nokogiri 1.0.0 → 1.6.8.1

data/ext/nokogiri/xml_syntax_error.c

@@ -1,175 +1,51 @@
 #include <xml_syntax_error.h>
 
-static void dealloc(xmlErrorPtr ptr)
+void Nokogiri_error_array_pusher(void * ctx, xmlErrorPtr error)
 {
-  NOKOGIRI_DEBUG_START(ptr);
-  free(ptr);
-  NOKOGIRI_DEBUG_END(ptr);
+  VALUE list = (VALUE)ctx;
+  Check_Type(list, T_ARRAY);
+  rb_ary_push(list,  Nokogiri_wrap_xml_syntax_error(error));
 }
 
-/*
- * call-seq:
- *  column
- *
- * Column number or 0 if not available
- */
-static VALUE column(VALUE self)
+void Nokogiri_error_raise(void * ctx, xmlErrorPtr error)
 {
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->int2);
+  rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
 }
 
-/*
- * call-seq:
- *  int1
- *
- * Extra number information
- */
-static VALUE int1(VALUE self)
+VALUE Nokogiri_wrap_xml_syntax_error(xmlErrorPtr error)
 {
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->int1);
-}
-
-/*
- * call-seq:
- *  str3
- *
- * Extra string information
- */
-static VALUE str3(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  if(error->str3)
-    return rb_str_new2(error->str3);
-  return Qnil;
-}
-
-/*
- * call-seq:
- *  str2
- *
- * Extra string information
- */
-static VALUE str2(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  if(error->str2)
-    return rb_str_new2(error->str2);
-  return Qnil;
-}
-
-/*
- * call-seq:
- *  str1
- *
- * Extra string information
- */
-static VALUE str1(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  if(error->str1)
-    return rb_str_new2(error->str1);
-  return Qnil;
-}
-
-/*
- * call-seq:
- *  line
- *
- * Get the line number of the error
- */
-static VALUE line(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->line);
-}
-
-/*
- * call-seq:
- *  file
- *
- * Get the filename for the error
- */
-static VALUE file(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  if(error->file)
-    return rb_str_new2(error->file);
-
-  return Qnil;
-}
-
-/*
- * call-seq:
- *  level
- *
- * Get the error level
- */
-static VALUE level(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM((short)error->level);
-}
-
-/*
- * call-seq:
- *  code
- *
- * Get the error code
- */
-static VALUE code(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->code);
-}
-
-/*
- * call-seq:
- *  domain
- *
- * Get the part of the library that raised this exception
- */
-static VALUE domain(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->domain);
-}
-
-/*
- * call-seq:
- *  message
- *
- * Get the human readable message.
- */
-static VALUE message(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return rb_str_new2(error->message);
-}
-
-void Nokogiri_error_handler(void * ctx, xmlErrorPtr error)
-{
-  // FIXME: I'm interneting this.  I *think* the pointer passed in here gets
-  // freed on its own, thats why I copy it.
-  // The files are *in* the computer.
-  xmlErrorPtr ptr = calloc(1, sizeof(xmlError));
-  xmlCopyError(error, ptr);
-
-  VALUE err = Data_Wrap_Struct(cNokogiriXmlSyntaxError, NULL, dealloc, ptr);
-  VALUE block = rb_funcall(mNokogiri, rb_intern("error_handler"), 0);
-  rb_funcall(block, rb_intern("call"), 1, err);
+  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"));
+  }
+
+  msg = (error && error->message) ? NOKOGIRI_STR_NEW2(error->message) : Qnil;
+
+  e = rb_class_new_instance(
+      1,
+      &msg,
+      klass
+  );
+
+  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));
+    rb_iv_set(e, "@file", RBSTR_OR_QNIL(error->file));
+    rb_iv_set(e, "@line", INT2NUM(error->line));
+    rb_iv_set(e, "@str1", RBSTR_OR_QNIL(error->str1));
+    rb_iv_set(e, "@str2", RBSTR_OR_QNIL(error->str2));
+    rb_iv_set(e, "@str3", RBSTR_OR_QNIL(error->str3));
+    rb_iv_set(e, "@int1", INT2NUM(error->int1));
+    rb_iv_set(e, "@column", INT2NUM(error->int2));
+  }
+
+  return e;
 }
 
