nokogiri 1.11.1 → 1.12.0.rc1

data/ext/nokogiri/extconf.rb

@@ -1,5 +1,5 @@
-# :stopdoc:
-ENV['RC_ARCHS'] = '' if RUBY_PLATFORM =~ /darwin/
+# frozen_string_literal: true
+ENV["RC_ARCHS"] = "" if RUBY_PLATFORM =~ /darwin/
 
 require "mkmf"
 require "rbconfig"
@@ -7,16 +7,14 @@ require "fileutils"
 require "shellwords"
 require "pathname"
 
-#
-#  helpful constants
-#
-PACKAGE_ROOT_DIR = File.expand_path(File.join(File.dirname(__FILE__), '..', '..'))
+# helpful constants
+PACKAGE_ROOT_DIR = File.expand_path(File.join(File.dirname(__FILE__), "..", ".."))
 REQUIRED_LIBXML_VERSION = "2.6.21"
 RECOMMENDED_LIBXML_VERSION = "2.9.3"
 
 # The gem version constraint in the Rakefile is not respected at install time.
 # Keep this version in sync with the one in the Rakefile !
-REQUIRED_MINI_PORTILE_VERSION = "~> 2.5.0"
+REQUIRED_MINI_PORTILE_VERSION = "~> 2.6.1"
 REQUIRED_PKG_CONFIG_VERSION = "~> 1.1"
 
 # Keep track of what versions of what libraries we build against
@@ -28,10 +26,16 @@ NOKOGIRI_HELP_MESSAGE = <<~HELP
     Flags that are always valid:
 
       --use-system-libraries
-          Use system libraries instead of building and using the packaged libraries
+      --enable-system-libraries
+          Use system libraries instead of building and using the packaged libraries.
+
+      --disable-system-libraries
+          Use the packaged libraries, and ignore the system libraries. This is the default on most
+          platforms, and overrides `--use-system-libraries` and the environment variable
+          `NOKOGIRI_USE_SYSTEM_LIBRARIES`.
 
       --disable-clean
-          Do not clean out intermediate files after successful build
+          Do not clean out intermediate files after successful build.
 
       --prevent-strip
           Take steps to prevent stripping the symbol table and debugging info from the shared
@@ -40,47 +44,88 @@ NOKOGIRI_HELP_MESSAGE = <<~HELP
 
     Flags only used when using system libraries:
 
-      --with-opt-dir=DIRECTORY
-          Look for headers and libraries in DIRECTORY
+      General:
+
+        --with-opt-dir=DIRECTORY
+            Look for headers and libraries in DIRECTORY.
+
+        --with-opt-lib=DIRECTORY
+            Look for libraries in DIRECTORY.
+
+        --with-opt-include=DIRECTORY
+            Look for headers in DIRECTORY.
+
+
+      Related to zlib:
+
+        --with-zlib-dir=DIRECTORY
+            Look for zlib headers and library in DIRECTORY.
+
+        --with-zlib-lib=DIRECTORY
+            Look for zlib library in DIRECTORY.
+
+        --with-zlib-include=DIRECTORY
+            Look for zlib headers in DIRECTORY.
+
+
+      Related to iconv:
+
+        --with-iconv-dir=DIRECTORY
+            Look for iconv headers and library in DIRECTORY.
+
+        --with-iconv-lib=DIRECTORY
+            Look for iconv library in DIRECTORY.
+
+        --with-iconv-include=DIRECTORY
+            Look for iconv headers in DIRECTORY.
+
+
+      Related to libxml2:
+
+        --with-xml2-dir=DIRECTORY
+            Look for xml2 headers and library in DIRECTORY.
+
+        --with-xml2-lib=DIRECTORY
+            Look for xml2 library in DIRECTORY.
+
+        --with-xml2-include=DIRECTORY
+            Look for xml2 headers in DIRECTORY.
+
+        --with-xml2-source-dir=DIRECTORY
+            (dev only) Build libxml2 from the source code in DIRECTORY
 
-      --with-zlib-dir=DIR
-          Look for zlib header and library in DIRECTORY
 
-      --with-iconv-dir=DIRECTORY
-          Look for iconv header and library in DIRECTORY
+      Related to libxslt:
 
-      --with-xml2-dir=DIRECTORY
-          Look for xml2 headers and library in DIRECTORY
+        --with-xslt-dir=DIRECTORY
+            Look for xslt headers and library in DIRECTORY.
 
-      --with-xml2-lib=DIRECTORY
-          Look for xml2 library in DIRECTORY
+        --with-xslt-lib=DIRECTORY
+            Look for xslt library in DIRECTORY.
 
-      --with-xslt-include=DIRECTORY
-          Look for xslt headers in DIRECTORY
+        --with-xslt-include=DIRECTORY
+            Look for xslt headers in DIRECTORY.
 
-      --with-xslt-dir=DIRECTORY
-          Look for xslt headers and library in DIRECTORY
+        --with-xslt-source-dir=DIRECTORY
+            (dev only) Build libxslt from the source code in DIRECTORY
 
-      --with-xslt-lib=DIRECTORY
-          Look for xslt library in DIRECTORY
 
-      --with-xslt-include=DIRECTORY
-          Look for xslt headers in DIRECTORY
+      Related to libexslt:
 
-      --with-exslt-dir=DIRECTORY
-          Look for exslt headers and library in DIRECTORY
+        --with-exslt-dir=DIRECTORY
+            Look for exslt headers and library in DIRECTORY.
 
-      --with-exslt-lib=DIRECTORY
-          Look for exslt library in DIRECTORY
+        --with-exslt-lib=DIRECTORY
+            Look for exslt library in DIRECTORY.
 
-      --with-exslt-include=DIRECTORY
-          Look for exslt headers in DIRECTORY
+        --with-exslt-include=DIRECTORY
+            Look for exslt headers in DIRECTORY.
 
 
     Flags only used when building and using the packaged libraries:
 
       --disable-static
-          Do not statically link packaged libraries, instead use shared libraries
+          Do not statically link packaged libraries, instead use shared libraries.
 
