resolves #3331 fix parsing of wrapped link and xref text PR (#3334)

4 files changed, 94 insertions, 6 deletions

diff --git a/CHANGELOG.adoc b/CHANGELOG.adoc
index e07d7fb3..8f6e5334 100644
--- a/CHANGELOG.adoc
+++ b/CHANGELOG.adoc
@@ -17,6 +17,7 @@ For a detailed view of what has changed, refer to the {uri-repo}/commits/master[
 
 Bug Fixes::
 
+  * Fix parsing of wrapped link and xref text, including when an attrlist signature is detected (#3331)
   * Restore deprecated writable number property on AbstractBlock
 
 Improvements::
diff --git a/features/xref.feature b/features/xref.feature
index e7f2bfde..5ddcf308 100644
--- a/features/xref.feature
+++ b/features/xref.feature
@@ -950,6 +950,63 @@ Feature: Cross References
           a< href='#_section_one' "The Premier Section"
       """
 
+    Scenario: Does not parse text of xref macro as attribute if no attributes found
+    Given the AsciiDoc source
+      """
+      == Section One
+
+      content
+
+      == Section Two
+
+      refer to xref:_section_one[Section One
+      = First Section]
+      """
+    When it is converted to html
+    Then the result should match the HTML structure
+      """
+      .sect1
+        h2#_section_one
+          |Section One
+        .sectionbody: .paragraph: p content
+      .sect1
+        h2#_section_two Section Two
+        .sectionbody: .paragraph: p
+          |refer to
+          a< href='#_section_one'
+            |Section One
+            |= First Section
+      """
+
+    Scenario: Does not parse formatted text of xref macro as attributes
+    Given the AsciiDoc source
+      """
+      == Section One
+
+      content
+
+      == Section Two
+
+      refer to xref:_section_one[[.role]#Section
+      One#]
+      """
+    When it is converted to html
+    Then the result should match the HTML structure
+      """
+      .sect1
+        h2#_section_one
+          |Section One
+        .sectionbody: .paragraph: p content
+      .sect1
+        h2#_section_two Section Two
+        .sectionbody: .paragraph: p
+          |refer to
+          a< href='#_section_one'
+            span.role
+              |Section
+              |One
+      """
+
     Scenario: Can escape double quotes in text of xref macro using backslashes when text is parsed as attributes
     Given the AsciiDoc source
       """
diff --git a/lib/asciidoctor/substitutors.rb b/lib/asciidoctor/substitutors.rb
index 5f4d21bb..e47f0292 100644
--- a/lib/asciidoctor/substitutors.rb
+++ b/lib/asciidoctor/substitutors.rb
@@ -591,7 +591,8 @@ module Substitutors
         unless text.empty?
           text = text.gsub ESC_R_SB, R_SB if text.include? R_SB
           if !doc.compat_mode && (text.include? '=')
-            text = (attrs = (AttributeList.new text, self).parse)[1] || ''
+            # NOTE if an equals sign (=) is present, extract attributes from text
+            text, attrs = extract_attributes_from_text text, ''
             link_opts[:id] = attrs['id']
           end
 
@@ -637,7 +638,8 @@ module Substitutors
           text = text.gsub ESC_R_SB, R_SB if text.include? R_SB
           if mailto
             if !doc.compat_mode && (text.include? ',')
-              text = (attrs = (AttributeList.new text, self).parse)[1] || ''
+              # NOTE if a comma (,) is present, extract attributes from text
+              text, attrs = extract_attributes_from_text text, ''
               link_opts[:id] = attrs['id']
               if attrs.key? 2
                 if attrs.key? 3
@@ -648,7 +650,8 @@ module Substitutors
               end
             end
           elsif !doc.compat_mode && (text.include? '=')
-            text = (attrs = (AttributeList.new text, self).parse)[1] || ''
+            # NOTE if an equals sign (=) is present, extract attributes from text
+            text, attrs = extract_attributes_from_text text, ''
             link_opts[:id] = attrs['id']
           end
 
