Autotools gumbo

.github/workflows/ci.yml

@@ -46,6 +46,7 @@ jobs:
       - uses: actions/checkout@v1 # v1 because of https://github.com/actions/checkout/issues/334
         with:
           submodules: true
+      - run: apt install -y autoconf
       - run: bundle install --local || bundle install
       - run: bundle exec rake compile -- --${{matrix.sys}}-system-libraries
       - run: bundle exec rake test

ext/nokogiri/extconf.rb

@@ -1055,50 +1055,8 @@ def configure
   $VPATH << "$(srcdir)/../../gumbo-parser/src"
   find_header("nokogiri_gumbo.h") || abort("nokogiri_gumbo.h not found")
 else
-  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", "-O2", "-g")
-
-        env = { "CC" => gcc_cmd, "CFLAGS" => cflags }
-        if config_cross_build?
-          if host.include?("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
+  libgumbo_recipe = process_recipe("libgumbo", "1.0.0-nokogiri", true, cross_build_p, false) do |recipe|
+    recipe.source_directory = File.join(PACKAGE_ROOT_DIR, "gumbo-parser")
   end
   append_cppflags("-I#{File.join(libgumbo_recipe.path, "include")}")
   $libs = $libs + " " + File.join(libgumbo_recipe.path, "lib", "libgumbo.a")
@@ -1142,6 +1100,7 @@ def compile
   File.open("Makefile", "at") do |mk|
     mk.print(<<~EOF)
 
+      .PHONY: clean-ports
       all: clean-ports
       clean-ports: $(DLLIB)
       \t-$(Q)$(RUBY) $(srcdir)/extconf.rb --clean --#{static_p ? "enable" : "disable"}-static

gumbo-parser/.gitignore

@@ -1,5 +1,10 @@
-build
-googletest
-src/*.o
-fuzzer/build
-src/libgumbo.a
+/build
+/googletest
+/src/*.o
+/fuzzer/build
+/src/libgumbo.a
+/aclocal.m4
+/autom4te.cache/
+/configure
+/configure.in
+Makefile.in

gumbo-parser/Makefile.am

@@ -0,0 +1,5 @@
+if HAS_TESTS
+  SUBDIRS = src test
+else
+  SUBDIRS = src
+endif

gumbo-parser/build-aux/.gitignore

@@ -0,0 +1,4 @@
+# Ignore everything in this directory except for this file.
+# Credit: https://stackoverflow.com/a/932982
+*
+!/.gitignore

gumbo-parser/configure.ac

@@ -0,0 +1,11 @@
+AC_INIT([Gumbo], [1.0])
+AC_CONFIG_SRCDIR([src/nokogiri_gumbo.h])
+AC_CONFIG_AUX_DIR([build-aux])
+AM_INIT_AUTOMAKE([subdir-objects foreign serial-tests no-dist no-installinfo no-installman -Wall])
+AC_PROG_RANLIB
+AC_PROG_CC
+AC_PROG_CXX
+AM_PROG_AR
+AM_CONDITIONAL([HAS_TESTS], [test -d "${srcdir}/test"])
+AC_CONFIG_FILES([Makefile src/Makefile test/Makefile])
+AC_OUTPUT

gumbo-parser/src/Makefile.am

@@ -0,0 +1,20 @@
+lib_LIBRARIES = libgumbo.a
+libgumbo_a_SOURCES = \
+	ascii.c \
+	attribute.c \
+	char_ref.c \
+	error.c \
+	foreign_attrs.c \
+	parser.c \
+	string_buffer.c \
+	string_piece.c \
+	svg_attrs.c \
+	svg_tags.c \
+	tag.c \
+	tag_lookup.c \
+	token_buffer.c \
+	tokenizer.c \
+	utf8.c \
+	util.c \
+	vector.c
+include_HEADERS = nokogiri_gumbo.h

gumbo-parser/test/Makefile.am

@@ -0,0 +1,21 @@
+check_LIBRARIES = libgtest_main.a
+libgtest_main_a_SOURCES = ../googletest/src/gtest-all.cc ../googletest/src/gtest_main.cc
+libgtest_main_a_CPPFLAGS = -I$(top_srcdir)/googletest/include -I$(top_srcdir)/googletest
+libgtest_main_a_CXXFLAGS = -pthread
+
+check_PROGRAMS = gumbotest
+gumbotest_SOURCES = attribute.cc \
+    parser.cc \
+    string_buffer.cc \
+    string_piece.cc \
+    test_utils.cc \
+    token_buffer.cc \
+    tokenizer.cc \
+    utf8.cc \
+    vector.cc
+
+gumbotest_LDADD = libgtest_main.a ../src/libgumbo.a
+gumbotest_LDFLAGS = -pthread
+gumbotest_CPPFLAGS = -I$(top_srcdir)/googletest/include -I$(top_srcdir)/src
+
+TESTS = gumbotest

