superfeedr-nokogiri 1.4.0.20091116183308

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 +1,272 @@
+#include <html_element_description.h>
+
+/*
+ * call-seq:
+ *  required_attributes
+ *
+ * A list of required attributes for this element
+ */
+static VALUE required_attributes(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  VALUE list = rb_ary_new();
+
+  if(NULL == description->attrs_req) return list;
+
+  int i = 0;
+  while(description->attrs_req[i]) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
+    i++;
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  deprecated_attributes
+ *
+ * A list of deprecated attributes for this element
+ */
+static VALUE deprecated_attributes(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  VALUE list = rb_ary_new();
+
+  if(NULL == description->attrs_depr) return list;
+
+  int i = 0;
+  while(description->attrs_depr[i]) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
+    i++;
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  optional_attributes
+ *
+ * A list of optional attributes for this element
+ */
+static VALUE optional_attributes(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  VALUE list = rb_ary_new();
+
+  if(NULL == description->attrs_opt) return list;
+
+  int i = 0;
+  while(description->attrs_opt[i]) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
+    i++;
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  default_sub_element
+ *
+ * The default sub element for this element
+ */
+static VALUE default_sub_element(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  return NOKOGIRI_STR_NEW2(description->defaultsubelt);
+}
+
+/*
+ * call-seq:
+ *  sub_elements
+ *
+ * A list of allowed sub elements for this element.
+ */
+static VALUE sub_elements(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  VALUE list = rb_ary_new();
+
+  if(NULL == description->subelts) return list;
+
+  int i = 0;
+  while(description->subelts[i]) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
+    i++;
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  description
+ *
+ * The description for this element
+ */
+static VALUE description(VALUE self)
+{
+  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)
+{
+  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)
+{
+  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)
+{
+  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)
+{
+  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)
+{
+  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)
+{
+  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)
+{
+  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 *)StringValuePtr(tag_name)
+  );
+
+  if(NULL == description) return Qnil;
+  return Data_Wrap_Struct(klass, 0, 0, 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);
+}

data/ext/nokogiri/html_element_description.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_HTML_ELEMENT_DESCRIPTION
+#define NOKOGIRI_HTML_ELEMENT_DESCRIPTION
+
+#include <nokogiri.h>
+
+void init_html_element_description();
+
+extern VALUE cNokogiriHtmlElementDescription ;
+
+#endif

data/ext/nokogiri/html_entity_lookup.c

@@ -0,0 +1,32 @@
+#include <html_entity_lookup.h>
+
+/*
+ * call-seq:
+ *  get(key)
+ *
+ * Get the HTML::EntityDescription for +key+
+ */
+static VALUE get(VALUE self, VALUE key)
+{
+  const htmlEntityDesc * desc =
+    htmlEntityLookup((const xmlChar *)StringValuePtr(key));
+
+  if(NULL == desc) return Qnil;
+  VALUE klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
+
+  VALUE args[3];
+  args[0] = INT2NUM((long)desc->value);
+  args[1] = NOKOGIRI_STR_NEW2(desc->name);
+  args[2] = NOKOGIRI_STR_NEW2(desc->desc);
+
+  return rb_class_new_instance(3, args, klass);
+}
+
+void init_html_entity_lookup()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE html = rb_define_module_under(nokogiri, "HTML");
+  VALUE klass = rb_define_class_under(html, "EntityLookup", rb_cObject);
+
+  rb_define_method(klass, "get", get, 1);
+}

data/ext/nokogiri/html_entity_lookup.h

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

data/ext/nokogiri/html_sax_parser_context.c

@@ -0,0 +1,92 @@
+#include <html_sax_parser_context.h>
+
+VALUE cNokogiriHtmlSaxParserContext ;
+
+static void deallocate(xmlParserCtxtPtr ctxt)
+{
+  NOKOGIRI_DEBUG_START(handler);
+
+  ctxt->sax = NULL;
+
+  htmlFreeParserCtxt(ctxt);
+
+  NOKOGIRI_DEBUG_END(handler);
+}
+
+static VALUE parse_memory(VALUE klass, VALUE data, VALUE encoding)
+{
+  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");
+
+  htmlParserCtxtPtr ctxt = htmlCreateMemoryParserCtxt(
+      StringValuePtr(data),
+      (int)RSTRING_LEN(data)
+  );
+
+  if(RTEST(encoding)) {
+    xmlCharEncoding enc = xmlParseCharEncoding(StringValuePtr(encoding));
+    if(enc != XML_CHAR_ENCODING_ERROR) {
+      xmlSwitchEncoding(ctxt, enc);
+      if(ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
+        rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
+            StringValuePtr(encoding));
+      }
+    }
+  }
+
+  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+}
+
+static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
+{
+  htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
+      StringValuePtr(filename),
+      StringValuePtr(encoding)
+  );
+  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+}
+
+static VALUE parse_with(VALUE self, VALUE sax_handler)
+{
+  if(!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
+    rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
+
+  htmlParserCtxtPtr ctxt;
+  Data_Get_Struct(self, htmlParserCtxt, ctxt);
+
+  htmlSAXHandlerPtr sax;
+  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);
+
+  htmlParseDocument(ctxt);
+
+  if(NULL != ctxt->myDoc) xmlFreeDoc(ctxt->myDoc);
+
+  NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
+  return self;
+}
+
+void init_html_sax_parser_context()
+{
+  VALUE nokogiri  = rb_define_module("Nokogiri");
+  VALUE xml       = rb_define_module_under(nokogiri, "XML");
+  VALUE html      = rb_define_module_under(nokogiri, "HTML");
+  VALUE sax       = rb_define_module_under(xml, "SAX");
+  VALUE hsax      = rb_define_module_under(html, "SAX");
+  VALUE pc        = rb_define_class_under(sax, "ParserContext", rb_cObject);
+  VALUE klass     = rb_define_class_under(hsax, "ParserContext", pc);
+
+  cNokogiriHtmlSaxParserContext = klass;
+
+  rb_define_singleton_method(klass, "memory", parse_memory, 2);
+  rb_define_singleton_method(klass, "file", parse_file, 2);
+
+  rb_define_method(klass, "parse_with", parse_with, 1);
+}

