canon 0.3.99 → 0.3.100

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4e993fe0bac445ec82e1a02983bfa2f297c1b7d9d112bcb7c724f3f010a1b353
-  data.tar.gz: 869324e74378af4f4cb9493fbfdc04ac1ba8e76c0378159547bc7e23ebf66084
+  metadata.gz: 47ebab684f05c7445139ea4b0ed5133ab734c1d16918f3dadc055b3ddbc5692f
+  data.tar.gz: d80c1ae4a66f43ffb92c3eb8a14ebca8e2bcd2eb373130797a71e7f79a748d9a
 SHA512:
-  metadata.gz: 70c6dcb6bb9da7ef764f9aed9e1e8d02fb95193c446102df62496a10dbe33010d29b5d0e90293d929cf3adc267b5128ad1dd7e9666dff38b892d221bc681f20c
-  data.tar.gz: dc45835bdab2c58479e8413eb87aee5c8615237ebf6bd2c435ae725b58b7df247c3740c25a3ab192fb0aaf86c6776d3634c5a66170ea6834a3f3bedf5582fd27
+  metadata.gz: 3555261e4b8280e6967bd63900dec75bb19d2225ff2da16748136ee7ddcb076fcee4f04b6502801afd92d344ef67b578c692868e63c3c181dff8008efc472986
+  data.tar.gz: 590b8d12f49526bd3965d78f2000b4d7e64452e0b026547285d0762f6fb0565311ffb00c595aaac85813c90658656842aca1a8970e25f88f20e4f6d09e6ddeb8

data/lib/canon/html/data_model.rb

@@ -50,11 +50,62 @@ module Canon
         root
       end
 
+      # Elements whose inner whitespace is always significant.
+      WHITESPACE_PRESERVING = %w[pre textarea script style].freeze
+
       # Deterministic single-line engine formatting for the :format
-      # string preprocessor — parse + re-serialize as a pure function
-      # of the parsed tree.
+      # string preprocessor: parse + strip whitespace-only text that
+      # is insignificant between block siblings + re-serialize — a
+      # pure function of the parsed tree, so equivalent inputs
+      # converge to identical bytes regardless of source indentation.
+      # Inline-significant whitespace (between two inline siblings,
+      # inside whitespace-preserving elements) survives.
       def self.engine_html_format(html)
-        Canon::XmlParsing.moxml_context.parse_html(html).to_html
+        doc = engine_html_parse(html)
+        strip_insignificant_whitespace!(doc.root)
+        doc.to_html
+      end
+
+      # Parse a full HTML document through the engine WHATWG lane.
+      def self.engine_html_parse(html)
+        Canon::XmlParsing.moxml_context.parse_html(html)
+      end
+
+      def self.strip_insignificant_whitespace!(node)
+        return unless node.is_a?(Moxml::Element)
+
+        preserve_parent = WHITESPACE_PRESERVING.include?(node.name)
+        node.children.to_a.each do |child|
+          if child.is_a?(Moxml::Text)
+            next unless child.content.strip.empty?
+            next if preserve_parent
+            next if inline_significant?(child, node)
+
+            child.remove
+          elsif child.is_a?(Moxml::Element)
+            strip_insignificant_whitespace!(child)
+          end
+        end
+      end
+
+      # Whitespace between two inline siblings (or anchored by content)
+      # is significant — the #112 rule. Edge or block-anchored runs are
+      # formatting artifacts.
+      def self.inline_significant?(text, parent)
+        siblings = parent.children.to_a
+        idx = siblings.index(text)
+        prev = siblings[0...idx].reverse_each.find { |s| !whitespace_only?(s) }
+        nxt = siblings[(idx + 1)..].find { |s| !whitespace_only?(s) }
+        inline_node?(prev) && inline_node?(nxt)
+      end
+
+      def self.whitespace_only?(node)
+        node.is_a?(Moxml::Text) && node.content.strip.empty?
+      end
+
+      def self.inline_node?(node)
+        !node.nil? && node.is_a?(Moxml::Element) &&
+          Canon::Comparison::WhitespaceSensitivity::INLINE_ELEMENTS.include?(node.name)
       end
 
