nokogiri 1.4.1 → 1.4.2

data/lib/nokogiri/xml/xpath/syntax_error.rb

@@ -2,6 +2,9 @@ module Nokogiri
   module XML
     class XPath
       class SyntaxError < XML::SyntaxError
+        def to_s
+          [super.chomp, str1].compact.join(': ')
+        end
       end
     end
   end

data/tasks/cross_compile.rb

@@ -0,0 +1,158 @@
+require 'rake/extensioncompiler'
+HOST = Rake::ExtensionCompiler.mingw_host
+TARGET = 'i386-pc-mingw32'
+
+ZLIB    = 'zlib-1.2.5'
+ICONV   = 'libiconv-1.13.1'
+LIBXML  = 'libxml2-2.7.7'
+LIBXSLT = 'libxslt-1.1.26'
+
+### Build zlib ###
+file "tmp/cross/download/#{ZLIB}" do |t|
+  FileUtils.mkdir_p('tmp/cross/download')
+
+  file = ZLIB
+  url  = "http://zlib.net/#{file}.tar.gz"
+
+  Dir.chdir('tmp/cross/download') do
+    sh "wget #{url} || curl -O #{url}"
+    sh "tar zxvf #{file}.tar.gz"
+  end
+
+  Dir.chdir t.name do
+    mk = File.read('win32/Makefile.gcc')
+    File.open('win32/Makefile.gcc', 'wb') do |f|
+      f.puts "BINARY_PATH = #{CROSS_DIR}/bin"
+      f.puts "LIBRARY_PATH = #{CROSS_DIR}/lib"
+      f.puts "INCLUDE_PATH = #{CROSS_DIR}/include"
+
+      # FIXME: need to make the cross compiler dynamic
+      f.puts mk.sub(/^PREFIX\s*=\s*$/, "PREFIX = #{HOST}-") #.
+        #sub(/^SHARED_MODE=0$/, 'SHARED_MODE=1').
+        #sub(/^IMPLIB\s*=.*$/, 'IMPLIB=libz.dll.a')
+    end
+  end
+end
+
+file 'tmp/cross/lib/libz.a' => "tmp/cross/download/#{ZLIB}" do |t|
+  Dir.chdir t.prerequisites.first do
+    sh 'make -f win32/Makefile.gcc'
+    sh 'make -f win32/Makefile.gcc install'
+  end
+end
+### End build zlib ###
+
+### Build iconv ###
+file "tmp/cross/download/#{ICONV}" do |t|
+  FileUtils.mkdir_p('tmp/cross/download')
+
+  file = ICONV
+  url  = "http://ftp.gnu.org/pub/gnu/libiconv/#{file}.tar.gz"
+
+  Dir.chdir('tmp/cross/download') do
+    sh "wget #{url} || curl -O #{url}"
+    sh "tar zxvf #{file}.tar.gz"
+  end
+
+  Dir.chdir t.name do
+    # FIXME: need to make the host dynamic
+    sh "./configure --disable-shared --enable-static --host=#{HOST} --target=#{TARGET} --prefix=#{CROSS_DIR} CPPFLAGS='-mno-cygwin -Wall' CFLAGS='-mno-cygwin -O2 -g' CXXFLAGS='-mno-cygwin -O2 -g' LDFLAGS=-mno-cygwin"
+  end
+end
+
+file 'tmp/cross/bin/iconv.exe' => "tmp/cross/download/#{ICONV}" do |t|
+  Dir.chdir t.prerequisites.first do
+    sh 'make'
+    sh 'make install'
+  end
+end
+### End build iconv ###
+
+### Build libxml2 ###
+file "tmp/cross/download/#{LIBXML}" do |t|
+  FileUtils.mkdir_p('tmp/cross/download')
+
+  file = LIBXML
+  url  = "ftp://ftp.xmlsoft.org/libxml2/#{file}.tar.gz"
+
+  Dir.chdir('tmp/cross/download') do
+    sh "wget #{url} || curl -O #{url}"
+    sh "tar zxvf #{file}.tar.gz"
+  end
+
+  Dir.chdir t.name do
+    # FIXME: need to make the host dynamic
+    sh "CFLAGS='-DIN_LIBXML' ./configure --host=#{HOST} --target=#{TARGET} --enable-static --disable-shared --prefix=#{CROSS_DIR} --with-zlib=#{CROSS_DIR} --with-iconv=#{CROSS_DIR} --without-python --without-readline"
+  end
+end
+
+file 'tmp/cross/bin/xml2-config' => "tmp/cross/download/#{LIBXML}" do |t|
+  Dir.chdir t.prerequisites.first do
+    sh 'make LDFLAGS="-avoid-version"'
+    sh 'make install'
+  end
+end
+### End build libxml2 ###
+
+### Build libxslt ###
+file "tmp/cross/download/#{LIBXSLT}" do |t|
+  FileUtils.mkdir_p('tmp/cross/download')
+
+  file = LIBXSLT
+  url  = "ftp://ftp.xmlsoft.org/libxml2/#{file}.tar.gz"
+
+  Dir.chdir('tmp/cross/download') do
+    sh "wget #{url} || curl -O #{url}"
+    sh "tar zxvf #{file}.tar.gz"
+  end
+
+  Dir.chdir t.name do
+    # FIXME: need to make the host dynamic
+    sh "CFLAGS='-DIN_LIBXML' ./configure --host=#{HOST} --target=#{TARGET} --enable-static --disable-shared --prefix=#{CROSS_DIR} --with-libxml-prefix=#{CROSS_DIR} --without-python"
+  end
+end
+
+file 'tmp/cross/bin/xslt-config' => "tmp/cross/download/#{LIBXSLT}" do |t|
+  Dir.chdir t.prerequisites.first do
+    sh 'make LDFLAGS="-avoid-version"'
+    sh 'make install'
+  end
+end
+### End build libxslt ###
+
+file 'lib/nokogiri/nokogiri.rb' => 'cross:libxslt' do
+  File.open("lib/#{HOE.name}/#{HOE.name}.rb", 'wb') do |f|
+    f.write <<-eoruby
+require "#{HOE.name}/\#{RUBY_VERSION.sub(/\\.\\d+$/, '')}/#{HOE.name}"
+    eoruby
+  end
+end
+
+namespace :cross do
+  task :iconv   => 'tmp/cross/bin/iconv.exe'
+  task :zlib    => 'tmp/cross/lib/libz.a'
+  task :libxml2 => ['cross:zlib', 'cross:iconv', 'tmp/cross/bin/xml2-config']
+  task :libxslt => ['cross:libxml2', 'tmp/cross/bin/xslt-config']
+
+  task :copy_dlls do
+    Dir['tmp/cross/bin/*.dll'].each do |file|
+      cp file, "ext/nokogiri"
+    end
+  end
+
+  task :file_list => 'cross:copy_dlls' do
+    HOE.spec.extensions = []
+    HOE.spec.files += Dir["lib/#{HOE.name}/#{HOE.name}.rb"]
+    HOE.spec.files += Dir["lib/#{HOE.name}/1.{8,9}/#{HOE.name}.so"]
+    HOE.spec.files += Dir["ext/nokogiri/*.dll"]
+  end
+end
+
+CLOBBER.include("lib/nokogiri/nokogiri.{so,dylib,rb,bundle}")
+CLOBBER.include("lib/nokogiri/1.{8,9}")
+CLOBBER.include("ext/nokogiri/*.dll")
+
+if Rake::Task.task_defined?(:cross)
+  Rake::Task[:cross].prerequisites << "lib/nokogiri/nokogiri.rb"
+  Rake::Task[:cross].prerequisites << "cross:file_list"
+end

