nokogiri 1.11.0.rc1-x86-linux

data/ext/nokogiri/html_document.c

@@ -0,0 +1,170 @@
+#include <html_document.h>
+
+static ID id_encoding_found;
+
+/*
+ * call-seq:
+ *  new
+ *
+ * Create a new document
+ */
+static VALUE new(int argc, VALUE *argv, VALUE klass)
+{
+  VALUE uri, external_id, rest, rb_doc;
+  htmlDocPtr doc;
+
+  rb_scan_args(argc, argv, "0*", &rest);
+  uri         = rb_ary_entry(rest, (long)0);
+  external_id = rb_ary_entry(rest, (long)1);
+
+  doc = htmlNewDoc(
+      RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
+      RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
+  );
+  rb_doc = Nokogiri_wrap_xml_document(klass, doc);
+  rb_obj_call_init(rb_doc, argc, argv);
+  return rb_doc ;
+}
+
+/*
+ * call-seq:
+ *  read_io(io, url, encoding, options)
+ *
+ * Read the HTML document from +io+ with given +url+, +encoding+,
+ * and +options+.  See Nokogiri::HTML.parse
+ */
+static VALUE read_io( VALUE klass,
+                      VALUE io,
+                      VALUE url,
+                      VALUE encoding,
+                      VALUE options )
+{
+  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
+  VALUE error_list      = rb_ary_new();
+  VALUE document;
+  htmlDocPtr doc;
+
+  xmlResetLastError();
+  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
+
+  doc = htmlReadIO(
+      io_read_callback,
+      io_close_callback,
+      (void *)io,
+      c_url,
+      c_enc,
+      (int)NUM2INT(options)
+  );
+  xmlSetStructuredErrorFunc(NULL, NULL);
+
+  /*
+   * If EncodingFound has occurred in EncodingReader, make sure to do
+   * a cleanup and propagate the error.
+   */
+  if (rb_respond_to(io, id_encoding_found)) {
+    VALUE encoding_found = rb_funcall(io, id_encoding_found, 0);
+    if (!NIL_P(encoding_found)) {
+      xmlFreeDoc(doc);
+      rb_exc_raise(encoding_found);
+    }
+  }
+
+  if(doc == NULL) {
+    xmlErrorPtr error;
+
+    xmlFreeDoc(doc);
+
+    error = xmlGetLastError();
+    if(error)
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
+    else
+      rb_raise(rb_eRuntimeError, "Could not parse document");
+
+    return Qnil;
+  }
+
+  document = Nokogiri_wrap_xml_document(klass, doc);
+  rb_iv_set(document, "@errors", error_list);
+  return document;
+}
+
+/*
+ * call-seq:
+ *  read_memory(string, url, encoding, options)
+ *
+ * Read the HTML document contained in +string+ with given +url+, +encoding+,
+ * and +options+.  See Nokogiri::HTML.parse
+ */
+static VALUE read_memory( VALUE klass,
+                          VALUE string,
+                          VALUE url,
+                          VALUE encoding,
+                          VALUE options )
+{
+  const char * c_buffer = StringValuePtr(string);
+  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
+  int len               = (int)RSTRING_LEN(string);
+  VALUE error_list      = rb_ary_new();
+  VALUE document;
+  htmlDocPtr doc;
+
+  xmlResetLastError();
+  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
+
+  doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
+  xmlSetStructuredErrorFunc(NULL, NULL);
+
+  if(doc == NULL) {
+    xmlErrorPtr error;
+
+    xmlFreeDoc(doc);
+
+    error = xmlGetLastError();
+    if(error)
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
+    else
+      rb_raise(rb_eRuntimeError, "Could not parse document");
+
+    return Qnil;
+  }
+
+  document = Nokogiri_wrap_xml_document(klass, doc);
+  rb_iv_set(document, "@errors", error_list);
+  return document;
+}
+
+/*
+ * call-seq:
+ *  type
+ *
+ * The type for this document
+ */
+static VALUE type(VALUE self)
+{
+  htmlDocPtr doc;
+  Data_Get_Struct(self, xmlDoc, doc);
+  return INT2NUM((long)doc->type);
+}
+
+VALUE cNokogiriHtmlDocument ;
+void init_html_document()
+{
+  VALUE nokogiri  = rb_define_module("Nokogiri");
+  VALUE html      = rb_define_module_under(nokogiri, "HTML");
+  VALUE xml       = rb_define_module_under(nokogiri, "XML");
+  VALUE node      = rb_define_class_under(xml, "Node", rb_cObject);
+  VALUE xml_doc   = rb_define_class_under(xml, "Document", node);
+  VALUE klass     = rb_define_class_under(html, "Document", xml_doc);
+
+  cNokogiriHtmlDocument = klass;
+
+  rb_define_singleton_method(klass, "read_memory", read_memory, 4);
+  rb_define_singleton_method(klass, "read_io", read_io, 4);
+  rb_define_singleton_method(klass, "new", new, -1);
+
+  rb_define_method(klass, "type", type, 0);
+
+  id_encoding_found = rb_intern("encoding_found");
+}

