moxml 0.5.128 → 0.5.129

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5b80e9601eb4a227d342aa484359a4724f3273a561c9f74cd8d375ffd5321186
-  data.tar.gz: 987e9faa827418fe3471456b1b939ced916467c895454670c3ba1273ee5fd616
+  metadata.gz: 9f040a5d5fc3e041e15a54cc42aeb10ca5d3cc882ecc22ca97faa7d45cbbb2d5
+  data.tar.gz: de59fd55dd4eff18451ddec5a20f246325be236313a55d0aeb4cab5af56ce4f9
 SHA512:
-  metadata.gz: 0f059f790d6a76ba2213d83ced97289d837ab0ad1b4448dd1f026bb991a5e90b8131fabc31d6f2f0eeb948564693231ea89c3899a542618d65de153db687bde7
-  data.tar.gz: edae98439cf51d03ea590709a22fa05848f9e7b62fa7f274bf6e1ffdd34f0115eca0622b165e3f1a884d8f9a73ba4047338061d6281c4c88161ce13bbd44a7ae
+  metadata.gz: 668587c18c1bf04415c99e715460b1dece9f5619206d6a7fb1ffa8260a32be81f6faa20853eeca322b3b016523a589550711647362f03ff79f64ccc507ef8e03
+  data.tar.gz: 7ff1f188edb97ee6f36c0294745232f1e1a4078bd6e94d567d56e649c01f5939617b345ebed2fe5b21bf325b6be1863f68b5af1acfb7171bd63b0a4bef58bd2a

data/lib/moxml/adapter/leptris/document_parts.rb

@@ -129,15 +129,37 @@ module Moxml
           end
 
           doctype = attachments.get(doc, :doctype)
-          parts << doctype.to_xml << "\n" if doctype.is_a?(CustomizedLeptris::Doctype)
-
-          native = native_doctype_xml(doc)
-          parts << native << "\n" if native
+          if doctype.is_a?(CustomizedLeptris::Doctype)
+            parts << doctype.to_xml << "\n"
+          else
+            # #364: the same DOCTYPE renders ONCE — the attachment
+            # is the parse-time capture; native_doctype_xml re-reads
+            # the engine's own doctype, and both firing doubled the
+            # line (moxml#364's \n\n\n).
+            native = native_doctype_xml(doc)
+            parts << native << "\n" if native
+          end
 
           # The libxml2-model document node: prolog PIs/comments, the
           # root, epilog PIs/comments — in document order, so epilog
           # parts serialize after the root (issue #130).
-          doc.children.each { |child| parts << raw_serialize(child, options) << "\n" }
+          # #364: document-level WHITESPACE-only text is layout
+          # residue (the input's newline after the doctype), not
+          # content — the newline-terminated part shape already
+          # carries the separation; rendering the text node stacked
+          # a second (or third) newline.
+          doc.children.each do |child|
+            # #364: document-level WHITESPACE-only text is layout
+            # residue (the input's newline after the doctype) — the
+            # newline-terminated part shape already carries the
+            # separation. Binding nodes walk here (native children).
+            if child.is_a?(::Leptris::XML::Text) &&
+                child.content.match?(/\A\s*\z/)
+              next
+            end
+
+            parts << raw_serialize(child, options) << "\n"
+          end
 
           texts = attachments.get(doc, :document_text)
           texts&.each { |text| parts << XmlEmitter.escape_text(text.content.to_s) }

data/lib/moxml/adapter/nokogiri.rb

@@ -160,7 +160,20 @@ module Moxml
         # malformed input never raises.
         def parse_html(html, _options = {}, _context = nil)
           html_string = html.is_a?(IO) || html.is_a?(StringIO) ? html.read : html.to_s
+          # #364: preserve-input-only — libxml2's HTML parser
+          # synthesizes an HTML 4.0 doctype when the input has none
+          # (leptris doesn't); dropping the synthesized node keeps
+          # the shared contract doctype-faithful. A doctype before
+          # the root (after optional BOM/xml-decl/whitespace) means
+          # the source had one.
+          had_doctype = html_string.match?(
+            /\A(\xEF\xBB\xBF|\s|<\?xml[^>]*\?>)*<!DOCTYPE/i,
+          )
           native_doc = ::Nokogiri::HTML(html_string, &:huge)
+          if !had_doctype &&
+              (dtd = native_doc.children.find { |n| n.is_a?(::Nokogiri::XML::DTD) })
+            dtd.unlink
+          end
           Wrappers::Document.new(native_doc, _context || Context.new(:nokogiri))
         end
 

