nokogiri 1.16.8-x86_64-linux → 1.17.0-x86_64-linux

data/ext/nokogiri/xml_reader.c

@@ -15,8 +15,8 @@ xml_reader_deallocate(void *data)
   }
 }
 
-static const rb_data_type_t xml_reader_type = {
-  .wrap_struct_name = "Nokogiri::XML::Reader",
+static const rb_data_type_t xml_text_reader_type = {
+  .wrap_struct_name = "xmlTextReader",
   .function = {
     .dfree = xml_reader_deallocate,
   },
@@ -84,7 +84,7 @@ default_eh(VALUE self)
   xmlTextReaderPtr reader;
   int eh;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   eh = xmlTextReaderIsDefault(reader);
   if (eh == 0) { return Qfalse; }
   if (eh == 1) { return Qtrue; }
@@ -104,7 +104,7 @@ value_eh(VALUE self)
   xmlTextReaderPtr reader;
   int eh;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   eh = xmlTextReaderHasValue(reader);
   if (eh == 0) { return Qfalse; }
   if (eh == 1) { return Qtrue; }
@@ -124,7 +124,7 @@ attributes_eh(VALUE self)
   xmlTextReaderPtr reader;
   int eh;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   eh = has_attributes(reader);
   if (eh == 0) { return Qfalse; }
   if (eh == 1) { return Qtrue; }
@@ -146,7 +146,7 @@ rb_xml_reader_namespaces(VALUE rb_reader)
   xmlNodePtr c_node;
   VALUE rb_errors;
 
-  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_reader_type, c_reader);
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
 
   if (! has_attributes(c_reader)) {
     return rb_namespaces ;
@@ -154,7 +154,7 @@ rb_xml_reader_namespaces(VALUE rb_reader)
 
   rb_errors = rb_funcall(rb_reader, rb_intern("errors"), 0);
 
-  xmlSetStructuredErrorFunc((void *)rb_errors, Nokogiri_error_array_pusher);
+  xmlSetStructuredErrorFunc((void *)rb_errors, noko__error_array_pusher);
   c_node = xmlTextReaderExpand(c_reader);
   xmlSetStructuredErrorFunc(NULL, NULL);
 
@@ -188,7 +188,7 @@ rb_xml_reader_attribute_hash(VALUE rb_reader)
   xmlAttrPtr c_property;
   VALUE rb_errors;
 
-  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_reader_type, c_reader);
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
 
   if (!has_attributes(c_reader)) {
     return rb_attributes;
@@ -196,7 +196,7 @@ rb_xml_reader_attribute_hash(VALUE rb_reader)
 
   rb_errors = rb_funcall(rb_reader, rb_intern("errors"), 0);
 
-  xmlSetStructuredErrorFunc((void *)rb_errors, Nokogiri_error_array_pusher);
+  xmlSetStructuredErrorFunc((void *)rb_errors, noko__error_array_pusher);
   c_node = xmlTextReaderExpand(c_reader);
   xmlSetStructuredErrorFunc(NULL, NULL);
 
@@ -241,7 +241,7 @@ attribute_at(VALUE self, VALUE index)
   xmlChar *value;
   VALUE rb_value;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
   if (NIL_P(index)) { return Qnil; }
   index = rb_Integer(index);
@@ -270,7 +270,7 @@ reader_attribute(VALUE self, VALUE name)
   xmlChar *value ;
   VALUE rb_value;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
   if (NIL_P(name)) { return Qnil; }
   name = StringValue(name) ;
@@ -295,7 +295,7 @@ attribute_count(VALUE self)
   xmlTextReaderPtr reader;
   int count;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   count = xmlTextReaderAttributeCount(reader);
   if (count == -1) { return Qnil; }
 
@@ -314,7 +314,7 @@ depth(VALUE self)
   xmlTextReaderPtr reader;
   int depth;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   depth = xmlTextReaderDepth(reader);
   if (depth == -1) { return Qnil; }
 
@@ -333,7 +333,7 @@ xml_version(VALUE self)
   xmlTextReaderPtr reader;
   const char *version;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   version = (const char *)xmlTextReaderConstXmlVersion(reader);
   if (version == NULL) { return Qnil; }
 
