test: scripts to assert on packaged gem files and installed gems

some tests are skipped because they assert new behavior which will be
implemented in the next few commits.

Closes #2076
Closes #2078

scripts/test-gem-file-contents

@@ -0,0 +1,210 @@
+#! /usr/bin/env ruby
+# frozen_string_literal: true
+#
+#  this script is intended to run as part of the CI test suite.
+#
+#  it inspects the contents of a nokogiri gem file -- both the files and the gemspec -- to ensure
+#  we're packaging what we expect, and that we're not packaging anything we don't expect.
+#
+#  this file isn't in the `test/` subdirectory because it's intended to be run standalone against a
+#  built gem file (and not against the source code or behavior of the gem itself).
+#
+
+require "bundler/inline"
+
+gemfile do
+  source "https://rubygems.org"
+  gem "minitest"
+  gem "minitest-reporters"
+end
+
+require "yaml"
+
+def usage_and_exit(message = nil)
+  puts "ERROR: #{message}" if message
+  puts "USAGE: #{File.basename(__FILE__)} <gemfile> [options]"
+  exit(1)
+end
+
+usage_and_exit if ARGV.include?("-h")
+usage_and_exit unless gemfile = ARGV[0]
+usage_and_exit("#{gemfile} does not exist") unless File.file?(gemfile)
+usage_and_exit("#{gemfile} is not a gem") unless /\.gem$/.match(gemfile)
+gemfile = File.expand_path(gemfile)
+
+gemfile_contents = Dir.mktmpdir do |dir|
+  Dir.chdir(dir) do
+    unless system("tar -xf #{gemfile} data.tar.gz")
+      raise "could not unpack gem #{gemfile}"
+    end
+    %x(tar -ztf data.tar.gz).split("\n")
+  end
+end
+
+gemspec = Dir.mktmpdir do |dir|
+  Dir.chdir(dir) do
+    unless system("tar -xf #{gemfile} metadata.gz")
+      raise "could not unpack gem #{gemfile}"
+    end
+    YAML.load(%x(gunzip -c metadata.gz))
+  end
+end
+
+if ARGV.include?("-v")
+  puts "---------- gemfile contents ----------"
+  puts gemfile_contents
+  puts
+  puts "---------- gemspec ----------"
+  puts gemspec.to_ruby
+  puts
+end
+
+require "minitest/autorun"
+require "minitest/reporters"
+Minitest::Reporters.use!([Minitest::Reporters::SpecReporter.new])
+
+puts "Testing '#{gemfile}' (#{gemspec.platform})"
+describe File.basename(gemfile) do
+  let(:cross_rubies_path) { File.join(File.dirname(__FILE__), "..", ".cross_rubies") }
+  let(:native_ruby_versions) do
+    File.read(cross_rubies_path).split("\n").map do |line|
+      line.split(":").first.split(".").take(2).join(".") # ugh
+    end.uniq.sort
+  end
+
+  describe "setup" do
+    it "gemfile contains some files" do
+      actual = gemfile_contents.length
+      assert_operator(actual, :>, 60, "expected gemfile to contain more than #{actual} files")
+    end
+
+    it "gemspec is a Gem::Specfication" do
+      assert_equal(Gem::Specification, gemspec.class)
+    end
+  end
+
+  describe "all platforms" do
+    it "contains every ruby file in lib/" do
+      expected = %x(git ls-files lib).split("\n").grep(/\.rb$/).sort
+      actual = gemfile_contents.grep(%r{^lib/}).grep(/\.rb$/).sort
+      assert_equal(expected, actual)
+    end
+  end
+
+  describe "ruby platform" do
+    it "contains the port files" do
+      actual_ports = gemfile_contents.grep(%r{^ports/})
+      assert(actual_ports.grep(/libxml2-\d+\.\d+\.\d+\.tar\.gz/).length == 1,
+             "expected #{actual_ports} to include libxml2")
+      assert(actual_ports.grep(/libxslt-\d+\.\d+\.\d+\.tar\.gz/).length == 1,
+             "expected #{actual_ports} to include libxslt")
+      assert_equal(2, actual_ports.length)
+    end
+
+    it "contains the patch files" do
+      actual_patches = gemfile_contents.grep(%r{^patches/})
+      assert_operator(actual_patches.length, :>, 0)
+    end
+
+    it "does not contain header files in lib/nokogiri/include" do
+      # these are copied to lib/nokogiri/include during compilation, not packaged
+      assert_empty(gemfile_contents.grep(%r{^lib/.*\.h$}))