data/ext/nokogiri/html_sax_parser_context.h

@@ -0,0 +1,11 @@
+#ifndef NOKOGIRI_HTML_SAX_PARSER_CONTEXT
+#define NOKOGIRI_HTML_SAX_PARSER_CONTEXT
+
+#include <nokogiri.h>
+
+extern VALUE cNokogiriHtmlSaxParserContext;
+
+void init_html_sax_parser_context();
+
+#endif
+

data/ext/nokogiri/nokogiri.c

@@ -0,0 +1,89 @@
+#include <nokogiri.h>
+
+VALUE mNokogiri ;
+VALUE mNokogiriXml ;
+VALUE mNokogiriHtml ;
+VALUE mNokogiriXslt ;
+VALUE mNokogiriXmlSax ;
+VALUE mNokogiriHtmlSax ;
+
+#ifdef USE_INCLUDED_VASPRINTF
+/*
+ * I srsly hate windows.  it doesn't have vasprintf.
+ * Thank you Geoffroy Couprie for this implementation of vasprintf!
+ */
+int vasprintf (char **strp, const char *fmt, va_list ap)
+{
+  int len = vsnprintf (NULL, 0, fmt, ap) + 1;
+  char *res = (char *)malloc((unsigned int)len);
+  if (res == NULL)
+      return -1;
+  *strp = res;
+  return vsnprintf(res, (unsigned int)len, fmt, ap);
+}
+#endif
+
+int is_2_6_16(void)
+{
+  return (strcmp(xmlParserVersion, "20616") <= 0) ? 1 : 0 ;
+}
+
+void Init_nokogiri()
+{
+  xmlMemSetup(
+      (xmlFreeFunc)ruby_xfree,
+      (xmlMallocFunc)ruby_xmalloc,
+      (xmlReallocFunc)ruby_xrealloc,
+      strdup
+  );
+
+  mNokogiri         = rb_define_module("Nokogiri");
+  mNokogiriXml      = rb_define_module_under(mNokogiri, "XML");
+  mNokogiriHtml     = rb_define_module_under(mNokogiri, "HTML");
+  mNokogiriXslt     = rb_define_module_under(mNokogiri, "XSLT");
+  mNokogiriXmlSax   = rb_define_module_under(mNokogiriXml, "SAX");
+  mNokogiriHtmlSax  = rb_define_module_under(mNokogiriHtml, "SAX");
+
+  rb_const_set( mNokogiri,
+                rb_intern("LIBXML_VERSION"),
+                NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION)
+              );
+  rb_const_set( mNokogiri,
+                rb_intern("LIBXML_PARSER_VERSION"),
+                NOKOGIRI_STR_NEW2(xmlParserVersion)
+              );
+
+  xmlInitParser();
+
+  init_xml_document();
+  init_html_document();
+  init_xml_node();
+  init_xml_document_fragment();
+  init_xml_text();
+  init_xml_cdata();
+  init_xml_processing_instruction();
+  init_xml_attr();
+  init_xml_entity_reference();
+  init_xml_comment();
+  init_xml_node_set();
+  init_xml_xpath_context();
+  init_xml_xpath();
+  init_xml_sax_parser_context();
+  init_xml_sax_parser();
+  init_xml_sax_push_parser();
+  init_xml_reader();
+  init_xml_dtd();
+  init_xml_element_content();
+  init_xml_attribute_decl();
+  init_xml_element_decl();
+  init_xml_entity_decl();
+  init_xml_namespace();
+  init_html_sax_parser_context();
+  init_xslt_stylesheet();
+  init_xml_syntax_error();
+  init_html_entity_lookup();
+  init_html_element_description();
+  init_xml_schema();
+  init_xml_relax_ng();
+  init_nokogiri_io();
+}