fix: namespace nodes behave with compaction

ext/nokogiri/html4_sax_parser_context.c

@@ -5,13 +5,8 @@ VALUE cNokogiriHtml4SaxParserContext ;
 static void
 deallocate(xmlParserCtxtPtr ctxt)
 {
-  NOKOGIRI_DEBUG_START(ctxt);
-
   ctxt->sax = NULL;
-
   htmlFreeParserCtxt(ctxt);
-
-  NOKOGIRI_DEBUG_END(ctxt);
 }
 
 static VALUE

ext/nokogiri/nokogiri.h

@@ -75,14 +75,6 @@ xmlNodePtr xmlLastElementChild(xmlNodePtr parent);
 #define NOKOGIRI_STR_NEW(str, len) rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
 #define RBSTR_OR_QNIL(_str) (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
 
-#ifdef DEBUG
-#  define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
-#  define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
-#else
-#  define NOKOGIRI_DEBUG_START(p)
-#  define NOKOGIRI_DEBUG_END(p)
-#endif
-
 #ifndef NORETURN_DECL
 #  if defined(__GNUC__)
 #    define NORETURN_DECL __attribute__ ((noreturn))
@@ -181,6 +173,7 @@ int noko_io_write(void *ctx, char *buffer, int len);
 int noko_io_close(void *ctx);
 
 #define Noko_Node_Get_Struct(obj,type,sval) ((sval) = (type*)DATA_PTR(obj))
+#define Noko_Namespace_Get_Struct(obj,type,sval) ((sval) = (type*)DATA_PTR(obj))
 
 VALUE noko_xml_node_wrap(VALUE klass, xmlNodePtr node) ;
 VALUE noko_xml_node_wrap_node_set_result(xmlNodePtr node, VALUE node_set) ;

ext/nokogiri/xml_document.c

@@ -65,8 +65,6 @@ dealloc(xmlDocPtr doc)
 {
   st_table *node_hash;
 
-  NOKOGIRI_DEBUG_START(doc);
-
   node_hash  = DOC_UNLINKED_NODE_HASH(doc);
 
   st_foreach(node_hash, dealloc_node_i, (st_data_t)doc);
@@ -84,8 +82,6 @@ dealloc(xmlDocPtr doc)
   }
 
   xmlFreeDoc(doc);
-
-  NOKOGIRI_DEBUG_END(doc);
 }
 
 static void
@@ -104,7 +100,11 @@ recursively_remove_namespaces_from_node(xmlNodePtr node)
        (node->type == XML_XINCLUDE_START) ||
        (node->type == XML_XINCLUDE_END)) &&
       node->nsDef) {
-    xmlFreeNsList(node->nsDef);
+    xmlNsPtr curr = node->nsDef;
+    while (curr) {
+      noko_xml_document_pin_namespace(curr, node->doc);
+      curr = curr->next;
+    }
     node->nsDef = NULL;
   }
 

ext/nokogiri/xml_namespace.c

@@ -25,23 +25,53 @@
 VALUE cNokogiriXmlNamespace ;
 
 static void
-dealloc_namespace(xmlNsPtr ns)
+_xml_namespace_dealloc(void *ptr)
 {
   /*
    * this deallocator is only used for namespace nodes that are part of an xpath
    * node set. see noko_xml_namespace_wrap().
    */
-  NOKOGIRI_DEBUG_START(ns) ;
+  xmlNsPtr ns = ptr;
+
   if (ns->href) {
     xmlFree(DISCARD_CONST_QUAL_XMLCHAR(ns->href));
   }
   if (ns->prefix) {
     xmlFree(DISCARD_CONST_QUAL_XMLCHAR(ns->prefix));
   }
   xmlFree(ns);
-  NOKOGIRI_DEBUG_END(ns) ;
 }
 
+#ifdef HAVE_RB_GC_LOCATION
+static void
+_xml_namespace_update_references(void *ptr)
+{
+  xmlNsPtr ns = ptr;
+  if (ns->_private) {
+    ns->_private = (void *)rb_gc_location((VALUE)ns->_private);
+  }
+}
+#else
+#  define _xml_namespace_update_references 0
+#endif
+
+static const rb_data_type_t nokogiri_xml_namespace_type_with_dealloc = {
+  "Nokogiri/XMLNamespace/WithDealloc",
+  {0, _xml_namespace_dealloc, 0, _xml_namespace_update_references},
+  0, 0,
+#ifdef RUBY_TYPED_FREE_IMMEDIATELY
+  RUBY_TYPED_FREE_IMMEDIATELY,
+#endif
+};
+
+static const rb_data_type_t nokogiri_xml_namespace_type_without_dealloc = {
+  "Nokogiri/XMLNamespace/WithoutDealloc",
+  {0, 0, 0, _xml_namespace_update_references},
+  0, 0,
+#ifdef RUBY_TYPED_FREE_IMMEDIATELY
+  RUBY_TYPED_FREE_IMMEDIATELY,
+#endif
+};
 