 VALUE cNokogiriXmlSyntaxError;
@@ -177,18 +53,12 @@ void init_xml_syntax_error()
 {
   VALUE nokogiri = rb_define_module("Nokogiri");
   VALUE xml = rb_define_module_under(nokogiri, "XML");
-  VALUE klass = rb_define_class_under(xml, "SyntaxError", rb_eSyntaxError);
+
+  /*
+   * 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;
 
-  rb_define_method(klass, "message", message, 0);
-  rb_define_method(klass, "domain", domain, 0);
-  rb_define_method(klass, "code", code, 0);
-  rb_define_method(klass, "level", level, 0);
-  rb_define_method(klass, "file", file, 0);
-  rb_define_method(klass, "line", line, 0);
-  rb_define_method(klass, "str1", str1, 0);
-  rb_define_method(klass, "str2", str2, 0);
-  rb_define_method(klass, "str3", str3, 0);
-  rb_define_method(klass, "int1", int1, 0);
-  rb_define_method(klass, "column", column, 0);
 }

data/ext/nokogiri/xml_syntax_error.h

@@ -1,10 +1,12 @@
 #ifndef NOKOGIRI_XML_SYNTAX_ERROR
 #define NOKOGIRI_XML_SYNTAX_ERROR
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xml_syntax_error();
-void Nokogiri_error_handler(void * ctx, xmlErrorPtr error);
+VALUE Nokogiri_wrap_xml_syntax_error(xmlErrorPtr error);
+void Nokogiri_error_array_pusher(void * ctx, xmlErrorPtr error);
+NORETURN(void Nokogiri_error_raise(void * ctx, xmlErrorPtr error));
 
 extern VALUE cNokogiriXmlSyntaxError;
 #endif

data/ext/nokogiri/xml_text.c

@@ -1,19 +1,31 @@
 #include <xml_text.h>
 
-static void dealloc(xmlNodePtr node)
+/*
+ * call-seq:
+ *  new(content, document)
+ *
+ * Create a new Text element on the +document+ with +content+
+ */
+static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
-  if (node->doc == NULL) {
-    NOKOGIRI_DEBUG_START_TEXT(node);
-    xmlFreeNode(node);
-    NOKOGIRI_DEBUG_END(node);
-  }
-}
+  xmlDocPtr doc;
+  xmlNodePtr node;
+  VALUE string;
+  VALUE document;
+  VALUE rest;
+  VALUE rb_node;
 
-static VALUE new(VALUE klass, VALUE string)
-{
-  xmlNodePtr node = xmlNewText((xmlChar *)StringValuePtr(string));
-  VALUE rb_node = Data_Wrap_Struct(klass, NULL, dealloc, node);
-  node->_private = (void *)rb_node;
+  rb_scan_args(argc, argv, "2*", &string, &document, &rest);
+
+  Data_Get_Struct(document, xmlDoc, doc);
+
+  node = xmlNewText((xmlChar *)StringValueCStr(string));
+  node->doc = 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);
 
@@ -23,7 +35,18 @@ static VALUE new(VALUE klass, VALUE string)
 VALUE cNokogiriXmlText ;
 void init_xml_text()
 {
-  VALUE klass = cNokogiriXmlText = rb_const_get(mNokogiriXml, rb_intern("Text"));
+  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);
+
+  /*
+   * Wraps Text nodes.
+   */
+  VALUE klass = rb_define_class_under(xml, "Text", char_data);
+
+  cNokogiriXmlText = klass;
 