       --enable-cross-build
           Enable cross-build mode. (You probably do not want to set this manually.)
@@ -89,8 +134,7 @@ NOKOGIRI_HELP_MESSAGE = <<~HELP
     Environment variables used:
 
       NOKOGIRI_USE_SYSTEM_LIBRARIES
-          When set, even if nil or blank, use system libraries instead of building and using the
-          packaged libraries. Equivalent to `--use-system-libraries`.
+          Equivalent to `--enable-system-libraries` when set, even if nil or blank.
 
       CC
           Use this path to invoke the compiler instead of `RbConfig::CONFIG['CC']`
@@ -111,20 +155,39 @@ HELP
 #
 #  utility functions
 #
+def config_clean?
+  enable_config("clean", true)
+end
+
+def config_static?
+  default_static = !truffle?
+  enable_config("static", default_static)
+end
+
+def config_cross_build?
+  enable_config("cross-build")
+end
+
+def config_system_libraries?
+  enable_config("system-libraries", ENV.key?("NOKOGIRI_USE_SYSTEM_LIBRARIES")) do |_, default|
+    arg_config("--use-system-libraries", default)
+  end
+end
+
 def windows?
-  RbConfig::CONFIG['target_os'] =~ /mingw32|mswin/
+  RbConfig::CONFIG["target_os"] =~ /mingw32|mswin/
 end
 
 def solaris?
-  RbConfig::CONFIG['target_os'] =~ /solaris/
+  RbConfig::CONFIG["target_os"] =~ /solaris/
 end
 
 def darwin?
-  RbConfig::CONFIG['target_os'] =~ /darwin/
+  RbConfig::CONFIG["target_os"] =~ /darwin/
 end
 
 def openbsd?
-  RbConfig::CONFIG['target_os'] =~ /openbsd/
+  RbConfig::CONFIG["target_os"] =~ /openbsd/
 end
 
 def aix?
@@ -132,15 +195,19 @@ def aix?
 end
 
 def nix?
-  ! (windows? || solaris? || darwin?)
+  !(windows? || solaris? || darwin?)
+end
+
+def truffle?
+  ::RUBY_ENGINE == "truffleruby"
 end
 
-def concat_flags *args
+def concat_flags(*args)
   args.compact.join(" ")
 end
 
-def local_have_library(lib, func=nil, headers=nil)
-  have_library(lib, func, headers) or have_library("lib#{lib}", func, headers)
+def local_have_library(lib, func = nil, headers = nil)
+  have_library(lib, func, headers) || have_library("lib#{lib}", func, headers)
 end
 
 LOCAL_PACKAGE_RESPONSE = Object.new
@@ -150,7 +217,7 @@ end
 
 # wrapper around MakeMakefil#pkg_config and the PKGConfig gem
 def try_package_configuration(pc)
-  if !ENV.key?("NOKOGIRI_TEST_PKG_CONFIG_GEM")
+  unless ENV.key?("NOKOGIRI_TEST_PKG_CONFIG_GEM")
     # try MakeMakefile#pkg_config, which uses the system utility `pkg-config`.
     return if checking_for("#{pc} using `pkg_config`", LOCAL_PACKAGE_RESPONSE) do
       pkg_config(pc)
@@ -161,9 +228,9 @@ def try_package_configuration(pc)
   # let's fall back to the pkg-config gem, which knows how to parse .pc files, and wrap it with the
   # same logic as MakeMakefile#pkg_config
   begin
-    require 'rubygems'
-    gem 'pkg-config', REQUIRED_PKG_CONFIG_VERSION
-    require 'pkg-config'
+    require "rubygems"
+    gem("pkg-config", REQUIRED_PKG_CONFIG_VERSION)
+    require "pkg-config"
 
     checking_for("#{pc} using pkg-config gem version #{PKGConfig::VERSION}", LOCAL_PACKAGE_RESPONSE) do
       if PKGConfig.have_package(pc)
@@ -171,14 +238,14 @@ def try_package_configuration(pc)
         ldflags = PKGConfig.libs_only_L(pc)
         libs    = PKGConfig.libs_only_l(pc)
 
-        Logging::message "pkg-config gem found package configuration for %s\n", pc
-        Logging::message "cflags: %s\nldflags: %s\nlibs: %s\n\n", cflags, ldflags, libs
+        Logging.message("pkg-config gem found package configuration for %s\n", pc)
+        Logging.message("cflags: %s\nldflags: %s\nlibs: %s\n\n", cflags, ldflags, libs)
 
         [cflags, ldflags, libs]
       end
     end
   rescue LoadError
-    message "Please install either the `pkg-config` utility or the `pkg-config` rubygem.\n"
+    message("Please install either the `pkg-config` utility or the `pkg-config` rubygem.\n")
   end
 end
 
@@ -190,7 +257,7 @@ def have_package_configuration(opt: nil, pc: nil, lib:, func:, headers:)
   end
 
   # see if we have enough path info to do this without trying any harder
-  if !ENV.key?("NOKOGIRI_TEST_PKG_CONFIG")
+  unless ENV.key?("NOKOGIRI_TEST_PKG_CONFIG")
     return true if local_have_library(lib, func, headers)
   end
 
@@ -201,12 +268,12 @@ def have_package_configuration(opt: nil, pc: nil, lib:, func:, headers:)
 end
 
 def ensure_package_configuration(opt: nil, pc: nil, lib:, func:, headers:)
-  have_package_configuration(opt: opt, pc: pc, lib: lib, func: func, headers: headers) or
+  have_package_configuration(opt: opt, pc: pc, lib: lib, func: func, headers: headers) ||
     abort_could_not_find_library(lib)
 end
 
