Add Nokogiri XML parser (#55)

This one is similar to the LibXML parser, but uses the Nokogiri SAX
parser under the hood. This library is already used in a lot of web
projects and is easier to install in Windows than LibXML.
There is no direct dependency on the nokogiri gem, if the gem is not
available the code will only error in case it is used.

Unfortunately, I could not find a way to enable the gem in the CI for
the Windows tasks. It passes on the regular Windows runners, but fails
on mingw/mswin, and I could not find a way to separate those two. If
someone knows how to fix that, or how to install/compile Nokogiri on the
latter two platforms, please go ahead and fix it.

This has been suggested/requested in
https://bugs.ruby-lang.org/issues/9379, all the way back in 2014.

---------

Co-authored-by: Herwin <herwinw@users.noreply.github.com>

Author: herwinw

Gemfile

@@ -5,6 +5,7 @@ gemspec
 group :development do
   gem "bundler"
   gem "libxml-ruby", platforms: [:ruby, :jruby]
+  gem "nokogiri", platforms: [:ruby, :jruby]
   gem "rake"
   gem "test-unit"
 end

lib/xmlrpc.rb

@@ -59,9 +59,12 @@
 #   * libxml (LibXMLStreamParser)
 #     * Compiled
 #     * See https://rubygems.org/gems/libxml-ruby/
+#   * nokogiri (NokogiriStreamParser)
+#     * Compiled
+#     * See https://nokogiri.org
 #
 # * General
-#   * possible to choose between XMLParser module (Expat wrapper) and REXML (pure Ruby) parsers
+#   * possible to choose between REXML (pure Ruby) and LibXML/Nokogiri (compiled) parsers
 #   * Marshalling Ruby objects to Hashes and reconstruct them later from a Hash
 #   * SandStorm component architecture XMLRPC::Client interface
 #

lib/xmlrpc/config.rb

@@ -15,6 +15,7 @@ module Config
     #
     # * XMLParser::REXMLStreamParser
     # * XMLParser::LibXMLStreamParser
+    # * XMLParser::NokogiriStreamParser
     DEFAULT_PARSER = XMLParser::REXMLStreamParser
 
     # enable <code><nil/></code> tag

lib/xmlrpc/parser.rb

@@ -624,7 +624,51 @@ def parse(str)
       end
     end
 
-    Classes = [REXMLStreamParser, LibXMLStreamParser]
+    class NokogiriStreamParser < AbstractStreamParser
+      def initialize
+        require 'nokogiri'
+        @parser_class = NokogiriStreamListener
+      end
+
+      class NokogiriStreamListener
+        include StreamParserMixin
+
+        def self.handler
+          # We need to construct this on first use, since we cannot be sure Nokogiri is available
+          @handler ||= begin
+            Class.new(Nokogiri::XML::SAX::Document) do
+              def initialize(parent)
+                super()
+                @parent = parent
+              end
+
+              def start_element_namespace(name, attrs = [], prefix = nil, uri = nil, ns = [])
+                @parent.startElement(name, attrs)
+              end
+
+              def end_element_namespace(name, prefix = nil, uri = nil)
+                @parent.endElement(name)
+              end
+
+              def characters(string)
+                @parent.character(string)
+              end
+
+              def cdata_block(string)
+                @parent.character(string)
+              end
+            end
+          end
+        end
+
+        def parse(str)
+          parser = Nokogiri::XML::SAX::Parser.new(self.class.handler.new(self))
+          parser.parse(str)
+        end
+      end
+    end
+
+    Classes = [REXMLStreamParser, LibXMLStreamParser, NokogiriStreamParser]
 
     # yields an instance of each installed parser
     def self.each_installed_parser