nokogiri 1.0.0 → 1.6.8.1

data/ext/nokogiri/xml_document_fragment.c

@@ -0,0 +1,48 @@
+#include <xml_document_fragment.h>
+
+/*
+ * call-seq:
+ *  new(document)
+ *
+ * Create a new DocumentFragment element on the +document+
+ */
+static VALUE new(int argc, VALUE *argv, VALUE klass)
+{
+  xmlDocPtr xml_doc;
+  xmlNodePtr node;
+  VALUE document;
+  VALUE rest;
+  VALUE rb_node;
+
+  rb_scan_args(argc, argv, "1*", &document, &rest);
+
+  Data_Get_Struct(document, xmlDoc, xml_doc);
+
+  node = xmlNewDocFragment(xml_doc->doc);
+
+  nokogiri_root_node(node);
+
+  rb_node = Nokogiri_wrap_xml_node(klass, node);
+  rb_obj_call_init(rb_node, argc, argv);
+
+  if(rb_block_given_p()) rb_yield(rb_node);
+
+  return rb_node;
+}
+
+VALUE cNokogiriXmlDocumentFragment;
+void init_xml_document_fragment()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE xml = rb_define_module_under(nokogiri, "XML");
+  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
+
+  /*
+   * DocumentFragment represents a DocumentFragment node in an xml document.
+   */
+  VALUE klass = rb_define_class_under(xml, "DocumentFragment", node);
+
+  cNokogiriXmlDocumentFragment = klass;
+
+  rb_define_singleton_method(klass, "new", new, -1);
+}

data/ext/nokogiri/xml_document_fragment.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_XML_DOCUMENT_FRAGMENT
+#define NOKOGIRI_XML_DOCUMENT_FRAGMENT
+
+#include <nokogiri.h>
+
+void init_xml_document_fragment();
+
+extern VALUE cNokogiriXmlDocumentFragment;
+#endif
+

data/ext/nokogiri/xml_dtd.c

@@ -6,22 +6,25 @@ static void notation_copier(void *payload, void *data, xmlChar *name)
   VALUE klass = rb_const_get(mNokogiriXml, rb_intern("Notation"));
 
   xmlNotationPtr c_notation = (xmlNotationPtr)payload;
+  VALUE notation;
+  VALUE argv[3];
+  argv[0] = (c_notation->name ? NOKOGIRI_STR_NEW2(c_notation->name) : Qnil);
+  argv[1] = (c_notation->PublicID ? NOKOGIRI_STR_NEW2(c_notation->PublicID) : Qnil);
+  argv[2] = (c_notation->SystemID ? NOKOGIRI_STR_NEW2(c_notation->SystemID) : Qnil);
 
-  VALUE notation = rb_funcall(klass, rb_intern("new"), 3,
-      c_notation->name ? rb_str_new2((const char *)c_notation->name) : Qnil,
-      c_notation->PublicID ? rb_str_new2((const char *)c_notation->PublicID) : Qnil,
-      c_notation->SystemID ? rb_str_new2((const char *)c_notation->SystemID) : Qnil);
+  notation = rb_class_new_instance(3, argv, klass);
 
-  rb_hash_aset(hash, rb_str_new2((const char *)name), notation);
+  rb_hash_aset(hash, NOKOGIRI_STR_NEW2(name),notation);
 }
 
-static void element_copier(void *payload, void *data, xmlChar *name)
+static void element_copier(void *_payload, void *data, xmlChar *name)
 {
   VALUE hash = (VALUE)data;
+  xmlNodePtr payload = (xmlNodePtr)_payload;
 
-  VALUE element = Nokogiri_wrap_xml_node((xmlNodePtr)payload);
+  VALUE element = Nokogiri_wrap_xml_node(Qnil, payload);
 
-  rb_hash_aset(hash, rb_str_new2((const char *)name), element);
+  rb_hash_aset(hash, NOKOGIRI_STR_NEW2(name), element);
 }
 
 /*
@@ -33,11 +36,13 @@ static void element_copier(void *payload, void *data, xmlChar *name)
 static VALUE entities(VALUE self)
 {
   xmlDtdPtr dtd;
+  VALUE hash;
+
   Data_Get_Struct(self, xmlDtd, dtd);
 
   if(!dtd->entities) return Qnil;
 
-  VALUE hash = rb_hash_new();
+  hash = rb_hash_new();
 
   xmlHashScan((xmlHashTablePtr)dtd->entities, element_copier, (void *)hash);
 
@@ -46,40 +51,44 @@ static VALUE entities(VALUE self)
 
