nokogiri 1.11.1 → 1.12.0.rc1

data/ext/nokogiri/{html_document.c → html4_document.c}

@@ -1,4 +1,6 @@
-#include <html_document.h>
+#include <nokogiri.h>
+
+VALUE cNokogiriHtml4Document ;
 
 static ID id_encoding_found;
 static ID id_to_s;
@@ -23,8 +25,7 @@ rb_html_document_s_new(int argc, VALUE *argv, VALUE klass)
           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);
+  rb_doc = noko_xml_document_wrap_with_init_args(klass, doc, argc, argv);
   return rb_doc ;
 }
 
@@ -33,7 +34,7 @@ rb_html_document_s_new(int argc, VALUE *argv, VALUE klass)
  *  read_io(io, url, encoding, options)
  *
  * Read the HTML document from +io+ with given +url+, +encoding+,
- * and +options+.  See Nokogiri::HTML.parse
+ * and +options+.  See Nokogiri::HTML4.parse
  */
 static VALUE
 rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
@@ -47,7 +48,7 @@ rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_enco
 
   xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
 
-  c_doc = htmlReadIO(io_read_callback, io_close_callback, (void *)rb_io, c_url, c_encoding, options);
+  c_doc = htmlReadIO(noko_io_read, noko_io_close, (void *)rb_io, c_url, c_encoding, options);
 
   xmlSetStructuredErrorFunc(NULL, NULL);
 
@@ -81,7 +82,7 @@ rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_enco
     return Qnil;
   }
 
-  rb_doc = Nokogiri_wrap_xml_document(klass, c_doc);
+  rb_doc = noko_xml_document_wrap(klass, c_doc);
   rb_iv_set(rb_doc, "@errors", rb_error_list);
   return rb_doc;
 }
@@ -91,7 +92,7 @@ rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_enco
  *  read_memory(string, url, encoding, options)
  *
  * Read the HTML document contained in +string+ with given +url+, +encoding+,
- * and +options+.  See Nokogiri::HTML.parse
+ * and +options+.  See Nokogiri::HTML4.parse
  */
 static VALUE
 rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
@@ -129,7 +130,7 @@ rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE r
     return Qnil;
   }
 
-  rb_doc = Nokogiri_wrap_xml_document(klass, c_doc);
+  rb_doc = noko_xml_document_wrap(klass, c_doc);
   rb_iv_set(rb_doc, "@errors", rb_error_list);
   return rb_doc;
 }
@@ -148,23 +149,17 @@ rb_html_document_type(VALUE self)
   return INT2NUM((long)doc->type);
 }
 
-VALUE cNokogiriHtmlDocument ;
-
 void
-init_html_document()
+noko_init_html_document()
 {
-  VALUE nokogiri = rb_define_module("Nokogiri");
-  VALUE nokogiri_xml = rb_define_module_under(nokogiri, "XML");
-  VALUE nokogiri_xml_node = rb_define_class_under(nokogiri_xml, "Node", rb_cObject);
-  VALUE nokogiri_xml_document = rb_define_class_under(nokogiri_xml, "Document", nokogiri_xml_node);
-  VALUE nokogiri_html = rb_define_module_under(nokogiri, "HTML");
-  cNokogiriHtmlDocument = rb_define_class_under(nokogiri_html, "Document", nokogiri_xml_document);
-
-  rb_define_singleton_method(cNokogiriHtmlDocument, "read_memory", rb_html_document_s_read_memory, 4);
-  rb_define_singleton_method(cNokogiriHtmlDocument, "read_io", rb_html_document_s_read_io, 4);
-  rb_define_singleton_method(cNokogiriHtmlDocument, "new", rb_html_document_s_new, -1);
-
-  rb_define_method(cNokogiriHtmlDocument, "type", rb_html_document_type, 0);
+  assert(cNokogiriXmlDocument);
+  cNokogiriHtml4Document = rb_define_class_under(mNokogiriHtml4, "Document", cNokogiriXmlDocument);
+
+  rb_define_singleton_method(cNokogiriHtml4Document, "read_memory", rb_html_document_s_read_memory, 4);
+  rb_define_singleton_method(cNokogiriHtml4Document, "read_io", rb_html_document_s_read_io, 4);
+  rb_define_singleton_method(cNokogiriHtml4Document, "new", rb_html_document_s_new, -1);
+
+  rb_define_method(cNokogiriHtml4Document, "type", rb_html_document_type, 0);
 
   id_encoding_found = rb_intern("encoding_found");
   id_to_s = rb_intern("to_s");

data/ext/nokogiri/html4_element_description.c

@@ -0,0 +1,294 @@
+#include <nokogiri.h>
+
+VALUE cNokogiriHtml4ElementDescription ;
+
+/*
+ * 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);
+}
+
+void
+noko_init_html_element_description()
+{
+  cNokogiriHtml4ElementDescription = rb_define_class_under(mNokogiriHtml4, "ElementDescription", rb_cObject);
+
+  rb_undef_alloc_func(cNokogiriHtml4ElementDescription);
+
+  rb_define_singleton_method(cNokogiriHtml4ElementDescription, "[]", get_description, 1);
+
+  rb_define_method(cNokogiriHtml4ElementDescription, "name", name, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "implied_start_tag?", implied_start_tag_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "implied_end_tag?", implied_end_tag_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "save_end_tag?", save_end_tag_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "empty?", empty_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "deprecated?", deprecated_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "inline?", inline_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "description", description, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "sub_elements", sub_elements, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "default_sub_element", default_sub_element, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "optional_attributes", optional_attributes, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "deprecated_attributes", deprecated_attributes, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "required_attributes", required_attributes, 0);
+}

