nokogiri 1.13.8 → 1.15.7

data/ext/nokogiri/gumbo.c

@@ -23,13 +23,13 @@
 //
 // Processing starts by calling gumbo_parse_with_options. The resulting document tree
 // is then walked, a parallel libxml2 tree is constructed, and the final document is
-// then wrapped using Nokogiri_wrap_xml_document. This approach reduces memory and CPU
+// then wrapped using noko_xml_document_wrap. This approach reduces memory and CPU
 // requirements as Ruby objects are only built when necessary.
 //
 
 #include <nokogiri.h>
 
-#include "gumbo.h"
+#include "nokogiri_gumbo.h"
 
 VALUE cNokogiriHtml5Document;
 
@@ -281,12 +281,12 @@ add_errors(const GumboOutput *output, VALUE rdoc, VALUE input, VALUE url)
       rb_iv_set(syntax_error, "@code", INT2NUM(1));   // XML_ERR_INTERNAL_ERROR
       rb_iv_set(syntax_error, "@level", INT2NUM(2));  // XML_ERR_ERROR
       rb_iv_set(syntax_error, "@file", url);
-      rb_iv_set(syntax_error, "@line", INT2NUM(position.line));
+      rb_iv_set(syntax_error, "@line", SIZET2NUM(position.line));
       rb_iv_set(syntax_error, "@str1", str1);
       rb_iv_set(syntax_error, "@str2", Qnil);
       rb_iv_set(syntax_error, "@str3", Qnil);
       rb_iv_set(syntax_error, "@int1", INT2NUM(0));
-      rb_iv_set(syntax_error, "@column", INT2NUM(position.column));
+      rb_iv_set(syntax_error, "@column", SIZET2NUM(position.column));
       rb_ary_push(rerrors, syntax_error);
     }
     rb_iv_set(rdoc, "@errors", rerrors);
@@ -297,6 +297,7 @@ typedef struct {
   GumboOutput *output;
   VALUE input;
   VALUE url_or_frag;
+  VALUE klass;
   xmlDocPtr doc;
 } ParseArgs;
 
@@ -321,7 +322,7 @@ static VALUE parse_continue(VALUE parse_args);
  *  @!visibility protected
  */
 static VALUE
-parse(VALUE self, VALUE input, VALUE url, VALUE max_attributes, VALUE max_errors, VALUE max_depth)
+parse(VALUE self, VALUE input, VALUE url, VALUE max_attributes, VALUE max_errors, VALUE max_depth, VALUE klass)
 {
   GumboOptions options = kGumboDefaultOptions;
   options.max_attributes = NUM2INT(max_attributes);
@@ -333,6 +334,7 @@ parse(VALUE self, VALUE input, VALUE url, VALUE max_attributes, VALUE max_errors
     .output = output,
     .input = input,
     .url_or_frag = url,
+    .klass = klass,
     .doc = NULL,
   };
 
@@ -357,7 +359,9 @@ 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(cNokogiriHtml5Document, doc);
+  VALUE rdoc = noko_xml_document_wrap(args->klass, doc);
+  rb_iv_set(rdoc, "@url", args->url_or_frag);
+  rb_iv_set(rdoc, "@quirks_mode", INT2NUM(output->document->v.document.doc_type_quirks_mode));
   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;
@@ -498,9 +502,11 @@ error:
     }
 
     // Encoding.
-    if (RSTRING_LEN(tag_name) == 14
+    if (ctx_ns == GUMBO_NAMESPACE_MATHML
+        && RSTRING_LEN(tag_name) == 14
         && !st_strcasecmp(ctx_tag, "annotation-xml")) {
       VALUE enc = rb_funcall(ctx, rb_intern_const("[]"),
+                             1,
                              rb_utf8_str_new_static("encoding", 8));
       if (RTEST(enc)) {
         Check_Type(enc, T_STRING);
@@ -512,8 +518,11 @@ error:
   // Quirks mode.
   VALUE doc = rb_funcall(doc_fragment, rb_intern_const("document"), 0);
   VALUE dtd = rb_funcall(doc, internal_subset, 0);
-  if (NIL_P(dtd)) {
+  VALUE doc_quirks_mode = rb_iv_get(doc, "@quirks_mode");
+  if (NIL_P(ctx) || NIL_P(doc_quirks_mode)) {
     quirks_mode = GUMBO_DOCTYPE_NO_QUIRKS;
+  } else if (NIL_P(dtd)) {
+    quirks_mode = GUMBO_DOCTYPE_QUIRKS;
   } else {
     VALUE dtd_name = rb_funcall(dtd, name, 0);
     VALUE pubid = rb_funcall(dtd, rb_intern_const("external_id"), 0);
@@ -560,13 +569,14 @@ fragment_continue(VALUE parse_args)
   args->doc = NULL; // The Ruby runtime owns doc so make sure we don't delete it.
   xmlNodePtr xml_frag = extract_xml_node(doc_fragment);
   build_tree(xml_doc, xml_frag, output->root);
+  rb_iv_set(doc_fragment, "@quirks_mode", INT2NUM(output->document->v.document.doc_type_quirks_mode));
   add_errors(output, doc_fragment, args->input, rb_utf8_str_new_static("#fragment", 9));
   return Qnil;
 }
 