 /*
  * call-seq:
- *   attributes
+ *   notations
  *
- * Get a hash of the attributes for this DTD.
+ * Get a hash of the notations for this DTD.
  */
-static VALUE attributes(VALUE self)
+static VALUE notations(VALUE self)
 {
   xmlDtdPtr dtd;
+  VALUE hash;
+
   Data_Get_Struct(self, xmlDtd, dtd);
 
-  if(!dtd->attributes) return Qnil;
+  if(!dtd->notations) return Qnil;
 
-  VALUE hash = rb_hash_new();
+  hash = rb_hash_new();
 
-  xmlHashScan((xmlHashTablePtr)dtd->attributes, element_copier, (void *)hash);
+  xmlHashScan((xmlHashTablePtr)dtd->notations, notation_copier, (void *)hash);
 
   return hash;
 }
 
 /*
  * call-seq:
- *   notations
+ *   attributes
  *
- * Get a hash of the notations for this DTD.
+ * Get a hash of the attributes for this DTD.
  */
-static VALUE notations(VALUE self)
+static VALUE attributes(VALUE self)
 {
   xmlDtdPtr dtd;
+  VALUE hash;
+
   Data_Get_Struct(self, xmlDtd, dtd);
 
-  if(!dtd->notations) return Qnil;
+  hash = rb_hash_new();
 
-  VALUE hash = rb_hash_new();
+  if(!dtd->attributes) return hash;
 
-  xmlHashScan((xmlHashTablePtr)dtd->notations, notation_copier, (void *)hash);
+  xmlHashScan((xmlHashTablePtr)dtd->attributes, element_copier, (void *)hash);
 
   return hash;
 }
@@ -93,25 +102,101 @@ static VALUE notations(VALUE self)
 static VALUE elements(VALUE self)
 {
   xmlDtdPtr dtd;
+  VALUE hash;
+
   Data_Get_Struct(self, xmlDtd, dtd);
 
   if(!dtd->elements) return Qnil;
 
-  VALUE hash = rb_hash_new();
+  hash = rb_hash_new();
 
   xmlHashScan((xmlHashTablePtr)dtd->elements, element_copier, (void *)hash);
 
   return hash;
 }
 
+/*
+ * call-seq:
+ *   validate(document)
+ *
+ * Validate +document+ returning a list of errors
+ */
+static VALUE validate(VALUE self, VALUE document)
+{
+  xmlDocPtr doc;
+  xmlDtdPtr dtd;
+  xmlValidCtxtPtr ctxt;
+  VALUE error_list;
+
+  Data_Get_Struct(self, xmlDtd, dtd);
+  Data_Get_Struct(document, xmlDoc, doc);
+  error_list = rb_ary_new();
+
+  ctxt = xmlNewValidCtxt();
+
+  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
+
+  xmlValidateDtd(ctxt, doc, dtd);
+
+  xmlSetStructuredErrorFunc(NULL, NULL);
+
+  xmlFreeValidCtxt(ctxt);
+
+  return error_list;
+}
+
+/*
+ * call-seq:
+ *   system_id
+ *
+ * Get the System ID for this DTD
+ */
+static VALUE system_id(VALUE self)
+{
+  xmlDtdPtr dtd;
+  Data_Get_Struct(self, xmlDtd, dtd);
+
+  if(!dtd->SystemID) return Qnil;
+
+  return NOKOGIRI_STR_NEW2(dtd->SystemID);
+}
+
+/*
+ * call-seq:
+ *   external_id
+ *
+ * Get the External ID for this DTD
+ */
+static VALUE external_id(VALUE self)
+{
+  xmlDtdPtr dtd;
+  Data_Get_Struct(self, xmlDtd, dtd);
+
+  if(!dtd->ExternalID) return Qnil;
+
+  return NOKOGIRI_STR_NEW2(dtd->ExternalID);
+}
+
+VALUE cNokogiriXmlDtd;
+
 void init_xml_dtd()
 {
   VALUE nokogiri = rb_define_module("Nokogiri");
   VALUE xml = rb_define_module_under(nokogiri, "XML");
-  VALUE klass = rb_define_class_under(xml, "DTD", cNokogiriXmlNode);
+  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
+
+  /*
+   * Nokogiri::XML::DTD wraps DTD nodes in an XML document
+   */
+  VALUE klass = rb_define_class_under(xml, "DTD", node);
+
+  cNokogiriXmlDtd = klass;
 