 /*
  * call-seq:
@@ -54,7 +84,7 @@ prefix(VALUE self)
 {
   xmlNsPtr ns;
 
-  Data_Get_Struct(self, xmlNs, ns);
+  Noko_Namespace_Get_Struct(self, xmlNs, ns);
   if (!ns->prefix) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(ns->prefix);
@@ -71,7 +101,7 @@ href(VALUE self)
 {
   xmlNsPtr ns;
 
-  Data_Get_Struct(self, xmlNs, ns);
+  Noko_Namespace_Get_Struct(self, xmlNs, ns);
   if (!ns->href) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(ns->href);
@@ -87,14 +117,18 @@ noko_xml_namespace_wrap(xmlNsPtr c_namespace, xmlDocPtr c_document)
   }
 
   if (c_document) {
-    rb_namespace = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, c_namespace);
+    rb_namespace = TypedData_Wrap_Struct(cNokogiriXmlNamespace,
+                                         &nokogiri_xml_namespace_type_without_dealloc,
+                                         c_namespace);
 
     if (DOC_RUBY_OBJECT_TEST(c_document)) {
       rb_iv_set(rb_namespace, "@document", DOC_RUBY_OBJECT(c_document));
       rb_ary_push(DOC_NODE_CACHE(c_document), rb_namespace);
     }
   } else {
-    rb_namespace = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, dealloc_namespace, c_namespace);
+    rb_namespace = TypedData_Wrap_Struct(cNokogiriXmlNamespace,
+                                         &nokogiri_xml_namespace_type_with_dealloc,
+                                         c_namespace);
   }
 
   c_namespace->_private = (void *)rb_namespace;

ext/nokogiri/xml_node.c

@@ -9,20 +9,11 @@ static ID id_decorate, id_decorate_bang;
 
 typedef xmlNodePtr(*pivot_reparentee_func)(xmlNodePtr, xmlNodePtr);
 
-#ifdef DEBUG
 static void
-_xml_node_dealloc(xmlNodePtr x)
+_xml_node_mark(void *ptr)
 {
-  NOKOGIRI_DEBUG_START(x)
-  NOKOGIRI_DEBUG_END(x)
-}
-#else
-#  define _xml_node_dealloc 0
-#endif
+  xmlNodePtr node = ptr;
 
-static void
-_xml_node_mark(xmlNodePtr node)
-{
   if (!DOC_RUBY_OBJECT_TEST(node->doc)) {
     return;
   }
@@ -39,24 +30,21 @@ _xml_node_mark(xmlNodePtr node)
 
 #ifdef HAVE_RB_GC_LOCATION
 static void
-_xml_node_update_references(xmlNodePtr node)
+_xml_node_update_references(void *ptr)
 {
+  xmlNodePtr node = ptr;
+
   if (node->_private) {
     node->_private = (void *)rb_gc_location((VALUE)node->_private);
   }
 }
+#else
+#  define _xml_node_update_references 0
 #endif
 
-typedef void (*gc_callback_t)(void *);
-
 static const rb_data_type_t nokogiri_node_type = {
   "Nokogiri/XMLNode",
-  {
-    (gc_callback_t)_xml_node_mark, (gc_callback_t)_xml_node_dealloc, 0,
-#ifdef HAVE_RB_GC_LOCATION
-    (gc_callback_t)_xml_node_update_references
-#endif
-  },
+  {_xml_node_mark, 0, 0, _xml_node_update_references},
   0, 0,
 #ifdef RUBY_TYPED_FREE_IMMEDIATELY
   RUBY_TYPED_FREE_IMMEDIATELY,
@@ -1355,7 +1343,7 @@ set_namespace(VALUE self, VALUE namespace)
   Noko_Node_Get_Struct(self, xmlNode, node);
 
   if (!NIL_P(namespace)) {
-    Data_Get_Struct(namespace, xmlNs, ns);
+    Noko_Namespace_Get_Struct(namespace, xmlNs, ns);
   }
 
   xmlSetNs(node, ns);

