nokogiri 1.13.6 → 1.16.7

data/ext/nokogiri/xml_sax_parser.c

@@ -195,40 +195,48 @@ comment_func(void *ctx, const xmlChar *value)
   rb_funcall(doc, id_comment, 1, str);
 }
 
+PRINTFLIKE_DECL(2, 3)
 static void
 warning_func(void *ctx, const char *msg, ...)
 {
   VALUE self = NOKOGIRI_SAX_SELF(ctx);
   VALUE doc = rb_iv_get(self, "@document");
-  char *message;
-  VALUE ruby_message;
+  VALUE rb_message;
 
+#ifdef TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+  /* It is not currently possible to pass var args from native
+     functions to sulong, so we work around the issue here. */
+  rb_message = rb_sprintf("warning_func: %s", msg);
+#else
   va_list args;
   va_start(args, msg);
-  vasprintf(&message, msg, args);
+  rb_message = rb_vsprintf(msg, args);
   va_end(args);
+#endif
 
-  ruby_message = NOKOGIRI_STR_NEW2(message);
-  free(message);
-  rb_funcall(doc, id_warning, 1, ruby_message);
+  rb_funcall(doc, id_warning, 1, rb_message);
 }
 
+PRINTFLIKE_DECL(2, 3)
 static void
 error_func(void *ctx, const char *msg, ...)
 {
   VALUE self = NOKOGIRI_SAX_SELF(ctx);
   VALUE doc = rb_iv_get(self, "@document");
-  char *message;
-  VALUE ruby_message;
+  VALUE rb_message;
 
+#ifdef TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+  /* It is not currently possible to pass var args from native
+     functions to sulong, so we work around the issue here. */
+  rb_message = rb_sprintf("error_func: %s", msg);
+#else
   va_list args;
   va_start(args, msg);
-  vasprintf(&message, msg, args);
+  rb_message = rb_vsprintf(msg, args);
   va_end(args);
+#endif
 
-  ruby_message = NOKOGIRI_STR_NEW2(message);
-  free(message);
-  rb_funcall(doc, id_error, 1, ruby_message);
+  rb_funcall(doc, id_error, 1, rb_message);
 }
 
 static void
@@ -257,18 +265,27 @@ processing_instruction(void *ctx, const xmlChar *name, const xmlChar *content)
             );
 }
 
-static void
-deallocate(xmlSAXHandlerPtr handler)
+static size_t
+memsize(const void *data)
 {
-  NOKOGIRI_DEBUG_START(handler);
-  free(handler);
-  NOKOGIRI_DEBUG_END(handler);
+  return sizeof(xmlSAXHandler);
 }
 
+/* Used by Nokogiri::XML::SAX::Parser and Nokogiri::HTML::SAX::Parser */
+static const rb_data_type_t noko_sax_handler_type = {
+  .wrap_struct_name = "Nokogiri::SAXHandler",
+  .function = {
+    .dfree = RUBY_TYPED_DEFAULT_FREE,
+    .dsize = memsize
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED
+};
+
 static VALUE
 allocate(VALUE klass)
 {
-  xmlSAXHandlerPtr handler = calloc((size_t)1, sizeof(xmlSAXHandler));
+  xmlSAXHandlerPtr handler;
+  VALUE self = TypedData_Make_Struct(klass, xmlSAXHandler, &noko_sax_handler_type, handler);
 
   handler->startDocument = start_document;
   handler->endDocument = end_document;
@@ -284,11 +301,19 @@ allocate(VALUE klass)
   handler->processingInstruction = processing_instruction;
   handler->initialized = XML_SAX2_MAGIC;
 
-  return Data_Wrap_Struct(klass, NULL, deallocate, handler);
+  return self;
+}
+
+xmlSAXHandlerPtr
+noko_sax_handler_unwrap(VALUE rb_sax_handler)
+{
+  xmlSAXHandlerPtr c_sax_handler;
+  TypedData_Get_Struct(rb_sax_handler, xmlSAXHandler, &noko_sax_handler_type, c_sax_handler);
+  return c_sax_handler;
 }
 
 void
-noko_init_xml_sax_parser()
+noko_init_xml_sax_parser(void)
 {
   cNokogiriXmlSaxParser = rb_define_class_under(mNokogiriXmlSax, "Parser", rb_cObject);
 

data/ext/nokogiri/xml_sax_parser_context.c

@@ -5,17 +5,40 @@ VALUE cNokogiriXmlSaxParserContext ;
 static ID id_read;
 
 static void
-deallocate(xmlParserCtxtPtr ctxt)
+xml_sax_parser_context_free(void *data)
 {
-  NOKOGIRI_DEBUG_START(ctxt);
-
+  xmlParserCtxtPtr ctxt = data;
   ctxt->sax = NULL;
-
   xmlFreeParserCtxt(ctxt);
+}
 
