nokogiri 1.14.3 → 1.15.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 55cb24234324fd658259d4835cfb45f0585efd60ca7f76a40f4955843ff35a09
-  data.tar.gz: 3531c056ef67437b5b104fa5d2553c924757d910578a1ef975fb2116c40e62d9
+  metadata.gz: cad4c790fb66865a2b47c27345cdb2785c61b9a599490fcbef62150e83078547
+  data.tar.gz: da81f866cd1451d0583b8fd376625185052c8d17f7a3b6250c43fec7d906daf8
 SHA512:
-  metadata.gz: aa0062b110af74e67a8cc3d1aa84d60bc30d788ce1eb8a9698db675b204f40f59054326fc7ede103dae16c5768c8482af24a9038b91b25a98e765a4ad7c3c7bb
-  data.tar.gz: '0783632d61fd733cc634a2fd82c02b46c6265d8e4db1ed3853891147eedaee18e825b102923b86919408b6fe33ad5c9762555d53ee39b05bacb27bedfbef298f'
+  metadata.gz: 5b7e4a97225c6a4516fc298a9826d890fd652d1adcd7a2b90e02979e46015114ba9f63d9fdd34b5d398b562a23745386c77ba90165f8e214354a149758b88bfa
+  data.tar.gz: a7127084785dbd6761e87d14b57fe17fa338babba7a05e8f724bfeb749b7ce4344681bb9429eca6c9bc6defd27f6238d4c785a0657656a7d9ca7e531926d1423

data/Gemfile

@@ -20,19 +20,20 @@ group :development do
   gem "rexical", "= 1.0.7"
 
   # tests
-  gem "minitest", "5.17.0"
-  gem "minitest-reporters", "= 1.5.0"
-  gem "ruby_memcheck", "1.2.0" unless RUBY_PLATFORM == "java"
-  gem "simplecov", "= 0.21.2"
+  gem "minitest", "5.18.0"
+  gem "minitest-reporters", "1.6.0"
+  gem "ruby_memcheck", git: "https://github.com/flavorjones/ruby_memcheck", ref: "flavorjones-nokogiri-huge-parse-option"
   gem "rubyzip", "~> 2.3.2"
+  gem "simplecov", "= 0.21.2"
 
   # rubocop
   if Gem::Requirement.new("~> 3.0").satisfied_by?(Gem::Version.new(RUBY_VERSION))
-    gem "rubocop", "1.44.1"
-    gem "rubocop-minitest", "0.27.0"
-    gem "rubocop-performance", "1.15.2"
+    gem "rubocop", "1.50.2"
+    gem "rubocop-minitest", "0.31.0"
+    gem "rubocop-packaging", "0.5.2"
+    gem "rubocop-performance", "1.17.1"
     gem "rubocop-rake", "= 0.6.0"
-    gem "rubocop-shopify", "2.10.1"
+    gem "rubocop-shopify", "2.13.0"
   end
 end
 

data/dependencies.yml

@@ -1,12 +1,12 @@
 libxml2:
-  version: "2.10.4"
-  sha256: "ed0c91c5845008f1936739e4eee2035531c1c94742c6541f44ee66d885948d45"
-  # sha-256 hash provided in https://download.gnome.org/sources/libxml2/2.10/libxml2-2.10.4.sha256sum
+  version: "2.11.3"
+  sha256: "f1acae1664bda006cd81bfc238238217043d586d06659d5c0e3d1bcebe040870"
+  # sha-256 hash provided in https://download.gnome.org/sources/libxml2/2.11/libxml2-2.11.3.sha256sum
 
 libxslt:
-  version: "1.1.37"
-  sha256: "3a4b27dc8027ccd6146725950336f1ec520928f320f144eb5fa7990ae6123ab4"
-  # sha-256 hash provided in https://download.gnome.org/sources/libxslt/1.1/libxslt-1.1.37.sha256sum
+  version: "1.1.38"
+  sha256: "1f32450425819a09acaff2ab7a5a7f8a2ec7956e505d7beeb45e843d0e1ecab1"
+  # sha-256 hash provided in https://download.gnome.org/sources/libxslt/1.1/libxslt-1.1.38.sha256sum
 
 zlib:
   version: "1.2.13"

data/ext/nokogiri/extconf.rb

@@ -15,7 +15,7 @@ PACKAGE_ROOT_DIR = File.expand_path(File.join(File.dirname(__FILE__), "..", ".."
 REQUIRED_LIBXML_VERSION = "2.6.21"
 RECOMMENDED_LIBXML_VERSION = "2.9.3"
 
