Replace Nokogiri with REXML in the JUnit formatter

In #1454, we welcomed a newly-revamped JUnit formatter which has
a dependency on Nokogiri. Unfortunately, this had led us to problems
getting InSpec included in Chef omnibus builds (see chef/chef#5937)
because Chef is using Ruby 2.4.1 and the Nokogiri maintainers have
not yet released a windows binary gem that supports Ruby 2.4.x.
This has led to breaking builds in Chef's CI platform and would
block the acceptance of chef/chef#5937.

This change replaces Nokogiri use with REXML instead. While REXML
can be slower than Nokogiri, it does not require native extensions
and is supported on all Chef platforms.

Signed-off-by: Adam Leff <adam@leff.co>

habitat/plan.sh

@@ -51,18 +51,11 @@ do_build() {
   local _libxslt_dir
 
   _bundler_dir="$(pkg_path_for bundler)"
-  _libxml2_dir="$(pkg_path_for libxml2)"
-  _libxslt_dir="$(pkg_path_for libxslt)"
 
   export GEM_HOME=${pkg_path}/vendor/bundle
   export GEM_PATH=${_bundler_dir}:${GEM_HOME}
   export BUNDLE_SILENCE_ROOT_WARNING=1
 
-  # don't let bundler split up the nokogiri config string (it breaks
-  # the build), so specify it as an env var instead
-  export NOKOGIRI_CONFIG="--use-system-libraries --with-zlib-dir=${_zlib_dir} --with-xslt-dir=${_libxslt_dir} --with-xml2-include=${_libxml2_dir}/include/libxml2 --with-xml2-lib=${_libxml2_dir}/lib"
-  bundle config build.nokogiri "${NOKOGIRI_CONFIG}"
-
   bundle install --jobs "$(nproc)" --retry 5 --standalone \
     --path "$pkg_prefix/bundle" \
     --binstubs "$pkg_prefix/bin"

inspec.gemspec

@@ -39,7 +39,6 @@ Gem::Specification.new do |spec|
   spec.add_dependency 'mixlib-log'
   spec.add_dependency 'sslshake', '~> 1'
   spec.add_dependency 'parallel', '~> 1.9'
-  spec.add_dependency 'nokogiri', '~> 1.6'
   spec.add_dependency 'faraday', '>=0.9.0'
   spec.add_dependency 'toml', '~> 0.1'
   spec.add_dependency 'addressable', '~> 2.5'

lib/inspec/rspec_json_formatter.rb

@@ -795,51 +795,61 @@ class InspecRspecJUnit < InspecRspecJson
 
   #
   # This is the last method is invoked through the formatter interface.
-  # Converts the junit formatter constructed output_hash into nokogiri generated
+  # Converts the junit formatter constructed output_hash into REXML generated
   # XML and writes it to output.
   #
   def close(_notification)
-    require 'nokogiri'
-    xml_output = Nokogiri::XML::Builder.new { |xml|
-      xml.testsuites do
-        @output_hash[:profiles].each do |profile|
-          build_profile_xml(xml, profile)
-        end
-      end
-    }.to_xml
-    output.puts xml_output
+    require 'rexml/document'
+    xml_output = REXML::Document.new
+    xml_output.add(REXML::XMLDecl.new)
+
+    testsuites = REXML::Element.new('testsuites')
+    xml_output.add(testsuites)
+
+    @output_hash[:profiles].each do |profile|
+      testsuites.add(build_profile_xml(profile))
+    end
+
+    formatter = REXML::Formatters::Pretty.new
+    formatter.compact = true
+    output.puts formatter.write(xml_output.xml_decl, '')
+    output.puts formatter.write(xml_output.root, '')
   end
 
   private
 
-  def build_profile_xml(xml, profile)
-    xml.testsuite(
-      name: profile[:name],
-      tests: count_profile_tests(profile),
-      failed: count_profile_failed_tests(profile),
-    ) do
-      profile[:controls].each do |control|
-        build_control_xml(xml, control)
+  def build_profile_xml(profile)
+    profile_xml = REXML::Element.new('testsuite')
+    profile_xml.add_attribute('name', profile[:name])
+    profile_xml.add_attribute('tests', count_profile_tests(profile))
+    profile_xml.add_attribute('failed', count_profile_failed_tests(profile))
+
+    profile[:controls].each do |control|
+      next if control[:results].nil?
+
+      control[:results].each do |result|
+        profile_xml.add(build_result_xml(control, result))
       end
     end
+
+    profile_xml
   end
 
