rubyjedi-nokogiri_java 1.4.0.20100513161003-java

data/test/files/valid_bar.xml

@@ -0,0 +1,2 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<bar />

data/test/helper.rb

@@ -0,0 +1,137 @@
+#Process.setrlimit(Process::RLIMIT_CORE, Process::RLIM_INFINITY) unless RUBY_PLATFORM =~ /(java|mswin|mingw)/i
+$VERBOSE = true
+require 'test/unit'
+require 'fileutils'
+require 'tempfile'
+require 'pp'
+
+require 'nokogiri'
+
+if Nokogiri.uses_libxml?
+  warn "#{__FILE__}:#{__LINE__}: libxml version info: #{Nokogiri::VERSION_INFO.inspect}"
+else
+  warn "#{__FILE__}:#{__LINE__}: pure java version"
+end
+
+
+module Nokogiri
+  class TestCase < Test::Unit::TestCase
+
+    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')
+    EXSLT_FILE      = File.join(ASSETS_DIR, 'exslt.xslt')
+    EXML_FILE       = File.join(ASSETS_DIR, 'exslt.xml')
+    HTML_FILE       = File.join(ASSETS_DIR, 'tlm.html')
+    NICH_FILE       = File.join(ASSETS_DIR, '2ch.html')
+    SHIFT_JIS_XML   = File.join(ASSETS_DIR, 'shift_jis.xml')
+    SHIFT_JIS_HTML  = File.join(ASSETS_DIR, 'shift_jis.html')
+    PO_XML_FILE     = File.join(ASSETS_DIR, 'po.xml')
+    PO_SCHEMA_FILE  = File.join(ASSETS_DIR, 'po.xsd')
+    ADDRESS_SCHEMA_FILE = File.join(ASSETS_DIR, 'address_book.rlx')
+    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
+      end
+    end
+
+    def assert_indent amount, doc, message = nil
+      nodes = []
+      doc.traverse do |node|
+        nodes << node if node.text? && node.blank?
+      end
+      assert nodes.length > 0
+      nodes.each do |node|
+        len = node.content.gsub(/[\r\n]/, '').length
+        assert_equal(0, len % amount, message)
+      end
+    end
+  end
+
+  module SAX
+    class TestCase < Nokogiri::TestCase
+      class Doc < XML::SAX::Document
+        attr_reader :start_elements, :start_document_called
+        attr_reader :end_elements, :end_document_called
+        attr_reader :data, :comments, :cdata_blocks, :start_elements_namespace
+        attr_reader :errors, :warnings, :end_elements_namespace
+        attr_reader :xmldecls
+
+        def xmldecl version, encoding, standalone
+          @xmldecls = [version, encoding, standalone].compact
+          super
+        end
+
+        def start_document
+          @start_document_called = true
+          super
+        end
+
+        def end_document
+          @end_document_called = true
+          super
+        end
+
+        def error error
+          (@errors ||= []) << error
+          super
+        end
+
+        def warning warning
+          (@warning ||= []) << warning
+          super
+        end
+
+        def start_element *args
+          (@start_elements ||= []) << args
+          super
+        end
+
+        def start_element_namespace *args
+          (@start_elements_namespace ||= []) << args
+          super
+        end
+
+        def end_element *args
+          (@end_elements ||= []) << args
+          super
+        end
+
+        def end_element_namespace *args
+          (@end_elements_namespace ||= []) << args
+          super
+        end
+
+        def characters string
+          @data ||= []
+          @data += [string]
+          super
+        end
+
+        def comment string
+          @comments ||= []
+          @comments += [string]
+          super
+        end
+
+        def cdata_block string
+          @cdata_blocks ||= []
+          @cdata_blocks += [string]
+          super
+        end
+      end
+    end
+  end
+end