   rb_define_method(klass, "notations", notations, 0);
   rb_define_method(klass, "elements", elements, 0);
-  rb_define_method(klass, "attributes", attributes, 0);
   rb_define_method(klass, "entities", entities, 0);
+  rb_define_method(klass, "validate", validate, 1);
+  rb_define_method(klass, "attributes", attributes, 0);
+  rb_define_method(klass, "system_id", system_id, 0);
+  rb_define_method(klass, "external_id", external_id, 0);
 }

data/ext/nokogiri/xml_dtd.h

@@ -1,7 +1,9 @@
 #ifndef NOKOGIRI_XML_DTD
 #define NOKOGIRI_XML_DTD
 
-#include <native.h>
+#include <nokogiri.h>
+
+extern VALUE cNokogiriXmlDtd;
 
 void init_xml_dtd();
 

data/ext/nokogiri/xml_element_content.c

@@ -0,0 +1,123 @@
+#include <xml_element_content.h>
+
+VALUE cNokogiriXmlElementContent;
+
+/*
+ * call-seq:
+ *  name
+ *
+ * Get the require element +name+
+ */
+static VALUE get_name(VALUE self)
+{
+  xmlElementContentPtr elem;
+  Data_Get_Struct(self, xmlElementContent, elem);
+
+  if(!elem->name) return Qnil;
+  return NOKOGIRI_STR_NEW2(elem->name);
+}
+
+/*
+ * call-seq:
+ *  type
+ *
+ * Get the element content +type+.  Possible values are PCDATA, ELEMENT, SEQ,
+ * or OR.
+ */
+static VALUE get_type(VALUE self)
+{
+  xmlElementContentPtr elem;
+  Data_Get_Struct(self, xmlElementContent, elem);
+
+  return INT2NUM((long)elem->type);
+}
+
+/*
+ * call-seq:
+ *  c1
+ *
+ * Get the first child.
+ */
+static VALUE get_c1(VALUE self)
+{
+  xmlElementContentPtr elem;
+  Data_Get_Struct(self, xmlElementContent, elem);
+
+  if(!elem->c1) return Qnil;
+  return Nokogiri_wrap_element_content(rb_iv_get(self, "@document"), elem->c1);
+}
+
+/*
+ * call-seq:
+ *  c2
+ *
+ * Get the first child.
+ */
+static VALUE get_c2(VALUE self)
+{
+  xmlElementContentPtr elem;
+  Data_Get_Struct(self, xmlElementContent, elem);
+
+  if(!elem->c2) return Qnil;
+  return Nokogiri_wrap_element_content(rb_iv_get(self, "@document"), elem->c2);
+}
+
+/*
+ * call-seq:
+ *  occur
+ *
+ * Get the element content +occur+ flag.  Possible values are ONCE, OPT, MULT
+ * or PLUS.
+ */
+static VALUE get_occur(VALUE self)
+{
+  xmlElementContentPtr elem;
+  Data_Get_Struct(self, xmlElementContent, elem);
+
+  return INT2NUM((long)elem->ocur);
+}
+
+/*
+ * call-seq:
+ *  prefix
+ *
+ * Get the element content namespace +prefix+.
+ */
+static VALUE get_prefix(VALUE self)
+{
+  xmlElementContentPtr elem;
+  Data_Get_Struct(self, xmlElementContent, elem);
+
+  if(!elem->prefix) return Qnil;
+
+  return NOKOGIRI_STR_NEW2(elem->prefix);
+}
+
+VALUE Nokogiri_wrap_element_content(VALUE doc, xmlElementContentPtr element)
+{
+  VALUE elem = Data_Wrap_Struct(cNokogiriXmlElementContent, 0, 0, element);
+
+  /* Setting the document is necessary so that this does not get GC'd until */
+  /* the document is GC'd */
+  rb_iv_set(elem, "@document", doc);
+
+  return elem;
+}
+
+void init_xml_element_content()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE xml = rb_define_module_under(nokogiri, "XML");
+
+  VALUE klass = rb_define_class_under(xml, "ElementContent", rb_cObject);
+
+  cNokogiriXmlElementContent = klass;
+
+  rb_define_method(klass, "name", get_name, 0);
+  rb_define_method(klass, "type", get_type, 0);
+  rb_define_method(klass, "occur", get_occur, 0);
+  rb_define_method(klass, "prefix", get_prefix, 0);
+
+  rb_define_private_method(klass, "c1", get_c1, 0);
+  rb_define_private_method(klass, "c2", get_c2, 0);
+}

