Correctly patch host ruby 3.2.3 and 3.2.4 when cross compiling

configs/components/_base-rubygem.rb

@@ -39,13 +39,14 @@
 
 # If a gem needs more command line options to install set the :gem_install_options
 # in its component file rubygem-<compoment>, before the instance_eval of this file.
-if settings[:gem_install_options].nil?
+gem_install_options = settings["#{pkg.get_name}_gem_install_options".to_sym]
+if gem_install_options.nil?
   pkg.install do
     "#{settings[:gem_install]} #{name}-#{version}.gem"
   end
 else
   pkg.install do
-    "#{settings[:gem_install]} #{name}-#{version}.gem #{settings[:gem_install_options]}"
+    "#{settings[:gem_install]} #{name}-#{version}.gem #{gem_install_options}"
   end
 end
 

configs/components/pl-ruby-patch.rb

@@ -8,70 +8,33 @@
 # This component should also be present in the puppet-agent project
 component "pl-ruby-patch" do |pkg, settings, platform|
   if platform.is_cross_compiled?
-    if platform.is_macos?
-      pkg.build_requires 'gnu-sed'
-      pkg.environment "PATH", "/usr/local/opt/gnu-sed/libexec/gnubin:$(PATH)"
-    end
 
-    ruby_api_version = settings[:ruby_version].gsub(/\.\d*$/, '.0')
     ruby_version_y = settings[:ruby_version].gsub(/(\d+)\.(\d+)\.(\d+)/, '\1.\2')
 
-    base_ruby = case platform.name
-                when /solaris-10/
-                  "/opt/csw/lib/ruby/2.0.0"
-                when /osx/
-                  "/usr/local/opt/ruby@#{ruby_version_y}/lib/ruby/#{ruby_api_version}"
-                else
-                  "/opt/pl-build-tools/lib/ruby/2.1.0"
-                end
+    pkg.add_source("file://resources/files/ruby/patch-hostruby.rb")
 
     # The `target_triple` determines which directory native extensions are stored in the
     # compiled ruby and must match ruby's naming convention.
-    #
-    # solaris 10 uses ruby 2.0 which doesn't install native extensions based on architecture
-    unless platform.name =~ /solaris-10/
-      # weird architecture naming conventions...
-      target_triple = if platform.architecture =~ /ppc64el|ppc64le/
-                        "powerpc64le-linux"
-                      elsif platform.name == 'solaris-11-sparc'
-                        "sparc-solaris-2.11"
-                      elsif platform.is_macos?
-                        if ruby_version_y.start_with?('2')
-                          "aarch64-darwin"
-                        else
-                          "arm64-darwin"
-                        end
+    # weird architecture naming conventions...
+    target_triple = if platform.architecture =~ /ppc64el|ppc64le/
+                      "powerpc64le-linux"
+                    elsif platform.name == 'solaris-11-sparc'
+                      "sparc-solaris-2.11"
+                    elsif platform.name =~ /solaris-10/
+                      "sparc-solaris"
+                    elsif platform.is_macos?
+                      if ruby_version_y.start_with?('2')
+                        "aarch64-darwin"
                       else
-                        "#{platform.architecture}-linux"
+                        "arm64-darwin"
                       end
+                    else
+                      "#{platform.architecture}-linux"
+                    end
 
-      pkg.build do
-        [
-          %(#{platform[:sed]} -i 's/Gem::Platform.local.to_s/"#{target_triple}"/' #{base_ruby}/rubygems/basic_specification.rb),
-          %(#{platform[:sed]} -i 's/Gem.extension_api_version/"#{ruby_api_version}"/' #{base_ruby}/rubygems/basic_specification.rb)
-        ]
-      end
-    end
-
-    # make rubygems use our target rbconfig when installing gems
-    case File.basename(base_ruby)
-    when '2.0.0', '2.1.0'
-      sed_command = %(s|Gem.ruby|&, '-r/opt/puppetlabs/puppet/share/doc/rbconfig-#{settings[:ruby_version]}-orig.rb'|)
-    else
-      sed_command = %(s|Gem.ruby.shellsplit|& << '-r/opt/puppetlabs/puppet/share/doc/rbconfig-#{settings[:ruby_version]}-orig.rb'|)
-    end
-
-    # rubygems switched which file has the command we need to patch starting in rubygems 3.4.10, which we install in our formula
-    # for ruby in homebrew-puppet
-    if Gem::Version.new(settings[:ruby_version]) >= Gem::Version.new('3.2.2') || platform.is_macos? && ruby_version_y.start_with?('2')
-      filename = 'builder.rb'
-    else
-      filename = 'ext_conf_builder.rb'
-    end
-
-    pkg.build do
+    pkg.install do
       [
-        %(#{platform[:sed]} -i "#{sed_command}" #{base_ruby}/rubygems/ext/#{filename})
+        "#{settings[:host_ruby]} patch-hostruby.rb #{settings[:ruby_version]} #{target_triple}"
       ]
     end
   end

