nokogiri 1.10.7 → 1.16.0

data/ext/nokogiri/xml_schema.c

@@ -1,33 +1,44 @@
-#include <xml_schema.h>
+#include <nokogiri.h>
 
-static void dealloc(xmlSchemaPtr schema)
+VALUE cNokogiriXmlSchema;
+
+static void
+xml_schema_deallocate(void *data)
 {
-  NOKOGIRI_DEBUG_START(schema);
+  xmlSchemaPtr schema = data;
   xmlSchemaFree(schema);
-  NOKOGIRI_DEBUG_END(schema);
 }
 
+static const rb_data_type_t xml_schema_type = {
+  .wrap_struct_name = "Nokogiri::XML::Schema",
+  .function = {
+    .dfree = xml_schema_deallocate,
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
 /*
  * call-seq:
  *  validate_document(document)
  *
  * Validate a Nokogiri::XML::Document against this Schema.
  */
-static VALUE validate_document(VALUE self, VALUE document)
+static VALUE
+validate_document(VALUE self, VALUE document)
 {
   xmlDocPtr doc;
   xmlSchemaPtr schema;
   xmlSchemaValidCtxtPtr valid_ctxt;
   VALUE errors;
 
-  Data_Get_Struct(self, xmlSchema, schema);
-  Data_Get_Struct(document, xmlDoc, doc);
+  TypedData_Get_Struct(self, xmlSchema, &xml_schema_type, schema);
+  doc = noko_xml_document_unwrap(document);
 
   errors = rb_ary_new();
 
   valid_ctxt = xmlSchemaNewValidCtxt(schema);
 
-  if(NULL == valid_ctxt) {
+  if (NULL == valid_ctxt) {
     /* we have a problem */
     rb_raise(rb_eRuntimeError, "Could not create a validation context");
   }
@@ -53,21 +64,22 @@ static VALUE validate_document(VALUE self, VALUE document)
  *
  * Validate a file against this Schema.
  */
-static VALUE validate_file(VALUE self, VALUE rb_filename)
+static VALUE
+validate_file(VALUE self, VALUE rb_filename)
 {
   xmlSchemaPtr schema;
   xmlSchemaValidCtxtPtr valid_ctxt;
   const char *filename ;
   VALUE errors;
 
-  Data_Get_Struct(self, xmlSchema, schema);
-  filename = (const char*)StringValueCStr(rb_filename) ;
+  TypedData_Get_Struct(self, xmlSchema, &xml_schema_type, schema);
+  filename = (const char *)StringValueCStr(rb_filename) ;
 
   errors = rb_ary_new();
 
   valid_ctxt = xmlSchemaNewValidCtxt(schema);
 
-  if(NULL == valid_ctxt) {
+  if (NULL == valid_ctxt) {
     /* we have a problem */
     rb_raise(rb_eRuntimeError, "Could not create a validation context");
   }
@@ -87,119 +99,163 @@ static VALUE validate_file(VALUE self, VALUE rb_filename)
   return errors;
 }
 
-/*
- * call-seq:
- *  read_memory(string)
- *
- * Create a new Schema from the contents of +string+
- */
-static VALUE read_memory(VALUE klass, VALUE content)
+static VALUE
+xml_schema_parse_schema(
+  VALUE klass,
+  xmlSchemaParserCtxtPtr c_parser_context,
+  VALUE rb_parse_options
+)
 {
-  xmlSchemaPtr schema;
-  xmlSchemaParserCtxtPtr ctx = xmlSchemaNewMemParserCtxt(
-      (const char *)StringValuePtr(content),
-      (int)RSTRING_LEN(content)
-  );
+  VALUE rb_errors;
+  int parse_options_int;
+  xmlSchemaPtr c_schema;
+  xmlExternalEntityLoader old_loader = 0;
   VALUE rb_schema;
-  VALUE errors = rb_ary_new();
-  xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
+
+  if (NIL_P(rb_parse_options)) {
+    rb_parse_options = rb_const_get_at(
+                         rb_const_get_at(mNokogiriXml, rb_intern("ParseOptions")),
+                         rb_intern("DEFAULT_SCHEMA")
+                       );
+  }
+
+  rb_errors = rb_ary_new();
+  xmlSetStructuredErrorFunc((void *)rb_errors, Nokogiri_error_array_pusher);
 