data/ext/nokogiri/xml_element_content.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_XML_ELEMENT_CONTENT
+#define NOKOGIRI_XML_ELEMENT_CONTENT
+
+#include <nokogiri.h>
+
+
+VALUE Nokogiri_wrap_element_content(VALUE doc, xmlElementContentPtr element);
+void init_xml_element_content();
+
+#endif

data/ext/nokogiri/xml_element_decl.c

@@ -0,0 +1,69 @@
+#include <xml_element_decl.h>
+
+static ID id_document;
+
+/*
+ * call-seq:
+ *  element_type
+ *
+ * The element_type
+ */
+static VALUE element_type(VALUE self)
+{
+  xmlElementPtr node;
+  Data_Get_Struct(self, xmlElement, node);
+  return INT2NUM((long)node->etype);
+}
+
+/*
+ * call-seq:
+ *  content
+ *
+ * The allowed content for this ElementDecl
+ */
+static VALUE content(VALUE self)
+{
+  xmlElementPtr node;
+  Data_Get_Struct(self, xmlElement, node);
+
+  if(!node->content) return Qnil;
+
+  return Nokogiri_wrap_element_content(
+      rb_funcall(self, id_document, 0),
+      node->content
+  );
+}
+
+/*
+ * call-seq:
+ *  prefix
+ *
+ * The namespace prefix for this ElementDecl
+ */
+static VALUE prefix(VALUE self)
+{
+  xmlElementPtr node;
+  Data_Get_Struct(self, xmlElement, node);
+
+  if(!node->prefix) return Qnil;
+
+  return NOKOGIRI_STR_NEW2(node->prefix);
+}
+
+VALUE cNokogiriXmlElementDecl;
+
+void init_xml_element_decl()
+{
+  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 klass = rb_define_class_under(xml, "ElementDecl", node);
+
+  cNokogiriXmlElementDecl = klass;
+
+  rb_define_method(klass, "element_type", element_type, 0);
+  rb_define_method(klass, "content", content, 0);
+  rb_define_method(klass, "prefix", prefix, 0);
+
+  id_document = rb_intern("document");
+}

data/ext/nokogiri/xml_element_decl.h

@@ -0,0 +1,9 @@
+#ifndef NOKOGIRI_XML_ELEMENT_DECL
+#define NOKOGIRI_XML_ELEMENT_DECL
+
+#include <nokogiri.h>
+
+void init_xml_element_decl();
+
+extern VALUE cNokogiriXmlElementDecl;
+#endif

data/ext/nokogiri/xml_encoding_handler.c

@@ -0,0 +1,79 @@
+#include <xml_encoding_handler.h>
+
+/*
+ * call-seq: Nokogiri::EncodingHandler.[](name)
+ *
+ * Get the encoding handler for +name+
+ */
+static VALUE get(VALUE klass, VALUE key)
+{
+  xmlCharEncodingHandlerPtr handler;
+
+  handler = xmlFindCharEncodingHandler(StringValueCStr(key));
+  if(handler)
+    return Data_Wrap_Struct(klass, NULL, NULL, handler);
+
+  return Qnil;
+}
+
+/*
+ * call-seq: Nokogiri::EncodingHandler.delete(name)
+ *
+ * Delete the encoding alias named +name+
+ */
+static VALUE delete(VALUE klass, VALUE name)
+{
+  if(xmlDelEncodingAlias(StringValueCStr(name))) return Qnil;
+
+  return Qtrue;
+}
+
+/*
+ * call-seq: Nokogiri::EncodingHandler.alias(from, to)
+ *
+ * Alias encoding handler with name +from+ to name +to+
+ */
+static VALUE alias(VALUE klass, VALUE from, VALUE to)
+{
+  xmlAddEncodingAlias(StringValueCStr(from), StringValueCStr(to));
+
+  return to;
+}
+
+/*
+ * call-seq: Nokogiri::EncodingHandler.clear_aliases!
+ *
+ * Remove all encoding aliases.
+ */
+static VALUE clear_aliases(VALUE klass)
+{
+  xmlCleanupEncodingAliases();
+
+  return klass;
+}
+
+/*
+ * call-seq: name
+ *
+ * Get the name of this EncodingHandler
+ */
+static VALUE name(VALUE self)
+{
+  xmlCharEncodingHandlerPtr handler;
+
+  Data_Get_Struct(self, xmlCharEncodingHandler, handler);
+
+  return NOKOGIRI_STR_NEW2(handler->name);
+}
+
+void init_xml_encoding_handler()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE klass = rb_define_class_under(nokogiri, "EncodingHandler", rb_cObject);
+
+  rb_define_singleton_method(klass, "[]", get, 1);
+  rb_define_singleton_method(klass, "delete", delete, 1);
+  rb_define_singleton_method(klass, "alias", alias, 2);
+  rb_define_singleton_method(klass, "clear_aliases!", clear_aliases, 0);
+  rb_define_method(klass, "name", name, 0);
+}