data/test/html/sax/test_parser.rb

@@ -0,0 +1,83 @@
+require "helper"
+
+module Nokogiri
+  module HTML
+    module SAX
+      class TestParser < Nokogiri::SAX::TestCase
+        def setup
+          super
+          @parser = HTML::SAX::Parser.new(Doc.new)
+        end
+
+        def test_parse_empty_document
+          # This caused a segfault in libxml 2.6.x
+          assert_nothing_raised { @parser.parse '' }
+        end
+
+        def test_parse_empty_file
+          # Make sure empty files don't break stuff
+          empty_file_name =  File.join(Dir.tmpdir, 'bogus.xml')
+          FileUtils.touch empty_file_name
+          assert_nothing_raised { @parser.parse_file empty_file_name }
+        end
+
+        def test_parse_file
+          @parser.parse_file(HTML_FILE)
+          
+          # Take a look at the comment in test_parse_document to know
+          # a possible reason to this difference.
+          if Nokogiri.uses_libxml?
+            assert_equal 1110, @parser.document.end_elements.length
+          else
+            assert_equal 1119, @parser.document.end_elements.length
+          end
+        end
+
+        def test_parse_file_nil_argument
+          assert_raises(ArgumentError) {
+            @parser.parse_file(nil)
+          }
+        end
+
+        def test_parse_file_non_existant
+          assert_raise Errno::ENOENT do
+            @parser.parse_file('foo')
+          end
+        end
+
+        def test_parse_file_with_dir
+          assert_raise Errno::EISDIR do
+            @parser.parse_file(File.dirname(__FILE__))
+          end
+        end
+
+        def test_parse_memory_nil
+          assert_raise ArgumentError do
+            @parser.parse_memory(nil)
+          end
+        end
+
+        def test_parse_document
+          @parser.parse_memory(<<-eoxml)
+            <p>Paragraph 1</p>
+            <p>Paragraph 2</p>
+          eoxml
+
+          # JRuby version is different because of the internal implementation
+          # JRuby version uses NekoHTML which inserts empty "head" elements.
+          #
+          # Currently following features are set:
+          # "http://cyberneko.org/html/properties/names/elems" => "lower"
+          # "http://cyberneko.org/html/properties/names/attrs" => "lower"
+          if Nokogiri.uses_libxml?
+            assert_equal([["html", []], ["body", []], ["p", []], ["p", []]],
+                         @parser.document.start_elements)
+          else
+            assert_equal([["html", []], ["head", []], ["body", []], ["p", []], ["p", []]],
+                         @parser.document.start_elements)
+          end
+        end
+      end
+    end
+  end
+end

data/test/html/sax/test_parser_context.rb

@@ -0,0 +1,48 @@
+# -*- coding: utf-8 -*-
+
+require "helper"
+
+module Nokogiri
+  module HTML
+    module SAX
+      class TestParserContext < Nokogiri::SAX::TestCase
+        def test_from_io
+          assert_nothing_raised do
+            ctx = ParserContext.new StringIO.new('fo'), 'UTF-8'
+          end
+        end
+
+        def test_from_string
+          assert_nothing_raised do
+            ctx = ParserContext.new 'blah blah'
+          end
+        end
+
+        def test_parse_with
+          ctx = ParserContext.new 'blah'
+          assert_raises ArgumentError do
+            ctx.parse_with nil
+          end
+        end
+
+        def test_parse_with_sax_parser
+          assert_nothing_raised do
+            xml = "<root />"
+            ctx = ParserContext.new xml
+            parser = Parser.new Doc.new
+            ctx.parse_with parser
+          end
+        end
+
+        def test_from_file
+          assert_nothing_raised do
+            ctx = ParserContext.file HTML_FILE, 'UTF-8'
+            parser = Parser.new Doc.new
+            ctx.parse_with parser
+          end
+        end
+      end
+    end
+  end
+end
+