-  NOKOGIRI_DEBUG_END(ctxt);
+/*
+ *  note that htmlParserCtxtPtr == xmlParserCtxtPtr and xmlFreeParserCtxt() == htmlFreeParserCtxt()
+ *  so we use this type for both XML::SAX::ParserContext and HTML::SAX::ParserContext
+ */
+static const rb_data_type_t xml_sax_parser_context_type = {
+  .wrap_struct_name = "Nokogiri::XML::SAX::ParserContext",
+  .function = {
+    .dfree = xml_sax_parser_context_free,
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
+xmlParserCtxtPtr
+noko_xml_sax_parser_context_unwrap(VALUE rb_context)
+{
+  xmlParserCtxtPtr c_context;
+  TypedData_Get_Struct(rb_context, xmlParserCtxt, &xml_sax_parser_context_type, c_context);
+  return c_context;
 }
 
+VALUE
+noko_xml_sax_parser_context_wrap(VALUE klass, xmlParserCtxtPtr c_context)
+{
+  return TypedData_Wrap_Struct(klass, &xml_sax_parser_context_type, c_context);
+}
+
+
 /*
  * call-seq:
  *  parse_io(io, encoding)
@@ -36,12 +59,16 @@ parse_io(VALUE klass, VALUE io, VALUE encoding)
                                (xmlInputReadCallback)noko_io_read,
                                (xmlInputCloseCallback)noko_io_close,
                                (void *)io, enc);
+  if (!ctxt) {
+    rb_raise(rb_eRuntimeError, "failed to create xml sax parser context");
+  }
+
   if (ctxt->sax) {
     xmlFree(ctxt->sax);
     ctxt->sax = NULL;
   }
 
-  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+  return noko_xml_sax_parser_context_wrap(klass, ctxt);
 }
 
 /*
@@ -54,7 +81,13 @@ static VALUE
 parse_file(VALUE klass, VALUE filename)
 {
   xmlParserCtxtPtr ctxt = xmlCreateFileParserCtxt(StringValueCStr(filename));
-  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);
 }
 
 /*
@@ -81,7 +114,7 @@ parse_memory(VALUE klass, VALUE data)
     ctxt->sax = NULL;
   }
 
-  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+  return noko_xml_sax_parser_context_wrap(klass, ctxt);
 }
 
 static VALUE
@@ -121,13 +154,8 @@ parse_with(VALUE self, VALUE sax_handler)
     rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
   }
 
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
-  Data_Get_Struct(sax_handler, xmlSAXHandler, 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);
@@ -149,8 +177,7 @@ parse_with(VALUE self, VALUE sax_handler)
 static VALUE
 set_replace_entities(VALUE self, VALUE value)
 {
-  xmlParserCtxtPtr ctxt;
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
   if (Qfalse == value) {
     ctxt->replaceEntities = 0;
@@ -171,8 +198,7 @@ set_replace_entities(VALUE self, VALUE value)
 static VALUE
 get_replace_entities(VALUE self)
 {
-  xmlParserCtxtPtr ctxt;
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
   if (0 == ctxt->replaceEntities) {
     return Qfalse;
@@ -189,10 +215,8 @@ get_replace_entities(VALUE self)
 static VALUE
 line(VALUE self)
 {
-  xmlParserCtxtPtr ctxt;
   xmlParserInputPtr io;
-
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
   io = ctxt->input;
   if (io) {
@@ -210,11 +234,9 @@ line(VALUE self)
 static VALUE
 column(VALUE self)
 {
-  xmlParserCtxtPtr ctxt;
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
   xmlParserInputPtr io;
 
