canon 0.3.77 → 0.3.78

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 07b15f2de75b50f25cccb2d078d9dddc6c3d2fe79c5cef1a177ac288e8d667fb
-  data.tar.gz: d80d886320e5f3a00af951d2685429c48f4e0cce8b5c518c1523e55c9ca06a4b
+  metadata.gz: 031d0a1ecb48c26cb773e79d791d7ae090fa1231eb3132fbdf3d9138b0afcab6
+  data.tar.gz: 63dd55aa45a123cc1c39e6072033363def2bd0f6b6fc8de9ab88edd8a7f16ae4
 SHA512:
-  metadata.gz: 5fd91e21dce05280d14c4e9fcecf87df9b3693d261ed7837d7a5f9a156706950737a9d1038aa70ee6854b36fed88252c81d4d1e8ccefe5b9db9caf99c59bf633
-  data.tar.gz: dfe2e5ca7caacc830e27ac0d4f135081bb09d5c34139f8658d21662571e90d47bbfa246435719b784cf176ad6437c39fc5b596eca3246410216f6584f01e7c4b
+  metadata.gz: a77a02d18d89482275d8e288951b7bda5f4bb144fea76e08e21ce484df42ee4928d666cb45b7cba0f63d025b3bbda59f4e2550ed81e6190f7154cbe4c2c79931
+  data.tar.gz: 4bb93c91e98b9f04ff0513a17559a4eed805c5f7d8a3385dfd3f11e5af929c31d327eab1228c616b5712ace68cdcd8fdabcd49dbb25e5af14b3fb2160c40cc0f

data/lib/canon/pretty_printer/xml.rb

@@ -28,6 +28,21 @@ module Canon
       def nokogiri_format(xml_string)
         Canon::NokogiriLoader.require!("Nokogiri-engine XML pretty-printing")
         doc = Nokogiri::XML(xml_string, &:noblanks)
+
+        # The loud/tolerant contract, Nokogiri-lane edition: the
+        # non-strict parse recovers malformed input and records the
+        # errors — non-HTML malformed input stays loud, HTML-shaped
+        # input falls back to the HTML parser (mirrors
+        # XmlParsing.parse_with_html_fallback).
+        if doc.errors.any?
+          unless Canon::XmlParsing.html_shaped?(xml_string)
+            raise Moxml::ParseError,
+                  doc.errors.map(&:message).join("; ")
+          end
+
+          doc = Nokogiri::HTML4(xml_string, &:noblanks)
+        end
+
         if @indent_type == "tab"
           doc.to_xml(indent: 1, indent_text: "\t", encoding: "UTF-8")
         else
@@ -37,8 +52,12 @@ module Canon
 
       def moxml_format(xml_string)
         # noblanks mutates the tree (strips whitespace-only text), so
-        # the document cannot be readonly.
-        doc = Canon::XmlParsing.moxml_context.parse(xml_string, noblanks: true)
+        # the document cannot be readonly. HTML producers' output
+        # (Asciidoctor cover pages) routes through the tolerant HTML
+        # mode instead of failing the comparison (leptris#1564).
+        doc = Canon::XmlParsing.parse_with_html_fallback(
+          xml_string, noblanks: true
+        )
         if @indent_type == "tab"
           doc.to_xml(declaration: true, encoding: "UTF-8",
                      indent: 1, indent_text: "\t", expand_empty: false)

data/lib/canon/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Canon
-  VERSION = "0.3.77"
+  VERSION = "0.3.78"
 end