@@ -739,8 +742,8 @@ module Substitutors
           refid = $2
           if (text = $3)
             text = text.gsub ESC_R_SB, R_SB if text.include? R_SB
-            # NOTE if an equal sign (=) is present, parse text as attributes
-            text = ((AttributeList.new text, self).parse_into attrs)[1] if !doc.compat_mode && (text.include? '=')
+            # NOTE if an equals sign (=) is present, extract attributes from text
+            text, attrs = extract_attributes_from_text text if !doc.compat_mode && (text.include? '=')
           end
         end
 
@@ -1321,6 +1324,19 @@ module Substitutors
 
   private
 
+  # This method is used in cases when the attrlist can be mixed with the text of a macro.
+  # If no attributes are detected aside from the first positional attribute, and the first positional
+  # attribute matches the attrlist, then the original text is returned.
+  def extract_attributes_from_text text, default_text = nil
+    attrlist = (text.include? LF) ? (text.tr LF, ' ') : text
+    if (resolved_text = (attrs = (AttributeList.new attrlist, self).parse)[1])
+      # NOTE if resolved text remains unchanged, clear attributes and return unparsed text
+      resolved_text == attrlist ? [text, attrs.clear] : [resolved_text, attrs]
+    else
+      [default_text, attrs]
+    end
+  end
+
   # Internal: Extract the callout numbers from the source to prepare it for syntax highlighting.
   def extract_callouts source
     callout_marks = {}
diff --git a/test/links_test.rb b/test/links_test.rb
index 56ddd1e5..30ae3774 100644
--- a/test/links_test.rb
+++ b/test/links_test.rb
@@ -212,6 +212,10 @@ context 'Links' do
     assert_xpath '//a[@href="http://search.example.com"][text()="Google, Yahoo, Bing = Search Engines"]', convert_string_to_embedded('http://search.example.com["Google, Yahoo, Bing = Search Engines"]'), 1
   end
 
+  test 'should leave link text as is if it contains an equals sign but no attributes are found' do
+    assert_xpath %(//a[@href="https://example.com"][text()="What You Need\n= What You Get"]), convert_string_to_embedded(%(https://example.com[What You Need\n= What You Get])), 1
+  end
+
   test 'link with quoted text but no equal sign should carry quotes over to output' do
     assert_xpath %(//a[@href="http://search.example.com"][text()='"Google, Yahoo, Bing"']), convert_string_to_embedded('http://search.example.com["Google, Yahoo, Bing"]'), 1
   end
@@ -220,10 +224,20 @@ context 'Links' do
     assert_xpath '//a[@href="http://search.example.com"][text()="Google, Yahoo, Bing"]', convert_string_to_embedded('http://search.example.com[Google, Yahoo, Bing]'), 1
   end
 
-  test 'role and window attributes on link are processed' do
+  test 'link with formatted wrapped text should not be separated into attributes' do
+    result = convert_string_to_embedded %(https://example.com[[.role]#Foo\nBar#])
+    assert_include %(<a href="https://example.com"><span class="role">Foo\nBar</span></a>), result
+  end
+
+  test 'should process role and window attributes on link' do
     assert_xpath '//a[@href="http://google.com"][@class="external"][@target="_blank"]', convert_string_to_embedded('http://google.com[Google, role=external, window="_blank"]'), 1
   end
 
+  test 'should parse link with wrapped text that includes attributes' do
+    result = convert_string_to_embedded %(https://example.com[Foo\nBar,role=foobar])
+    assert_include %(<a href="https://example.com" class="foobar">Foo Bar</a>), result
+  end
+
   test 'link macro with attributes but no text should use URL as text' do
     url = 'https://fonts.googleapis.com/css?family=Roboto:400,400italic,'
     assert_xpath %(//a[@href="#{url}"][text()="#{url}"]), convert_string_to_embedded(%(link:#{url}[family=Roboto,weight=400])), 1