data/lib/moxml/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Moxml
-  VERSION = "0.5.128"
+  VERSION = "0.5.129"
 end

data/spec/consistency/html_doctype_spec.rb

@@ -0,0 +1,44 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+# #364: DOCTYPE semantics converge across HTML-capable adapters —
+# preserve-input-only (never synthesize: libxml2's HTML parser
+# invents an HTML 4.0 doctype, which the nokogiri adapter drops),
+# and the post-DOCTYPE separator is one newline (the leptris
+# document-part walk used to stack the input's whitespace residue
+# on top of the part newlines).
+RSpec.describe "HTML DOCTYPE convergence (#364)" do
+  let(:body) do
+    "<html><head><title>t</title></head><body><p>x</p></body></html>"
+  end
+
+  %i[leptris nokogiri].each do |adapter_name|
+    describe "on #{adapter_name}" do
+      let(:ctx) { Moxml.new(adapter_name) }
+
+      it "never synthesizes a DOCTYPE the input lacked" do
+        out = ctx.parse_html(body).to_s
+        expect(out).not_to include("<!DOCTYPE")
+        expect(out).to include("<html")
+      end
+
+      it "preserves the input DOCTYPE with a single separator" do
+        out = ctx.parse_html("<!DOCTYPE html>\n#{body}").to_s
+        expect(out).to start_with("<!DOCTYPE html>\n")
+        expect(out.scan("<!DOCTYPE").size).to eq(1)
+        expect(out).not_to include("\n\n\n")
+      end
+
+      it "preserves public/system identifiers verbatim" do
+        src =
+          '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" ' \
+          '"http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">' \
+          "\n#{body}"
+        out = ctx.parse_html(src).to_s
+        expect(out).to include("XHTML 1.0 Strict")
+        expect(out.scan("<!DOCTYPE").size).to eq(1)
+      end
+    end
+  end
+end

data/spec/moxml/to_html_spec.rb

@@ -13,20 +13,30 @@ RSpec.describe "Moxml::Document#to_html" do
   end
 
   describe "nokogiri" do
+    # The oracle mirrors the adapter's #364 preserve-input-only
+    # parse: the synthesized HTML 4.0 doctype is dropped the same
+    # way, so pass-through equality stays meaningful.
+    let(:oracle) do
+      d = Nokogiri::HTML(src)
+      if (dtd = d.children.find { |n| n.is_a?(Nokogiri::XML::DTD) })
+        dtd.unlink
+      end
+      d
+    end
     let(:doc) { Moxml.new(:nokogiri).parse_html(src) }
 
     it "serializes the engine's HTML bytes by default" do
-      expect(doc.to_html).to eq(Nokogiri::HTML(src).to_html)
+      expect(doc.to_html).to eq(oracle.to_html)
     end
 
     it "threads indent through to the engine" do
       expect(doc.to_html(indent: 2))
-        .to eq(Nokogiri::HTML(src).to_html(indent: 2))
+        .to eq(oracle.to_html(indent: 2))
     end
 
     it "threads the indent unit through to the engine" do
       expect(doc.to_html(indent: 1, indent_text: "\t"))
-        .to eq(Nokogiri::HTML(src).to_html(indent: 1, indent_text: "\t"))
+        .to eq(oracle.to_html(indent: 1, indent_text: "\t"))
     end
   end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: moxml
 version: !ruby/object:Gem::Version
-  version: 0.5.128
+  version: 0.5.129
 platform: ruby
 authors:
 - Ribose Inc.
@@ -312,6 +312,7 @@ files:
 - spec/consistency/README.md
 - spec/consistency/adapter_parity_spec.rb
 - spec/consistency/eol_normalization_spec.rb
+- spec/consistency/html_doctype_spec.rb
 - spec/consistency/round_trip_spec.rb
 - spec/consistency/xslt_spec.rb
 - spec/examples/README.md