+    end
+
+    it "contains ext/nokogiri" do
+      assert_operator(gemfile_contents.grep(%r{^ext/nokogiri/.*\.[ch]}).length, :>, 20)
+    end
+
+    it "does not contain the java jar files" do
+      assert_empty(gemfile_contents.grep(/.*\.jar$/))
+    end
+
+    it "depends on mini_portile2" do
+      assert(gemspec.dependencies.find { |d| d.name == "mini_portile2" })
+    end
+
+    it "includes C files in extra_rdoc_files" do
+      assert_operator(gemspec.extra_rdoc_files.grep(%r{ext/nokogiri/.*\.c$}).length, :>, 10)
+    end
+  end if gemspec.platform == Gem::Platform::RUBY
+
+  describe "native platform" do
+    it "does not contain the port files" do
+      actual_ports = gemfile_contents.grep(%r{^ports/})
+      assert_empty(actual_ports)
+    end
+
+    it "does not contain the patch files" do
+      actual_patches = gemfile_contents.grep(%r{^patches/})
+      assert_empty(actual_patches)
+    end
+
+    it "contains nokogiri header files in lib/nokogiri/include" do
+      skip "until we fix https://github.com/sparklemotion/nokogiri/pull/1788"
+      assert_includes(gemfile_contents, "lib/nokogiri/include/nokogiri.h")
+    end
+
+    it "contains packaged libraries' header files in lib/nokogiri/include" do
+      skip "until we fix https://github.com/sparklemotion/nokogiri/pull/1788"
+      assert_includes(gemfile_contents, "lib/nokogiri/include/libxml2/libxml/tree.h")
+      assert_includes(gemfile_contents, "lib/nokogiri/include/libxslt/xslt.h")
+      assert_includes(gemfile_contents, "lib/nokogiri/include/libexslt/exslt.h")
+    end
+
+    it "does not contain ext/nokogiri" do
+      skip "until we fix https://github.com/sparklemotion/nokogiri/issues/2077"
+      assert_empty(gemfile_contents.grep(%r{^ext/nokogiri/}))
+    end
+
+    it "does not contain the java jar files" do
+      assert_empty(gemfile_contents.grep(/.*\.jar$/))
+    end
+
+    it "does not depend on mini_portile2" do
+      # https://github.com/sparklemotion/nokogiri/issues/2078
+      refute(gemspec.dependencies.find { |d| d.name == "mini_portile2" })
+    end
+
+    it "packages only ruby-version-specific libraries" do
+      # https://github.com/sparklemotion/nokogiri/issues/2076
+      native_ruby_versions.each do |version|
+        assert_includes(gemfile_contents, "lib/nokogiri/#{version}/nokogiri.so")
+      end
+      refute_includes(gemfile_contents, "lib/nokogiri/nokogiri.so")
+    end
+  end if gemspec.platform.is_a?(Gem::Platform) && gemspec.platform.cpu
+
+  describe "java platform" do
+    it "does not contain the port files" do
+      actual_ports = gemfile_contents.grep(%r{^ports/})
+      assert_empty(actual_ports)
+    end
+
+    it "does not contain the patch files" do
+      actual_patches = gemfile_contents.grep(%r{^patches/})
+      assert_empty(actual_patches)
+    end
+
+    it "does not contain header files in lib/nokogiri/include" do
+      skip "until we fix https://github.com/sparklemotion/nokogiri/pull/1788"
+      assert_empty(gemfile_contents.grep(%r{^lib/.*\.h$}))
+    end
+
+    it "does not contain ext/nokogiri" do
+      skip "until we fix https://github.com/sparklemotion/nokogiri/issues/2099"
+      assert_empty(gemfile_contents.grep(%r{^ext/nokogiri/}))
+    end
+
+    it "does contains the java jar files" do
+      actual_jars = gemfile_contents.grep(/.*\.jar$/)
+      ["nokogiri", "isorelax", "jing", "nekodtd", "nekohtml", "serializer", "xalan", "xercesImpl", "xml-apis"].each do |jar|
+        assert_equal(1, actual_jars.grep(%r{/#{jar}\.jar$}).length, "expected to contain #{jar}.jar")
+      end
+    end
+
+    it "does not depend on mini_portile2" do
+      # https://github.com/sparklemotion/nokogiri/issues/2078
+      refute(gemspec.dependencies.find { |d| d.name == "mini_portile2" })
+    end
+  end if gemspec.platform == Gem::Platform.new("java")
+end