data/lib/canon/xml_backend.rb

@@ -6,8 +6,11 @@ module Canon
   # Two independent concerns, deliberately separated (MECE):
   # - XML engine — :nokogiri (raw) or :moxml (adapter-mediated). Drives DOM
   #   parsing and serialization of XML (see Canon::XmlParsing).
-  # - HTML support — always Nokogiri on CRuby; moxml has no HTML adapter and
-  #   leptris no HTML parser (see Canon::Html::NokogiriSupport).
+  # - HTML support — the dedicated HTML pipeline stays on Nokogiri
+  #   (see Canon::Html::NokogiriSupport); the XML lane additionally
+  #   accepts HTML-shaped content through moxml's Context#parse_html
+  #   (engine HTML modes: leptris >= 1.9.80, nokogiri) — see
+  #   XmlParsing.parse_with_html_fallback (leptris#1564).
   #
   # SSOT: moxml owns adapter preference (Moxml::Config prefers leptris
   # when installed; see XmlParsing.moxml_adapter_name). Both engines

data/lib/canon/xml_parsing.rb

@@ -44,6 +44,36 @@ module Canon
         end
       end
 
+      # leptris#1564: HTML producers' output (void elements, unclosed
+      # tags, inline JS with bare <) is not well-formed XML, but the
+      # comparison pretty-printer still needs a tree. HTML-SHAPED
+      # content — a <!DOCTYPE html> declaration or an <html> root —
+      # parses through the engine's tolerant HTML mode (moxml
+      # Context#parse_html; adapters with an engine HTML mode, leptris
+      # >= 1.9.80 and nokogiri alike). Anything else re-raises:
+      # malformed XML stays loud.
+      def parse_with_html_fallback(xml_string, options = {})
+        doc = moxml_context.parse(xml_string, options)
+
+        # Engines differ in how strict:false treats malformed input —
+        # some raise, some recover and record. The loud/tolerant
+        # contract keys off the RECORD, not the raise: non-HTML
+        # malformed input stays loud; HTML-shaped input falls back to
+        # the HTML parser. Engine-independent by construction.
+        recorded = doc.respond_to?(:parse_errors) ? doc.parse_errors : []
+        if recorded.any?
+          raise Moxml::ParseError, recorded.join("; ") unless html_shaped?(xml_string)
+
+          return moxml_context.parse_html(xml_string)
+        end
+
+        doc
+      rescue Moxml::ParseError
+        raise unless html_shaped?(xml_string)
+
+        moxml_context.parse_html(xml_string)
+      end
+
       def parse_fragment(xml_string)
         if XmlBackend.nokogiri?
           Nokogiri::XML.fragment(xml_string).children.to_a
@@ -53,6 +83,27 @@ module Canon
         end
       end
 
+      # Conservative HTML sniff on the document head: the fallback
+      # must never swallow a genuinely malformed XML comparison —
+      # only content that declares itself HTML. Index-based scan,
+      # no regex: the previous regex form was polynomial on
+      # uncontrolled data (CodeQL).
+      def html_shaped?(xml_string)
+        head = xml_string.to_s.slice(0, 2048).downcase
+        return true if head.include?("<!doctype html")
+
+        root = head.lstrip
+        return false unless root.start_with?("<?xml")
+
+        close = root.index("?>")
+        return false unless close
+
+        root = root[(close + 2)..].lstrip
+        delimiter = root[5]
+        root.start_with?("<html") &&
+          (delimiter.nil? || " \t\n\r\f>/'\"".include?(delimiter))
+      end
+
       # --- Serialization ---
 
       def serialize(node)

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: canon
 version: !ruby/object:Gem::Version
-  version: 0.3.77
+  version: 0.3.78
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-10-01 00:00:00.000000000 Z
+date: 2026-10-07 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: diff-lcs
@@ -44,28 +44,28 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.9.282
+        version: 1.9.314
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.9.282
+        version: 1.9.314
 - !ruby/object:Gem::Dependency
   name: moxml
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.5.98
+        version: 0.5.116
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.5.98
+        version: 0.5.116
 - !ruby/object:Gem::Dependency
   name: paint
   requirement: !ruby/object:Gem::Requirement
@@ -142,14 +142,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.6.28
+        version: 0.6.29
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.6.28
+        version: 0.6.29
 description: |-
   Canon provides canonicalization and pretty-printing for various serialization
   formats (XML, HTML, JSON, YAML), producing standardized forms suitable for