-def ensure_func(func, headers=nil)
-  have_func(func, headers) or abort_could_not_find_library(lib)
+def ensure_func(func, headers = nil)
+  have_func(func, headers) || abort_could_not_find_library(func)
 end
 
 def preserving_globals
@@ -217,20 +284,20 @@ ensure
 end
 
 def abort_could_not_find_library(lib)
-  abort "-----\n#{caller[0]}\n#{lib} is missing. Please locate mkmf.log to investigate how it is failing.\n-----"
+  abort("-----\n#{caller[0]}\n#{lib} is missing. Please locate mkmf.log to investigate how it is failing.\n-----")
 end
 
 def chdir_for_build
   # When using rake-compiler-dock on Windows, the underlying Virtualbox shared
   # folders don't support symlinks, but libiconv expects it for a build on
   # Linux. We work around this limitation by using the temp dir for cooking.
-  build_dir = ENV['RCD_HOST_RUBY_PLATFORM'].to_s =~ /mingw|mswin|cygwin/ ? '/tmp' : '.'
+  build_dir = ENV["RCD_HOST_RUBY_PLATFORM"].to_s =~ /mingw|mswin|cygwin/ ? "/tmp" : "."
   Dir.chdir(build_dir) do
     yield
   end
 end
 
-def sh_export_path path
+def sh_export_path(path)
   # because libxslt 1.1.29 configure.in uses AC_PATH_TOOL which treats ":"
   # as a $PATH separator, we need to convert windows paths from
   #
@@ -242,7 +309,7 @@ def sh_export_path path
   #
   # which is sh-compatible, in order to find things properly during
   # configuration
-  return path if !windows?
+  return path unless windows?
 
   match = Regexp.new("^([A-Z]):(/.*)").match(path)
   if match && match.length == 3
@@ -255,36 +322,31 @@ end
 def libflag_to_filename(ldflag)
   case ldflag
   when /\A-l(.+)/
-    "lib#{$1}.#{$LIBEXT}"
+    "lib#{Regexp.last_match(1)}.#{$LIBEXT}"
   end
 end
 
-def using_system_libraries?
-  # NOTE: TruffleRuby uses this env var as it does not support using static libraries yet.
-  arg_config('--use-system-libraries', ENV.key?("NOKOGIRI_USE_SYSTEM_LIBRARIES"))
-end
-
-def have_libxml_headers?(version=nil)
+def have_libxml_headers?(version = nil)
   source = if version.nil?
-             <<~SRC
-               #include <libxml/xmlversion.h>
-             SRC
-           else
-             version_int = sprintf "%d%2.2d%2.2d", *(version.split("."))
-             <<~SRC
-               #include <libxml/xmlversion.h>
-               #if LIBXML_VERSION < #{version_int}
-               #  error libxml2 is older than #{version}
-               #endif
-             SRC
-           end
-
-  try_cpp source
+    <<~SRC
+      #include <libxml/xmlversion.h>
+    SRC
+  else
+    version_int = format("%d%2.2d%2.2d", *version.split("."))
+    <<~SRC
+      #include <libxml/xmlversion.h>
+      #if LIBXML_VERSION < #{version_int}
+      #  error libxml2 is older than #{version}
+      #endif
+    SRC
+  end
+
+  try_cpp(source)
 end
 
 def try_link_iconv(using = nil)
-  checking_for(using ? "iconv using #{using}" : 'iconv') do
-    ['', '-liconv'].any? do |opt|
+  checking_for(using ? "iconv using #{using}" : "iconv") do
+    ["", "-liconv"].any? do |opt|
       preserving_globals do
         yield if block_given?
 
@@ -307,62 +369,60 @@ def iconv_configure_flags
   # give --with-iconv-dir and --with-opt-dir first priority
   ["iconv", "opt"].each do |target|
     config = preserving_globals { dir_config(target) }
-    if config.any? && try_link_iconv("--with-#{target}-* flags") { dir_config(target) }
-      idirs, ldirs = config.map do |dirs|
-        Array(dirs).flat_map do |dir|
-          dir.split(File::PATH_SEPARATOR)
-        end if dirs
-      end
-
-      return [
-        '--with-iconv=yes',
-        *("CPPFLAGS=#{idirs.map { |dir| '-I' + dir }.join(' ')}" if idirs),
-        *("LDFLAGS=#{ldirs.map { |dir| '-L' + dir }.join(' ')}" if ldirs),
-      ]
+    next unless config.any? && try_link_iconv("--with-#{target}-* flags") { dir_config(target) }
+    idirs, ldirs = config.map do |dirs|
+      Array(dirs).flat_map do |dir|
+        dir.split(File::PATH_SEPARATOR)
+      end if dirs
     end
+
+    return [
+      "--with-iconv=yes",
+      *("CPPFLAGS=#{idirs.map { |dir| "-I" + dir }.join(" ")}" if idirs),
+      *("LDFLAGS=#{ldirs.map { |dir| "-L" + dir }.join(" ")}" if ldirs),
+    ]
   end
 
   if try_link_iconv
-    return ['--with-iconv=yes']
+    return ["--with-iconv=yes"]
   end
 
-  config = preserving_globals { have_package_configuration('libiconv') }
-  if config && try_link_iconv('pkg-config libiconv') { have_package_configuration('libiconv') }
+  config = preserving_globals { have_package_configuration("libiconv") }
+  if config && try_link_iconv("pkg-config libiconv") { have_package_configuration("libiconv") }
     cflags, ldflags, libs = config
 
     return [
-      '--with-iconv=yes',
+      "--with-iconv=yes",
       "CPPFLAGS=#{cflags}",
       "LDFLAGS=#{ldflags}",
       "LIBS=#{libs}",
     ]
   end
 
-  abort_could_not_find_library "libiconv"
+  abort_could_not_find_library("libiconv")
 end
 
-def process_recipe(name, version, static_p, cross_p)
-  require 'rubygems'
-  gem 'mini_portile2', REQUIRED_MINI_PORTILE_VERSION
-  require 'mini_portile2'
-  message "Using mini_portile version #{MiniPortile::VERSION}\n"
+def process_recipe(name, version, static_p, cross_p, cacheable_p=true)
+  require "rubygems"
+  gem("mini_portile2", REQUIRED_MINI_PORTILE_VERSION)
+  require "mini_portile2"
+  message("Using mini_portile version #{MiniPortile::VERSION}\n")
 
