nokogiri 1.2.3 → 1.3.0

data/bin/nokogiri

@@ -0,0 +1,47 @@
+#!/usr/bin/env ruby
+require 'optparse'
+require 'open-uri'
+require 'irb'
+require 'uri'
+require 'rubygems'
+require 'nokogiri'
+
+opts = OptionParser.new do |opts|
+  opts.banner = "Nokogiri: an HTML, XML, SAX, and Reader parser"
+  opts.define_head "Usage: nokogiri <uri|path> [options]"
+  opts.separator ""
+  opts.separator "Examples:"
+  opts.separator "  nokogiri http://www.ruby-lang.org/"
+  opts.separator "  nokogiri ./public/index.html"
+  opts.separator ""
+  opts.separator "Options:"
+
+  opts.on_tail("-?", "--help", "Show this message") do
+    puts opts
+    exit
+  end
+
+  opts.on_tail("-v", "--version", "Show version") do
+    require 'yaml'
+    puts Nokogiri::VERSION_INFO.to_yaml
+    exit
+  end
+end
+opts.parse!
+
+uri = ARGV.shift
+
+if uri.to_s.strip.empty?
+  puts opts
+  exit 1
+end
+
+if URI.parse(uri).scheme
+  @doc = Nokogiri(open(uri).read)
+else
+  @doc = Nokogiri(File.read(uri))
+end
+
+puts "Your document is stored in @doc..."
+IRB.start
+

data/ext/nokogiri/extconf.rb

@@ -12,7 +12,9 @@ end
 
 $CFLAGS << " #{ENV["CFLAGS"]}"
 if Config::CONFIG['target_os'] == 'mingw32'
-  $CFLAGS << " -DXP_WIN -DXP_WIN32"
+  $CFLAGS << " -DXP_WIN -DXP_WIN32 -DUSE_INCLUDED_VASPRINTF"
+elsif Config::CONFIG['target_os'] == 'solaris2'
+  $CFLAGS << " -DUSE_INCLUDED_VASPRINTF"
 else
   $CFLAGS << " -g -DXP_UNIX"
 end
@@ -33,16 +35,10 @@ LIB_DIRS = [
   '/usr/lib'
 ]
 
-if Config::CONFIG['target_os'] == 'mingw32'
-  header = File.join(ROOT, 'cross', 'iconv-1.9.2.win32', 'include')
-  unless find_header('iconv.h', header)
-    abort "need iconv"
-  end
-else
-  unless find_header('iconv.h', INCLUDEDIR, '/opt/local/include',
-                     '/usr/local/include', '/usr/include')
-    abort "iconv is missing.  try 'port install iconv' or 'yum install iconv'"
-  end
+iconv_dirs = dir_config('iconv', '/opt/local/include', '/opt/local/lib')
+unless [nil, nil] == iconv_dirs
+  HEADER_DIRS.unshift iconv_dirs.first
+  LIB_DIRS.unshift iconv_dirs[1]
 end
 
 xml2_dirs = dir_config('xml2', '/opt/local/include/libxml2', '/opt/local/lib')
@@ -57,8 +53,12 @@ unless [nil, nil] == xslt_dirs
   LIB_DIRS.unshift xslt_dirs[1]
 end
 
+unless find_header('iconv.h', *HEADER_DIRS)
+  abort "iconv is missing.  try 'port install iconv' or 'yum install iconv'"
+end
+
 unless find_header('libxml/parser.h', *HEADER_DIRS)
-  abort "libxml2 is missing.  try 'port install libxml2' or 'yum install libxml2'"
+  abort "libxml2 is missing.  try 'port install libxml2' or 'yum install libxml2-devel'"
 end
 
 unless find_header('libxslt/xslt.h', *HEADER_DIRS)
@@ -80,4 +80,10 @@ unless find_library('exslt', 'exsltFuncRegister', *LIB_DIRS)
   abort "libxslt is missing.  try 'port install libxslt' or 'yum install libxslt-devel'"
 end
 
-create_makefile('nokogiri/native')
+if ENV['CPUPROFILE']
+  unless find_library('profiler', 'ProfilerEnable', *LIB_DIRS)
+    abort "google performance tools are not installed"
+  end
+end
+
+create_makefile('nokogiri/nokogiri')

data/ext/nokogiri/html_document.c

@@ -8,15 +8,19 @@
  */
 static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
-  VALUE uri, external_id;
+  VALUE uri, external_id, rest, rb_doc;
 
-  rb_scan_args(argc, argv, "02", &uri, &external_id);
+  rb_scan_args(argc, argv, "0*", &rest);
+  uri         = rb_ary_entry(rest, 0);
+  external_id = rb_ary_entry(rest, 1);
 
   htmlDocPtr doc = htmlNewDoc(
       RTEST(uri) ? (const xmlChar *)StringValuePtr(uri) : NULL,
       RTEST(external_id) ? (const xmlChar *)StringValuePtr(external_id) : NULL
   );
-  return Nokogiri_wrap_xml_document(klass, doc);
+  rb_doc = Nokogiri_wrap_xml_document(klass, doc);
+  rb_funcall2(rb_doc, rb_intern("initialize"), argc, argv);
+  return rb_doc ;
 }
 
 /*
@@ -127,6 +131,36 @@ static VALUE type(VALUE self)
   return INT2NUM((int)doc->type);
 }
 
+/*
+ * call-seq:
+ *  meta_encoding=
+ *
+ * Set the meta tag encoding for this document.
+ */
+static VALUE set_meta_encoding(VALUE self, VALUE encoding)
+{
+  htmlDocPtr doc;
+  Data_Get_Struct(self, xmlDoc, doc);
+
+  htmlSetMetaEncoding(doc, (const xmlChar *)StringValuePtr(encoding));
+
+  return encoding;
+}
+
+/*
+ * call-seq:
+ *  meta_encoding
+ *
+ * Get the meta tag encoding for this document.
+ */
+static VALUE meta_encoding(VALUE self)
+{
+  htmlDocPtr doc;
+  Data_Get_Struct(self, xmlDoc, doc);
+
+  return NOKOGIRI_STR_NEW2(htmlGetMetaEncoding(doc), doc->encoding);
+}
+
 VALUE cNokogiriHtmlDocument ;
 void init_html_document()
 {
@@ -144,4 +178,6 @@ void init_html_document()
   rb_define_singleton_method(klass, "new", new, -1);
 
   rb_define_method(klass, "type", type, 0);
+  rb_define_method(klass, "meta_encoding", meta_encoding, 0);
+  rb_define_method(klass, "meta_encoding=", set_meta_encoding, 1);
 }

data/ext/nokogiri/html_document.h

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

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], "UTF-8"));
+    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], "UTF-8"));
+    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], "UTF-8"));
+    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, "UTF-8");
+}
+
+/*
+ * 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], "UTF-8"));
+    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, "UTF-8");
+}
+
+/*
+ * 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, "UTF-8");
+}
+
+/*
+ * 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);
+}