Keep link targets out of meta description excerpts

lib/rdoc/generator/darkfish.rb

@@ -714,6 +714,17 @@ def template_for(file, page = true, klass = ERB)
       ParagraphExcerptRegexpOther = %r[\b\w[^./:]++\.]
       # use \p/\P{letter} instead of \w/\W in Unicode
       ParagraphExcerptRegexpUnicode = %r[\b\p{letter}[^./:]++\.]
+
+      # Link markup to replace with its text (or remove) before taking an excerpt
+      ExcerptMarkdownLinks = [
+        [/(?<![\w`])!\[[^\]]*\]\([^)]*\)/, ''],    # ![alt](image)
+        [/(?<![\w`])\[([^\]]*)\]\([^)]*\)/, '\\1'], # [text](url)
+      ]
+      ExcerptRDocLinks = [
+        [/\{([^}]*)\}\[[^\]]*\]/, '\\1'],  # {text}[url]
+        [/(\S+)\[(?:https?|ftp|mailto|irc|www|link|rdoc-[a-z]+):[^\]]*\]/, '\\1'], # text[url]
+      ]
+      ExcerptURL = %r[<?\b(?:https?|ftp)://[^\s>]*[^\s>.,;:!?)]>?]
       # :startdoc:
 
       # Returns an excerpt of the comment for usage in meta description tags
@@ -725,6 +736,8 @@ def excerpt(comment)
           comment
         end
 
+        text = strip_links(text, comment.respond_to?(:format) ? comment.format : nil)
+
         # Match from a capital letter to the first period, discarding any links, so
         # that we don't end up matching badges in the README
         pattern = ParagraphExcerptRegexpUnicode
@@ -745,6 +758,15 @@ def excerpt(comment)
         extracted_text[0...150].tr_s("\n", " ").squeeze(" ")
       end
 
+      # Replaces links in +text+ written in +format+ markup with their text, and
+      # removes images and bare URLs, so that link targets don't end up in excerpts
+      def strip_links(text, format) # :nodoc:
+        links = format == 'markdown' ? ExcerptMarkdownLinks : ExcerptRDocLinks
+
+        text = links.inject(text) { |t, (regexp, replacement)| t.gsub(regexp, replacement) }
+        text.gsub(ExcerptURL, '')
+      end
+
       def generate_ancestor_list(ancestors, klass)
         return '' if ancestors.empty?
 

test/rdoc/generator/darkfish_test.rb

@@ -581,6 +581,49 @@ def test_meta_tags_for_markdown_files
     )
   end
 
+  def test_meta_tags_for_markdown_links
+    top_level = @store.add_file("file.rb")
+    klass = top_level.add_class(RDoc::NormalClass, "Linked")
+    comment = RDoc::Comment.new(<<~MARKDOWN, top_level)
+      Foo.
+
+      [Example Link](https://example.com) and ![Badge](https://example.com/badge.svg)
+    MARKDOWN
+    comment.format = "markdown"
+    klass.add_comment(comment, top_level)
+
+    @g.generate
+
+    content = File.binread("Linked.html")
+    assert_include(content, '<meta name="description" content="class Linked: Foo.">')
+  end
+
+  def test_meta_tags_for_rdoc_links
+    top_level = @store.add_file("file.rb")
+    klass = top_level.add_class(RDoc::NormalClass, "Linked")
+    klass.add_comment(<<~RDOC, top_level)
+      See {the docs}[https://example.com/docs.html] or RDoc[https://ruby.github.io/rdoc/] for more.
+    RDOC
+
+    @g.generate
+
+    content = File.binread("Linked.html")
+    assert_include(content, '<meta name="description" content="class Linked: See the docs or RDoc for more.">')
+  end
+
+  def test_meta_tags_for_bare_urls
+    top_level = @store.add_file("file.rb")
+    klass = top_level.add_class(RDoc::NormalClass, "Linked")
+    klass.add_comment(<<~RDOC, top_level)
+      Details are at https://example.com/docs. Use it.
+    RDOC
+
+    @g.generate
+
+    content = File.binread("Linked.html")
+    assert_include(content, '<meta name="description" content="class Linked: Details are at . Use it.">')
+  end
+
   def test_meta_tags_for_raw_pages
     top_level = @store.add_file("MyPage", parser: RDoc::Parser::Simple)
     comment = RDoc::Comment.new('this is a comment')