-  rb_define_singleton_method(klass, "new", new, 1);
+  rb_define_singleton_method(klass, "new", new, -1);
 }

data/ext/nokogiri/xml_text.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_XML_TEXT
 #define NOKOGIRI_XML_TEXT
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xml_text();
 

data/ext/nokogiri/xml_xpath_context.c

@@ -1,5 +1,7 @@
 #include <xml_xpath_context.h>
 
+int vasprintf (char **strp, const char *fmt, va_list ap);
+
 static void deallocate(xmlXPathContextPtr ctx)
 {
   NOKOGIRI_DEBUG_START(ctx);
@@ -19,29 +21,234 @@ static VALUE register_ns(VALUE self, VALUE prefix, VALUE uri)
   Data_Get_Struct(self, xmlXPathContext, ctx);
 
   xmlXPathRegisterNs( ctx,
-                      (const xmlChar *)StringValuePtr(prefix),
-                      (const xmlChar *)StringValuePtr(uri)
+                      (const xmlChar *)StringValueCStr(prefix),
+                      (const xmlChar *)StringValueCStr(uri)
   );
   return self;
 }
 
 /*
  * call-seq:
- *  evaluate(search_path)
+ *  register_variable(name, value)
+ *
+ * Register the variable +name+ with +value+.
+ */
+static VALUE register_variable(VALUE self, VALUE name, VALUE value)
+{
+   xmlXPathContextPtr ctx;
+   xmlXPathObjectPtr xmlValue;
+   Data_Get_Struct(self, xmlXPathContext, ctx);
+
+   xmlValue = xmlXPathNewCString(StringValueCStr(value));
+
+   xmlXPathRegisterVariable( ctx,
+      (const xmlChar *)StringValueCStr(name),
+      xmlValue
+   );
+
+   return self;
+}
+
+void Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr ctx, int nargs, VALUE handler, const char* function_name)
+{
+  int i;
+  VALUE result, doc;
+  VALUE *argv;
+  VALUE node_set = Qnil;
+  xmlNodeSetPtr xml_node_set = NULL;
+  xmlXPathObjectPtr obj;
+
+  assert(ctx->context->doc);
+  assert(DOC_RUBY_OBJECT_TEST(ctx->context->doc));
+
+  argv = (VALUE *)calloc((size_t)nargs, sizeof(VALUE));
+  for (i = 0 ; i < nargs ; ++i) {
+    rb_gc_register_address(&argv[i]);
+  }
+
+  doc = DOC_RUBY_OBJECT(ctx->context->doc);
+
+  if (nargs > 0) {
+    i = nargs - 1;
+    do {
+      obj = valuePop(ctx);
+      switch(obj->type) {
+        case XPATH_STRING:
+          argv[i] = NOKOGIRI_STR_NEW2(obj->stringval);
+          break;
+        case XPATH_BOOLEAN:
+          argv[i] = obj->boolval == 1 ? Qtrue : Qfalse;
+          break;
+        case XPATH_NUMBER:
+          argv[i] = rb_float_new(obj->floatval);
+          break;
+        case XPATH_NODESET:
+          argv[i] = Nokogiri_wrap_xml_node_set(obj->nodesetval, doc);
+          break;
+        default:
+          argv[i] = NOKOGIRI_STR_NEW2(xmlXPathCastToString(obj));
+      }
+      xmlXPathFreeNodeSetList(obj);
+    } while(i-- > 0);
+  }
+
+  result = rb_funcall2(handler, rb_intern((const char*)function_name), nargs, argv);
+