-REQUIRED_MINI_PORTILE_VERSION = "~> 2.8.0" # keep this version in sync with the one in the gemspec
+REQUIRED_MINI_PORTILE_VERSION = "~> 2.8.2" # keep this version in sync with the one in the gemspec
 REQUIRED_PKG_CONFIG_VERSION = "~> 1.1"
 
 # Keep track of what versions of what libraries we build against
@@ -229,7 +229,7 @@ def gnome_source
   if ENV["NOKOGIRI_USE_CANONICAL_GNOME_SOURCE"]
     "https://download.gnome.org"
   else
-    "https://mirror.csclub.uwaterloo.ca/gnome" # old reliable
+    "https://muug.ca/mirror/gnome" # old reliable
   end
 end
 
@@ -372,7 +372,7 @@ def try_link_iconv(using = nil)
       preserving_globals do
         yield if block_given?
 
-        try_link(<<~'SRC', opt)
+        try_link(<<~SRC, opt)
           #include <stdlib.h>
           #include <iconv.h>
           int main(void)
@@ -695,14 +695,34 @@ append_cppflags(' "-Idummypath"') if windows?
 
 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")
+  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!")
@@ -785,8 +805,12 @@ 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: "https://ftp.gnu.org/pub/gnu/libiconv/#{recipe.name}-#{recipe.version}.tar.gz",
           sha256: dependencies["libiconv"]["sha256"],
@@ -824,15 +848,25 @@ else
   if zlib_recipe
     append_cppflags("-I#{zlib_recipe.path}/include")
     $LIBPATH = ["#{zlib_recipe.path}/lib"] | $LIBPATH
-    ensure_package_configuration(opt: "zlib", pc: "zlib", lib: "z",
-      headers: "zlib.h", func: "gzdopen")
+    ensure_package_configuration(
+      opt: "zlib",
+      pc: "zlib",
+      lib: "z",
+      headers: "zlib.h",
+      func: "gzdopen",
+    )
   end
 
   if libiconv_recipe
     append_cppflags("-I#{libiconv_recipe.path}/include")
     $LIBPATH = ["#{libiconv_recipe.path}/lib"] | $LIBPATH
-    ensure_package_configuration(opt: "iconv", pc: "iconv", lib: "iconv",
-      headers: "iconv.h", func: "iconv_open")
+    ensure_package_configuration(
+      opt: "iconv",
+      pc: "iconv",
+      lib: "iconv",
+      headers: "iconv.h",
+      func: "iconv_open",
+    )
   end
 
   libxml2_recipe = process_recipe("libxml2", dependencies["libxml2"]["version"], static_p, cross_build_p) do |recipe|
@@ -848,8 +882,13 @@ else
       recipe.patch_files = Dir[File.join(PACKAGE_ROOT_DIR, "patches", "libxml2", "*.patch")].sort
     end
 
+    cppflags = concat_flags(ENV["CPPFLAGS"])
     cflags = concat_flags(ENV["CFLAGS"], "-O2", "-U_FORTIFY_SOURCE", "-g")
 
+    if cross_build_p
+      cppflags = concat_flags(cppflags, "-DNOKOGIRI_PRECOMPILED_LIBRARIES")
+    end
+
     if zlib_recipe
       recipe.configure_options << "--with-zlib=#{zlib_recipe.path}"
     end
@@ -880,6 +919,7 @@ else
       "--with-c14n",
       "--with-debug",
       "--with-threads",
+      "CPPFLAGS=#{cppflags}",
       "CFLAGS=#{cflags}",
     ]
   end
@@ -1023,7 +1063,7 @@ libgumbo_recipe = process_recipe("libgumbo", "1.0.0-nokogiri", static_p, cross_b
 