configs/components/rubygem-ffi.rb

@@ -15,38 +15,16 @@
     pkg.sha256sum '6f2ed2fa68047962d6072b964420cba91d82ce6fa8ee251950c17fca6af3c2a0'
   end
 
-  instance_eval File.read('configs/components/_base-rubygem.rb')
-
   rb_major_minor_version = settings[:ruby_version].to_f
 
-  # Prior to ruby 3.2, both ruby and the ffi gem vendored a version of libffi.
-  # If libffi happened to be installed in /usr/lib, then the ffi gem preferred
-  # that instead of building libffi itself. To ensure consistency, we use
-  # --disable-system-ffi so that the ffi gem *always* builds libffi, then
-  # builds the ffi_c native extension and links it against libffi.so.
-  #
-  # In ruby 3.2 and up, libffi is no longer vendored. So we created a separate
-  # libffi vanagon component which is built before ruby. The ffi gem still
-  # vendors libffi, so we use the --enable-system-ffi option to ensure the ffi
-  # gem *always* uses the libffi.so we already built. Note the term "system" is
-  # misleading, because we override PKG_CONFIG_PATH below so that our libffi.so
-  # is preferred, not the one in /usr/lib.
-  if rb_major_minor_version > 2.7
-    pkg.install do
-      "#{settings[:gem_install]} ffi-#{pkg.get_version}.gem -- --enable-system-ffi"
-    end
-  else
-    pkg.install do
-      "#{settings[:gem_install]} ffi-#{pkg.get_version}.gem -- --disable-system-ffi"
-    end
-  end
-
   # Windows versions of the FFI gem have custom filenames, so we overwite the
   # defaults that _base-rubygem provides here, just for Windows for Ruby < 3.2
   if platform.is_windows? && rb_major_minor_version < 3.2
     # Pin this if lower than Ruby 2.7
     pkg.version '1.9.25' if rb_major_minor_version < 2.7
 
+    instance_eval File.read('configs/components/_base-rubygem.rb')
+
     # Vanagon's `pkg.mirror` is additive, and the _base_rubygem sets the
     # non-Windows gem as the first mirror, which is incorrect. We need to unset
     # the list of mirrors before adding the Windows-appropriate ones here:
@@ -81,6 +59,26 @@
     pkg.install do
       "#{settings[:gem_install]} ffi-#{pkg.get_version}-#{platform.architecture}-mingw32.gem"
     end
+  else
+    # Prior to ruby 3.2, both ruby and the ffi gem vendored a version of libffi.
+    # If libffi happened to be installed in /usr/lib, then the ffi gem preferred
+    # that instead of building libffi itself. To ensure consistency, we use
+    # --disable-system-libffi so that the ffi gem *always* builds libffi, then
+    # builds the ffi_c native extension and links it against libffi.so.
+    #
+    # In ruby 3.2 and up, libffi is no longer vendored. So we created a separate
+    # libffi vanagon component which is built before ruby. The ffi gem still
+    # vendors libffi, so we use the --enable-system-libffi option to ensure the ffi
+    # gem *always* uses the libffi.so we already built. Note the term "system" is
+    # misleading, because we override PKG_CONFIG_PATH below so that our libffi.so
+    # is preferred, not the one in /usr/lib.
+    settings["#{pkg.get_name}_gem_install_options".to_sym] =
+      if rb_major_minor_version > 2.7
+        "-- --enable-system-libffi"
+      else
+        "-- --disable-system-libffi"
+      end
+    instance_eval File.read('configs/components/_base-rubygem.rb')
   end
 