-  if name != "libxml2" && name != "libxslt"
+  unless ["libxml2", "libxslt"].include?(name)
     OTHER_LIBRARY_VERSIONS[name] = version
   end
 
   MiniPortile.new(name, version).tap do |recipe|
-    recipe.target = File.join(PACKAGE_ROOT_DIR, "ports")
+    recipe.target = File.join(PACKAGE_ROOT_DIR, "ports") if cacheable_p
     # Prefer host_alias over host in order to use i586-mingw32msvc as
     # correct compiler prefix for cross build, but use host if not set.
     recipe.host = RbConfig::CONFIG["host_alias"].empty? ? RbConfig::CONFIG["host"] : RbConfig::CONFIG["host_alias"]
-    recipe.patch_files = Dir[File.join(PACKAGE_ROOT_DIR, "patches", name, "*.patch")].sort
     recipe.configure_options << "--libdir=#{File.join(recipe.path, "lib")}"
 
     yield recipe
 
     env = Hash.new do |hash, key|
-      hash[key] = "#{ENV[key]}"  # (ENV[key].dup rescue '')
+      hash[key] = (ENV[key]).to_s
     end
 
     recipe.configure_options.flatten!
@@ -370,10 +430,10 @@ def process_recipe(name, version, static_p, cross_p)
     recipe.configure_options.delete_if do |option|
       case option
       when /\A(\w+)=(.*)\z/
-        if env.key?($1)
-          env[$1] = concat_flags(env[$1], $2)
+        env[Regexp.last_match(1)] = if env.key?(Regexp.last_match(1))
+          concat_flags(env[Regexp.last_match(1)], Regexp.last_match(2))
         else
-          env[$1] = $2
+          Regexp.last_match(2)
         end
         true
       else
@@ -401,10 +461,10 @@ def process_recipe(name, version, static_p, cross_p)
       ]
     end
 
-    if RbConfig::CONFIG['target_cpu'] == 'universal'
+    if RbConfig::CONFIG["target_cpu"] == "universal"
       %w[CFLAGS LDFLAGS].each do |key|
-        unless env[key].include?('-arch')
-          env[key] = concat_flags(env[key], RbConfig::CONFIG['ARCH_FLAG'])
+        unless env[key].include?("-arch")
+          env[key] = concat_flags(env[key], RbConfig::CONFIG["ARCH_FLAG"])
         end
       end
     end
@@ -414,44 +474,42 @@ def process_recipe(name, version, static_p, cross_p)
     end
 
     checkpoint = "#{recipe.target}/#{recipe.name}-#{recipe.version}-#{recipe.host}.installed"
-    if File.exist?(checkpoint)
-      message "Building Nokogiri with a packaged version of #{name}-#{version}.\n"
+    if File.exist?(checkpoint) && !recipe.source_directory
+      message("Building Nokogiri with a packaged version of #{name}-#{version}.\n")
     else
-      message <<~EOM
+      message(<<~EOM)
         ---------- IMPORTANT NOTICE ----------
         Building Nokogiri with a packaged version of #{name}-#{version}.
         Configuration options: #{recipe.configure_options.shelljoin}
       EOM
 
       unless recipe.patch_files.empty?
-        message "The following patches are being applied:\n"
+        message("The following patches are being applied:\n")
 
         recipe.patch_files.each do |patch|
-          message "  - %s\n" % File.basename(patch)
+          message("  - %s\n" % File.basename(patch))
         end
       end
 
-      message <<~EOM
+      message(<<~EOM) if name != "libgumbo"
 
         The Nokogiri maintainers intend to provide timely security updates, but if
         this is a concern for you and want to use your OS/distro system library
         instead, then abort this installation process and install nokogiri as
         instructed at:
 
-          https://nokogiri.org/tutorials/installing_nokogiri.html#install-with-system-libraries
+          https://nokogiri.org/tutorials/installing_nokogiri.html#installing-using-standard-system-libraries
 
       EOM
 
-      message <<~EOM if name == 'libxml2'
-        Note, however, that nokogiri cannot guarantee compatiblity with every
+      message(<<~EOM) if name == "libxml2"
+        Note, however, that nokogiri cannot guarantee compatibility with every
         version of libxml2 that may be provided by OS/package vendors.
 
       EOM
 
-      chdir_for_build do
-        recipe.cook
-      end
-      FileUtils.touch checkpoint
+      chdir_for_build { recipe.cook }
+      FileUtils.touch(checkpoint)
     end
     recipe.activate
   end
@@ -461,13 +519,13 @@ def copy_packaged_libraries_headers(to_path:, from_recipes:)
   FileUtils.rm_rf(to_path, secure: true)
   FileUtils.mkdir(to_path)
   from_recipes.each do |recipe|
-    FileUtils.cp_r(Dir[File.join(recipe.path, 'include/*')], to_path)
+    FileUtils.cp_r(Dir[File.join(recipe.path, "include/*")], to_path)
   end
 end
 
 def do_help
-  print NOKOGIRI_HELP_MESSAGE
-  exit! 0
+  print(NOKOGIRI_HELP_MESSAGE)
+  exit!(0)
 end
 
 def do_clean
@@ -475,50 +533,50 @@ def do_clean
   pwd  = Pathname(Dir.pwd)
 
   # Skip if this is a development work tree
-  unless (root + '.git').exist?
-    message "Cleaning files only used during build.\n"
+  unless (root + ".git").exist?
+    message("Cleaning files only used during build.\n")
 
     # (root + 'tmp') cannot be removed at this stage because
     # nokogiri.so is yet to be copied to lib.
 
     # clean the ports build directory
-    Pathname.glob(pwd.join('tmp', '*', 'ports')) do |dir|
+    Pathname.glob(pwd.join("tmp", "*", "ports")) do |dir|
       FileUtils.rm_rf(dir, verbose: true)
     end
 