       env = { "CC" => gcc_cmd, "CFLAGS" => cflags }
       if config_cross_build?
-        if /darwin/.match?(host)
+        if host.include?("darwin")
           env["AR"] = "#{host}-libtool"
           env["ARFLAGS"] = "-o"
         else
@@ -1057,12 +1097,16 @@ 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])
+    copy_packaged_libraries_headers(
+      to_path: File.join(PACKAGE_ROOT_DIR, "ext/nokogiri/include"),
+      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])
+    copy_packaged_libraries_headers(
+      to_path: "include",
+      from_recipes: [libxml2_recipe, libxslt_recipe],
+    )
     $INSTALLFILES << ["include/**/*.h", "$(rubylibdir)"]
   end
 end

data/ext/nokogiri/html4_document.c

@@ -144,8 +144,7 @@ rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE r
 static VALUE
 rb_html_document_type(VALUE self)
 {
-  htmlDocPtr doc;
-  Data_Get_Struct(self, xmlDoc, doc);
+  htmlDocPtr doc = noko_xml_document_unwrap(self);
   return INT2NUM(doc->type);
 }
 

data/ext/nokogiri/html4_element_description.c

@@ -1,5 +1,10 @@
 #include <nokogiri.h>
 
+static const rb_data_type_t html4_element_description_type = {
+  .wrap_struct_name = "Nokogiri::HTML4::ElementDescription",
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
 VALUE cNokogiriHtml4ElementDescription ;
 
 /*
@@ -15,7 +20,7 @@ required_attributes(VALUE self)
   VALUE list;
   int i;
 
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   list = rb_ary_new();
 
@@ -41,7 +46,7 @@ deprecated_attributes(VALUE self)
   VALUE list;
   int i;
 
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   list = rb_ary_new();
 
@@ -67,7 +72,7 @@ optional_attributes(VALUE self)
   VALUE list;
   int i;
 
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   list = rb_ary_new();
 
@@ -90,7 +95,7 @@ static VALUE
 default_sub_element(VALUE self)
 {
   const htmlElemDesc *description;
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   if (description->defaultsubelt) {
     return NOKOGIRI_STR_NEW2(description->defaultsubelt);
@@ -112,7 +117,7 @@ sub_elements(VALUE self)
   VALUE list;
   int i;
 
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   list = rb_ary_new();
 
@@ -135,7 +140,7 @@ static VALUE
 description(VALUE self)
 {
   const htmlElemDesc *description;
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   return NOKOGIRI_STR_NEW2(description->desc);
 }
@@ -150,7 +155,7 @@ static VALUE
 inline_eh(VALUE self)
 {
   const htmlElemDesc *description;
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   if (description->isinline) { return Qtrue; }
   return Qfalse;
@@ -166,7 +171,7 @@ static VALUE
 deprecated_eh(VALUE self)
 {
   const htmlElemDesc *description;
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   if (description->depr) { return Qtrue; }
   return Qfalse;
@@ -182,7 +187,7 @@ static VALUE
 empty_eh(VALUE self)
 {
   const htmlElemDesc *description;
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   if (description->empty) { return Qtrue; }
   return Qfalse;
@@ -198,7 +203,7 @@ static VALUE
 save_end_tag_eh(VALUE self)
 {
   const htmlElemDesc *description;
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   if (description->saveEndTag) { return Qtrue; }
   return Qfalse;
@@ -214,7 +219,7 @@ static VALUE
 implied_end_tag_eh(VALUE self)
 {
   const htmlElemDesc *description;
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   if (description->endTag) { return Qtrue; }
   return Qfalse;
@@ -230,7 +235,7 @@ static VALUE
 implied_start_tag_eh(VALUE self)
 {
   const htmlElemDesc *description;
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   if (description->startTag) { return Qtrue; }
   return Qfalse;
@@ -246,7 +251,7 @@ static VALUE
 name(VALUE self)
 {
   const htmlElemDesc *description;
-  Data_Get_Struct(self, htmlElemDesc, description);
+  TypedData_Get_Struct(self, htmlElemDesc, &html4_element_description_type, description);
 
   if (NULL == description->name) { return Qnil; }
   return NOKOGIRI_STR_NEW2(description->name);
@@ -266,7 +271,7 @@ get_description(VALUE klass, VALUE tag_name)
                                     );
 
   if (NULL == description) { return Qnil; }
-  return Data_Wrap_Struct(klass, 0, 0, DISCARD_CONST_QUAL(void *, description));
+  return TypedData_Wrap_Struct(klass, &html4_element_description_type, DISCARD_CONST_QUAL(void *, description));
 }
 
 void

data/ext/nokogiri/html4_sax_parser_context.c

@@ -2,13 +2,6 @@
 
 VALUE cNokogiriHtml4SaxParserContext ;
 