   # due to contrib/make_sunver.pl missing on solaris 11 we cannot compile libffi, so we provide the opencsw library
@@ -157,4 +155,4 @@
       %(#{platform[:sed]} -i '0,/ensure_required_ruby_version_met/b; /ensure_required_ruby_version_met/d' #{base_ruby}/rubygems/installer.rb)
     end
   end
-end
+end

configs/components/rubygem-nokogiri.rb

@@ -1,14 +1,13 @@
-component 'rubygem-nokogiri' do |pkg, _settings, _platform|
+component 'rubygem-nokogiri' do |pkg, settings, _platform|
   pkg.version '1.14.2'
   pkg.sha256sum 'c765a74aac6cf430a710bb0b6038b8ee11f177393cd6ae8dadc7a44a6e2658b6'
-  # On macOS when we are not cross compiling we need to use runtime's libxml2 and libxslt
-  if platform.is_macos? && !platform.is_cross_compiled?
-    settings[:gem_install_options] = "-- --use-system-libraries \
+
+  settings["#{pkg.get_name}_gem_install_options".to_sym] = "--platform=ruby -- \
+        --use-system-libraries \
         --with-xml2-lib=#{settings[:libdir]} \
         --with-xml2-include=#{settings[:includedir]}/libxml2 \
         --with-xslt-lib=#{settings[:libdir]} \
         --with-xslt-include=#{settings[:includedir]}"
-  end
   instance_eval File.read('configs/components/_base-rubygem.rb')
   pkg.build_requires 'rubygem-mini_portile2'
   gem_home = settings[:gem_home]