nokogiri.gemspec

@@ -182,9 +182,13 @@ Gem::Specification.new do |spec|
     "ext/nokogiri/xml_xpath_context.c",
     "ext/nokogiri/xslt_stylesheet.c",
     "gumbo-parser/CHANGES.md",
-    "gumbo-parser/Makefile",
+    "gumbo-parser/Makefile.am",
+    "gumbo-parser/Makefile.in",
     "gumbo-parser/THANKS",
-    "gumbo-parser/src/Makefile",
+    "gumbo-parser/configure",
+    "gumbo-parser/configure.ac",
+    "gumbo-parser/src/Makefile.am",
+    "gumbo-parser/src/Makefile.in",
     "gumbo-parser/src/README.md",
     "gumbo-parser/src/ascii.c",
     "gumbo-parser/src/ascii.h",

rakelib/check-manifest.rake

@@ -56,7 +56,12 @@ task :check_manifest, [:verbose] do |_, args|
     [0-9]*
     appveyor.yml
     **/compile_commands.json
+    gumbo-parser/Makefile.old
+    gumbo-parser/aclocal.m4
+    gumbo-parser/autom4te.cache/*
+    gumbo-parser/build-aux/*
     gumbo-parser/fuzzer/*
+    gumbo-parser/googletest/*
     gumbo-parser/test/*
     gumbo-parser/gperf-filter.sed
     lib/nokogiri/**/nokogiri.{jar,so}

rakelib/extensions.rake

@@ -444,7 +444,7 @@ else
   end
 
   Rake::ExtensionTask.new("nokogiri", NOKOGIRI_SPEC.dup) do |ext|
-    ext.source_pattern = "*.{c,cc,cpp,h}"
+    ext.source_pattern = "{.,../../gumbo-parser/src}/*.{c,cc,cpp,h}"
     ext.gem_spec.files.reject! { |path| File.fnmatch?("**/*.{java,jar}", path, File::FNM_EXTGLOB) }
 
     ext.lib_dir = File.join(*["lib", "nokogiri", ENV["FAT_DIR"]].compact)

rakelib/gumbo.rake

@@ -1,13 +1,18 @@
 # frozen_string_literal: true
 
+# We want to run the gumbo test suite using exactly the same compiled gumbo-parser
+# that Nokogiri uses.
+#
+# To that end, we first need to get the Rake ExtensionTask to run extconf.rb which will
+# run the gumbo-parser's configure script. We don't want to compile the extension
+# at this point, so we make `gumbo:test` depend on the Nokogiri Makefile.
 namespace "gumbo" do
-  gtest_pkg = "gumbo-parser/googletest"
-  gtest_lib = File.join(gtest_pkg, "make/gtest_main.a")
-
-  file gtest_lib => gtest_pkg do
-    sh("make -C gumbo-parser/googletest/make gtest_main.a")
-  end
+  host = RbConfig::CONFIG["host"]
+  nokogiri_makefile = File.join("tmp/#{RUBY_PLATFORM}/nokogiri/#{RUBY_VERSION}/Makefile")
+  gumbotest_builddir = "tmp/#{RUBY_PLATFORM}/nokogiri/#{RUBY_VERSION}/tmp/#{host}/ports/libgumbo/1.0.0-nokogiri/libgumbo-1.0.0-nokogiri"
+  gumbotest_configure = File.absolute_path("gumbo-parser/configure")
 
+  gtest_pkg = "gumbo-parser/googletest"
   file gtest_pkg do
     sh(<<~EOF)
       curl -L https://github.com/google/googletest/archive/release-1.8.0.tar.gz | \
@@ -16,19 +21,26 @@ namespace "gumbo" do
   end
 
   desc "Run the gumbo parser test suite"
-  task "test" => gtest_lib do
-    sh("make -j2 -C gumbo-parser")
+  task "test" => [nokogiri_makefile, gtest_pkg] do
+    sh("make", "-j2", "-C", gumbotest_builddir, "check")
   end
 
-  desc "Clean up after the gumbo parser test suite"
-  task "clean" do
-    sh("make -j2 -C gumbo-parser clean")
+  file gumbotest_configure do
+    sh("autoreconf", "-fiv", chdir: "gumbo-parser")
   end
 
+  # Make sure the libgumbo configure script is created before trying to compile the extension.
+  file nokogiri_makefile => gumbotest_configure
+
   CLOBBER.add(gtest_pkg)
+  CLOBBER.add(gumbotest_configure)
+  CLOBBER.add("gumbo-parser/Makefile.in")
+  CLOBBER.add("gumbo-parser/src/Makefile.in")
+  CLOBBER.add("gumbo-parser/test/Makefile.in")
+  CLOBBER.add("gumbo-parser/build-aux/*")
+  CLOBBER.add("gumbo-parser/autom4te.cache")
+  CLOBBER.add("gumbo-parser/aclocal.m4")
 end
 
 desc "Run the gumbo parser test suite"
 task "gumbo" => "gumbo:test"
-
-task "clean" => "gumbo:clean" # rubocop:disable Rake/Desc