-static void
-deallocate(xmlParserCtxtPtr ctxt)
-{
-  ctxt->sax = NULL;
-  htmlFreeParserCtxt(ctxt);
-}
-
 static VALUE
 parse_memory(VALUE klass, VALUE data, VALUE encoding)
 {
@@ -38,7 +31,7 @@ parse_memory(VALUE klass, VALUE data, VALUE encoding)
     }
   }
 
-  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+  return noko_xml_sax_parser_context_wrap(klass, ctxt);
 }
 
 static VALUE
@@ -48,7 +41,13 @@ parse_file(VALUE klass, VALUE filename, VALUE encoding)
                              StringValueCStr(filename),
                              StringValueCStr(encoding)
                            );
-  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+
+  if (ctxt->sax) {
+    xmlFree(ctxt->sax);
+    ctxt->sax = NULL;
+  }
+
+  return noko_xml_sax_parser_context_wrap(klass, ctxt);
 }
 
 static VALUE
@@ -82,13 +81,8 @@ parse_with(VALUE self, VALUE sax_handler)
     rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
   }
 
-  Data_Get_Struct(self, htmlParserCtxt, ctxt);
-  Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
-
-  /* Free the sax handler since we'll assign our own */
-  if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) {
-    xmlFree(ctxt->sax);
-  }
+  ctxt = noko_xml_sax_parser_context_unwrap(self);
+  sax = noko_sax_handler_unwrap(sax_handler);
 
   ctxt->sax = sax;
   ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);

data/ext/nokogiri/html4_sax_push_parser.c

@@ -17,7 +17,7 @@ native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
   int status = 0;
   libxmlStructuredErrorHandlerState handler_state;
 
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   if (Qnil != _chunk) {
     chunk = StringValuePtr(_chunk);
@@ -54,7 +54,7 @@ initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
   htmlParserCtxtPtr ctx;
   xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
 
-  Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
+  sax = noko_sax_handler_unwrap(_xml_sax);
 
   if (_filename != Qnil) { filename = StringValueCStr(_filename); }
 

data/ext/nokogiri/nokogiri.c

@@ -137,6 +137,48 @@ noko_io_close(void *io)
 }
 
 
+#if defined(_WIN32) && !defined(NOKOGIRI_PACKAGED_LIBRARIES)
+#  define NOKOGIRI_WINDOWS_DLLS 1
+#else
+#  define NOKOGIRI_WINDOWS_DLLS 0
+#endif
+
+//
+//   |      dlls || true    | false   |
+//   | nlmm      ||         |         |
+//   |-----------++---------+---------|
+//   | NULL      || default | ruby    |
+//   | "random"  || default | ruby    |
+//   | "ruby"    || ruby    | ruby    |
+//   | "default" || default | default |
+//
+//   We choose *not* to use Ruby's memory management functions with windows DLLs because of this
+//   issue: https://github.com/sparklemotion/nokogiri/issues/2241
+//
+static void
+set_libxml_memory_management(void)
+{
+  const char *nlmm = getenv("NOKOGIRI_LIBXML_MEMORY_MANAGEMENT");
+  if (nlmm) {
+    if (strcmp(nlmm, "default") == 0) {
+      goto libxml_uses_default_memory_management;
+    } else if (strcmp(nlmm, "ruby") == 0) {
+      goto libxml_uses_ruby_memory_management;
+    }
+  }
+  if (NOKOGIRI_WINDOWS_DLLS) {
+libxml_uses_default_memory_management:
+    rb_const_set(mNokogiri, rb_intern("LIBXML_MEMORY_MANAGEMENT"), NOKOGIRI_STR_NEW2("default"));
+    return;
+  } else {
+libxml_uses_ruby_memory_management:
+    rb_const_set(mNokogiri, rb_intern("LIBXML_MEMORY_MANAGEMENT"), NOKOGIRI_STR_NEW2("ruby"));
+    xmlMemSetup((xmlFreeFunc)ruby_xfree, (xmlMallocFunc)ruby_xmalloc, (xmlReallocFunc)ruby_xrealloc, ruby_strdup);
+    return;
+  }
+}
+
+
 void
 Init_nokogiri(void)
 {
@@ -150,6 +192,10 @@ Init_nokogiri(void)
   mNokogiriXmlXpath = rb_define_module_under(mNokogiriXml, "XPath");
   mNokogiriXslt     = rb_define_module_under(mNokogiri, "XSLT");
 