+  for (i = 0 ; i < nargs ; ++i) {
+    rb_gc_unregister_address(&argv[i]);
+  }
+  free(argv);
+
+  switch(TYPE(result)) {
+    case T_FLOAT:
+    case T_BIGNUM:
+    case T_FIXNUM:
+      xmlXPathReturnNumber(ctx, NUM2DBL(result));
+      break;
+    case T_STRING:
+      xmlXPathReturnString(
+          ctx,
+          xmlCharStrdup(StringValueCStr(result))
+      );
+      break;
+    case T_TRUE:
+      xmlXPathReturnTrue(ctx);
+      break;
+    case T_FALSE:
+      xmlXPathReturnFalse(ctx);
+      break;
+    case T_NIL:
+      break;
+    case T_ARRAY:
+      {
+        VALUE args[2];
+        args[0] = doc;
+        args[1] = result;
+        node_set = rb_class_new_instance(2, args, cNokogiriXmlNodeSet);
+        Data_Get_Struct(node_set, xmlNodeSet, xml_node_set);
+        xmlXPathReturnNodeSet(ctx, xmlXPathNodeSetMerge(NULL, xml_node_set));
+      }
+    break;
+    case T_DATA:
+      if(rb_obj_is_kind_of(result, cNokogiriXmlNodeSet)) {
+        Data_Get_Struct(result, xmlNodeSet, xml_node_set);
+        /* Copy the node set, otherwise it will get GC'd. */
+        xmlXPathReturnNodeSet(ctx, xmlXPathNodeSetMerge(NULL, xml_node_set));
+        break;
+      }
+    default:
+      rb_raise(rb_eRuntimeError, "Invalid return type");
+    }
+}
+
+static void ruby_funcall(xmlXPathParserContextPtr ctx, int nargs)
+{
+  VALUE handler = Qnil;
+  const char *function = NULL ;
+
+  assert(ctx);
+  assert(ctx->context);
+  assert(ctx->context->userData);
+  assert(ctx->context->function);
+
+  handler = (VALUE)(ctx->context->userData);
+  function = (const char*)(ctx->context->function);
+
+  Nokogiri_marshal_xpath_funcall_and_return_values(ctx, nargs, handler, function);
+}
+
+static xmlXPathFunction lookup( void *ctx,
+                                const xmlChar * name,
+                                const xmlChar* ns_uri )
+{
+  VALUE xpath_handler = (VALUE)ctx;
+  if(rb_respond_to(xpath_handler, rb_intern((const char *)name)))
+    return ruby_funcall;
+
+  return NULL;
+}
+
+NORETURN(static void xpath_generic_exception_handler(void * ctx, const char *msg, ...));
+static void xpath_generic_exception_handler(void * ctx, const char *msg, ...)
+{
+  char * message;
+
+  va_list args;
+  va_start(args, msg);
+  vasprintf(&message, msg, args);
+  va_end(args);
+
+  rb_raise(rb_eRuntimeError, "%s", message);
+}
+
+/*
+ * call-seq:
+ *  evaluate(search_path, handler = nil)
  *
  * Evaluate the +search_path+ returning an XML::XPath object.
  */
-static VALUE evaluate(VALUE self, VALUE search_path)
+static VALUE evaluate(int argc, VALUE *argv, VALUE self)
 {
+  VALUE search_path, xpath_handler;
+  VALUE thing = Qnil;
   xmlXPathContextPtr ctx;
+  xmlXPathObjectPtr xpath;
+  xmlChar *query;
+
   Data_Get_Struct(self, xmlXPathContext, ctx);
 