 // Initialize the Nokogumbo class and fetch constants we will use later.
 void
-noko_init_gumbo()
+noko_init_gumbo(void)
 {
   // Class constants.
   cNokogiriHtml5Document = rb_define_class_under(mNokogiriHtml5, "Document", cNokogiriHtml4Document);
@@ -577,7 +587,7 @@ noko_init_gumbo()
   parent = rb_intern_const("parent");
 
   // Define Nokogumbo module with parse and fragment methods.
-  rb_define_singleton_method(mNokogiriGumbo, "parse", parse, 5);
+  rb_define_singleton_method(mNokogiriGumbo, "parse", parse, 6);
   rb_define_singleton_method(mNokogiriGumbo, "fragment", fragment, 6);
 }
 

data/ext/nokogiri/html4_document.c

@@ -144,13 +144,12 @@ 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);
-  return INT2NUM((long)doc->type);
+  htmlDocPtr doc = noko_xml_document_unwrap(self);
+  return INT2NUM(doc->type);
 }
 
 void
-noko_init_html_document()
+noko_init_html_document(void)
 {
   assert(cNokogiriXmlDocument);
   cNokogiriHtml4Document = rb_define_class_under(mNokogiriHtml4, "Document", cNokogiriXmlDocument);

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);
@@ -266,11 +271,11 @@ 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
-noko_init_html_element_description()
+noko_init_html_element_description(void)
 {
   cNokogiriHtml4ElementDescription = rb_define_class_under(mNokogiriHtml4, "ElementDescription", rb_cObject);
 

data/ext/nokogiri/html4_entity_lookup.c

@@ -20,7 +20,7 @@ get(VALUE _, VALUE rb_entity_name)
     return Qnil;
   }
 
-  rb_constructor_args[0] = INT2NUM((long)c_entity_desc->value);
+  rb_constructor_args[0] = UINT2NUM(c_entity_desc->value);
   rb_constructor_args[1] = NOKOGIRI_STR_NEW2(c_entity_desc->name);
   rb_constructor_args[2] = NOKOGIRI_STR_NEW2(c_entity_desc->desc);
 
@@ -29,7 +29,7 @@ get(VALUE _, VALUE rb_entity_name)
 }
 
 void
-noko_init_html_entity_lookup()
+noko_init_html_entity_lookup(void)
 {
   cNokogiriHtml4EntityLookup = rb_define_class_under(mNokogiriHtml4, "EntityLookup", rb_cObject);
 

data/ext/nokogiri/html4_sax_parser_context.c

@@ -2,18 +2,6 @@
 
 VALUE cNokogiriHtml4SaxParserContext ;
 
-static void
-deallocate(xmlParserCtxtPtr ctxt)
-{
-  NOKOGIRI_DEBUG_START(ctxt);
-
-  ctxt->sax = NULL;
-
-  htmlFreeParserCtxt(ctxt);
-
-  NOKOGIRI_DEBUG_END(ctxt);
-}
-
 static VALUE
 parse_memory(VALUE klass, VALUE data, VALUE encoding)
 {
@@ -43,7 +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
@@ -53,7 +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
@@ -87,13 +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);
@@ -106,7 +95,7 @@ parse_with(VALUE self, VALUE sax_handler)
 }
 
 void
-noko_init_html_sax_parser_context()
+noko_init_html_sax_parser_context(void)
 {
   assert(cNokogiriXmlSaxParserContext);
   cNokogiriHtml4SaxParserContext = rb_define_class_under(mNokogiriHtml4Sax, "ParserContext",

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); }
 
@@ -85,7 +85,7 @@ initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
 }
 
 void
-noko_init_html_sax_push_parser()
+noko_init_html_sax_push_parser(void)
 {
   assert(cNokogiriXmlSaxPushParser);
   cNokogiriHtml4SaxPushParser = rb_define_class_under(mNokogiriHtml4Sax, "PushParser", cNokogiriXmlSaxPushParser);

data/ext/nokogiri/nokogiri.c

@@ -49,34 +49,11 @@ void noko_init_html_sax_push_parser(void);
 void noko_init_gumbo(void);
 void noko_init_test_global_handlers(void);
 