+  set_libxml_memory_management(); /* must be before any function calls that might invoke xmlInitParser() */
+  xmlInitParser();
+  exsltRegisterAll();
+
   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));
 
@@ -182,30 +228,6 @@ Init_nokogiri(void)
   rb_const_set(mNokogiri, rb_intern("OTHER_LIBRARY_VERSIONS"), NOKOGIRI_STR_NEW2(NOKOGIRI_OTHER_LIBRARY_VERSIONS));
 #endif
 
-#if defined(_WIN32) && !defined(NOKOGIRI_PACKAGED_LIBRARIES)
-  /*
-   *  We choose *not* to do use Ruby's memory management functions with windows DLLs because of this
-   *  issue in libxml 2.9.12:
-   *
-   *    https://github.com/sparklemotion/nokogiri/issues/2241
-   *
-   *  If the atexit() issue gets fixed in a future version of libxml2, then we may be able to skip
-   *  this config only for the specific libxml2 versions 2.9.12.
-   *
-   *  Alternatively, now that Ruby has a generational GC, it might be OK to let libxml2 use its
-   *  default memory management functions (recall that this config was introduced to reduce memory
-   *  bloat and allow Ruby to GC more often); but we should *really* test with production workloads
-   *  before making that kind of a potentially-invasive change.
-   */
-  rb_const_set(mNokogiri, rb_intern("LIBXML_MEMORY_MANAGEMENT"), NOKOGIRI_STR_NEW2("default"));
-#else
-  rb_const_set(mNokogiri, rb_intern("LIBXML_MEMORY_MANAGEMENT"), NOKOGIRI_STR_NEW2("ruby"));
-  xmlMemSetup((xmlFreeFunc)ruby_xfree, (xmlMallocFunc)ruby_xmalloc, (xmlReallocFunc)ruby_xrealloc, ruby_strdup);
-#endif
-
-  xmlInitParser();
-  exsltRegisterAll();
-
   if (xsltExtModuleFunctionLookup((const xmlChar *)"date-time", EXSLT_DATE_NAMESPACE)) {
     rb_const_set(mNokogiri, rb_intern("LIBXSLT_DATETIME_ENABLED"), Qtrue);
   } else {

data/ext/nokogiri/nokogiri.h

@@ -51,6 +51,7 @@
 #include <libxslt/xsltconfig.h>
 #include <libxslt/xsltutils.h>
 #include <libxslt/transform.h>
+#include <libxslt/imports.h>
 #include <libxslt/xsltInternals.h>
 
 #include <libexslt/exslt.h>
@@ -168,6 +169,7 @@ typedef struct _nokogiriXsltStylesheetTuple {
 
 void noko_xml_document_pin_node(xmlNodePtr);
 void noko_xml_document_pin_namespace(xmlNsPtr, xmlDocPtr);
+int noko_xml_document_has_wrapped_blank_nodes_p(xmlDocPtr c_document);
 
 int noko_io_read(void *ctx, char *buffer, int len);
 int noko_io_write(void *ctx, char *buffer, int len);
@@ -186,12 +188,21 @@ VALUE noko_xml_namespace_wrap_xpath_copy(xmlNsPtr node);
 VALUE noko_xml_element_content_wrap(VALUE doc, xmlElementContentPtr element);
 
 VALUE noko_xml_node_set_wrap(xmlNodeSetPtr node_set, VALUE document) ;
+xmlNodeSetPtr noko_xml_node_set_unwrap(VALUE rb_node_set) ;
 