@@ -352,7 +352,7 @@ lang(VALUE self)
   xmlTextReaderPtr reader;
   const char *lang;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   lang = (const char *)xmlTextReaderConstXmlLang(reader);
   if (lang == NULL) { return Qnil; }
 
@@ -371,7 +371,7 @@ value(VALUE self)
   xmlTextReaderPtr reader;
   const char *value;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   value = (const char *)xmlTextReaderConstValue(reader);
   if (value == NULL) { return Qnil; }
 
@@ -390,7 +390,7 @@ prefix(VALUE self)
   xmlTextReaderPtr reader;
   const char *prefix;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   prefix = (const char *)xmlTextReaderConstPrefix(reader);
   if (prefix == NULL) { return Qnil; }
 
@@ -409,7 +409,7 @@ namespace_uri(VALUE self)
   xmlTextReaderPtr reader;
   const char *uri;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   uri = (const char *)xmlTextReaderConstNamespaceUri(reader);
   if (uri == NULL) { return Qnil; }
 
@@ -428,7 +428,7 @@ local_name(VALUE self)
   xmlTextReaderPtr reader;
   const char *name;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   name = (const char *)xmlTextReaderConstLocalName(reader);
   if (name == NULL) { return Qnil; }
 
@@ -447,7 +447,7 @@ name(VALUE self)
   xmlTextReaderPtr reader;
   const char *name;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   name = (const char *)xmlTextReaderConstName(reader);
   if (name == NULL) { return Qnil; }
 
@@ -467,7 +467,7 @@ rb_xml_reader_base_uri(VALUE rb_reader)
   xmlTextReaderPtr c_reader;
   xmlChar *c_base_uri;
 
-  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_reader_type, c_reader);
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
 
   c_base_uri = xmlTextReaderBaseUri(c_reader);
   if (c_base_uri == NULL) {
@@ -490,7 +490,7 @@ static VALUE
 state(VALUE self)
 {
   xmlTextReaderPtr reader;
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   return INT2NUM(xmlTextReaderReadState(reader));
 }
 
@@ -504,7 +504,7 @@ static VALUE
 node_type(VALUE self)
 {
   xmlTextReaderPtr reader;
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   return INT2NUM(xmlTextReaderNodeType(reader));
 }
 
@@ -515,44 +515,41 @@ node_type(VALUE self)
  * Move the Reader forward through the XML document.
  */
 static VALUE
-read_more(VALUE self)
+read_more(VALUE rb_reader)
 {
-  xmlTextReaderPtr reader;
-  xmlErrorConstPtr error;
-  VALUE error_list;
-  int ret;
-  xmlDocPtr c_document;
+  xmlTextReaderPtr c_reader;
+  libxmlStructuredErrorHandlerState handler_state;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
 
-  error_list = rb_funcall(self, rb_intern("errors"), 0);
+  VALUE rb_errors = rb_funcall(rb_reader, rb_intern("errors"), 0);
+  noko__structured_error_func_save_and_set(&handler_state, (void *)rb_errors, noko__error_array_pusher);
 
-  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
-  ret = xmlTextReaderRead(reader);
-  xmlSetStructuredErrorFunc(NULL, NULL);
+  int status = xmlTextReaderRead(c_reader);
+
+  noko__structured_error_func_restore(&handler_state);
 
-  c_document = xmlTextReaderCurrentDoc(reader);
+  xmlDocPtr c_document = xmlTextReaderCurrentDoc(c_reader);
   if (c_document && c_document->encoding == NULL) {
-    VALUE constructor_encoding = rb_iv_get(self, "@encoding");
+    VALUE constructor_encoding = rb_iv_get(rb_reader, "@encoding");
     if (RTEST(constructor_encoding)) {
       c_document->encoding = xmlStrdup(BAD_CAST StringValueCStr(constructor_encoding));
     } else {
-      rb_iv_set(self, "@encoding", NOKOGIRI_STR_NEW2("UTF-8"));
+      rb_iv_set(rb_reader, "@encoding", NOKOGIRI_STR_NEW2("UTF-8"));
       c_document->encoding = xmlStrdup(BAD_CAST "UTF-8");
     }
   }
 