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
-
   io = ctxt->input;
   if (io) {
     return INT2NUM(io->col);
@@ -233,8 +255,7 @@ column(VALUE self)
 static VALUE
 set_recovery(VALUE self, VALUE value)
 {
-  xmlParserCtxtPtr ctxt;
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
   if (value == Qfalse) {
     ctxt->recovery = 0;
@@ -255,8 +276,7 @@ set_recovery(VALUE self, VALUE value)
 static VALUE
 get_recovery(VALUE self)
 {
-  xmlParserCtxtPtr ctxt;
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
   if (ctxt->recovery == 0) {
     return Qfalse;
@@ -266,7 +286,7 @@ get_recovery(VALUE self)
 }
 
 void
-noko_init_xml_sax_parser_context()
+noko_init_xml_sax_parser_context(void)
 {
   cNokogiriXmlSaxParserContext = rb_define_class_under(mNokogiriXmlSax, "ParserContext", rb_cObject);
 

data/ext/nokogiri/xml_sax_push_parser.c

@@ -3,20 +3,35 @@
 VALUE cNokogiriXmlSaxPushParser ;
 
 static void
-deallocate(xmlParserCtxtPtr ctx)
+xml_sax_push_parser_free(void *data)
 {
-  NOKOGIRI_DEBUG_START(ctx);
+  xmlParserCtxtPtr ctx = data;
   if (ctx != NULL) {
     NOKOGIRI_SAX_TUPLE_DESTROY(ctx->userData);
     xmlFreeParserCtxt(ctx);
   }
-  NOKOGIRI_DEBUG_END(ctx);
 }
 
+static const rb_data_type_t xml_sax_push_parser_type = {
+  .wrap_struct_name = "Nokogiri::XML::SAX::PushParser",
+  .function = {
+    .dfree = xml_sax_push_parser_free,
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
 static VALUE
 allocate(VALUE klass)
 {
-  return Data_Wrap_Struct(klass, NULL, deallocate, NULL);
+  return TypedData_Wrap_Struct(klass, &xml_sax_push_parser_type, NULL);
+}
+
+xmlParserCtxtPtr
+noko_xml_sax_push_parser_unwrap(VALUE rb_parser)
+{
+  xmlParserCtxtPtr c_parser;
+  TypedData_Get_Struct(rb_parser, xmlParserCtxt, &xml_sax_push_parser_type, c_parser);
+  return c_parser;
 }
 
 /*
@@ -33,7 +48,7 @@ native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
   int size            = 0;
 
 
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   if (Qnil != _chunk) {
     chunk = StringValuePtr(_chunk);
@@ -44,7 +59,7 @@ native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
 
   if (xmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
     if (!(ctx->options & XML_PARSE_RECOVER)) {
-      xmlErrorPtr e = xmlCtxtGetLastError(ctx);
+      xmlErrorConstPtr e = xmlCtxtGetLastError(ctx);
       Nokogiri_error_raise(NULL, e);
     }
   }
@@ -65,7 +80,7 @@ initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
   const char *filename = NULL;
   xmlParserCtxtPtr ctx;
 
-  Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
+  sax = noko_sax_handler_unwrap(_xml_sax);
 
   if (_filename != Qnil) { filename = StringValueCStr(_filename); }
 
@@ -91,7 +106,8 @@ static VALUE
 get_options(VALUE self)
 {
   xmlParserCtxtPtr ctx;
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   return INT2NUM(ctx->options);
 }
@@ -100,7 +116,8 @@ static VALUE
 set_options(VALUE self, VALUE options)
 {
   xmlParserCtxtPtr ctx;
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   if (xmlCtxtUseOptions(ctx, (int)NUM2INT(options)) != 0) {
     rb_raise(rb_eRuntimeError, "Cannot set XML parser context options");
@@ -120,7 +137,8 @@ static VALUE
 get_replace_entities(VALUE self)
 {
   xmlParserCtxtPtr ctx;
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   if (0 == ctx->replaceEntities) {
     return Qfalse;
@@ -140,7 +158,8 @@ static VALUE
 set_replace_entities(VALUE self, VALUE value)
 {
   xmlParserCtxtPtr ctx;
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   if (Qfalse == value) {
     ctx->replaceEntities = 0;
@@ -152,7 +171,7 @@ set_replace_entities(VALUE self, VALUE value)
 }
 
 void
-noko_init_xml_sax_push_parser()
+noko_init_xml_sax_push_parser(void)
 {
   cNokogiriXmlSaxPushParser = rb_define_class_under(mNokogiriXmlSax, "PushParser", rb_cObject);