-  def build_control_xml(xml, control)
-    return if control[:results].nil?
-    control[:results].each do |result|
-      build_result_xml(xml, control, result)
-    end
-  end
+  def build_result_xml(control, result)
+    result_xml = REXML::Element.new('testcase')
+    result_xml.add_attribute('name', result[:code_desc])
+    result_xml.add_attribute('class', control[:title].nil? ? 'Anonymous' : control[:id])
+    result_xml.add_attribute('time', result[:run_time])
 
-  def build_result_xml(xml, control, result)
-    test_class = control[:title].nil? ? 'Anonymous' : control[:id]
-    xml.testcase(name: result[:code_desc], class: test_class, time: result[:run_time]) do
-      if result[:status] == 'failed'
-        xml.failure(message: result[:message])
-      elsif result[:status] == 'skipped'
-        xml.skipped
-      end
+    if result[:status] == 'failed'
+      failure_element = REXML::Element.new('failure')
+      failure_element.add_attribute('message', result[:message])
+      result_xml.add(failure_element)
+    elsif result[:status] == 'skipped'
+      result_xml.add_element('skipped')
     end
+
+    result_xml
   end
 
   def count_profile_tests(profile)

omnibus/config/software/inspec.rb

@@ -23,7 +23,6 @@ dependency 'ruby'
 dependency 'rubygems'
 dependency 'bundler'
 dependency 'rb-readline'
-dependency 'nokogiri'
 dependency 'appbundler'
 
 license :project_license

test/functional/inspec_exec_junit_test.rb

@@ -2,7 +2,7 @@
 # author: John Kerry
 
 require 'functional/helper'
-require 'nokogiri'
+require 'rexml/document'
 
 describe 'inspec exec with junit formatter' do
   include FunctionalHelper
@@ -11,47 +11,47 @@ describe 'inspec exec with junit formatter' do
     out = inspec('exec ' + example_control + ' --format junit --no-create-lockfile')
     out.stderr.must_equal ''
     out.exit_status.must_equal 0
-    doc = Nokogiri::XML(out.stdout)
-    doc.errors.length.must_equal 0
+    doc = REXML::Document.new(out.stdout)
+    doc.has_elements?.must_equal true
   end
 
   it 'can execute the profile with the junit formatter' do
     out = inspec('exec ' + example_profile + ' --format junit --no-create-lockfile')
     out.stderr.must_equal ''
     out.exit_status.must_equal 0
-    doc = Nokogiri::XML(out.stdout)
-    doc.errors.length.must_equal 0
+    doc = REXML::Document.new(out.stdout)
+    doc.has_elements?.must_equal true
   end
 
   describe 'execute a profile with junit formatting' do
-    let(:doc) { Nokogiri::XML(inspec('exec ' + example_profile + ' --format junit --no-create-lockfile').stdout) }
+    let(:doc) { REXML::Document.new(inspec('exec ' + example_profile + ' --format junit --no-create-lockfile').stdout) }
 
     describe 'the document' do
       it 'has only one testsuite' do
-        doc.xpath("//testsuite").length.must_equal 1
+        doc.elements.to_a("//testsuite").length.must_equal 1
       end
     end
     describe 'the test suite' do
-      let(:suite) { doc.xpath("//testsuites/testsuite").first}
+      let(:suite) { doc.elements.to_a("//testsuites/testsuite").first }
 