resources/files/ruby/patch-hostruby.rb

@@ -0,0 +1,145 @@
+# When cross compiling we need to run gem install using the host ruby, but
+# force ruby to use our overridden rbconfig.rb. To do that, we insert a
+# require statement between the ruby executable and it's first argument,
+# thereby hooking the ruby process.
+#
+# In the future we could use the --target-rbconfig=<path> option to point
+# to our rbconfig.rb. But that option is only available in newer ruby versions.
+require 'rbconfig'
+require 'tempfile'
+
+if ARGV.length < 2
+  warn <<USAGE
+USAGE: patch-hostruby.rb <target_ruby_version> <target_triple>
+
+example: patch-hostruby.rb 3.2.2 arm64-darwin
+USAGE
+  exit(1)
+end
+
+# target ruby versions (what we're trying to build)
+target_ruby_version = ARGV[0]
+target_triple = ARGV[1]
+target_api_version = target_ruby_version.gsub(/\.\d*$/, '.0')
+
+# host ruby (the ruby we execute to build the target)
+host_rubylibdir = RbConfig::CONFIG['rubylibdir']
+GEM_VERSION = Gem::Version.new(Gem::VERSION)
+
+# Rewrite the file in-place securely, yielding each line to the caller
+def rewrite(file)
+  # create temp file in the same directory as the file we're patching,
+  # so rename doesn't cross filesystems
+  tmpfile = Tempfile.new(File.basename(file), File.dirname(file))
+  begin
+    File.open("#{file}.orig", "w") do |orig|
+      File.open(file, 'r').readlines.each do |line|
+        orig.write(line)
+        yield line
+        tmpfile.write(line)
+      end
+    end
+  ensure
+    tmpfile.close
+    File.unlink(file)
+    File.rename(tmpfile.path, file)
+    tmpfile.unlink
+  end
+end
+
+# Based on the RUBYGEMS version of the host ruby, the line and file that needs patching is different
+# Note the RUBY version doesn't matter (for either the host or target ruby).
+#
+# Here we define different intervals. For each interval, we specify the regexp to match, what to
+# replace it with, and which file to edit in-place. Note `\&` is a placeholder for whatever the regexp
+# was, that way we can easily append to it. And since it's in a double quoted string, it's escaped
+# as `\\&`
+#
+if GEM_VERSION <= Gem::Version.new('2.0.0')
+  # $ git show v2.0.0:lib/rubygems/ext/ext_conf_builder.rb
+  #   cmd = "#{Gem.ruby} #{File.basename extension}"
+  regexp  = /{Gem\.ruby}/
+  replace = "\\& -r/opt/puppetlabs/puppet/share/doc/rbconfig-#{target_ruby_version}-orig.rb"
+  builder = 'rubygems/ext/ext_conf_builder.rb'
+elsif GEM_VERSION < Gem::Version.new('3.0.0') # there weren't any tags between >= 2.7.11 and < 3.0.0
+  # $ git show v2.0.1:lib/rubygems/ext/ext_conf_builder.rb
+  #   cmd = [Gem.ruby, File.basename(extension), *args].join ' '
+  #
+  # $ git show v2.7.11:lib/rubygems/ext/ext_conf_builder.rb
+  #   cmd = [Gem.ruby, "-r", get_relative_path(siteconf.path), File.basename(extension), *args].join ' '
+  regexp  = /Gem\.ruby/
+  replace = "\\&, '-r/opt/puppetlabs/puppet/share/doc/rbconfig-#{target_ruby_version}-orig.rb'"
+  builder = 'rubygems/ext/ext_conf_builder.rb'
+elsif GEM_VERSION <= Gem::Version.new('3.4.8')
+  # $ git show v3.0.0:lib/rubygems/ext/ext_conf_builder.rb
+  #   cmd = Gem.ruby.shellsplit << "-I" << File.expand_path("../../..", __FILE__) <<
+  #
+  # $ git show v3.4.8:lib/rubygems/ext/ext_conf_builder.rb
+  #   cmd = Gem.ruby.shellsplit << "-I" << File.expand_path("../..", __dir__) << File.basename(extension)
+  regexp  = /Gem\.ruby\.shellsplit/
+  replace = "\\& << '-r/opt/puppetlabs/puppet/share/doc/rbconfig-#{target_ruby_version}-orig.rb'"
+  builder = 'rubygems/ext/ext_conf_builder.rb'
+elsif GEM_VERSION <= Gem::Version.new('3.4.14')
+  # NOTE: rubygems 3.4.9 moved the code to builder.rb
+  #
+  # $ git show v3.4.9:lib/rubygems/ext/builder.rb
+  #   cmd = Gem.ruby.shellsplit
+  #
+  # $ git show v3.4.14:lib/rubygems/ext/builder.rb
+  #   cmd = Gem.ruby.shellsplit
+  regexp  = /Gem\.ruby\.shellsplit/
+  replace = "\\& << '-r/opt/puppetlabs/puppet/share/doc/rbconfig-#{target_ruby_version}-orig.rb'"
+  builder = 'rubygems/ext/builder.rb'
+elsif GEM_VERSION <= Gem::Version.new('3.5.10')
+  # $ git show v3.4.9:lib/rubygems/ext/builder.rb
+  #     cmd = Shellwords.split(Gem.ruby)
+  #
+  # $ git show v3.5.10:lib/rubygems/ext/builder.rb
+  #     cmd = Shellwords.split(Gem.ruby)
+  regexp  = /Shellwords\.split\(Gem\.ruby\)/
+  replace = "\\& << '-r/opt/puppetlabs/puppet/share/doc/rbconfig-#{target_ruby_version}-orig.rb'"
+  builder = 'rubygems/ext/builder.rb'
+else
+  raise "We don't know how to patch rubygems #{GEM_VERSION}"
+end
+
+# path to the builder file on the HOST ruby
+builder = File.join(host_rubylibdir, builder)
+
+raise "We can't patch #{builder} because it doesn't exist" unless File.exist?(builder)
+
+# hook rubygems builder so it loads our rbconfig when building native gems
+patched = false
+rewrite(builder) do |line|
+  if line.gsub!(regexp, replace)
+    patched = true
+  end
+end
+
+raise "Failed to patch rubygems hook, because we couldn't match #{regexp} in #{builder}" unless patched
+
+puts "Patched '#{regexp.inspect}' in #{builder}"
+
+# solaris 10 uses ruby 2.0 which doesn't install native extensions based on architecture
+if RUBY_PLATFORM !~ /solaris2\.10$/ || RUBY_VERSION != '2.0.0'
+  # ensure native extensions are written to a directory that matches the
+  # architecture of the target ruby we're building for. To do that we
+  # patch the host ruby to pretend to be the target architecture.
+  triple_patched = false
+  api_version_patched = false
+  spec_file = "#{host_rubylibdir}/rubygems/basic_specification.rb"
+  rewrite(spec_file) do |line|
+    if line.gsub!(/Gem::Platform\.local\.to_s/, "'#{target_triple}'")
+      triple_patched = true
+    end
+    if line.gsub!(/Gem\.extension_api_version/, "'#{target_api_version}'")
+      api_version_patched = true
+    end
+  end
+
+  raise "Failed to patch '#{target_triple}' in #{spec_file}" unless triple_patched
+  puts "Patched '#{target_triple}' in #{spec_file}"
+
+  raise "Failed to patch '#{target_api_version}' in #{spec_file}" unless api_version_patched
+  puts "Patched '#{target_api_version}' in #{spec_file}"
+end