nokogiri 1.0.0-x86-mswin32-60

data/ext/nokogiri/xml_node.h

@@ -0,0 +1,15 @@
+#ifndef NOKOGIRI_XML_NODE
+#define NOKOGIRI_XML_NODE
+
+#include <native.h>
+
+void init_xml_node();
+VALUE Nokogiri_wrap_xml_node(xmlNodePtr root);
+
+extern VALUE cNokogiriXmlNode ;
+VALUE Nokogiri_wrap_xml_node(xmlNodePtr node) ;
+void Nokogiri_xml_node_properties(xmlNodePtr node, VALUE attr_hash) ;
+void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash) ;
+int Nokogiri_xml_node_owned_get(xmlNodePtr node) ;
+void Nokogiri_xml_node_owned_set(xmlNodePtr node) ;
+#endif

data/ext/nokogiri/xml_node_set.c

@@ -0,0 +1,124 @@
+#include <xml_node_set.h>
+#include <libxml/xpathInternals.h>
+/*
+ * call-seq:
+ *  length
+ *
+ * Get the length of the node set
+ */
+static VALUE length(VALUE self)
+{
+  xmlNodeSetPtr node_set;
+  Data_Get_Struct(self, xmlNodeSet, node_set);
+
+  if(node_set)
+    return INT2NUM(node_set->nodeNr);
+
+  return INT2NUM(0);
+}
+
+/*
+ * call-seq:
+ *  push(node)
+ *
+ * Append +node+ to the NodeSet.
+ */
+static VALUE push(VALUE self, VALUE rb_node)
+{
+  xmlNodeSetPtr node_set;
+  xmlNodePtr node;
+
+  Data_Get_Struct(self, xmlNodeSet, node_set);
+  Data_Get_Struct(rb_node, xmlNode, node);
+  xmlXPathNodeSetAdd(node_set, node);
+  return self;
+}
+
+/*
+ * call-seq:
+ *  [](i)
+ *
+ * Get the node at index +i+
+ */
+static VALUE index_at(VALUE self, VALUE number)
+{
+  int i = NUM2INT(number);
+  xmlNodeSetPtr node_set;
+  Data_Get_Struct(self, xmlNodeSet, node_set);
+
+  if(i >= node_set->nodeNr || abs(i) > node_set->nodeNr)
+    return Qnil;
+
+  if(i < 0)
+    i = i + node_set->nodeNr;
+
+  return Nokogiri_wrap_xml_node(node_set->nodeTab[i]);
+}
+
+static void gc_mark(xmlNodeSetPtr node_set)
+{
+    int j ;
+    for (j = 0 ; j < node_set->nodeNr ; ++j) {
+        if (node_set->nodeTab[j]->_private)
+            rb_gc_mark((VALUE)node_set->nodeTab[j]->_private);
+    }
+}
+
+static void deallocate(xmlNodeSetPtr node_set)
+{
+  /*
+   *  xmlXPathFreeNodeSet() contains an implicit assumption that it is being
+   *  called before any of its pointed-to nodes have been free()d. this
+   *  assumption lies in the operation where it dereferences nodeTab pointers
+   *  while searching for namespace nodes to free.
+   *
+   *  however, since Ruby's GC mechanism cannot guarantee the strict order in
+   *  which ruby objects will be GC'd, nodes may be garbage collected before a
+   *  nodeset containing pointers to those nodes. (this is true regardless of
+   *  how we declare dependencies between objects with rb_gc_mark().)
+   *
+   *  as a result, xmlXPathFreeNodeSet() will perform unsafe memory operations,
+   *  and calling it would be evil.
+   *
+   *  on the bright side, though, Nokogiri's API currently does not cause
+   *  namespace nodes to be included in node sets, ever.
+   *
+   *  armed with that fact, we examined xmlXPathFreeNodeSet() and related libxml
+   *  code and determined that, within the Nokogiri abstraction, we will not
+   *  leak memory if we simply free the node set's memory directly. that's only
+   *  quasi-evil!
+   *
+   *  there's probably a lesson in here somewhere about intermingling, within a
+   *  single array, structs with different memory-ownership semantics. or more
+   *  generally, a lesson about building an API in C/C++ that does not contain
+   *  assumptions about the strict order in which memory will be released. hey,
+   *  that sounds like a great idea for a blog post! get to it!
+   *
+   *  "In Valgrind We Trust." seriously.
+   */
+  NOKOGIRI_DEBUG_START(node_set) ;
+  if (node_set->nodeTab != NULL)
+    xmlFree(node_set->nodeTab);
+  xmlFree(node_set);
+  NOKOGIRI_DEBUG_END(node_set) ;
+}
+
+static VALUE allocate(VALUE klass)
+{
+  return Nokogiri_wrap_xml_node_set(xmlXPathNodeSetCreate(NULL));
+}
+
+VALUE Nokogiri_wrap_xml_node_set(xmlNodeSetPtr node_set)
+{
+    return Data_Wrap_Struct(cNokogiriXmlNodeSet, gc_mark, deallocate, node_set);
+}
+
+VALUE cNokogiriXmlNodeSet ;
+void init_xml_node_set(void)
+{
+  VALUE klass = cNokogiriXmlNodeSet = rb_eval_string("Nokogiri::XML::NodeSet");
+  rb_define_alloc_func(klass, allocate);
+  rb_define_method(klass, "length", length, 0);
+  rb_define_method(klass, "[]", index_at, 1);
+  rb_define_method(klass, "push", push, 1);
+}