data/tasks/test.rb

@@ -1,9 +1,3 @@
-def run_with_env(cmd)
-  cmd = "LD_LIBRARY_PATH='#{ENV['LD_LIBRARY_PATH']}' #{cmd}"
-  puts "=> #{cmd}"
-  system cmd
-end
-
 namespace :test do
   desc "run test suite with aggressive GC"
   task :gc => :build do

data/test/css/test_xpath_visitor.rb

@@ -8,6 +8,15 @@ module Nokogiri
         @parser = Nokogiri::CSS::Parser.new
       end
 
+      def test_not_simple_selector
+        assert_xpath('//ol/*[not(self::li)]', @parser.parse('ol > *:not(li)'))
+      end
+
+      def test_not_last_child
+        assert_xpath('//ol/*[not(position() = last())]',
+          @parser.parse('ol > *:not(:last-child)'))
+      end
+
       def test_function_calls_allow_at_params
         assert_xpath("//a[foo(., @href)]", @parser.parse('a:foo(@href)'))
         assert_xpath("//a[foo(., @a, b)]", @parser.parse('a:foo(@a, b)'))

data/test/helper.rb

@@ -1,6 +1,6 @@
 #Process.setrlimit(Process::RLIMIT_CORE, Process::RLIM_INFINITY) unless RUBY_PLATFORM =~ /(java|mswin|mingw)/i
 $VERBOSE = true
-require 'test/unit'
+require 'minitest/autorun'
 require 'fileutils'
 require 'tempfile'
 require 'pp'
@@ -10,7 +10,7 @@ require 'nokogiri'
 warn "#{__FILE__}:#{__LINE__}: libxml version info: #{Nokogiri::VERSION_INFO.inspect}"
 