 #ifdef HAVE_XMLSCHEMASETPARSERSTRUCTUREDERRORS
   xmlSchemaSetParserStructuredErrors(
-    ctx,
+    c_parser_context,
     Nokogiri_error_array_pusher,
-    (void *)errors
+    (void *)rb_errors
   );
 #endif
 
-   schema = xmlSchemaParse(ctx);
+  parse_options_int = (int)NUM2INT(rb_funcall(rb_parse_options, rb_intern("to_i"), 0));
+  if (parse_options_int & XML_PARSE_NONET) {
+    old_loader = xmlGetExternalEntityLoader();
+    xmlSetExternalEntityLoader(xmlNoNetExternalEntityLoader);
+  }
+
+  c_schema = xmlSchemaParse(c_parser_context);
+
+  if (old_loader) {
+    xmlSetExternalEntityLoader(old_loader);
+  }
 
   xmlSetStructuredErrorFunc(NULL, NULL);
-  xmlSchemaFreeParserCtxt(ctx);
+  xmlSchemaFreeParserCtxt(c_parser_context);
 
-  if(NULL == schema) {
-    xmlErrorPtr error = xmlGetLastError();
-    if(error)
+  if (NULL == c_schema) {
+    xmlErrorConstPtr error = xmlGetLastError();
+    if (error) {
       Nokogiri_error_raise(NULL, error);
-    else
+    } else {
       rb_raise(rb_eRuntimeError, "Could not parse document");
+    }
 
     return Qnil;
   }
 
-  rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
-  rb_iv_set(rb_schema, "@errors", errors);
+  rb_schema = TypedData_Wrap_Struct(klass, &xml_schema_type, c_schema);
+  rb_iv_set(rb_schema, "@errors", rb_errors);
+  rb_iv_set(rb_schema, "@parse_options", rb_parse_options);
 
   return rb_schema;
 }
 
 /*
  * call-seq:
- *  from_document(doc)
+ *   read_memory(string) → Nokogiri::XML::Schema
+ *
+ * Create a new schema parsed from the contents of +string+
+ *
+ * [Parameters]
+ * - +string+: String containing XML to be parsed as a schema
  *
- * Create a new Schema from the Nokogiri::XML::Document +doc+
+ * [Returns] Nokogiri::XML::Schema
  */
-static VALUE from_document(VALUE klass, VALUE document)
+static VALUE
+read_memory(int argc, VALUE *argv, VALUE klass)
 {
-  xmlDocPtr doc;
-  xmlSchemaParserCtxtPtr ctx;
-  xmlSchemaPtr schema;
-  VALUE errors;
-  VALUE rb_schema;
-
-  Data_Get_Struct(document, xmlDoc, doc);
+  VALUE rb_content;
+  VALUE rb_parse_options;
+  xmlSchemaParserCtxtPtr c_parser_context;
 
-  /* In case someone passes us a node. ugh. */
-  doc = doc->doc;
+  rb_scan_args(argc, argv, "11", &rb_content, &rb_parse_options);
 
-  ctx = xmlSchemaNewDocParserCtxt(doc);
+  c_parser_context = xmlSchemaNewMemParserCtxt(
+                       (const char *)StringValuePtr(rb_content),
+                       (int)RSTRING_LEN(rb_content)
+                     );
 
-  errors = rb_ary_new();
-  xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
+  return xml_schema_parse_schema(klass, c_parser_context, rb_parse_options);
+}
 