data/ext/nokogiri/xml_node_set.h

@@ -0,0 +1,9 @@
+#ifndef NOKOGIRI_XML_NODE_SET
+#define NOKOGIRI_XML_NODE_SET
+
+#include <native.h>
+void init_xml_node_set();
+
+extern VALUE cNokogiriXmlNodeSet ;
+VALUE Nokogiri_wrap_xml_node_set(xmlNodeSetPtr node_set) ;
+#endif

data/ext/nokogiri/xml_reader.c

@@ -0,0 +1,429 @@
+#include <xml_reader.h>
+
+static void dealloc(xmlTextReaderPtr reader)
+{
+  NOKOGIRI_DEBUG_START(reader);
+  xmlFreeTextReader(reader);
+  NOKOGIRI_DEBUG_END(reader);
+}
+
+static int has_attributes(xmlTextReaderPtr reader)
+{
+  /*
+   *  this implementation of xmlTextReaderHasAttributes explicitly includes
+   *  namespaces and properties, because some earlier versions ignore
+   *  namespaces.
+   */
+  xmlNodePtr node ;
+  node = xmlTextReaderCurrentNode(reader);
+  if (node == NULL)
+    return(0);
+
+  if ((node->type == XML_ELEMENT_NODE) &&
+      ((node->properties != NULL) || (node->nsDef != NULL)))
+    return(1);
+  return(0);
+}
+
+/*
+ * call-seq:
+ *   default?
+ *
+ * Was an attribute generated from the default value in the DTD or schema?
+ */
+static VALUE default_eh(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  int eh = xmlTextReaderIsDefault(reader);
+  if(eh == 0) return Qfalse;
+  if(eh == 1) return Qtrue;
+
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *   value?
+ *
+ * Does this node have a text value?
+ */
+static VALUE value_eh(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  int eh = xmlTextReaderHasValue(reader);
+  if(eh == 0) return Qfalse;
+  if(eh == 1) return Qtrue;
+
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *   attributes?
+ *
+ * Does this node have attributes?
+ */
+static VALUE attributes_eh(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  int eh = has_attributes(reader);
+  if(eh == 0) return Qfalse;
+  if(eh == 1) return Qtrue;
+
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *   attributes
+ *
+ * Get a Hash of attributes for this node
+ */
+static VALUE attributes(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  VALUE attr ;
+
+  Data_Get_Struct(self, xmlTextReader, reader);
+
+  attr = rb_hash_new() ;
+
+  if (! has_attributes(reader))
+    return attr ;
+
+  xmlNodePtr ptr = xmlTextReaderExpand(reader);
+  if(ptr == NULL) return Qnil;
+
+  Nokogiri_xml_node_namespaces(ptr, attr);
+  Nokogiri_xml_node_properties(ptr, attr);
+
+  return attr ;
+}
+
+/*
+ * call-seq:
+ *   attribute_at(index)
+ *
+ * Get the value of attribute at +index+
+ */
+static VALUE attribute_at(VALUE self, VALUE index)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+
+  if(index == Qnil) return Qnil;
+  index = rb_funcall(index, rb_intern("to_i"), 0);
+
+  xmlChar * value = xmlTextReaderGetAttributeNo(
+      reader,
+      NUM2INT(index)
+  );
+  if(value == NULL) return Qnil;
+
+  VALUE rb_value = rb_str_new2((const char *)value);
+  xmlFree(value);
+  return rb_value;
+}
+
+/*
+ * call-seq:
+ *   attribute(name)
+ *
+ * Get the value of attribute named +name+
+ */
+static VALUE reader_attribute(VALUE self, VALUE name)
+{
+  xmlTextReaderPtr reader;
+  xmlChar *value ;
+  Data_Get_Struct(self, xmlTextReader, reader);
+
+  if(name == Qnil) return Qnil;
+  name = StringValue(name) ;
+
+  value = xmlTextReaderGetAttribute(reader, (xmlChar*)StringValuePtr(name));