       # Alias for compatibility

data/lib/canon/pretty_printer/html.rb

@@ -1,6 +1,5 @@
 # frozen_string_literal: true
 
-Canon::NokogiriLoader.require!("HTML pretty-printing")
 require "stringio"
 
 module Canon
@@ -37,7 +36,14 @@ module Canon
       end
 
       def format(html_string)
-        return format_fixture_ready(html_string) if @fixture_ready
+        if @fixture_ready
+          # The fixture-ready display mode is the one Nokogiri lane
+          # left (libxml FORMAT|AS_XHTML save flags — moxml#390 filed
+          # for an engine XHTML mode); everything else is engine-
+          # served.
+          Canon::NokogiriLoader.require!("HTML fixture-ready display")
+          return format_fixture_ready(html_string)
+        end
 
         if xhtml?(html_string)
           format_as_xhtml(html_string)
@@ -73,13 +79,15 @@ module Canon
         expand_non_void_self_closing(out)
       end
 
+      # Engine serialization with the #354 indent knobs — the engine's
+      # own nested layout (a deliberate byte change vs the Nokogiri
+      # lane: synthesis shape differs, e.g. no empty <head></head>).
       def format_as_html(html_string)
-        doc = Nokogiri::HTML5(html_string)
-
+        doc = Canon::Html::DataModel.engine_html_parse(html_string)
         if @indent_type == "tab"
-          doc.to_html(indent: 1, indent_text: "\t", encoding: "UTF-8")
+          doc.to_html(indent: 1, indent_text: "\t")
         else
-          doc.to_html(indent: @indent, encoding: "UTF-8")
+          doc.to_html(indent: @indent)
         end
       end
 

data/lib/canon/validators/html_validator.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-Canon::NokogiriLoader.require!("HTML validation")
-
 module Canon
   module Validators
     # Validator for HTML input
@@ -44,27 +42,43 @@ module Canon
       # @param input [String] The XHTML string to validate
       # @raise [Canon::ValidationError] If XHTML is malformed
       # @return [void]
+      # XHTML is XML: the engine strict parse is the validator. The
+      # engine reports the failure position as a byte offset embedded
+      # in the error message; line/column derive from it.
       def self.validate_xhtml!(input)
-        Nokogiri::XML(input) do |config|
-          config.strict.nonet
-        end
-      rescue Nokogiri::XML::SyntaxError => e
-        location = extract_location(e)
+        Canon::XmlParsing.moxml_context.parse(input, strict: true)
+      rescue Moxml::ParseError => e
+        location = extract_engine_location(e, input)
         raise Canon::ValidationError.new(
           e.message.split("\n").first,
           format: :html,
           line: location[:line],
           column: location[:column],
-          details: "XHTML validation failed: #{extract_details(e)}",
+          details: "XHTML validation failed",
         )
       end
 
+      # Derive line/column from the engine error's "at byte N" offset.
+      def self.extract_engine_location(error, input)
+        byte = error.message[/at byte (\d+)/, 1].to_i
+        before = input.byteslice(0, byte) || ""
+        line = before.count("\n") + 1
+        last_nl = before.rindex("\n")
+        column = last_nl ? byte - (last_nl + 1) : byte
+        { line: line, column: column }
+      end
+
       # Validate HTML5 input
       #
       # @param input [String] The HTML5 string to validate
       # @raise [Canon::ValidationError] If HTML5 is malformed
       # @return [void]
       def self.validate_html5!(input)
+        # The HTML5 lint layer stays on Nokogiri/gumbo: the WHATWG
+        # parser never fails, so the engine has no parse-error token
+        # stream to classify (filed as leptris#1771 - the ask carries
+        # canon's acceptance protocol).
+        Canon::NokogiriLoader.require!("HTML5 validation (leptris#1771)")
         doc = Nokogiri::HTML5(input, max_errors: 100)
 
         # Check for parse errors
@@ -126,7 +140,7 @@ module Canon
 
       private_class_method :xhtml?, :validate_xhtml!, :validate_html5!,
                            :extract_details, :build_error_details,
-                           :doctype_or_warning?
+                           :doctype_or_warning?, :extract_engine_location
     end
   end
 end

data/lib/canon/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Canon
-  VERSION = "0.3.99"
+  VERSION = "0.3.100"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: canon
 version: !ruby/object:Gem::Version
-  version: 0.3.99
+  version: 0.3.100
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-10-18 00:00:00.000000000 Z
+date: 2026-10-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: diff-lcs