-#ifdef HAVE_XMLSCHEMASETPARSERSTRUCTUREDERRORS
-  xmlSchemaSetParserStructuredErrors(
-    ctx,
-    Nokogiri_error_array_pusher,
-    (void *)errors
-  );
-#endif
+/*
+ * call-seq:
+ *   from_document(document) → Nokogiri::XML::Schema
+ *
+ * Create a new schema parsed from the +document+.
+ *
+ * [Parameters]
+ * - +document+: Nokogiri::XML::Document to be parsed
+ *
+ * [Returns] Nokogiri::XML::Schema
+ */
+static VALUE
+rb_xml_schema_s_from_document(int argc, VALUE *argv, VALUE klass)
+{
+  VALUE rb_document;
+  VALUE rb_parse_options;
+  VALUE rb_schema;
+  xmlDocPtr c_document;
+  xmlSchemaParserCtxtPtr c_parser_context;
+  int defensive_copy_p = 0;
 
-  schema = xmlSchemaParse(ctx);
+  rb_scan_args(argc, argv, "11", &rb_document, &rb_parse_options);
 
-  xmlSetStructuredErrorFunc(NULL, NULL);
-  xmlSchemaFreeParserCtxt(ctx);
+  if (!rb_obj_is_kind_of(rb_document, cNokogiriXmlNode)) {
+    rb_raise(rb_eTypeError,
+             "expected parameter to be a Nokogiri::XML::Document, received %"PRIsVALUE,
+             rb_obj_class(rb_document));
+  }
 
-  if(NULL == schema) {
-    xmlErrorPtr error = xmlGetLastError();
-    if(error)
-      Nokogiri_error_raise(NULL, error);
-    else
-      rb_raise(rb_eRuntimeError, "Could not parse document");
+  if (!rb_obj_is_kind_of(rb_document, cNokogiriXmlDocument)) {
+    xmlNodePtr deprecated_node_type_arg;
+    NOKO_WARN_DEPRECATION("Passing a Node as the first parameter to Schema.from_document is deprecated. Please pass a Document instead. This will become an error in Nokogiri v1.17.0."); // TODO: deprecated in v1.15.3, remove in v1.17.0
+    Noko_Node_Get_Struct(rb_document, xmlNode, deprecated_node_type_arg);
+    c_document = deprecated_node_type_arg->doc;
+  } else {
+    c_document = noko_xml_document_unwrap(rb_document);
+  }
 
-    return Qnil;
+  if (noko_xml_document_has_wrapped_blank_nodes_p(c_document)) {
+    // see https://github.com/sparklemotion/nokogiri/pull/2001
+    c_document = xmlCopyDoc(c_document, 1);
+    defensive_copy_p = 1;
   }
 
-  rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
-  rb_iv_set(rb_schema, "@errors", errors);
+  c_parser_context = xmlSchemaNewDocParserCtxt(c_document);
+  rb_schema = xml_schema_parse_schema(klass, c_parser_context, rb_parse_options);
 
-  return rb_schema;
+  if (defensive_copy_p) {
+    xmlFreeDoc(c_document);
+    c_document = NULL;
+  }
 
-  return Qnil;
+  return rb_schema;
 }
 
-VALUE cNokogiriXmlSchema;
-void init_xml_schema()
+void
+noko_init_xml_schema(void)
 {
-  VALUE nokogiri = rb_define_module("Nokogiri");
-  VALUE xml = rb_define_module_under(nokogiri, "XML");
-  VALUE klass = rb_define_class_under(xml, "Schema", rb_cObject);
+  cNokogiriXmlSchema = rb_define_class_under(mNokogiriXml, "Schema", rb_cObject);
 
-  cNokogiriXmlSchema = klass;
+  rb_undef_alloc_func(cNokogiriXmlSchema);
 
-  rb_define_singleton_method(klass, "read_memory", read_memory, 1);
-  rb_define_singleton_method(klass, "from_document", from_document, 1);
+  rb_define_singleton_method(cNokogiriXmlSchema, "read_memory", read_memory, -1);
+  rb_define_singleton_method(cNokogiriXmlSchema, "from_document", rb_xml_schema_s_from_document, -1);
 