+  if(value == NULL) {
+    /* this section is an attempt to workaround older versions of libxml that
+       don't handle namespaces properly in all attribute-and-friends functions */
+    xmlChar *prefix = NULL ;
+    xmlChar *localname = xmlSplitQName2((xmlChar*)StringValuePtr(name), &prefix);
+    if (localname != NULL) {
+      value = xmlTextReaderLookupNamespace(reader, localname);
+      free(localname) ;
+    } else {
+      value = xmlTextReaderLookupNamespace(reader, prefix);
+    }
+  }
+  if(value == NULL) return Qnil;
+
+  VALUE rb_value = rb_str_new2((const char *)value);
+  xmlFree(value);
+  return rb_value;
+}
+
+/*
+ * call-seq:
+ *   attribute_count
+ *
+ * Get the number of attributes for the current node
+ */
+static VALUE attribute_count(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  int count = xmlTextReaderAttributeCount(reader);
+  if(count == -1) return Qnil;
+
+  return INT2NUM(count);
+}
+
+/*
+ * call-seq:
+ *   depth
+ *
+ * Get the depth of the node
+ */
+static VALUE depth(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  int depth = xmlTextReaderDepth(reader);
+  if(depth == -1) return Qnil;
+
+  return INT2NUM(depth);
+}
+
+/*
+ * call-seq:
+ *   encoding
+ *
+ * Get the encoding for the document
+ */
+static VALUE encoding(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  const char * encoding = (const char *)xmlTextReaderConstEncoding(reader);
+  if(encoding == NULL) return Qnil;
+
+  return rb_str_new2(encoding);
+}
+
+/*
+ * call-seq:
+ *   xml_version
+ *
+ * Get the XML version of the document being read
+ */
+static VALUE xml_version(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  const char * version = (const char *)xmlTextReaderConstXmlVersion(reader);
+  if(version == NULL) return Qnil;
+
+  return rb_str_new2(version);
+}
+
+/*
+ * call-seq:
+ *   lang
+ *
+ * Get the xml:lang scope within which the node resides.
+ */
+static VALUE lang(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  const char * lang = (const char *)xmlTextReaderConstXmlLang(reader);
+  if(lang == NULL) return Qnil;
+
+  return rb_str_new2(lang);
+}
+
+/*
+ * call-seq:
+ *   value
+ *
+ * Get the text value of the node if present
+ */
+static VALUE value(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  const char * value = (const char *)xmlTextReaderConstValue(reader);
+  if(value == NULL) return Qnil;
+
+  return rb_str_new2(value);
+}
+
+/*
+ * call-seq:
+ *   prefix
+ *
+ * Get the shorthand reference to the namespace associated with the node.
+ */
+static VALUE prefix(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  const char * prefix = (const char *)xmlTextReaderConstPrefix(reader);
+  if(prefix == NULL) return Qnil;
+
+  return rb_str_new2(prefix);
+}
+
+/*
+ * call-seq:
+ *   namespace_uri
+ *
+ * Get the URI defining the namespace associated with the node
+ */
+static VALUE namespace_uri(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  const char * uri = (const char *)xmlTextReaderConstNamespaceUri(reader);
+  if(uri == NULL) return Qnil;
+
+  return rb_str_new2(uri);
+}
+
+/*
+ * call-seq:
+ *   local_name
+ *
+ * Get the local name of the node
+ */
+static VALUE local_name(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  const char * name = (const char *)xmlTextReaderConstLocalName(reader);