data/ext/nokogiri/xml_encoding_handler.h

@@ -0,0 +1,8 @@
+#ifndef NOKOGIRI_XML_ENCODING_HANDLER
+#define NOKOGIRI_XML_ENCODING_HANDLER
+
+#include <nokogiri.h>
+
+void init_xml_encoding_handler();
+
+#endif

data/ext/nokogiri/xml_entity_decl.c

@@ -0,0 +1,110 @@
+#include <xml_entity_decl.h>
+
+/*
+ * call-seq:
+ *  original_content
+ *
+ * Get the original_content before ref substitution
+ */
+static VALUE original_content(VALUE self)
+{
+  xmlEntityPtr node;
+  Data_Get_Struct(self, xmlEntity, node);
+
+  if(!node->orig) return Qnil;
+
+  return NOKOGIRI_STR_NEW2(node->orig);
+}
+
+/*
+ * call-seq:
+ *  content
+ *
+ * Get the content
+ */
+static VALUE get_content(VALUE self)
+{
+  xmlEntityPtr node;
+  Data_Get_Struct(self, xmlEntity, node);
+
+  if(!node->content) return Qnil;
+
+  return NOKOGIRI_STR_NEW(node->content, node->length);
+}
+
+/*
+ * call-seq:
+ *  entity_type
+ *
+ * Get the entity type
+ */
+static VALUE entity_type(VALUE self)
+{
+  xmlEntityPtr node;
+  Data_Get_Struct(self, xmlEntity, node);
+
+  return INT2NUM((int)node->etype);
+}
+
+/*
+ * call-seq:
+ *  external_id
+ *
+ * Get the external identifier for PUBLIC
+ */
+static VALUE external_id(VALUE self)
+{
+  xmlEntityPtr node;
+  Data_Get_Struct(self, xmlEntity, node);
+
+  if(!node->ExternalID) return Qnil;
+
+  return NOKOGIRI_STR_NEW2(node->ExternalID);
+}
+
+/*
+ * call-seq:
+ *  system_id
+ *
+ * Get the URI for a SYSTEM or PUBLIC Entity
+ */
+static VALUE system_id(VALUE self)
+{
+  xmlEntityPtr node;
+  Data_Get_Struct(self, xmlEntity, node);
+
+  if(!node->SystemID) return Qnil;
+
+  return NOKOGIRI_STR_NEW2(node->SystemID);
+}
+
+VALUE cNokogiriXmlEntityDecl;
+
+void init_xml_entity_decl()
+{
+  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 klass = rb_define_class_under(xml, "EntityDecl", node);
+
+  cNokogiriXmlEntityDecl = klass;
+
+  rb_define_method(klass, "original_content", original_content, 0);
+  rb_define_method(klass, "content", get_content, 0);
+  rb_define_method(klass, "entity_type", entity_type, 0);
+  rb_define_method(klass, "external_id", external_id, 0);
+  rb_define_method(klass, "system_id", system_id, 0);
+
+  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("INTERNAL_GENERAL"),
+      INT2NUM(XML_INTERNAL_GENERAL_ENTITY));
+  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("EXTERNAL_GENERAL_PARSED"),
+      INT2NUM(XML_EXTERNAL_GENERAL_PARSED_ENTITY));
+  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("EXTERNAL_GENERAL_UNPARSED"),
+      INT2NUM(XML_EXTERNAL_GENERAL_UNPARSED_ENTITY));
+  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("INTERNAL_PARAMETER"),
+      INT2NUM(XML_INTERNAL_PARAMETER_ENTITY));
+  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("EXTERNAL_PARAMETER"),
+      INT2NUM(XML_EXTERNAL_PARAMETER_ENTITY));
+  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("INTERNAL_PREDEFINED"),
+      INT2NUM(XML_INTERNAL_PREDEFINED_ENTITY));
+}

data/ext/nokogiri/xml_entity_decl.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_XML_ENTITY_DECL
+#define NOKOGIRI_XML_ENTITY_DECL
+
+#include <nokogiri.h>
+
+void init_xml_entity_decl();
+
+extern VALUE cNokogiriXmlEntityDecl;
+#endif
+