[ruby/rdoc] Refine regexp usages and reduce substring allocations

https://github.com/ruby/rdoc/commit/a976fb9d39

3 files changed, 27 insertions, 31 deletions

diff --git a/lib/rdoc/generator/markup.rb b/lib/rdoc/generator/markup.rb
index 41e132450d..b54265717c 100644
--- a/lib/rdoc/generator/markup.rb
+++ b/lib/rdoc/generator/markup.rb
@@ -109,7 +109,7 @@ class RDoc::MethodAttr
     lines.shift if src =~ /\A.*#\ *File/i # remove '# File' comment
     lines.each do |line|
       if line =~ /^ *(?=\S)/
-        n = $&.length
+        n = $~.end(0)
         indent = n if n < indent
         break if n == 0
       end
diff --git a/lib/rdoc/markup/attribute_manager.rb b/lib/rdoc/markup/attribute_manager.rb
index 6ef5af8856..601e6bc189 100644
--- a/lib/rdoc/markup/attribute_manager.rb
+++ b/lib/rdoc/markup/attribute_manager.rb
@@ -2,6 +2,17 @@
 ##
 # Manages changes of attributes in a block of text
 
+unless MatchData.method_defined?(:match_length)
+  using Module.new {
+    refine(MatchData) {
+      def match_length(nth)
+        b, e = offset(nth)
+        e - b if b
+      end
+    }
+  }
+end
+
 class RDoc::Markup::AttributeManager
 
   ##
@@ -153,16 +164,17 @@ class RDoc::Markup::AttributeManager
     tags = "[#{tags.join("")}](?!#{PROTECT_ATTR})"
     all_tags = "[#{@matching_word_pairs.keys.join("")}](?!#{PROTECT_ATTR})"
 
-    re = /(^|\W|#{all_tags})(#{tags})(\2*[#\\]?[\w:#{PROTECT_ATTR}.\/\[\]-]+?\S?)\2(?!\2)(#{all_tags}|\W|$)/
+    re = /(?:^|\W|#{all_tags})\K(#{tags})(\1*[#\\]?[\w:#{PROTECT_ATTR}.\/\[\]-]+?\S?)\1(?!\1)(?=#{all_tags}|\W|$)/
 
     1 while str.gsub!(re) { |orig|
-      attr = @matching_word_pairs[$2]
-      attr_updated = attrs.set_attrs($`.length + $1.length + $2.length, $3.length, attr)
-      if attr_updated
-        $1 + NULL * $2.length + $3 + NULL * $2.length + $4
+      a, w = (m = $~).values_at(1, 2)
+      attr = @matching_word_pairs[a]
+      if attrs.set_attrs(m.begin(2), w.length, attr)
+        a = NULL * a.length
       else
-        $1 + NON_PRINTING_START + $2 + NON_PRINTING_END + $3 + NON_PRINTING_START + $2 + NON_PRINTING_END + $4
+        a = NON_PRINTING_START + a + NON_PRINTING_END
       end
+      a + w + a
     }
     str.delete!(NON_PRINTING_START + NON_PRINTING_END)
   end
@@ -173,9 +185,10 @@ class RDoc::Markup::AttributeManager
       @word_pair_map.each do |regexp, attr|
         next unless exclusive == exclusive?(attr)
         1 while str.gsub!(regexp) { |orig|
-          updated = attrs.set_attrs($`.length + $1.length, $2.length, attr)
+          w = (m = ($~))[2]
+          updated = attrs.set_attrs(m.begin(2), w.length, attr)
           if updated
-            NULL * $1.length + $2 + NULL * $3.length
+            NULL * m.match_length(1) + w + NULL * m.match_length(3)
           else
             orig
           end
@@ -194,9 +207,9 @@ class RDoc::Markup::AttributeManager
 
     1 while str.gsub!(/<(#{tags})>(.*?)<\/\1>/i) { |orig|
       attr = @html_tags[$1.downcase]
-      html_length = $1.length + 2
+      html_length = $~.match_length(1) + 2 # "<>".length
       seq = NULL * html_length
-      attrs.set_attrs($`.length + html_length, $2.length, attr)
+      attrs.set_attrs($~.begin(2), $~.match_length(2), attr)
       seq + $2 + seq + NULL
     }
   end
diff --git a/lib/rdoc/markup/to_rdoc.rb b/lib/rdoc/markup/to_rdoc.rb
index 3cdf4fd08b..2a9b05625c 100644
--- a/lib/rdoc/markup/to_rdoc.rb
+++ b/lib/rdoc/markup/to_rdoc.rb
@@ -330,31 +330,14 @@ class RDoc::Markup::ToRdoc < RDoc::Markup::Formatter
 
     text_len = 20 if text_len < 20
 
-    re = /^(.{0,#{text_len}})[ \n]/
     next_prefix = ' ' * @indent
 
     prefix = @prefix || next_prefix
     @prefix = nil
 
-    @res << prefix
-
-    while text.length > text_len
-      if text =~ re then
-        @res << $1
-        text.slice!(0, $&.length)
-      else
-        @res << text.slice!(0, text_len)
-      end
-
-      @res << "\n" << next_prefix
-    end
-
-    if text.empty? then
-      @res.pop
-      @res.pop
-    else
-      @res << text
-      @res << "\n"
+    text.scan(/\G(?:([^ \n]{#{text_len}})(?=[^ \n])|(.{1,#{text_len}})(?:[ \n]|\z))/) do
+      @res << prefix << ($1 || $2) << "\n"
+      prefix = next_prefix
     end
   end