Prevent style rendering in code blocks

Author: st0012

lib/rdoc/markup/attribute_manager.rb

@@ -95,6 +95,12 @@ def initialize
     add_html "b",  :BOLD, true
     add_html "tt",   :TT, true
     add_html "code", :TT, true
+
+    @word_pair_chars = @matching_word_pairs.keys.join
+
+    # Matches a word pair delimiter (*, _, +) that is NOT already protected.
+    # Used by #protect_code_markup to escape delimiters inside <code>/<tt> tags.
+    @unprotected_word_pair_regexp = /([#{@word_pair_chars}])(?!#{PROTECT_ATTR})/
   end
 
   ##
@@ -164,7 +170,7 @@ def convert_attrs_matching_word_pairs(str, attrs, exclusive)
     }.keys
     return if tags.empty?
     tags = "[#{tags.join("")}](?!#{PROTECT_ATTR})"
-    all_tags = "[#{@matching_word_pairs.keys.join("")}](?!#{PROTECT_ATTR})"
+    all_tags = "[#{@word_pair_chars}](?!#{PROTECT_ATTR})"
 
     re = /(?:^|\W|#{all_tags})\K(#{tags})(\1*[#\\]?[\w:#{PROTECT_ATTR}.\/\[\]-]+?\S?)\1(?!\1)(?=#{all_tags}|\W|$)/
 
@@ -245,6 +251,20 @@ def mask_protected_sequences
     @str.gsub!(/\\(\\[#{Regexp.escape @protectable.join}])/m, "\\1")
   end
 
+  ##
+  # Protects word pair delimiters (*, _, +) inside
+  # <code> and <tt> tags from being processed as inline formatting.
+  # For example, *bold* in +*bold*+ will NOT be rendered as bold.
+
+  def protect_code_markup
+    @str.gsub!(/<(code|tt)>(.*?)<\/\1>/im) do
+      tag = $1
+      content = $2
+      escaped = content.gsub(@unprotected_word_pair_regexp, "\\1#{PROTECT_ATTR}")
+      "<#{tag}>#{escaped}</#{tag}>"
+    end
+  end
+
   ##
   # Unescapes regexp handling sequences of text
 
@@ -308,6 +328,7 @@ def flow(str)
     @str = str.dup
 
     mask_protected_sequences
+    protect_code_markup
 
     @attrs = RDoc::Markup::AttrSpan.new @str.length, @exclusive_bitmap
 

test/rdoc/markup/attribute_manager_test.rb

@@ -202,6 +202,30 @@ def test_convert_attrs_ignores_code
     assert_equal 'foo <CODE>__send__</CODE> bar', output('foo <code>__send__</code> bar')
   end
 
+  def test_convert_attrs_ignores_bold_inside_code
+    assert_equal 'foo <CODE>*bold*</CODE> bar', output('foo <code>*bold*</code> bar')
+  end
+
+  def test_convert_attrs_ignores_em_inside_code
+    assert_equal 'foo <CODE>_em_</CODE> bar', output('foo <code>_em_</code> bar')
+  end
+
+  def test_convert_attrs_ignores_tt_inside_code
+    assert_equal 'foo <CODE>+tt+</CODE> bar', output('foo <code>+tt+</code> bar')
+  end
+
+  def test_convert_attrs_ignores_bold_inside_tt
+    assert_equal 'foo <CODE>*bold*</CODE> bar', output('foo <tt>*bold*</tt> bar')
+  end
+
+  def test_convert_attrs_ignores_em_inside_tt
+    assert_equal 'foo <CODE>_em_</CODE> bar', output('foo <tt>_em_</tt> bar')
+  end
+
+  def test_convert_attrs_ignores_tt_inside_tt
+    assert_equal 'foo <CODE>+tt+</CODE> bar', output('foo <tt>+tt+</tt> bar')
+  end
+
   def test_convert_attrs_ignores_tt
     assert_equal 'foo <CODE>__send__</CODE> bar', output('foo <tt>__send__</tt> bar')
   end

test/rdoc/rdoc_markdown_test.rb

@@ -1283,6 +1283,16 @@ def test_markdown_link_with_styled_label
     assert_includes html, '<a href="https://example.com">Link to <code>Foo</code> and <code>Bar</code> and <code>Baz</code></a>'
   end
 
+  def test_code_span_preserves_inline_formatting_chars
+    # Code spans should display formatting characters literally, not as styling
+    doc = parse "Code: `*bold*` and `_em_` and `+tt+`"
+    html = @to_html.convert doc
+
+    assert_includes html, '<code>*bold*</code>'
+    assert_includes html, '<code>_em_</code>'
+    assert_includes html, '<code>+tt+</code>'
+  end
+
   def parse(text)
     @parser.parse text
   end