scripts/test-gem-installation

@@ -0,0 +1,114 @@
+#! /usr/bin/env ruby
+# frozen_string_literal: true
+#
+#  this script is intended to run as part of the CI test suite.
+#
+#  it inspects the filesystem of a nokogiri gem installation to ensure it's complete and sane, and
+#  doesn't install anything we don't expect.
+#
+#  this file isn't in the `test/` subdirectory because it's intended to be run standalone against an
+#  installed gem (and not against the source code or behavior of the gem itself).
+#
+
+# this line needs to come before the bundler bit, to assert that we're running against an
+# already-installed version (and not some other version that bundler/inline might install if it came
+# first)
+gemspec = Gem::Specification.find_by_name('nokogiri')
+raise "could not find installed gem" unless gemspec
+
+require "bundler/inline"
+
+gemfile do
+  source "https://rubygems.org"
+  gem "minitest"
+  gem "minitest-reporters"
+  gem "nokogiri"
+end
+
+require 'nokogiri'
+require 'yaml'
+
+if ARGV.include?("-v")
+  puts "---------- Nokogiri version info ----------"
+  puts Nokogiri::VERSION_INFO.to_yaml
+  puts
+  puts "---------- Nokogiri installed gemspec ----------"
+  puts gemspec.to_ruby
+  puts
+end
+
+require "minitest/autorun"
+require "minitest/reporters"
+Minitest::Reporters.use!([Minitest::Reporters::SpecReporter.new])
+
+puts "Testing #{gemspec.full_name} installed in #{gemspec.base_dir}"
+describe gemspec.full_name do
+  let(:nokogiri_include_dir) { File.join(gemspec.gem_dir, "lib/nokogiri/include") }
+
+  # representative sample of the files
+  let(:nokogiri_header_files) { ["nokogiri.h", "xml_document.h", "xml_node.h"] }
+  let(:packaged_library_header_files) { ["libxml2/libxml/tree.h", "libxslt/xslt.h", "libexslt/exslt.h"] }
+
+  it "loads the same version as the spec we've loaded" do
+    assert_equal(Nokogiri::VERSION, gemspec.version.to_s)
+  end
+
+  describe "cruby" do
+    it "installs nokogiri headers" do
+      skip "until we fix https://github.com/sparklemotion/nokogiri/pull/1788"
+      assert(File.directory?(nokogiri_include_dir),
+             "expected directory to exist: #{nokogiri_include_dir}")
+
+      nokogiri_header_files.each do |header|
+        assert(File.file?(File.join(nokogiri_include_dir, header)),
+               "expected #{header} to be installed in #{nokogiri_include_dir}")
+      end
+    end
+
+    describe "native platform" do
+      it "acknowledges it is using packaged, precompiled libraries" do
+        assert(Nokogiri::VersionInfo.instance.libxml2_using_packaged?)
+        assert(Nokogiri::VERSION_INFO["libxml"].key?("source"))
+        assert_equal("packaged", Nokogiri::VERSION_INFO["libxml"]["source"])
+
+        assert(Nokogiri::VersionInfo.instance.libxml2_precompiled?)
+        assert(Nokogiri::VERSION_INFO["libxml"].key?("precompiled"))
+        assert(Nokogiri::VERSION_INFO["libxml"]["precompiled"])
+      end
+    end if gemspec.platform.is_a?(Gem::Platform) && gemspec.platform.cpu
+
+    describe "library" do
+      describe "packaged" do
+        it "declares where headers are installed" do
+          skip "until we fix https://github.com/sparklemotion/nokogiri/pull/1788"
+          # this is for nokogumbo and shouldn't be forever
+          nokogiri_include_parent_dir = File.dirname(nokogiri_include_dir)
+          assert_equal(nokogiri_include_parent_dir, Nokogiri::VERSION_INFO["libxml"]["libxml2_path"],
+                       "expected Nokogiri::VERSION_INFO to point to #{nokogiri_include_parent_dir}")
+        end
+
+        it "installs packaged libraries' headers" do
+          skip "until we fix https://github.com/sparklemotion/nokogiri/pull/1788"
+          packaged_library_header_files.each do |header|
+            assert(File.file?(File.join(nokogiri_include_dir, header)),
+                   "expected #{header} to be installed in #{nokogiri_include_dir}")
+          end
+        end
+      end if Nokogiri::VersionInfo.instance.libxml2_using_packaged?
+
+      describe "using system libraries" do
+        it "doesn't declare where headers are installed" do
+          refute(Nokogiri::VERSION_INFO["libxml"].key?("libxml2_path"))
+        end
+
+        it "does not install packaged libraries' headers" do
+          packaged_library_header_files.each do |header|
+            dir = File.join(nokogiri_include_dir, File.dirname(header))
+            refute(File.directory?(dir),
+                   "expected directory #{dir} to not exist")
+          end
+        end
+      end unless Nokogiri::VersionInfo.instance.libxml2_using_packaged?
+    end
+  end unless gemspec.platform == Gem::Platform.new("java")
+end