data/ext/nokogiri/html_document.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_HTML_DOCUMENT
+#define NOKOGIRI_HTML_DOCUMENT
+
+#include <nokogiri.h>
+
+void init_html_document();
+
+extern VALUE cNokogiriHtmlDocument ;
+
+#endif

data/ext/nokogiri/html_element_description.c

@@ -0,0 +1,279 @@
+#include <html_element_description.h>
+
+/*
+ * call-seq:
+ *  required_attributes
+ *
+ * A list of required attributes for this element
+ */
+static VALUE required_attributes(VALUE self)
+{
+  const htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->attrs_req) return list;
+
+  for(i = 0; description->attrs_depr[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  deprecated_attributes
+ *
+ * A list of deprecated attributes for this element
+ */
+static VALUE deprecated_attributes(VALUE self)
+{
+  const htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->attrs_depr) return list;
+
+  for(i = 0; description->attrs_depr[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  optional_attributes
+ *
+ * A list of optional attributes for this element
+ */
+static VALUE optional_attributes(VALUE self)
+{
+  const htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->attrs_opt) return list;
+
+  for(i = 0; description->attrs_opt[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  default_sub_element
+ *
+ * The default sub element for this element
+ */
+static VALUE default_sub_element(VALUE self)
+{
+  const htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if (description->defaultsubelt)
+    return NOKOGIRI_STR_NEW2(description->defaultsubelt);
+
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *  sub_elements
+ *
+ * A list of allowed sub elements for this element.
+ */
+static VALUE sub_elements(VALUE self)
+{
+  const htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->subelts) return list;
+
+  for(i = 0; description->subelts[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  description
+ *
+ * The description for this element
+ */
+static VALUE description(VALUE self)
+{
+  const htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  return NOKOGIRI_STR_NEW2(description->desc);
+}
+
+/*
+ * call-seq:
+ *  inline?
+ *
+ * Is this element an inline element?
+ */
+static VALUE inline_eh(VALUE self)
+{
+  const htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->isinline) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  deprecated?
+ *
+ * Is this element deprecated?
+ */
+static VALUE deprecated_eh(VALUE self)
+{
+  const htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->depr) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  empty?
+ *
+ * Is this an empty element?
+ */
+static VALUE empty_eh(VALUE self)
+{
+  const htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->empty) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  save_end_tag?
+ *
+ * Should the end tag be saved?
+ */
+static VALUE save_end_tag_eh(VALUE self)
+{
+  const htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->saveEndTag) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  implied_end_tag?
+ *
+ * Can the end tag be implied for this tag?
+ */
+static VALUE implied_end_tag_eh(VALUE self)
+{
+  const htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->endTag) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  implied_start_tag?
+ *
+ * Can the start tag be implied for this tag?
+ */
+static VALUE implied_start_tag_eh(VALUE self)
+{
+  const htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->startTag) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  name
+ *
+ * Get the tag name for this ElemementDescription
+ */
+static VALUE name(VALUE self)
+{
+  const htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(NULL == description->name) return Qnil;
+  return NOKOGIRI_STR_NEW2(description->name);
+}
+
+/*
+ * call-seq:
+ *  [](tag_name)
+ *
+ * Get ElemementDescription for +tag_name+
+ */
+static VALUE get_description(VALUE klass, VALUE tag_name)
+{
+  const htmlElemDesc * description = htmlTagLookup(
+      (const xmlChar *)StringValueCStr(tag_name)
+  );
+
+  if(NULL == description) return Qnil;
+  return Data_Wrap_Struct(klass, 0, 0, (void *)(uintptr_t)description);
+}
+
+VALUE cNokogiriHtmlElementDescription ;
+void init_html_element_description()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE html     = rb_define_module_under(nokogiri, "HTML");
+  VALUE klass    = rb_define_class_under(html, "ElementDescription",rb_cObject);
+
+  cNokogiriHtmlElementDescription = klass;
+
+  rb_define_singleton_method(klass, "[]", get_description, 1);
+
+  rb_define_method(klass, "name", name, 0);
+  rb_define_method(klass, "implied_start_tag?", implied_start_tag_eh, 0);
+  rb_define_method(klass, "implied_end_tag?", implied_end_tag_eh, 0);
+  rb_define_method(klass, "save_end_tag?", save_end_tag_eh, 0);
+  rb_define_method(klass, "empty?", empty_eh, 0);
+  rb_define_method(klass, "deprecated?", deprecated_eh, 0);
+  rb_define_method(klass, "inline?", inline_eh, 0);
+  rb_define_method(klass, "description", description, 0);
+  rb_define_method(klass, "sub_elements", sub_elements, 0);
+  rb_define_method(klass, "default_sub_element", default_sub_element, 0);
+  rb_define_method(klass, "optional_attributes", optional_attributes, 0);
+  rb_define_method(klass, "deprecated_attributes", deprecated_attributes, 0);
+  rb_define_method(klass, "required_attributes", required_attributes, 0);
+}