ext/nokogiri/xml_node_set.c

@@ -88,13 +88,11 @@ deallocate(xmlNodeSetPtr node_set)
    * For reasons outlined in xml_namespace.c, here we reproduce xmlXPathFreeNodeSet() except for the
    * offending call to xmlXPathNodeSetFreeNs().
    */
-  NOKOGIRI_DEBUG_START(node_set) ;
   if (node_set->nodeTab != NULL) {
     xmlFree(node_set->nodeTab);
   }
 
   xmlFree(node_set);
-  NOKOGIRI_DEBUG_END(node_set) ;
 }
 
 

ext/nokogiri/xml_reader.c

@@ -5,9 +5,7 @@ VALUE cNokogiriXmlReader;
 static void
 dealloc(xmlTextReaderPtr reader)
 {
-  NOKOGIRI_DEBUG_START(reader);
   xmlFreeTextReader(reader);
-  NOKOGIRI_DEBUG_END(reader);
 }
 
 static int

ext/nokogiri/xml_relax_ng.c

@@ -5,9 +5,7 @@ VALUE cNokogiriXmlRelaxNG;
 static void
 dealloc(xmlRelaxNGPtr schema)
 {
-  NOKOGIRI_DEBUG_START(schema);
   xmlRelaxNGFree(schema);
-  NOKOGIRI_DEBUG_END(schema);
 }
 
 /*

ext/nokogiri/xml_sax_parser.c

@@ -268,9 +268,7 @@ processing_instruction(void *ctx, const xmlChar *name, const xmlChar *content)
 static void
 deallocate(xmlSAXHandlerPtr handler)
 {
-  NOKOGIRI_DEBUG_START(handler);
   ruby_xfree(handler);
-  NOKOGIRI_DEBUG_END(handler);
 }
 
 static VALUE

ext/nokogiri/xml_sax_parser_context.c

@@ -7,13 +7,8 @@ static ID id_read;
 static void
 deallocate(xmlParserCtxtPtr ctxt)
 {
-  NOKOGIRI_DEBUG_START(ctxt);
-
   ctxt->sax = NULL;
-
   xmlFreeParserCtxt(ctxt);
-
-  NOKOGIRI_DEBUG_END(ctxt);
 }
 
 /*

ext/nokogiri/xml_sax_push_parser.c

@@ -5,12 +5,10 @@ VALUE cNokogiriXmlSaxPushParser ;
 static void
 deallocate(xmlParserCtxtPtr ctx)
 {
-  NOKOGIRI_DEBUG_START(ctx);
   if (ctx != NULL) {
     NOKOGIRI_SAX_TUPLE_DESTROY(ctx->userData);
     xmlFreeParserCtxt(ctx);
   }
-  NOKOGIRI_DEBUG_END(ctx);
 }
 
 static VALUE

ext/nokogiri/xml_schema.c

@@ -5,9 +5,7 @@ VALUE cNokogiriXmlSchema;
 static void
 dealloc(xmlSchemaPtr schema)
 {
-  NOKOGIRI_DEBUG_START(schema);
   xmlSchemaFree(schema);
-  NOKOGIRI_DEBUG_END(schema);
 }
 
 /*

ext/nokogiri/xml_xpath_context.c

@@ -14,9 +14,7 @@ static const xmlChar *NOKOGIRI_BUILTIN_URI = (const xmlChar *)"https://www.nokog
 static void
 xml_xpath_context_deallocate(xmlXPathContextPtr ctx)
 {
-  NOKOGIRI_DEBUG_START(ctx);
   xmlXPathFreeContext(ctx);
-  NOKOGIRI_DEBUG_END(ctx);
 }
 
 /* find a CSS class in an HTML element's `class` attribute */

ext/nokogiri/xslt_stylesheet.c

@@ -12,11 +12,7 @@ static void
 dealloc(nokogiriXsltStylesheetTuple *wrapper)
 {
   xsltStylesheetPtr doc = wrapper->ss;
-
-  NOKOGIRI_DEBUG_START(doc);
-  xsltFreeStylesheet(doc); /* commented out for now. */
-  NOKOGIRI_DEBUG_END(doc);
-
+  xsltFreeStylesheet(doc);
   ruby_xfree(wrapper);
 }
 

rakelib/debug.rake

@@ -1,12 +1,5 @@
 # frozen_string_literal: true
 
-desc "Set environment variables to build and/or test with debug options"
-task :debug do
-  ENV["NOKOGIRI_DEBUG"] = "true"
-  ENV["CFLAGS"] ||= ""
-  ENV["CFLAGS"] += " -DDEBUG"
-end
-
 task :java_debug do # rubocop:disable Rake/Desc
   ENV["JRUBY_OPTS"] = "#{ENV["JRUBY_OPTS"]} --debug --dev"
   if ENV["JAVA_DEBUG"]