-  if (ret == 1) { return self; }
-  if (ret == 0) { return Qnil; }
+  if (status == 1) { return rb_reader; }
+  if (status == 0) { return Qnil; }
 
-  error = xmlGetLastError();
-  if (error) {
-    rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
+  /* if we're here, there was an error */
+  VALUE exception = rb_funcall(cNokogiriXmlSyntaxError, rb_intern("aggregate"), 1, rb_errors);
+  if (RB_TEST(exception)) {
+    rb_exc_raise(exception);
   } else {
-    rb_raise(rb_eRuntimeError, "Error pulling: %d", ret);
+    rb_raise(rb_eRuntimeError, "Error pulling: %d", status);
   }
-
-  return Qnil;
 }
 
 /*
@@ -569,7 +566,7 @@ inner_xml(VALUE self)
   xmlChar *value;
   VALUE str;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
   value = xmlTextReaderReadInnerXml(reader);
 
@@ -596,7 +593,7 @@ outer_xml(VALUE self)
   xmlChar *value;
   VALUE str = Qnil;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
   value = xmlTextReaderReadOuterXml(reader);
 
@@ -611,11 +608,13 @@ outer_xml(VALUE self)
  * call-seq:
  *   from_memory(string, url = nil, encoding = nil, options = 0)
  *
- * Create a new reader that parses +string+
+ * Create a new Reader to parse a String.
  */
 static VALUE
 from_memory(int argc, VALUE *argv, VALUE klass)
 {
+  /* TODO: deprecate this method, since Reader.new can handle both memory and IO. It can then
+   * become private. */
   VALUE rb_buffer, rb_url, encoding, rb_options;
   xmlTextReaderPtr reader;
   const char *c_url      = NULL;
@@ -643,7 +642,7 @@ from_memory(int argc, VALUE *argv, VALUE klass)
     rb_raise(rb_eRuntimeError, "couldn't create a parser");
   }
 
-  rb_reader = TypedData_Wrap_Struct(klass, &xml_reader_type, reader);
+  rb_reader = TypedData_Wrap_Struct(klass, &xml_text_reader_type, reader);
   args[0] = rb_buffer;
   args[1] = rb_url;
   args[2] = encoding;
@@ -656,11 +655,13 @@ from_memory(int argc, VALUE *argv, VALUE klass)
  * call-seq:
  *   from_io(io, url = nil, encoding = nil, options = 0)
  *
- * Create a new reader that parses +io+
+ * Create a new Reader to parse an IO stream.
  */
 static VALUE
 from_io(int argc, VALUE *argv, VALUE klass)
 {
+  /* TODO: deprecate this method, since Reader.new can handle both memory and IO. It can then
+   * become private. */
   VALUE rb_io, rb_url, encoding, rb_options;
   xmlTextReaderPtr reader;
   const char *c_url      = NULL;
@@ -689,7 +690,7 @@ from_io(int argc, VALUE *argv, VALUE klass)
     rb_raise(rb_eRuntimeError, "couldn't create a parser");
   }
 
-  rb_reader = TypedData_Wrap_Struct(klass, &xml_reader_type, reader);
+  rb_reader = TypedData_Wrap_Struct(klass, &xml_text_reader_type, reader);
   args[0] = rb_io;
   args[1] = rb_url;
   args[2] = encoding;
@@ -709,7 +710,7 @@ empty_element_p(VALUE self)
 {
   xmlTextReaderPtr reader;
 
-  TypedData_Get_Struct(self, xmlTextReader, &xml_reader_type, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
   if (xmlTextReaderIsEmptyElement(reader)) {
     return Qtrue;
@@ -725,7 +726,7 @@ rb_xml_reader_encoding(VALUE rb_reader)
   const char *parser_encoding;
   VALUE constructor_encoding;
 