 module Nokogiri
-  class TestCase < Test::Unit::TestCase
+  class TestCase < MiniTest::Spec
     ASSETS_DIR      = File.expand_path File.join(File.dirname(__FILE__), 'files')
     XML_FILE        = File.join(ASSETS_DIR, 'staff.xml')
     XSLT_FILE       = File.join(ASSETS_DIR, 'staff.xslt')
@@ -26,18 +26,15 @@ module Nokogiri
     ADDRESS_XML_FILE = File.join(ASSETS_DIR, 'address_book.xml')
     SNUGGLES_FILE   = File.join(ASSETS_DIR, 'snuggles.xml')
 
-    unless RUBY_VERSION >= '1.9'
-      undef :default_test
-    end
-
-    def setup
-      warn "#{name}" if ENV['TESTOPTS'] == '-v'
-    end
-
     def teardown
       if ENV['NOKOGIRI_GC']
         STDOUT.putc '!'
-        GC.start
+        if RUBY_PLATFORM =~ /java/
+          require 'java'
+          java.lang.System.gc
+        else
+          GC.start
+        end
       end
     end
 
@@ -52,6 +49,47 @@ module Nokogiri
         assert_equal(0, len % amount, message)
       end
     end
+
+    def util_decorate(document, decorator_module)
+      document.decorators(XML::Node) << decorator_module
+      document.decorators(XML::NodeSet) << decorator_module
+      document.decorate!
+    end
+
+    #
+    #  Test::Unit backwards compatibility section
+    #
+    alias :assert_no_match      :refute_match
+    alias :assert_not_nil       :refute_nil
+    alias :assert_raise         :assert_raises
+    alias :assert_not_equal     :refute_equal
+
+    def assert_nothing_raised(*args)
+      self._assertions += 1
+      if Module === args.last
+        msg = nil
+      else
+        msg = args.pop
+      end
+      begin
+        line = __LINE__; yield
+      rescue Exception => e
+        bt = e.backtrace
+        as = e.instance_of?(MiniTest::Assertion)
+        if as
+          ans = /\A#{Regexp.quote(__FILE__)}:#{line}:in /o
+          bt.reject! {|ln| ans =~ ln}
+        end
+        if ((args.empty? && !as) ||
+            args.any? {|a| a.instance_of?(Module) ? e.is_a?(a) : e.class == a })
+          msg = message(msg) { "Exception raised:\n<#{mu_pp(e)}>" }
+          raise MiniTest::Assertion, msg.call, bt
+        else
+          raise
+        end
+      end
+      nil
+    end
   end
 
   module SAX

data/test/html/sax/test_parser.rb

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 require "helper"
 
 module Nokogiri
@@ -34,7 +35,7 @@ module Nokogiri
 
         def test_parse_file_non_existant
           assert_raise Errno::ENOENT do
-            @parser.parse_file('foo')
+            @parser.parse_file('there_is_no_reasonable_way_this_file_exists')
           end
         end
 
@@ -50,6 +51,15 @@ module Nokogiri
           end
         end
 
+        def test_parse_force_encoding
+          @parser.parse_memory(<<-HTML, 'UTF-8')
+          <meta http-equiv="Content-Type" content="text/html; charset=windows-1251">
+          Информация
+          HTML
+          assert_equal("Информация",
+                       @parser.document.data.join.strip)
+        end
+
         def test_parse_document
           @parser.parse_memory(<<-eoxml)
             <p>Paragraph 1</p>

data/test/html/test_document.rb

@@ -8,6 +8,14 @@ module Nokogiri
         @html = Nokogiri::HTML.parse(File.read(HTML_FILE))
       end
 
+      def test_exceptions_remove_newlines
+        errors = @html.errors
+        assert errors.length > 0, 'has errors'
+        errors.each do |error|
+          assert_equal(error.to_s.chomp, error.to_s)
+        end
+      end
+
       def test_fragment
         fragment = @html.fragment
         assert_equal 0, fragment.children.length

data/test/html/test_document_fragment.rb

@@ -9,6 +9,18 @@ module Nokogiri
         @html = Nokogiri::HTML.parse(File.read(HTML_FILE), HTML_FILE)
       end
 