       it 'must have 5 testcase children' do
-        suite.xpath("//testcase").length.must_equal 5
+        suite.elements.to_a("//testcase").length.must_equal 5
       end
 
       it 'has the tests attribute with 5 total tests' do
-        suite["tests"].must_equal "5"
+        suite.attribute('tests').value.must_equal "5"
       end
 
       it 'has the failures attribute with 0 total tests' do
-        suite["failed"].must_equal "0"
+        suite.attribute('failed').value.must_equal "0"
       end
 
       it 'has 2 elements named "File /tmp should be directory"' do
-        suite.xpath("//testcase[@name='File /tmp should be directory']").length.must_equal 2
+        REXML::XPath.match(suite, "//testcase[@name='File /tmp should be directory']").length.must_equal 2
       end
 
       describe 'the testcase named "gordon_config Can\'t find file ..."' do
-        let(:gordon_yml_tests) { suite.xpath("//testcase[@class='gordon-1.0' and @name='gordon_config']") }
+        let(:gordon_yml_tests) { REXML::XPath.match(suite, "//testcase[@class='gordon-1.0' and @name='gordon_config']") }
         let(:first_gordon_test) {gordon_yml_tests.first}
 
         it 'should be unique' do
@@ -59,7 +59,7 @@ describe 'inspec exec with junit formatter' do
         end
 
         it 'should be skipped' do
-          first_gordon_test.xpath("//skipped").length.must_equal 1
+          first_gordon_test.elements.to_a('//skipped').length.must_equal 1
         end
       end
     end

www/Gemfile.lock

@@ -8,7 +8,6 @@ PATH
       json (>= 1.8, < 3.0)
       method_source (~> 0.8)
       mixlib-log
-      nokogiri (~> 1.6)
       parallel (~> 1.9)
       pry (~> 0)
       rainbow (~> 2)
@@ -56,7 +55,7 @@ GEM
     concurrent-ruby (1.0.5)
     contracts (0.13.0)
     diff-lcs (1.3)
-    docker-api (1.33.2)
+    docker-api (1.33.3)
       excon (>= 0.38.0)
       json
     dotenv (2.2.0)
@@ -67,13 +66,12 @@ GEM
     eventmachine (1.2.3)
     excon (0.55.0)
     execjs (2.7.0)
-    faraday (0.11.0)
+    faraday (0.12.0.1)
       multipart-post (>= 1.2, < 3)
     fast_blank (1.0.0)
     fastimage (2.1.0)
     ffi (1.9.18)
-    github-markup (1.5.0)
-      rinku
+    github-markup (1.6.0)
     gssapi (1.2.0)
       ffi (>= 1.0.1)
     gyoku (1.3.1)
@@ -148,7 +146,6 @@ GEM
     middleman-syntax (3.0.0)
       middleman-core (>= 3.2)
       rouge (~> 2.0)
-    mini_portile2 (2.1.0)
     minitest (5.10.1)
     mixlib-log (1.7.1)
     mixlib-shellout (2.2.7)
@@ -157,8 +154,6 @@ GEM
     net-scp (1.2.1)
       net-ssh (>= 2.6.5)
     net-ssh (4.1.0)
-    nokogiri (1.7.1)
-      mini_portile2 (~> 2.1.0)
     nori (2.6.0)
     padrino-helpers (0.13.3.3)
       i18n (~> 0.6, >= 0.6.7)
@@ -183,7 +178,6 @@ GEM
     rb-inotify (0.9.8)
       ffi (>= 0.5.0)
     redcarpet (3.4.0)
-    rinku (2.0.2)
     rouge (2.0.7)
     rspec (3.5.0)
       rspec-core (~> 3.5.0)
@@ -233,7 +227,7 @@ GEM
       winrm-fs (~> 1.0)
     tzinfo (1.2.3)
       thread_safe (~> 0.1)
-    uglifier (3.1.11)
+    uglifier (3.1.13)
       execjs (>= 0.3.0, < 3)
     winrm (2.1.3)
       builder (>= 2.1.2)