute order, comments). The caller excludes those;
+    # this module answers the digest question only.
+    #
+    # A digest miss costs two readonly engine parses (no field
+    # materialization, no canon tree) — a fraction of the comparison
+    # it precedes. Any parse failure answers false and lets the full
+    # pipeline surface the error.
+    module DigestGate
+      module_function
+
+      def available?
+        return false if RUBY_ENGINE == "opal"
+        return false unless Canon::XmlBackend.moxml? &&
+          Canon::XmlParsing.moxml_adapter_name == :leptris
+
+        # Node#digest is the 1.9.144 surface; feature-detect it on a
+        # throwaway document rather than probing the class.
+        doc = Canon::XmlParsing.moxml_context.parse("<r/>", readonly: true,
+                                                            strict: false)
+        root = doc.root
+        digestable = !root.native.digest(drop_ws: true).nil?
+        doc.free
+        digestable
+      rescue StandardError
+        false
+      end
+
+      # True when both documents' root subtrees digest identically
+      # AND their document-level skeletons (prolog/epilog comments and
+      # PIs — the only doc-level nodes canon compares; doctype and
+      # whitespace-only text are excluded) serialize identically.
+      # Anything else — parse failure, missing root, digest
+      # unavailable — is false: the caller falls through to the full
+      # pipeline, which surfaces the real parse errors.
+      def equal?(xml1, xml2)
+        return false unless xml1.is_a?(String) && xml2.is_a?(String)
+
+        begin
+          # xml:space documents carry attribute-scoped whitespace canon
+          # makes normative and the digest cannot see — decline them.
+          # The scan also declines non-ASCII-compatible encodings
+          # (include? raises) — those are the full pipeline's to
+          # normalize.
+          return false if xml1.include?("xml:space") || xml2.include?("xml:space")
+        rescue Encoding::CompatibilityError
+          return false
+        end
+
+        context = Canon::XmlParsing.moxml_context
+        begin
+          left = fingerprint(context, xml1)
+          right = fingerprint(context, xml2)
+          !left.nil? && left == right
+        rescue StandardError
+          # Any engine-level surprise (encodings, broken input) is the
+          # full pipeline's domain — it normalizes and surfaces errors.
+          false
+        end
+      end
+
+      # [root digest, doc-level skeleton] or nil when unparseable.
+      def fingerprint(context, xml)
+        doc = context.parse(xml, readonly: true, strict: false)
+        root = doc.root
+        return nil unless root
+
+        skeleton = doc.children.filter_map do |child|
+          next if child.equal?(root)
+
+          case child
+          when Moxml::Comment, Moxml::ProcessingInstruction then child.to_s
+          when Moxml::Text then child.content.strip.empty? ? nil : child.to_s
+          end
+        end
+        [root.native.digest(drop_ws: true), skeleton]
+      ensure
+        doc&.free
+      end
+    end
+  end
+end

data/lib/canon/xml.rb

@@ -16,6 +16,7 @@ module Canon
   module Xml
     autoload :AttributeHandler, "canon/xml/attribute_handler"
     autoload :C14n, "canon/xml/c14n"
+    autoload :DigestGate, "canon/xml/digest_gate"
     autoload :CharacterEncoder, "canon/xml/character_encoder"
     autoload :DataModel, "canon/xml/data_model"
     autoload :ElementMatcher, "canon/xml/element_matcher"

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: canon
 version: !ruby/object:Gem::Version
-  version: 0.3.30
+  version: 0.3.31
 platform: ruby
 authors:
 - Ribose Inc.
@@ -391,6 +391,7 @@ files:
 - lib/canon/xml/c14n.rb
 - lib/canon/xml/character_encoder.rb
 - lib/canon/xml/data_model.rb
+- lib/canon/xml/digest_gate.rb
 - lib/canon/xml/element_matcher.rb
 - lib/canon/xml/line_range_mapper.rb
 - lib/canon/xml/namespace_handler.rb