-  xmlChar* query = (xmlChar *)StringValuePtr(search_path);
-  xmlXPathObjectPtr xpath = xmlXPathEvalExpression(query, ctx);
+  if(rb_scan_args(argc, argv, "11", &search_path, &xpath_handler) == 1)
+    xpath_handler = Qnil;
+
+  query = (xmlChar *)StringValueCStr(search_path);
+
+  if(Qnil != xpath_handler) {
+    /* FIXME: not sure if this is the correct place to shove private data. */
+    ctx->userData = (void *)xpath_handler;
+    xmlXPathRegisterFuncLookup(ctx, lookup, (void *)xpath_handler);
+  }
+
+  xmlResetLastError();
+  xmlSetStructuredErrorFunc(NULL, Nokogiri_error_raise);
+
+  /* For some reason, xmlXPathEvalExpression will blow up with a generic error */
+  /* when there is a non existent function. */
+  xmlSetGenericErrorFunc(NULL, xpath_generic_exception_handler);
+
+  xpath = xmlXPathEvalExpression(query, ctx);
+  xmlSetStructuredErrorFunc(NULL, NULL);
+  xmlSetGenericErrorFunc(NULL, NULL);
+
   if(xpath == NULL) {
-    rb_raise(rb_eRuntimeError, "Couldn't evaluate expression '%s'", query);
+    xmlErrorPtr error = xmlGetLastError();
+    rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
   }
-  return Nokogiri_wrap_xml_xpath(xpath);
+
+  assert(ctx->doc);
+  assert(DOC_RUBY_OBJECT_TEST(ctx->doc));
+
+  switch(xpath->type) {
+    case XPATH_STRING:
+      thing = NOKOGIRI_STR_NEW2(xpath->stringval);
+      xmlFree(xpath->stringval);
+      break;
+    case XPATH_NODESET:
+      thing = Nokogiri_wrap_xml_node_set(xpath->nodesetval,
+                                         DOC_RUBY_OBJECT(ctx->doc));
+      break;
+    case XPATH_NUMBER:
+      thing = rb_float_new(xpath->floatval);
+      break;
+    case XPATH_BOOLEAN:
+      thing = xpath->boolval == 1 ? Qtrue : Qfalse;
+      break;
+    default:
+      thing = Nokogiri_wrap_xml_node_set(NULL, DOC_RUBY_OBJECT(ctx->doc));
+  }
+
+  xmlXPathFreeNodeSetList(xpath);
+
+  return thing;
 }
 
 /*
@@ -52,20 +259,29 @@ static VALUE evaluate(VALUE self, VALUE search_path)
  */
 static VALUE new(VALUE klass, VALUE nodeobj)
 {
+  xmlNodePtr node;
+  xmlXPathContextPtr ctx;
+  VALUE self;
+
   xmlXPathInit();
 
-  xmlNodePtr node ;
   Data_Get_Struct(nodeobj, xmlNode, node);
 
-  xmlXPathContextPtr ctx = xmlXPathNewContext(node->doc);
-  ctx->node = node ;
-  return Data_Wrap_Struct(klass, 0, deallocate, ctx);
+  ctx = xmlXPathNewContext(node->doc);
+  ctx->node = node;
+  self = Data_Wrap_Struct(klass, 0, deallocate, ctx);
+  /*rb_iv_set(self, "@xpath_handler", Qnil); */
+  return self;
 }
 
 VALUE cNokogiriXmlXpathContext;
 void init_xml_xpath_context(void)
 {
   VALUE module = rb_define_module("Nokogiri");
+
+  /*
+   * Nokogiri::XML
+   */
   VALUE xml = rb_define_module_under(module, "XML");
 
   /*
@@ -76,6 +292,7 @@ void init_xml_xpath_context(void)
   cNokogiriXmlXpathContext = klass;
 
   rb_define_singleton_method(klass, "new", new, 1);
-  rb_define_method(klass, "evaluate", evaluate, 1);
+  rb_define_method(klass, "evaluate", evaluate, -1);
+  rb_define_method(klass, "register_variable", register_variable, 2);
   rb_define_method(klass, "register_ns", register_ns, 2);
 }

data/ext/nokogiri/xml_xpath_context.h

@@ -1,9 +1,10 @@
 #ifndef NOKOGIRI_XML_XPATH_CONTEXT
 #define NOKOGIRI_XML_XPATH_CONTEXT
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xml_xpath_context();
+void Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr ctx, int nargs, VALUE handler, const char* function_name) ;
 
 extern VALUE cNokogiriXmlXpathContext;
 #endif