 VALUE noko_xml_document_wrap_with_init_args(VALUE klass, xmlDocPtr doc, int argc, VALUE *argv);
 VALUE noko_xml_document_wrap(VALUE klass, xmlDocPtr doc);
+xmlDocPtr noko_xml_document_unwrap(VALUE rb_document);
 NOKOPUBFUN VALUE Nokogiri_wrap_xml_document(VALUE klass,
     xmlDocPtr doc); /* deprecated. use noko_xml_document_wrap() instead. */
 
+xmlSAXHandlerPtr noko_sax_handler_unwrap(VALUE rb_sax_handler);
+
+xmlParserCtxtPtr noko_xml_sax_push_parser_unwrap(VALUE rb_parser);
+
+VALUE noko_xml_sax_parser_context_wrap(VALUE klass, xmlParserCtxtPtr c_context);
+xmlParserCtxtPtr noko_xml_sax_parser_context_unwrap(VALUE rb_context);
+
 #define DOC_RUBY_OBJECT_TEST(x) ((nokogiriTuplePtr)(x->_private))
 #define DOC_RUBY_OBJECT(x) (((nokogiriTuplePtr)(x->_private))->doc)
 #define DOC_UNLINKED_NODE_HASH(x) (((nokogiriTuplePtr)(x->_private))->unlinkedNodes)
@@ -207,9 +218,9 @@ NOKOPUBFUN VALUE Nokogiri_wrap_xml_document(VALUE klass,
 #define DISCARD_CONST_QUAL_XMLCHAR(v) DISCARD_CONST_QUAL(xmlChar *, v)
 
 #if HAVE_RB_CATEGORY_WARNING
-#  define NOKO_WARN_DEPRECATION(message) rb_category_warning(RB_WARN_CATEGORY_DEPRECATED, message)
+#  define NOKO_WARN_DEPRECATION(message...) rb_category_warning(RB_WARN_CATEGORY_DEPRECATED, message)
 #else
-#  define NOKO_WARN_DEPRECATION(message) rb_warning(message)
+#  define NOKO_WARN_DEPRECATION(message...) rb_warning(message)
 #endif
 
 void Nokogiri_structured_error_func_save(libxmlStructuredErrorHandlerState *handler_state);

data/ext/nokogiri/xml_attr.c

@@ -68,7 +68,7 @@ new (int argc, VALUE *argv, VALUE klass)
     rb_raise(rb_eArgError, "parameter must be a Nokogiri::XML::Document");
   }
 
-  Noko_Node_Get_Struct(document, xmlDoc, xml_doc);
+  xml_doc = noko_xml_document_unwrap(document);
 
   node = xmlNewDocProp(
            xml_doc,

data/ext/nokogiri/xml_cdata.c

@@ -25,14 +25,22 @@ new (int argc, VALUE *argv, VALUE klass)
 
   rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
 
-  Noko_Node_Get_Struct(doc, xmlDoc, xml_doc);
+  if (rb_obj_is_kind_of(doc, cNokogiriXmlDocument)) {
+    xml_doc = noko_xml_document_unwrap(doc);
+  } else {
+    xmlNodePtr deprecated_node_type_arg;
+    // TODO: deprecate allowing Node
+    NOKO_WARN_DEPRECATION("Passing a Node as the first parameter to CDATA.new is deprecated. Please pass a Document instead. This will become an error in a future release of Nokogiri.");
+    Noko_Node_Get_Struct(doc, xmlNode, deprecated_node_type_arg);
+    xml_doc = deprecated_node_type_arg->doc;
+  }
 
   if (!NIL_P(content)) {
     content_str = (xmlChar *)StringValuePtr(content);
     content_str_len = RSTRING_LENINT(content);
   }
 
-  node = xmlNewCDataBlock(xml_doc->doc, content_str, content_str_len);
+  node = xmlNewCDataBlock(xml_doc, content_str, content_str_len);
 
   noko_xml_document_pin_node(node);
 

data/ext/nokogiri/xml_comment.c

@@ -30,7 +30,7 @@ new (int argc, VALUE *argv, VALUE klass)
     rb_raise(rb_eArgError, "first argument must be a XML::Document or XML::Node");
   }
 
-  Data_Get_Struct(document, xmlDoc, xml_doc);
+  xml_doc = noko_xml_document_unwrap(document);
 
   node = xmlNewDocComment(
            xml_doc,