-  rb_define_private_method(klass, "validate_document", validate_document, 1);
-  rb_define_private_method(klass, "validate_file",     validate_file, 1);
+  rb_define_private_method(cNokogiriXmlSchema, "validate_document", validate_document, 1);
+  rb_define_private_method(cNokogiriXmlSchema, "validate_file",     validate_file, 1);
 }

data/ext/nokogiri/xml_syntax_error.c

@@ -1,38 +1,64 @@
-#include <xml_syntax_error.h>
+#include <nokogiri.h>
 
-void Nokogiri_error_array_pusher(void * ctx, xmlErrorPtr error)
+VALUE cNokogiriXmlSyntaxError;
+
+void
+Nokogiri_structured_error_func_save(libxmlStructuredErrorHandlerState *handler_state)
+{
+  /* this method is tightly coupled to the implementation of xmlSetStructuredErrorFunc */
+  handler_state->user_data = xmlStructuredErrorContext;
+  handler_state->handler = xmlStructuredError;
+}
+
+void
+Nokogiri_structured_error_func_save_and_set(libxmlStructuredErrorHandlerState *handler_state,
+    void *user_data,
+    xmlStructuredErrorFunc handler)
+{
+  Nokogiri_structured_error_func_save(handler_state);
+  xmlSetStructuredErrorFunc(user_data, handler);
+}
+
+void
+Nokogiri_structured_error_func_restore(libxmlStructuredErrorHandlerState *handler_state)
+{
+  xmlSetStructuredErrorFunc(handler_state->user_data, handler_state->handler);
+}
+
+void
+Nokogiri_error_array_pusher(void *ctx, xmlErrorConstPtr error)
 {
   VALUE list = (VALUE)ctx;
   Check_Type(list, T_ARRAY);
   rb_ary_push(list,  Nokogiri_wrap_xml_syntax_error(error));
 }
 
-void Nokogiri_error_raise(void * ctx, xmlErrorPtr error)
+void
+Nokogiri_error_raise(void *ctx, xmlErrorConstPtr error)
 {
   rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
 }
 
-VALUE Nokogiri_wrap_xml_syntax_error(xmlErrorPtr error)
+VALUE
+Nokogiri_wrap_xml_syntax_error(xmlErrorConstPtr error)
 {
   VALUE msg, e, klass;
 
   klass = cNokogiriXmlSyntaxError;
 
   if (error && error->domain == XML_FROM_XPATH) {
-    VALUE xpath = rb_const_get(mNokogiriXml, rb_intern("XPath"));
-    klass = rb_const_get(xpath, rb_intern("SyntaxError"));
+    klass = cNokogiriXmlXpathSyntaxError;
   }
 
   msg = (error && error->message) ? NOKOGIRI_STR_NEW2(error->message) : Qnil;
 
   e = rb_class_new_instance(
-      1,
-      &msg,
-      klass
-  );
+        1,
+        &msg,
+        klass
+      );
 
-  if (error)
-  {
+  if (error) {
     rb_iv_set(e, "@domain", INT2NUM(error->domain));
     rb_iv_set(e, "@code", INT2NUM(error->code));
     rb_iv_set(e, "@level", INT2NUM((short)error->level));
@@ -48,17 +74,12 @@ VALUE Nokogiri_wrap_xml_syntax_error(xmlErrorPtr error)
   return e;
 }
 