-    if enable_config('static')
+    if config_static?
       # ports installation can be safely removed if statically linked.
-      FileUtils.rm_rf(root + 'ports', verbose: true)
+      FileUtils.rm_rf(root + "ports", verbose: true)
     else
-      FileUtils.rm_rf(root + 'ports' + 'archives', verbose: true)
+      FileUtils.rm_rf(root + "ports" + "archives", verbose: true)
     end
   end
 
-  exit! 0
+  exit!(0)
 end
 
 #
 #  main
 #
-do_help if arg_config('--help')
-do_clean if arg_config('--clean')
+do_help if arg_config("--help")
+do_clean if arg_config("--clean")
 
-if openbsd? && !using_system_libraries?
-  if `#{ENV['CC'] || '/usr/bin/cc'} -v 2>&1` !~ /clang/
-    ENV['CC'] ||= find_executable('egcc') or
-      abort "Please install gcc 4.9+ from ports using `pkg_add -v gcc`"
+if openbsd? && !config_system_libraries?
+  if %x(#{ENV["CC"] || "/usr/bin/cc"} -v 2>&1) !~ /clang/
+    (ENV["CC"] ||= find_executable("egcc")) ||
+      abort("Please install gcc 4.9+ from ports using `pkg_add -v gcc`")
   end
   append_cppflags "-I/usr/local/include"
 end
 
-if ENV['CC']
-  RbConfig::CONFIG['CC'] = RbConfig::MAKEFILE_CONFIG['CC'] = ENV['CC']
+if ENV["CC"]
+  RbConfig::CONFIG["CC"] = RbConfig::MAKEFILE_CONFIG["CC"] = ENV["CC"]
 end
 
 # use same c compiler for libxml and libxslt
-ENV['CC'] = RbConfig::CONFIG['CC']
+ENV["CC"] = RbConfig::CONFIG["CC"]
 
-if arg_config('--prevent-strip')
+if arg_config("--prevent-strip")
   old_cflags = $CFLAGS.split.join(" ")
   old_ldflags = $LDFLAGS.split.join(" ")
   old_dldflags = $DLDFLAGS.split.join(" ")
@@ -531,20 +589,29 @@ if arg_config('--prevent-strip')
 end
 
 # adopt environment config
-append_cflags(ENV["CFLAGS"].split) if !ENV["CFLAGS"].nil?
-append_cppflags(ENV["CPPFLAGS"].split) if !ENV["CPPFLAGS"].nil?
-append_ldflags(ENV["LDFLAGS"].split) if !ENV["LDFLAGS"].nil?
+append_cflags(ENV["CFLAGS"].split) unless ENV["CFLAGS"].nil?
+append_cppflags(ENV["CPPFLAGS"].split) unless ENV["CPPFLAGS"].nil?
+append_ldflags(ENV["LDFLAGS"].split) unless ENV["LDFLAGS"].nil?
 $LIBS = concat_flags($LIBS, ENV["LIBS"])
 
-append_cflags("-g") # always include debugging information
-append_cflags("-Winline") # we use at least one inline function in the C extension
-append_cflags("-Wmissing-noreturn") # good to have no matter what Ruby was compiled with
+# always include debugging information
+append_cflags("-g")
+
+# we use at least one inline function in the C extension
+append_cflags("-Winline")
+
+# good to have no matter what Ruby was compiled with
+append_cflags("-Wmissing-noreturn")
+
+# handle clang variations, see #1101
 append_cflags("-Wno-error=unused-command-line-argument-hard-error-in-future") if darwin?
-# append_cflags(["-Wcast-qual", "-Wwrite-strings"]) # these tend to be noisy, but on occasion useful during development
+
+# these tend to be noisy, but on occasion useful during development
+# append_cflags(["-Wcast-qual", "-Wwrite-strings"])
 
 # Add SDK-specific include path for macOS and brew versions before v2.2.12 (2020-04-08) [#1851, #1801]
 macos_mojave_sdk_include_path = "/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/libxml2"
-if using_system_libraries? && darwin? && Dir.exist?(macos_mojave_sdk_include_path)
+if config_system_libraries? && darwin? && Dir.exist?(macos_mojave_sdk_include_path)
   append_cppflags("-I#{macos_mojave_sdk_include_path}")
 end
 
@@ -552,46 +619,50 @@ end
 # See https://sourceforge.net/p/mingw/bugs/2142
 append_cppflags(' "-Idummypath"') if windows?
 
-if using_system_libraries?
+if config_system_libraries?
   message "Building nokogiri using system libraries.\n"
-  ensure_package_configuration(opt: "zlib", pc: "zlib", lib: "z", headers: "zlib.h", func: "gzdopen")
-  ensure_package_configuration(opt: "xml2", pc: "libxml-2.0", lib: "xml2", headers: "libxml/parser.h", func: "xmlParseDoc")
-  ensure_package_configuration(opt: "xslt", pc: "libxslt", lib: "xslt", headers: "libxslt/xslt.h", func: "xsltParseStylesheetDoc")
-  ensure_package_configuration(opt: "exslt", pc: "libexslt", lib: "exslt", headers: "libexslt/exslt.h", func: "exsltFuncRegister")
-
-  have_libxml_headers?(REQUIRED_LIBXML_VERSION) or
-    abort "ERROR: libxml2 version #{REQUIRED_LIBXML_VERSION} or later is required!"
-  have_libxml_headers?(RECOMMENDED_LIBXML_VERSION) or
-    warn "WARNING: libxml2 version #{RECOMMENDED_LIBXML_VERSION} or later is highly recommended, but proceeding anyway."
+  ensure_package_configuration(opt: "zlib", pc: "zlib", lib: "z",
+    headers: "zlib.h", func: "gzdopen")
+  ensure_package_configuration(opt: "xml2", pc: "libxml-2.0", lib: "xml2",
+    headers: "libxml/parser.h", func: "xmlParseDoc")
+  ensure_package_configuration(opt: "xslt", pc: "libxslt", lib: "xslt",
+    headers: "libxslt/xslt.h", func: "xsltParseStylesheetDoc")
+  ensure_package_configuration(opt: "exslt", pc: "libexslt", lib: "exslt",
+    headers: "libexslt/exslt.h", func: "exsltFuncRegister")
+
+  have_libxml_headers?(REQUIRED_LIBXML_VERSION) ||
+    abort("ERROR: libxml2 version #{REQUIRED_LIBXML_VERSION} or later is required!")
+  have_libxml_headers?(RECOMMENDED_LIBXML_VERSION) ||
+    warn("WARNING: libxml2 version #{RECOMMENDED_LIBXML_VERSION} or later is highly recommended, but proceeding anyway.")
 
