From ab4962c7dd5ce30be7220d52d71fddf73f9205e1 Mon Sep 17 00:00:00 2001
From: TSUYUSATO Kitsune <make.just.on@gmail.com>
Date: Thu, 30 May 2024 21:45:25 +0900
Subject: [PATCH] Optimize implementation

---
 .../css_selector/adapters/rexml_adapter.rb     | 10 ++++++----
 lib/rexml/css_selector/base_adapter.rb         |  8 ++++++--
 tool/bench.rb                                  | 18 +++++++++++-------
 3 files changed, 23 insertions(+), 13 deletions(-)

diff --git a/lib/rexml/css_selector/adapters/rexml_adapter.rb b/lib/rexml/css_selector/adapters/rexml_adapter.rb
index 2c47de8..6376f4e 100644
--- a/lib/rexml/css_selector/adapters/rexml_adapter.rb
+++ b/lib/rexml/css_selector/adapters/rexml_adapter.rb
@@ -35,7 +35,10 @@ def get_attribute(element, name, namespace = nil, attribute_name_case = :sensiti
 
           case attribute_name_case
           in :sensitive
-            element.attribute(name, namespace)&.value
+            attrs = element.attributes
+            # NOTE: `REXML::Element.attribute` is too slow to use.
+            # Therefore, we call `REXML::Attributes#[]` instead.
+            namespace.nil? ? attrs[name] : attrs.get_attribute_ns(namespace, name)&.value
           in :insensitive
             name = name.downcase(:ascii)
             target_attr = nil
@@ -65,9 +68,8 @@ def each_child_element(element, &)
           element.each_child { yield _1 if element?(_1) }
         end
 
-        def each_recursive_element(element, &)
-          element.each_recursive(&)
-        end
+        # NOTE: `REXML::Element#each_recursive` is too slow.
+        # Therefore, we use our default implementation instead.
 
         # INSTANCE is the default instance.
         INSTANCE = new
diff --git a/lib/rexml/css_selector/base_adapter.rb b/lib/rexml/css_selector/base_adapter.rb
index 3def7dd..3f1cdb7 100644
--- a/lib/rexml/css_selector/base_adapter.rb
+++ b/lib/rexml/css_selector/base_adapter.rb
@@ -78,9 +78,13 @@ def get_id(element)
 
       # Enumerates the elements in +element+
       def each_recursive_element(element, &)
-        each_child_element(element) do |child|
+        stack = []
+        each_child_element(element) { stack.unshift _1 }
+        until stack.empty?
+          child = stack.pop
           yield child
-          each_recursive_element(child, &)
+          n = stack.size
+          each_child_element(child) { stack.insert n, _1 }
         end
       end
     end
diff --git a/tool/bench.rb b/tool/bench.rb
index 0abc2eb..165d5eb 100644
--- a/tool/bench.rb
+++ b/tool/bench.rb
@@ -10,6 +10,7 @@
 
 filepath = Fixture.filepath("sizzle.html")
 selector = "h2, #qunit-fixture p"
+bench_rexml_xpath = false
 n = 1000
 
 opt = OptionParser.new
@@ -17,30 +18,33 @@
 opt.on("-f <filepath>") { filepath = _1 }
 opt.on("-s <selector>") { selector = _1 }
 opt.on("-n <n>") { n = _1.to_i }
-opt.on("-o <filepath>") { _1 }
+opt.on("--bench-rexml-xpath") { bench_rexml_xpath = true }
 
 puts "==> Parse command-line options"
 opt.parse!(ARGV)
 
 puts <<~HERE
-  filepath: #{filepath.inspect}
-  selector: #{selector.inspect}
-         n: #{n}
+           filepath: #{filepath.inspect}
+           selector: #{selector.inspect}
+                  n: #{n}
+  bench_rexml_xpath: #{bench_rexml_xpath}
   HERE
 
 puts "==> Load and parse a XML file"
 
 content = File.read(filepath)
 nokogiri_doc = Nokogiri.HTML(content)
-selector_xpath = Nokogiri::CSS.xpath_for(selector).join(" | ")
 rexml_doc = REXML::Document.new(content)
 
-puts "   XPath: #{selector_xpath}"
+if bench_rexml_xpath
+  selector_xpath = Nokogiri::CSS.xpath_for(selector).join(" | ")
+  puts "   XPath: #{selector_xpath}"
+end
 
 puts "==> Start a benchmark"
 
 Benchmark.bm do |x|
   x.report("Nokogiri          ") { n.times { nokogiri_doc.css(selector) } }
-  x.report("REXML (XPath)     ") { n.times { rexml_doc.get_elements(selector_xpath) } }
+  x.report("REXML (XPath)     ") { n.times { rexml_doc.get_elements(selector_xpath) } } if bench_rexml_xpath
   x.report("REXML::CSSSelector") { n.times { REXML::CSSSelector.select_all(rexml_doc, selector) } }
 end