-static ID id_read, id_write;
-
-
-#ifndef HAVE_VASPRINTF
-/*
- * Thank you Geoffroy Couprie for this implementation of vasprintf!
- */
-int
-vasprintf(char **strp, const char *fmt, va_list ap)
-{
-  /* Mingw32/64 have a broken vsnprintf implementation that fails when
-   * using a zero-byte limit in order to retrieve the required size for malloc.
-   * So we use a one byte buffer instead.
-   */
-  char tmp[1];
-  int len = vsnprintf(tmp, 1, fmt, ap) + 1;
-  char *res = (char *)malloc((unsigned int)len);
-  if (res == NULL) {
-    return -1;
-  }
-  *strp = res;
-  return vsnprintf(res, (unsigned int)len, fmt, ap);
-}
-#endif
+static ID id_read, id_write, id_external_encoding;
 
 
 static VALUE
-read_check(VALUE val)
+noko_io_read_check(VALUE val)
 {
   VALUE *args = (VALUE *)val;
   return rb_funcall(args[0], id_read, 1, args[1]);
@@ -84,75 +61,126 @@ read_check(VALUE val)
 
 
 static VALUE
-read_failed(VALUE arg, VALUE exc)
+noko_io_read_failed(VALUE arg, VALUE exc)
 {
   return Qundef;
 }
 
 
 int
-noko_io_read(void *ctx, char *buffer, int len)
+noko_io_read(void *io, char *c_buffer, int c_buffer_len)
 {
-  VALUE string, args[2];
-  size_t str_len, safe_len;
+  VALUE rb_io = (VALUE)io;
+  VALUE rb_read_string, rb_args[2];
+  size_t n_bytes_read, safe_len;
 
-  args[0] = (VALUE)ctx;
-  args[1] = INT2NUM(len);
+  rb_args[0] = rb_io;
+  rb_args[1] = INT2NUM(c_buffer_len);
 
-  string = rb_rescue(read_check, (VALUE)args, read_failed, 0);
+  rb_read_string = rb_rescue(noko_io_read_check, (VALUE)rb_args, noko_io_read_failed, 0);
 
-  if (NIL_P(string)) { return 0; }
-  if (string == Qundef) { return -1; }
-  if (TYPE(string) != T_STRING) { return -1; }
+  if (NIL_P(rb_read_string)) { return 0; }
+  if (rb_read_string == Qundef) { return -1; }
+  if (TYPE(rb_read_string) != T_STRING) { return -1; }
 
-  str_len = (size_t)RSTRING_LEN(string);
-  safe_len = str_len > (size_t)len ? (size_t)len : str_len;
-  memcpy(buffer, StringValuePtr(string), safe_len);
+  n_bytes_read = (size_t)RSTRING_LEN(rb_read_string);
+  safe_len = (n_bytes_read > (size_t)c_buffer_len) ? (size_t)c_buffer_len : n_bytes_read;
+  memcpy(c_buffer, StringValuePtr(rb_read_string), safe_len);
 
   return (int)safe_len;
 }
 
 
 static VALUE
-write_check(VALUE val)
+noko_io_write_check(VALUE rb_args)
 {
-  VALUE *args = (VALUE *)val;
-  return rb_funcall(args[0], id_write, 1, args[1]);
+  VALUE rb_io = ((VALUE *)rb_args)[0];
+  VALUE rb_output = ((VALUE *)rb_args)[1];
+  return rb_funcall(rb_io, id_write, 1, rb_output);
 }
 
 
 static VALUE
-write_failed(VALUE arg, VALUE exc)
+noko_io_write_failed(VALUE arg, VALUE exc)
 {
   return Qundef;
 }
 
 
 int
-noko_io_write(void *ctx, char *buffer, int len)
+noko_io_write(void *io, char *c_buffer, int c_buffer_len)
 {
-  VALUE args[2], size;
+  VALUE rb_args[2], rb_n_bytes_written;
+  VALUE rb_io = (VALUE)io;
+  VALUE rb_enc = Qnil;
+  rb_encoding *io_encoding;
 
-  args[0] = (VALUE)ctx;
-  args[1] = rb_str_new(buffer, (long)len);
+  if (rb_respond_to(rb_io, id_external_encoding)) {
+    rb_enc = rb_funcall(rb_io, id_external_encoding, 0);
+  }
+  io_encoding = RB_NIL_P(rb_enc) ? rb_ascii8bit_encoding() : rb_to_encoding(rb_enc);
 
-  size = rb_rescue(write_check, (VALUE)args, write_failed, 0);
+  rb_args[0] = rb_io;
+  rb_args[1] = rb_enc_str_new(c_buffer, (long)c_buffer_len, io_encoding);
 
-  if (size == Qundef) { return -1; }
+  rb_n_bytes_written = rb_rescue(noko_io_write_check, (VALUE)rb_args, noko_io_write_failed, 0);
+  if (rb_n_bytes_written == Qundef) { return -1; }
 
-  return NUM2INT(size);
+  return NUM2INT(rb_n_bytes_written);
 }
 
 
 int
-noko_io_close(void *ctx)
+noko_io_close(void *io)
 {
   return 0;
 }
 
 
+#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()
+Init_nokogiri(void)
 {
   mNokogiri         = rb_define_module("Nokogiri");
   mNokogiriGumbo    = rb_define_module_under(mNokogiri, "Gumbo");
@@ -164,6 +192,10 @@ Init_nokogiri()
   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));
 
@@ -196,30 +228,6 @@ Init_nokogiri()
   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 {
@@ -275,4 +283,5 @@ Init_nokogiri()
 
   id_read = rb_intern("read");
   id_write = rb_intern("write");
+  id_external_encoding = rb_intern("external_encoding");
 }