 else
   message "Building nokogiri using packaged libraries.\n"
 
-  static_p = enable_config("static", true)
+  static_p = config_static?
   message "Static linking is #{static_p ? "enabled" : "disabled"}.\n"
 
-  cross_build_p = enable_config("cross-build")
+  cross_build_p = config_cross_build?
   message "Cross build is #{cross_build_p ? "enabled" : "disabled"}.\n"
 
-  require 'yaml'
+  require "yaml"
   dependencies = YAML.load_file(File.join(PACKAGE_ROOT_DIR, "dependencies.yml"))
 
-  dir_config('zlib')
+  dir_config("zlib")
 
   if cross_build_p || windows?
     zlib_recipe = process_recipe("zlib", dependencies["zlib"]["version"], static_p, cross_build_p) do |recipe|
       recipe.files = [{
-          url: "http://zlib.net/fossils/#{recipe.name}-#{recipe.version}.tar.gz",
-          sha256: dependencies["zlib"]["sha256"]
-        }]
+        url: "http://zlib.net/fossils/#{recipe.name}-#{recipe.version}.tar.gz",
+        sha256: dependencies["zlib"]["sha256"],
+      }]
       if windows?
         class << recipe
           attr_accessor :cross_build_p
 
           def configure
-            Dir.chdir work_path do
-              mk = File.read 'win32/Makefile.gcc'
-              File.open 'win32/Makefile.gcc', 'wb' do |f|
+            Dir.chdir(work_path) do
+              mk = File.read("win32/Makefile.gcc")
+              File.open("win32/Makefile.gcc", "wb") do |f|
                 f.puts "BINARY_PATH = #{path}/bin"
                 f.puts "LIBRARY_PATH = #{path}/lib"
                 f.puts "INCLUDE_PATH = #{path}/include"
@@ -602,17 +673,17 @@ else
           end
 
           def configured?
-            Dir.chdir work_path do
-              !! (File.read('win32/Makefile.gcc') =~ /^BINARY_PATH/)
+            Dir.chdir(work_path) do
+              !!(File.read("win32/Makefile.gcc") =~ /^BINARY_PATH/)
             end
           end
 
           def compile
-            execute "compile", "make -f win32/Makefile.gcc"
+            execute("compile", "make -f win32/Makefile.gcc")
           end
 
           def install
-            execute "install", "make -f win32/Makefile.gcc install"
+            execute("install", "make -f win32/Makefile.gcc install")
           end
         end
         recipe.cross_build_p = cross_build_p
@@ -620,12 +691,13 @@ else
         class << recipe
           def configure
             cflags = concat_flags(ENV["CFLAGS"], "-fPIC", "-g")
-            execute "configure", ["env", "CHOST=#{host}", "CFLAGS=#{cflags}", "./configure", "--static", configure_prefix]
+            execute("configure",
+              ["env", "CHOST=#{host}", "CFLAGS=#{cflags}", "./configure", "--static", configure_prefix])
           end
 
           def compile
-            if host=~/darwin/
-              execute "compile", "make AR=#{host}-libtool"
+            if host =~ /darwin/
+              execute("compile", "make AR=#{host}-libtool")
             else
               super
             end
@@ -635,49 +707,55 @@ else
     end
 
     unless nix?
-      libiconv_recipe = process_recipe("libiconv", dependencies["libiconv"]["version"], static_p, cross_build_p) do |recipe|
+      libiconv_recipe = process_recipe("libiconv", dependencies["libiconv"]["version"], static_p,
+        cross_build_p) do |recipe|
         recipe.files = [{
-            url: "http://ftp.gnu.org/pub/gnu/libiconv/#{recipe.name}-#{recipe.version}.tar.gz",
-            sha256: dependencies["libiconv"]["sha256"]
-          }]
+          url: "http://ftp.gnu.org/pub/gnu/libiconv/#{recipe.name}-#{recipe.version}.tar.gz",
+          sha256: dependencies["libiconv"]["sha256"],
+        }]
 
         cflags = concat_flags(ENV["CFLAGS"], "-O2", "-U_FORTIFY_SOURCE", "-g")
 
         recipe.configure_options += [
+          "--disable-dependency-tracking",
           "CPPFLAGS=-Wall",
           "CFLAGS=#{cflags}",
           "CXXFLAGS=#{cflags}",
-          "LDFLAGS="
+          "LDFLAGS=",
         ]
       end
     end
-  else
-    if darwin? && !have_header('iconv.h')
-      abort <<~EOM.chomp
-        -----
-        The file "iconv.h" is missing in your build environment,
-        which means you haven't installed Xcode Command Line Tools properly.
-
-        To install Command Line Tools, try running `xcode-select --install` on
-        terminal and follow the instructions.  If it fails, open Xcode.app,
-        select from the menu "Xcode" - "Open Developer Tool" - "More Developer
-        Tools" to open the developer site, download the installer for your OS
-        version and run it.
-        -----
+  elsif darwin? && !have_header("iconv.h")
+    abort(<<~EOM.chomp)
+      -----
+      The file "iconv.h" is missing in your build environment,
+      which means you haven't installed Xcode Command Line Tools properly.
+
+      To install Command Line Tools, try running `xcode-select --install` on
+      terminal and follow the instructions.  If it fails, open Xcode.app,
+      select from the menu "Xcode" - "Open Developer Tool" - "More Developer
+      Tools" to open the developer site, download the installer for your OS
+      version and run it.
+      -----
       EOM
-    end
   end
 