+  if(name == NULL) return Qnil;
+
+  return rb_str_new2(name);
+}
+
+/*
+ * call-seq:
+ *   name
+ *
+ * Get the name of the node
+ */
+static VALUE name(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  const char * name = (const char *)xmlTextReaderConstName(reader);
+  if(name == NULL) return Qnil;
+
+  return rb_str_new2(name);
+}
+
+/*
+ * call-seq:
+ *   state
+ *
+ * Get the state of the reader
+ */
+static VALUE state(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  return INT2NUM(xmlTextReaderReadState(reader));
+}
+
+/*
+ * call-seq:
+ *   read
+ *
+ * Move the Reader forward through the XML document.
+ */
+static VALUE read_more(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+
+  int ret = xmlTextReaderRead(reader);
+  if(ret == 1) return self;
+  if(ret == 0) return Qnil;
+
+  rb_raise(rb_eRuntimeError, "Error pulling: %d", ret);
+}
+
+/*
+ * call-seq:
+ *   from_memory(string, url = nil, encoding = nil, options = 0)
+ *
+ * Create a new reader that parses +string+
+ */
+static VALUE from_memory(int argc, VALUE *argv, VALUE klass)
+{
+  VALUE rb_buffer, rb_url, encoding, rb_options;
+
+  const char * c_url      = NULL;
+  const char * c_encoding = NULL;
+  int c_options           = 0; 
+
+  rb_scan_args(argc, argv, "13", &rb_buffer, &rb_url, &encoding, &rb_options);
+
+  rb_buffer = StringValue(rb_buffer) ;
+  if (RTEST(rb_url)) c_url = StringValuePtr(rb_url);
+  if (RTEST(encoding)) c_encoding = StringValuePtr(rb_url);
+  if (RTEST(rb_options)) c_options = NUM2INT(rb_options);
+
+  xmlTextReaderPtr reader = xmlReaderForMemory(
+      StringValuePtr(rb_buffer),
+      NUM2INT(rb_funcall(rb_buffer, rb_intern("length"), 0)),
+      c_url,
+      c_encoding,
+      c_options
+  );
+
+  if(reader == NULL) {
+    xmlFreeTextReader(reader);
+    rb_raise(rb_eRuntimeError, "couldn't create a parser");
+  }
+
+  return Data_Wrap_Struct(klass, NULL, dealloc, reader);
+}
+
+VALUE cNokogiriXmlReader;
+
+void init_xml_reader()
+{
+  VALUE module = rb_define_module("Nokogiri");
+  VALUE xml = rb_define_module_under(module, "XML");
+
+  /*
+   * 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!
+   */
+  VALUE klass = rb_define_class_under(xml, "Reader", rb_cObject);
+
+  cNokogiriXmlReader = klass;
+
+  rb_define_singleton_method(klass, "from_memory", from_memory, -1);
+  rb_define_method(klass, "read", read_more, 0);
+  rb_define_method(klass, "state", state, 0);
+  rb_define_method(klass, "name", name, 0);
+  rb_define_method(klass, "local_name", local_name, 0);
+  rb_define_method(klass, "namespace_uri", namespace_uri, 0);
+  rb_define_method(klass, "prefix", prefix, 0);
+  rb_define_method(klass, "value", value, 0);
+  rb_define_method(klass, "lang", lang, 0);
+  rb_define_method(klass, "xml_version", xml_version, 0);
+  rb_define_method(klass, "encoding", encoding, 0);
+  rb_define_method(klass, "depth", depth, 0);
+  rb_define_method(klass, "attribute_count", attribute_count, 0);
+  rb_define_method(klass, "attribute", reader_attribute, 1);
+  rb_define_method(klass, "attribute_at", attribute_at, 1);
+  rb_define_method(klass, "attributes", attributes, 0);
+  rb_define_method(klass, "attributes?", attributes_eh, 0);
+  rb_define_method(klass, "value?", value_eh, 0);
+  rb_define_method(klass, "default?", default_eh, 0);
+}