-  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_reader_type, c_reader);
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
   parser_encoding = (const char *)xmlTextReaderConstEncoding(c_reader);
   if (parser_encoding) {
     return NOKOGIRI_STR_NEW2(parser_encoding);
@@ -742,11 +743,6 @@ rb_xml_reader_encoding(VALUE rb_reader)
 void
 noko_init_xml_reader(void)
 {
-  /*
-   * The Reader parser allows you to effectively pull parse an XML document.
-   * Once instantiated, call Nokogiri::XML::Reader#each to iterate over each
-   * node.  Note that you may only iterate over the document once!
-   */
   cNokogiriXmlReader = rb_define_class_under(mNokogiriXml, "Reader", rb_cObject);
 
   rb_undef_alloc_func(cNokogiriXmlReader);

data/ext/nokogiri/xml_relax_ng.c

@@ -3,28 +3,22 @@
 VALUE cNokogiriXmlRelaxNG;
 
 static void
-xml_relax_ng_deallocate(void *data)
+_noko_xml_relax_ng_deallocate(void *data)
 {
   xmlRelaxNGPtr schema = data;
   xmlRelaxNGFree(schema);
 }
 
 static const rb_data_type_t xml_relax_ng_type = {
-  .wrap_struct_name = "Nokogiri::XML::RelaxNG",
+  .wrap_struct_name = "xmlRelaxNG",
   .function = {
-    .dfree = xml_relax_ng_deallocate,
+    .dfree = _noko_xml_relax_ng_deallocate,
   },
   .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
 };
 
-/*
- * call-seq:
- *  validate_document(document)
- *
- * Validate a Nokogiri::XML::Document against this RelaxNG schema.
- */
 static VALUE
-validate_document(VALUE self, VALUE document)
+noko_xml_relax_ng__validate_document(VALUE self, VALUE document)
 {
   xmlDocPtr doc;
   xmlRelaxNGPtr schema;
@@ -43,13 +37,11 @@ validate_document(VALUE self, VALUE document)
     rb_raise(rb_eRuntimeError, "Could not create a validation context");
   }
 
-#ifdef HAVE_XMLRELAXNGSETVALIDSTRUCTUREDERRORS
   xmlRelaxNGSetValidStructuredErrors(
     valid_ctxt,
-    Nokogiri_error_array_pusher,
+    noko__error_array_pusher,
     (void *)errors
   );
-#endif
 
   xmlRelaxNGValidateDoc(valid_ctxt, doc);
 
@@ -59,8 +51,8 @@ validate_document(VALUE self, VALUE document)
 }
 
 static VALUE
-xml_relax_ng_parse_schema(
-  VALUE klass,
+_noko_xml_relax_ng_parse_schema(
+  VALUE rb_class,
   xmlRelaxNGParserCtxtPtr c_parser_context,
   VALUE rb_parse_options
 )
@@ -68,6 +60,7 @@ xml_relax_ng_parse_schema(
   VALUE rb_errors;
   VALUE rb_schema;
   xmlRelaxNGPtr c_schema;
+  libxmlStructuredErrorHandlerState handler_state;
 
   if (NIL_P(rb_parse_options)) {
     rb_parse_options = rb_const_get_at(
@@ -77,33 +70,30 @@ xml_relax_ng_parse_schema(
   }
 
   rb_errors = rb_ary_new();
-  xmlSetStructuredErrorFunc((void *)rb_errors, Nokogiri_error_array_pusher);
 
-#ifdef HAVE_XMLRELAXNGSETPARSERSTRUCTUREDERRORS
+  noko__structured_error_func_save_and_set(&handler_state, (void *)rb_errors, noko__error_array_pusher);
   xmlRelaxNGSetParserStructuredErrors(
     c_parser_context,
-    Nokogiri_error_array_pusher,
+    noko__error_array_pusher,
     (void *)rb_errors
   );
-#endif
 
   c_schema = xmlRelaxNGParse(c_parser_context);
 
-  xmlSetStructuredErrorFunc(NULL, NULL);
   xmlRelaxNGFreeParserCtxt(c_parser_context);
+  noko__structured_error_func_restore(&handler_state);
 
   if (NULL == c_schema) {
-    xmlErrorConstPtr error = xmlGetLastError();
-    if (error) {
-      Nokogiri_error_raise(NULL, error);
+    VALUE exception = rb_funcall(cNokogiriXmlSyntaxError, rb_intern("aggregate"), 1, rb_errors);
+
+    if (RB_TEST(exception)) {
+      rb_exc_raise(exception);
     } else {
       rb_raise(rb_eRuntimeError, "Could not parse document");
     }
-
-    return Qnil;
   }
 