-VALUE cNokogiriXmlSyntaxError;
-void init_xml_syntax_error()
+void
+noko_init_xml_syntax_error(void)
 {
-  VALUE nokogiri = rb_define_module("Nokogiri");
-  VALUE xml = rb_define_module_under(nokogiri, "XML");
-
+  assert(cNokogiriSyntaxError);
   /*
    * The XML::SyntaxError is raised on parse errors
    */
-  VALUE syntax_error_mommy = rb_define_class_under(nokogiri, "SyntaxError", rb_eStandardError);
-  VALUE klass = rb_define_class_under(xml, "SyntaxError", syntax_error_mommy);
-  cNokogiriXmlSyntaxError = klass;
-
+  cNokogiriXmlSyntaxError = rb_define_class_under(mNokogiriXml, "SyntaxError", cNokogiriSyntaxError);
 }

data/ext/nokogiri/xml_text.c

@@ -1,4 +1,6 @@
-#include <xml_text.h>
+#include <nokogiri.h>
+
+VALUE cNokogiriXmlText ;
 
 /*
  * call-seq:
@@ -6,47 +8,54 @@
  *
  * Create a new Text element on the +document+ with +content+
  */
-static VALUE new(int argc, VALUE *argv, VALUE klass)
+static VALUE
+rb_xml_text_s_new(int argc, VALUE *argv, VALUE klass)
 {
-  xmlDocPtr doc;
-  xmlNodePtr node;
-  VALUE string;
-  VALUE document;
-  VALUE rest;
+  xmlDocPtr c_document;
+  xmlNodePtr c_node;
+  VALUE rb_string;
+  VALUE rb_document;
+  VALUE rb_rest;
   VALUE rb_node;
 
-  rb_scan_args(argc, argv, "2*", &string, &document, &rest);
+  rb_scan_args(argc, argv, "2*", &rb_string, &rb_document, &rb_rest);
+
+  if (!rb_obj_is_kind_of(rb_document, cNokogiriXmlNode)) {
+    rb_raise(rb_eTypeError,
+             "expected second parameter to be a Nokogiri::XML::Document, received %"PRIsVALUE,
+             rb_obj_class(rb_document));
+  }
 
-  Data_Get_Struct(document, xmlDoc, doc);
+  if (!rb_obj_is_kind_of(rb_document, cNokogiriXmlDocument)) {
+    xmlNodePtr deprecated_node_type_arg;
+    NOKO_WARN_DEPRECATION("Passing a Node as the second parameter to Text.new is deprecated. Please pass a Document instead. This will become an error in Nokogiri v1.17.0."); // TODO: deprecated in v1.15.3, remove in v1.17.0
+    Noko_Node_Get_Struct(rb_document, xmlNode, deprecated_node_type_arg);
+    c_document = deprecated_node_type_arg->doc;
+  } else {
+    c_document = noko_xml_document_unwrap(rb_document);
+  }
 
-  node = xmlNewText((xmlChar *)StringValueCStr(string));
-  node->doc = doc->doc;
+  c_node = xmlNewText((xmlChar *)StringValueCStr(rb_string));
+  c_node->doc = c_document;
 
-  nokogiri_root_node(node);
+  noko_xml_document_pin_node(c_node);
 
-  rb_node = Nokogiri_wrap_xml_node(klass, node) ;
+  rb_node = noko_xml_node_wrap(klass, c_node) ;
   rb_obj_call_init(rb_node, argc, argv);
 
-  if(rb_block_given_p()) rb_yield(rb_node);
+  if (rb_block_given_p()) { rb_yield(rb_node); }
 
   return rb_node;
 }
 
-VALUE cNokogiriXmlText ;
-void init_xml_text()
+void
+noko_init_xml_text(void)
 {
-  VALUE nokogiri = rb_define_module("Nokogiri");
-  VALUE xml = rb_define_module_under(nokogiri, "XML");
-  /* */
-  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
-  VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
-
+  assert(cNokogiriXmlCharacterData);
   /*
    * Wraps Text nodes.
    */
-  VALUE klass = rb_define_class_under(xml, "Text", char_data);
-
-  cNokogiriXmlText = klass;
+  cNokogiriXmlText = rb_define_class_under(mNokogiriXml, "Text", cNokogiriXmlCharacterData);
 
-  rb_define_singleton_method(klass, "new", new, -1);
+  rb_define_singleton_method(cNokogiriXmlText, "new", rb_xml_text_s_new, -1);
 }