   unless windows?
-    preserving_globals { local_have_library('z', 'gzdopen', 'zlib.h') } or
-      abort 'zlib is missing; necessary for building libxml2'
+    preserving_globals { local_have_library("z", "gzdopen", "zlib.h") } ||
+      abort("zlib is missing; necessary for building libxml2")
   end
 
   libxml2_recipe = process_recipe("libxml2", dependencies["libxml2"]["version"], static_p, cross_build_p) do |recipe|
-    recipe.files = [{
+    source_dir = arg_config("--with-xml2-source-dir")
+    if source_dir
+      recipe.source_directory = source_dir
+    else
+      recipe.files = [{
         url: "http://xmlsoft.org/sources/#{recipe.name}-#{recipe.version}.tar.gz",
-        sha256: dependencies["libxml2"]["sha256"]
+        sha256: dependencies["libxml2"]["sha256"],
       }]
+      recipe.patch_files = Dir[File.join(PACKAGE_ROOT_DIR, "patches", "libxml2", "*.patch")].sort
+    end
 
     cflags = concat_flags(ENV["CFLAGS"], "-O2", "-U_FORTIFY_SOURCE", "-g")
 
@@ -696,6 +774,16 @@ else
       recipe.configure_options += ["RANLIB=/usr/bin/ranlib", "AR=/usr/bin/ar"]
     end
 
+    if windows?
+      cflags = concat_flags(cflags, "-ULIBXML_STATIC", "-DIN_LIBXML")
+    end
+
+    recipe.configure_options << if source_dir
+      "--config-cache"
+    else
+      "--disable-dependency-tracking"
+    end
+
     recipe.configure_options += [
       "--without-python",
       "--without-readline",
@@ -707,10 +795,16 @@ else
   end
 
   libxslt_recipe = process_recipe("libxslt", dependencies["libxslt"]["version"], static_p, cross_build_p) do |recipe|
-    recipe.files = [{
+    source_dir = arg_config("--with-xslt-source-dir")
+    if source_dir
+      recipe.source_directory = source_dir
+    else
+      recipe.files = [{
         url: "http://xmlsoft.org/sources/#{recipe.name}-#{recipe.version}.tar.gz",
-        sha256: dependencies["libxslt"]["sha256"]
+        sha256: dependencies["libxslt"]["sha256"],
       }]
+      recipe.patch_files = Dir[File.join(PACKAGE_ROOT_DIR, "patches", "libxslt", "*.patch")].sort
+    end
 
     cflags = concat_flags(ENV["CFLAGS"], "-O2", "-U_FORTIFY_SOURCE", "-g")
 
@@ -718,6 +812,12 @@ else
       recipe.configure_options += ["RANLIB=/usr/bin/ranlib", "AR=/usr/bin/ar"]
     end
 
+    recipe.configure_options << if source_dir
+      "--config-cache"
+    else
+      "--disable-dependency-tracking"
+    end
+
     recipe.configure_options += [
       "--without-python",
       "--without-crypto",
@@ -738,39 +838,39 @@ else
       libname = recipe.name[/\Alib(.+)\z/, 1]
       File.join(recipe.path, "bin", "#{libname}-config").tap do |config|
         # call config scripts explicit with 'sh' for compat with Windows
-        $CPPFLAGS = `sh #{config} --cflags`.strip << ' ' << $CPPFLAGS
-        `sh #{config} --libs`.strip.shellsplit.each do |arg|
+        $CPPFLAGS = %x(sh #{config} --cflags).strip << " " << $CPPFLAGS
+        %x(sh #{config} --libs).strip.shellsplit.each do |arg|
           case arg
           when /\A-L(.+)\z/
             # Prioritize ports' directories
-            if $1.start_with?(PACKAGE_ROOT_DIR + '/')
-              $LIBPATH = [$1] | $LIBPATH
+            $LIBPATH = if Regexp.last_match(1).start_with?(PACKAGE_ROOT_DIR + "/")
+              [Regexp.last_match(1)] | $LIBPATH
             else
-              $LIBPATH = $LIBPATH | [$1]
+              $LIBPATH | [Regexp.last_match(1)]
             end
           when /\A-l./
             libs.unshift(arg)
           else
-            $LDFLAGS << ' ' << arg.shellescape
+            $LDFLAGS << " " << arg.shellescape
           end
         end
       end
 
-      patches_string = recipe.patch_files.map { |path| File.basename(path) }.join(' ')
-      append_cppflags(%Q[-DNOKOGIRI_#{recipe.name.upcase}_PATCHES="\\\"#{patches_string}\\\""])
+      patches_string = recipe.patch_files.map { |path| File.basename(path) }.join(" ")
+      append_cppflags(%[-DNOKOGIRI_#{recipe.name.upcase}_PATCHES="\\\"#{patches_string}\\\""])
 
       case libname
-      when 'xml2'
+      when "xml2"
         # xslt-config --libs or pkg-config libxslt --libs does not include
         # -llzma, so we need to add it manually when linking statically.
-        if static_p && preserving_globals { local_have_library('lzma') }
+        if static_p && preserving_globals { local_have_library("lzma") }
           # Add it at the end; GH #988
-          libs << '-llzma'
+          libs << "-llzma"
         end
-      when 'xslt'
+      when "xslt"
         # xslt-config does not have a flag to emit options including
         # -lexslt, so add it manually.
-        libs.unshift('-lexslt')
+        libs.unshift("-lexslt")
       end
     end
   end.shelljoin
@@ -778,10 +878,10 @@ else
   if static_p
     $libs = $libs.shellsplit.map do |arg|
       case arg
-      when '-lxml2'
-        File.join(libxml2_recipe.path, 'lib', libflag_to_filename(arg))
-      when '-lxslt', '-lexslt'
-        File.join(libxslt_recipe.path, 'lib', libflag_to_filename(arg))
+      when "-lxml2"
+        File.join(libxml2_recipe.path, "lib", libflag_to_filename(arg))
+      when "-lxslt", "-lexslt"
+        File.join(libxslt_recipe.path, "lib", libflag_to_filename(arg))
       else
         arg
       end
@@ -793,44 +893,92 @@ else
   ensure_func("exsltFuncRegister", "libexslt/exslt.h")
 end
 