-  rb_schema = TypedData_Wrap_Struct(klass, &xml_relax_ng_type, c_schema);
+  rb_schema = TypedData_Wrap_Struct(rb_class, &xml_relax_ng_type, c_schema);
   rb_iv_set(rb_schema, "@errors", rb_errors);
   rb_iv_set(rb_schema, "@parse_options", rb_parse_options);
 
@@ -111,37 +101,27 @@ xml_relax_ng_parse_schema(
 }
 
 /*
- * call-seq:
- *  read_memory(string)
+ * :call-seq:
+ *   from_document(document) → Nokogiri::XML::RelaxNG
+ *   from_document(document, parse_options) → Nokogiri::XML::RelaxNG
  *
- * Create a new RelaxNG from the contents of +string+
- */
-static VALUE
-read_memory(int argc, VALUE *argv, VALUE klass)
-{
-  VALUE rb_content;
-  VALUE rb_parse_options;
-  xmlRelaxNGParserCtxtPtr c_parser_context;
-
-  rb_scan_args(argc, argv, "11", &rb_content, &rb_parse_options);
-
-  c_parser_context = xmlRelaxNGNewMemParserCtxt(
-                       (const char *)StringValuePtr(rb_content),
-                       (int)RSTRING_LEN(rb_content)
-                     );
-
-  return xml_relax_ng_parse_schema(klass, c_parser_context, rb_parse_options);
-}
-
-/*
- * call-seq:
- *  from_document(doc)
+ * Parse a RELAX NG schema definition from a Document to create a new Nokogiri::XML::RelaxNG.
+ *
+ * [Parameters]
+ * - +document+ (XML::Document) A document containing the RELAX NG schema definition
+ * - +parse_options+ (Nokogiri::XML::ParseOptions)
+ *   Defaults to ParseOptions::DEFAULT_SCHEMA ⚠ Unused
+ *
+ * [Returns] Nokogiri::XML::RelaxNG
  *
- * Create a new RelaxNG schema from the Nokogiri::XML::Document +doc+
+ * ⚠ +parse_options+ is currently unused by this method and is present only as a placeholder for
+ * future functionality.
  */
 static VALUE
-from_document(int argc, VALUE *argv, VALUE klass)
+noko_xml_relax_ng_s_from_document(int argc, VALUE *argv, VALUE rb_class)
 {
+  /* TODO: deprecate this method and put file-or-string logic into .new so that becomes the
+   * preferred entry point, and this can become a private method */
   VALUE rb_document;
   VALUE rb_parse_options;
   xmlDocPtr c_document;
@@ -154,7 +134,7 @@ from_document(int argc, VALUE *argv, VALUE klass)
 
   c_parser_context = xmlRelaxNGNewDocParserCtxt(c_document);
 
-  return xml_relax_ng_parse_schema(klass, c_parser_context, rb_parse_options);
+  return _noko_xml_relax_ng_parse_schema(rb_class, c_parser_context, rb_parse_options);
 }
 
 void
@@ -163,8 +143,7 @@ noko_init_xml_relax_ng(void)
   assert(cNokogiriXmlSchema);
   cNokogiriXmlRelaxNG = rb_define_class_under(mNokogiriXml, "RelaxNG", cNokogiriXmlSchema);
 
-  rb_define_singleton_method(cNokogiriXmlRelaxNG, "read_memory", read_memory, -1);
-  rb_define_singleton_method(cNokogiriXmlRelaxNG, "from_document", from_document, -1);
+  rb_define_singleton_method(cNokogiriXmlRelaxNG, "from_document", noko_xml_relax_ng_s_from_document, -1);
 
-  rb_define_private_method(cNokogiriXmlRelaxNG, "validate_document", validate_document, 1);
+  rb_define_private_method(cNokogiriXmlRelaxNG, "validate_document", noko_xml_relax_ng__validate_document, 1);
 }