+      def test_no_contextual_parsing_on_unlinked_nodes
+        node = @html.css('body').first
+        node.unlink
+        assert_raises(RuntimeError) do
+          node.parse('<br />')
+        end
+      end
+
+      def test_parse_in_context
+        assert_equal('<br>', @html.root.parse('<br />').to_s)
+      end
+
       def test_ancestors_search
         html = %q{
           <div>
@@ -91,9 +103,9 @@ module Nokogiri
       end
 
       def test_html_fragment_case_insensitivity
-        doc = "<crazyDiv>b</crazyDiv>"
+        doc = "<Div>b</Div>"
         fragment = Nokogiri::HTML::Document.new.fragment(doc)
-        assert_equal "<crazydiv>b</crazydiv>", fragment.to_s
+        assert_equal "<div>b</div>", fragment.to_s
       end
 
       def test_html_fragment_with_leading_whitespace

data/test/html/test_element_description.rb

@@ -56,7 +56,11 @@ module Nokogiri
 
       def test_subelements
         sub_elements = ElementDescription['body'].sub_elements
-        assert_equal 61, sub_elements.length
+        if Nokogiri::LIBXML_VERSION == '2.7.7'
+          assert_equal 65, sub_elements.length
+        else
+          assert_equal 61, sub_elements.length
+        end
       end
 
       def test_default_sub_element

data/test/html/test_node.rb

@@ -17,6 +17,10 @@ module Nokogiri
         eohtml
       end
 
+      def test_to_a
+        assert_equal [['class', 'bar'], ['href', 'foo']],@html.at('a').to_a.sort
+      end
+
       def test_attr
         node = @html.at('div.baz')
         assert_equal node['class'], node.attr('class')
@@ -64,18 +68,6 @@ module Nokogiri
         assert_equal 'a', desc.name
       end
 
-      def test_add_next_sibling_with_empty_nodeset
-        assert_raises(ArgumentError) {
-          @html.at('a').add_next_sibling(@html.at('head').children)
-        }
-      end
-
-      def test_add_next_sibling_with_non_empty_nodeset
-        assert_raises(ArgumentError) {
-          @html.at('head').add_next_sibling(@html.at('div').children)
-        }
-      end
-
       def test_ancestors_with_selector
         assert node = @html.at('a.bar').child
         assert list = node.ancestors('.baz')
@@ -135,35 +127,9 @@ module Nokogiri
         assert_equal 'foo&bar&baz', node['href']
       end
 
-
-      def test_before_will_prepend_text_nodes
-        assert node = @html.at('//body').children.first
-        node.before "some text"
-        assert_equal 'some text', @html.at('//body').children[0].content.strip
-      end
-
-      def test_before
-        @html.at('//div').before('<a href="awesome">town</a>')
-        assert_equal 'awesome', @html.at('//div').previous['href']
-      end
-
       def test_fragment_handler_does_not_regurge_on_invalid_attributes
         iframe = %Q{<iframe style="width: 0%; height: 0px" src="http://someurl" allowtransparency></iframe>}
-        assert_nothing_raised { @html.at('div').before(iframe) }
-        assert_nothing_raised { @html.at('div').after(iframe) }
-        assert_nothing_raised { @html.at('div').inner_html=(iframe) }
-      end
-
-      def test_inner_html=
-        assert div = @html.at('//div')
-        div.inner_html = '1<span>2</span>3'
-        assert_equal '1',    div.children[0].to_s
-        assert_equal 'span', div.children[1].name
-        assert_equal '2',    div.children[1].inner_text
-        assert_equal '3',    div.children[2].to_s
-
-        div.inner_html = 'testing'
-        assert_equal 'testing', div.content
+        assert_nothing_raised { @html.at('div').fragment(iframe) }
       end
 
       def test_fragment
@@ -190,33 +156,6 @@ module Nokogiri
         assert_equal "<div>foo</div>", fragment.to_s
       end
 
-      def test_after_will_append_text_nodes
-        assert node = @html.at('//body/div')
-        node.after "some text"
-        assert_equal 'some text', node.next.text.strip
-      end
-
-      def test_after
-        @html.at('//div').after('<a href="awesome">town</a>')
-        assert_equal 'awesome', @html.at('//div').next['href']
-      end
-
-      def test_replace
-        doc = Nokogiri::HTML(<<-eohtml)
-          <html>
-            <head></head>
-            <body>
-              <center><img src='logo.gif' /></center>
-            </body>
-          </html>
-        eohtml
-        center = doc.at("//center")
-        img = center.search("//img")
-        assert_raises ArgumentError do
-          center.replace img
-        end
-      end
-
       def test_to_html_does_not_contain_entities
         return unless defined?(NKF) # NKF is not implemented on Rubinius as of 2009-11-23
         html = NKF.nkf("-e --msdos", <<-EOH)