-have_func('xmlHasFeature') or abort("xmlHasFeature() is missing.") # introduced in libxml 2.6.21
-have_func('xmlFirstElementChild') # introduced in libxml 2.7.3
-have_func('xmlRelaxNGSetParserStructuredErrors') # introduced in libxml 2.6.24
-have_func('xmlRelaxNGSetValidStructuredErrors') # introduced in libxml 2.6.21
-have_func('xmlSchemaSetValidStructuredErrors') # introduced in libxml 2.6.23
-have_func('xmlSchemaSetParserStructuredErrors') # introduced in libxml 2.6.23
+libgumbo_recipe = process_recipe("libgumbo", "1.0.0-nokogiri", static_p, cross_build_p, false) do |recipe|
+  recipe.configure_options = []
+
+  class << recipe
+    def downloaded?
+      true
+    end
+
+    def extract
+      target = File.join(tmp_path, "gumbo-parser")
+      output("Copying gumbo-parser files into #{target}...")
+      FileUtils.mkdir_p(target)
+      FileUtils.cp(Dir.glob(File.join(PACKAGE_ROOT_DIR, "gumbo-parser/src/*")), target)
+    end
+
+    def configured?
+      true
+    end
+
+    def install
+      lib_dir = File.join(port_path, "lib")
+      inc_dir = File.join(port_path, "include")
+      FileUtils.mkdir_p([lib_dir, inc_dir])
+      FileUtils.cp(File.join(work_path, "libgumbo.a"), lib_dir)
+      FileUtils.cp(Dir.glob(File.join(work_path, "*.h")), inc_dir)
+    end
+
+    def compile
+      cflags = concat_flags(ENV["CFLAGS"], "-fPIC", "-g")
+
+      env = { "CC" => gcc_cmd, "CFLAGS" => cflags }
+      if config_cross_build?
+        if host =~ /darwin/
+          env["AR"] = "#{host}-libtool"
+          env["ARFLAGS"] = "-o"
+        else
+          env["AR"] = "#{host}-ar"
+        end
+        env["RANLIB"] = "#{host}-ranlib"
+      end
+
+      execute("compile", make_cmd, { env: env })
+    end
+  end
+end
+append_cppflags("-I#{File.join(libgumbo_recipe.path, "include")}")
+$libs = $libs + " " + File.join(libgumbo_recipe.path, "lib", "libgumbo.a")
+$LIBPATH = $LIBPATH | [File.join(libgumbo_recipe.path, "lib")]
+ensure_func("gumbo_parse_with_options", "gumbo.h")
+
+have_func("xmlHasFeature") || abort("xmlHasFeature() is missing.") # introduced in libxml 2.6.21
+have_func("xmlFirstElementChild") # introduced in libxml 2.7.3
+have_func("xmlRelaxNGSetParserStructuredErrors") # introduced in libxml 2.6.24
+have_func("xmlRelaxNGSetValidStructuredErrors") # introduced in libxml 2.6.21
+have_func("xmlSchemaSetValidStructuredErrors") # introduced in libxml 2.6.23
+have_func("xmlSchemaSetParserStructuredErrors") # introduced in libxml 2.6.23
 
-have_func('vasprintf')
+have_func("vasprintf")
 
-other_library_versions_string = OTHER_LIBRARY_VERSIONS.map { |k,v| [k,v].join(":") }.join(",")
-append_cppflags(%Q[-DNOKOGIRI_OTHER_LIBRARY_VERSIONS="\\\"#{other_library_versions_string}\\\""])
+other_library_versions_string = OTHER_LIBRARY_VERSIONS.map { |k, v| [k, v].join(":") }.join(",")
+append_cppflags(%[-DNOKOGIRI_OTHER_LIBRARY_VERSIONS="\\\"#{other_library_versions_string}\\\""])
 
-unless using_system_libraries?
+unless config_system_libraries?
   if cross_build_p
     # When precompiling native gems, copy packaged libraries' headers to ext/nokogiri/include
     # These are packaged up by the cross-compiling callback in the ExtensionTask
     copy_packaged_libraries_headers(to_path: File.join(PACKAGE_ROOT_DIR, "ext/nokogiri/include"),
-                                    from_recipes: [libxml2_recipe, libxslt_recipe])
+      from_recipes: [libxml2_recipe, libxslt_recipe])
   else
     # When compiling during installation, install packaged libraries' header files into ext/nokogiri/include
     copy_packaged_libraries_headers(to_path: "include",
-                                    from_recipes: [libxml2_recipe, libxslt_recipe])
+      from_recipes: [libxml2_recipe, libxslt_recipe])
     $INSTALLFILES << ["include/**/*.h", "$(rubylibdir)"]
   end
 end
 
-create_makefile('nokogiri/nokogiri')
+create_makefile("nokogiri/nokogiri")
 
-if enable_config('clean', true)
+if config_clean?
   # Do not clean if run in a development work tree.
-  File.open('Makefile', 'at') do |mk|
-    mk.print <<~EOF
+  File.open("Makefile", "at") do |mk|
+    mk.print(<<~EOF)
 
       all: clean-ports
       clean-ports: $(DLLIB)
-      \t-$(Q)$(RUBY) $(srcdir)/extconf.rb --clean --#{static_p ? 'enable' : 'disable'}-static
+      \t-$(Q)$(RUBY) $(srcdir)/extconf.rb --clean --#{static_p ? "enable" : "disable"}-static
     EOF
   end
 end
-
-# :startdoc: