nokogumbo-import: restructure as a mini_portile recipe

- move nokogumbo.c to ext/nokogiri/gumbo.c
- remove the nokogumbo extension
- treat the gumbo parser as a mini_portile recipe

I'm not 100% sure this is the right thing to do, but let's see how far
I get on cross-compiling before I give up.

ext/nokogiri/extconf.rb

@@ -402,7 +402,7 @@ def process_recipe(name, version, static_p, cross_p)
   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
 
@@ -486,7 +486,7 @@ def process_recipe(name, version, static_p, cross_p)
         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
@@ -498,7 +498,7 @@ def process_recipe(name, version, static_p, cross_p)
       EOM
 
       message(<<~EOM) if name == 'libxml2'
-        Note, however, that nokogiri cannot guarantee compatiblity with every
+        Note, however, that nokogiri cannot guarantee compatibility with every
         version of libxml2 that may be provided by OS/package vendors.
 
       EOM
@@ -865,6 +865,45 @@ def compile
   ensure_func("exsltFuncRegister", "libexslt/exslt.h")
 end
 
+libgumbo_recipe = process_recipe("libgumbo", "1.0.0-nokogiri", static_p, cross_build_p) 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")
+      command = [make_cmd, "CC=#{gcc_cmd}", "CFLAGS=#{cflags}"]
+      execute("compile", command)
+    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

ext/nokogumbo/nokogumbo.c → ext/nokogiri/gumbo.c

@@ -27,14 +27,11 @@
 // requirements as Ruby objects are only built when necessary.
 //
 
-#include <assert.h>
-#include <ruby.h>
-#include <ruby/version.h>
+#include <nokogiri.h>
 
 #include "gumbo.h"
 
-// class constants
-static VALUE Document;
+VALUE cNokogiriHtml5Document;
 
 // Interned symbols
 static ID internal_subset;
@@ -384,7 +381,7 @@ parse_continue(VALUE parse_args)
   }
   args->doc = doc; // Make sure doc gets cleaned up if an error is thrown.
   build_tree(doc, (xmlNodePtr)doc, output->document);
-  VALUE rdoc = Nokogiri_wrap_xml_document(Document, doc);
+  VALUE rdoc = Nokogiri_wrap_xml_document(cNokogiriHtml5Document, doc);
   args->doc = NULL; // The Ruby runtime now owns doc so don't delete it.
   add_errors(output, rdoc, args->input, args->url_or_frag);
   return rdoc;
@@ -591,21 +588,19 @@ fragment_continue(VALUE parse_args)
 
 // Initialize the Nokogumbo class and fetch constants we will use later.
 void
-Init_nokogumbo()
+noko_init_gumbo()
 {
   // Class constants.
-  VALUE HTML5 = rb_const_get(mNokogiri, rb_intern_const("HTML5"));
-  Document = rb_const_get(HTML5, rb_intern_const("Document"));
-  rb_gc_register_mark_object(Document);
+  cNokogiriHtml5Document = rb_define_class_under(mNokogiriHtml5, "Document", cNokogiriHtmlDocument);
+  rb_gc_register_mark_object(cNokogiriHtml5Document);
 
   // Interned symbols.
   internal_subset = rb_intern_const("internal_subset");
   parent = rb_intern_const("parent");
 
   // Define Nokogumbo module with parse and fragment methods.
-  VALUE Gumbo = rb_define_module_under(mNokogiri, "Gumbo");
-  rb_define_singleton_method(Gumbo, "parse", parse, 5);
-  rb_define_singleton_method(Gumbo, "fragment", fragment, 6);
+  rb_define_singleton_method(mNokogiriGumbo, "parse", parse, 5);
+  rb_define_singleton_method(mNokogiriGumbo, "fragment", fragment, 6);
 }
 
 // vim: set shiftwidth=2 softtabstop=2 tabstop=8 expandtab:

ext/nokogiri/nokogiri.c

@@ -1,8 +1,10 @@
 #include <nokogiri.h>
 
 VALUE mNokogiri ;
+VALUE mNokogiriGumbo ;
 VALUE mNokogiriHtml ;
 VALUE mNokogiriHtmlSax ;
+VALUE mNokogiriHtml5 ;
 VALUE mNokogiriXml ;
 VALUE mNokogiriXmlSax ;
 VALUE mNokogiriXmlXpath ;
@@ -44,6 +46,7 @@ void noko_init_html_element_description();
 void noko_init_html_entity_lookup();
 void noko_init_html_sax_parser_context();
 void noko_init_html_sax_push_parser();
+void noko_init_gumbo();
 void noko_init_test_global_handlers();
 
 static ID id_read, id_write;
@@ -152,12 +155,14 @@ void
 Init_nokogiri()
 {
   mNokogiri         = rb_define_module("Nokogiri");
-  mNokogiriXml      = rb_define_module_under(mNokogiri, "XML");
+  mNokogiriGumbo    = rb_define_module_under(mNokogiri, "Gumbo");
   mNokogiriHtml     = rb_define_module_under(mNokogiri, "HTML");
-  mNokogiriXslt     = rb_define_module_under(mNokogiri, "XSLT");
-  mNokogiriXmlXpath = rb_define_module_under(mNokogiriXml, "XPath");
-  mNokogiriXmlSax   = rb_define_module_under(mNokogiriXml, "SAX");
   mNokogiriHtmlSax  = rb_define_module_under(mNokogiriHtml, "SAX");
+  mNokogiriHtml5    = rb_define_module_under(mNokogiri, "HTML5");
+  mNokogiriXml      = rb_define_module_under(mNokogiri, "XML");
+  mNokogiriXmlSax   = rb_define_module_under(mNokogiriXml, "SAX");
+  mNokogiriXmlXpath = rb_define_module_under(mNokogiriXml, "XPath");
+  mNokogiriXslt     = rb_define_module_under(mNokogiri, "XSLT");
 
   rb_const_set(mNokogiri, rb_intern("LIBXML_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION));
   rb_const_set(mNokogiri, rb_intern("LIBXML_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xmlParserVersion));
@@ -238,6 +243,7 @@ Init_nokogiri()
   noko_init_xml_document_fragment();
   noko_init_xml_document();
   noko_init_html_document();
+  noko_init_gumbo();
 
   noko_init_test_global_handlers();
 

ext/nokogiri/nokogiri.h

@@ -69,6 +69,7 @@ xmlNodePtr xmlLastElementChild(xmlNodePtr parent);
 #include <ruby/st.h>
 #include <ruby/encoding.h>
 #include <ruby/util.h>
+#include <ruby/version.h>
 
 #define NOKOGIRI_STR_NEW2(str) NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
 #define NOKOGIRI_STR_NEW(str, len) rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
@@ -92,10 +93,13 @@ xmlNodePtr xmlLastElementChild(xmlNodePtr parent);
 
 
 NOKOPUBVAR VALUE mNokogiri ;
+NOKOPUBVAR VALUE mNokogiriGumbo ;
 NOKOPUBVAR VALUE mNokogiriHtml ;
 NOKOPUBVAR VALUE mNokogiriHtmlSax ;
+NOKOPUBVAR VALUE mNokogiriHtml5 ;
 NOKOPUBVAR VALUE mNokogiriXml ;
 NOKOPUBVAR VALUE mNokogiriXmlSax ;
+NOKOPUBVAR VALUE mNokogiriXmlXpath ;
 NOKOPUBVAR VALUE mNokogiriXslt ;
 
 NOKOPUBVAR VALUE cNokogiriSyntaxError;
@@ -129,6 +133,7 @@ NOKOPUBVAR VALUE cNokogiriXmlXpathSyntaxError;
 NOKOPUBVAR VALUE cNokogiriXsltStylesheet ;
 
 NOKOPUBVAR VALUE cNokogiriHtmlDocument ;
+NOKOPUBVAR VALUE cNokogiriHtml5Document ;
 NOKOPUBVAR VALUE cNokogiriHtmlSaxPushParser ;
 NOKOPUBVAR VALUE cNokogiriHtmlElementDescription ;
 NOKOPUBVAR VALUE cNokogiriHtmlSaxParserContext;

gumbo-parser/src/Makefile

@@ -0,0 +1,13 @@
+# this Makefile is used by ext/nokogiri/extconf.rb
+# to enable a mini_portile2 recipe to build the gumbo parser
+.PHONY: clean
+
+override CFLAGS += -std=c99 -Wall
+
+gumbo_objs := $(patsubst %.c,%.o,$(wildcard *.c))
+
+libgumbo.a: $(gumbo_objs)
+	ar rcs $@ $^
+
+clean:
+	rm -f $(gumbo_objs) libgumbo.a

lib/nokogiri.rb

@@ -21,7 +21,6 @@
 
 if Nokogiri.uses_gumbo?
   require 'nokogiri/html5'
-  require 'nokogumbo/nokogumbo'
 end
 
 # Nokogiri parses and searches XML/HTML very quickly, and also has

rakelib/extensions.rake

@@ -372,9 +372,4 @@ else
       end
     end
   end
-
-  Rake::ExtensionTask.new 'nokogumbo' do |e|
-    e.lib_dir = 'lib/nokogumbo'
-    e.source_pattern = '{,../../gumbo-parser/src/}*.[hc]'
-  end
 end