data/ext/nokogiri/html4_entity_lookup.c

@@ -0,0 +1,37 @@
+#include <nokogiri.h>
+
+static VALUE cNokogiriHtml4EntityLookup;
+
+/*
+ * call-seq:
+ *  get(key)
+ *
+ * Get the HTML4::EntityDescription for +key+
+ */
+static VALUE
+get(VALUE _, VALUE rb_entity_name)
+{
+  VALUE cNokogiriHtml4EntityDescription;
+  const htmlEntityDesc *c_entity_desc;
+  VALUE rb_constructor_args[3];
+
+  c_entity_desc = htmlEntityLookup((const xmlChar *)StringValueCStr(rb_entity_name));
+  if (NULL == c_entity_desc) {
+    return Qnil;
+  }
+
+  rb_constructor_args[0] = INT2NUM((long)c_entity_desc->value);
+  rb_constructor_args[1] = NOKOGIRI_STR_NEW2(c_entity_desc->name);
+  rb_constructor_args[2] = NOKOGIRI_STR_NEW2(c_entity_desc->desc);
+
+  cNokogiriHtml4EntityDescription = rb_const_get_at(mNokogiriHtml4, rb_intern("EntityDescription"));
+  return rb_class_new_instance(3, rb_constructor_args, cNokogiriHtml4EntityDescription);
+}
+
+void
+noko_init_html_entity_lookup()
+{
+  cNokogiriHtml4EntityLookup = rb_define_class_under(mNokogiriHtml4, "EntityLookup", rb_cObject);
+
+  rb_define_method(cNokogiriHtml4EntityLookup, "get", get, 1);
+}

data/ext/nokogiri/html4_sax_parser_context.c

@@ -0,0 +1,119 @@
+#include <nokogiri.h>
+
+VALUE cNokogiriHtml4SaxParserContext ;
+
+static void
+deallocate(xmlParserCtxtPtr ctxt)
+{
+  NOKOGIRI_DEBUG_START(ctxt);
+
+  ctxt->sax = NULL;
+
+  htmlFreeParserCtxt(ctxt);
+
+  NOKOGIRI_DEBUG_END(ctxt);
+}
+
+static VALUE
+parse_memory(VALUE klass, VALUE data, VALUE encoding)
+{
+  htmlParserCtxtPtr ctxt;
+
+  if (NIL_P(data)) {
+    rb_raise(rb_eArgError, "data cannot be nil");
+  }
+  if (!(int)RSTRING_LEN(data)) {
+    rb_raise(rb_eRuntimeError, "data cannot be empty");
+  }
+
+  ctxt = htmlCreateMemoryParserCtxt(StringValuePtr(data),
+                                    (int)RSTRING_LEN(data));
+  if (ctxt->sax) {
+    xmlFree(ctxt->sax);
+    ctxt->sax = NULL;
+  }
+
+  if (RTEST(encoding)) {
+    xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValueCStr(encoding));
+    if (enc != NULL) {
+      xmlSwitchToEncoding(ctxt, enc);
+      if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
+        rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
+                 StringValueCStr(encoding));
+      }
+    }
+  }
+
+  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+}
+
+static VALUE
+parse_file(VALUE klass, VALUE filename, VALUE encoding)
+{
+  htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
+                             StringValueCStr(filename),
+                             StringValueCStr(encoding)
+                           );
+  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+}
+
+static VALUE
+parse_doc(VALUE ctxt_val)
+{
+  htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
+  htmlParseDocument(ctxt);
+  return Qnil;
+}
+
+static VALUE
+parse_doc_finalize(VALUE ctxt_val)
+{
+  htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
+
+  if (ctxt->myDoc) {
+    xmlFreeDoc(ctxt->myDoc);
+  }
+
+  NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
+  return Qnil;
+}
+
+static VALUE
+parse_with(VALUE self, VALUE sax_handler)
+{
+  htmlParserCtxtPtr ctxt;
+  htmlSAXHandlerPtr sax;
+
+  if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser)) {
+    rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
+  }
+
+  Data_Get_Struct(self, htmlParserCtxt, ctxt);
+  Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
+
+  /* Free the sax handler since we'll assign our own */
+  if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) {
+    xmlFree(ctxt->sax);
+  }
+
+  ctxt->sax = sax;
+  ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
+
+  xmlSetStructuredErrorFunc(NULL, NULL);
+
+  rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
+
+  return self;
+}
+
+void
+noko_init_html_sax_parser_context()
+{
+  assert(cNokogiriXmlSaxParserContext);
+  cNokogiriHtml4SaxParserContext = rb_define_class_under(mNokogiriHtml4Sax, "ParserContext", cNokogiriXmlSaxParserContext);
+
+  rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "memory", parse_memory, 2);
+  rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "file", parse_file, 2);
+
+  rb_define_method(cNokogiriHtml4SaxParserContext, "parse_with", parse_with, 1);
+}