data/test/html/test_builder.rb

@@ -0,0 +1,164 @@
+require "helper"
+
+module Nokogiri
+  module HTML
+    class TestBuilder < Nokogiri::TestCase
+      def test_top_level_function_builds
+        foo = nil
+        Nokogiri() { |xml| foo = xml }
+        assert_instance_of Nokogiri::HTML::Builder, foo
+      end
+
+      def test_builder_with_explicit_tags
+        html_doc = Nokogiri::HTML::Builder.new {
+          div.slide(:class => 'another_class') {
+            node = Nokogiri::XML::Node.new("id", doc)
+            node.content = "hello"
+            insert(node)
+          }
+        }.doc
+        assert_equal 1, html_doc.css('div.slide > id').length
+        assert_equal 'hello', html_doc.at('div.slide > id').content
+      end
+
+      def test_hash_as_attributes_for_attribute_method
+        html = Nokogiri::HTML::Builder.new { ||
+          div.slide(:class => 'another_class') {
+            span 'Slide 1'
+          }
+        }.to_html
+        assert_match 'class="slide another_class"', html
+      end
+
+      def test_hash_as_attributes
+        builder = Nokogiri::HTML::Builder.new do
+          div(:id => 'awesome') {
+            h1 "america"
+          }
+        end
+        assert_equal('<div id="awesome"><h1>america</h1></div>',
+                     builder.doc.root.to_html.gsub(/\n/, '').gsub(/>\s*</, '><'))
+      end
+
+      def test_href_with_attributes
+        uri = 'http://tenderlovemaking.com/'
+        built = Nokogiri::XML::Builder.new {
+          div {
+            a('King Khan & The Shrines', :href => uri)
+          }
+        }
+        assert_equal 'http://tenderlovemaking.com/',
+          built.doc.at('a')[:href]
+      end
+
+      def test_tag_nesting
+        builder = Nokogiri::HTML::Builder.new do
+          body {
+            span.left ''
+            span.middle {
+              div.icon ''
+            }
+            span.right ''
+          }
+        end
+        assert node = builder.doc.css('span.right').first
+        assert_equal 'middle', node.previous_sibling['class']
+      end
+
+      def test_has_ampersand
+        builder = Nokogiri::HTML::Builder.new do
+          div.rad.thing! {
+            text "<awe&some>"
+            b "hello & world"
+          }
+        end
+        assert_equal(
+          '<div class="rad" id="thing">&lt;awe&amp;some&gt;<b>hello &amp; world</b></div>',
+                     builder.doc.root.to_html.gsub(/\n/, ''))
+      end
+
+      def test_multi_tags
+        builder = Nokogiri::HTML::Builder.new do
+          div.rad.thing! {
+            text "<awesome>"
+            b "hello"
+          }
+        end
+        assert_equal(
+          '<div class="rad" id="thing">&lt;awesome&gt;<b>hello</b></div>',
+                     builder.doc.root.to_html.gsub(/\n/, ''))
+      end
+
+      def test_attributes_plus_block
+        builder = Nokogiri::HTML::Builder.new do
+          div.rad.thing! {
+            text "<awesome>"
+          }
+        end
+        assert_equal('<div class="rad" id="thing">&lt;awesome&gt;</div>',
+                     builder.doc.root.to_html.chomp)
+      end
+
+      def test_builder_adds_attributes
+        builder = Nokogiri::HTML::Builder.new do
+          div.rad.thing! "tender div"
+        end
+        assert_equal('<div class="rad" id="thing">tender div</div>',
+                     builder.doc.root.to_html.chomp)
+      end
+
+      def test_bold_tag
+        builder = Nokogiri::HTML::Builder.new do
+          b "bold tag"
+        end
+        assert_equal('<b>bold tag</b>', builder.doc.root.to_html.chomp)
+      end
+
+      def test_html_then_body_tag
+        builder = Nokogiri::HTML::Builder.new do
+          html {
+            body {
+              b "bold tag"
+            }
+          }
+        end
+        assert_equal('<html><body><b>bold tag</b></body></html>',
+                     builder.doc.root.to_html.chomp.gsub(/>\s*</, '><'))
+      end
+
+      def test_instance_eval_with_delegation_to_block_context
+        class << self
+          def foo
+            "foo!"
+          end
+        end
+
+        builder = Nokogiri::HTML::Builder.new { text foo }
+        assert builder.to_html.include?("foo!")
+      end
+
+      def test_builder_with_param
+        doc = Nokogiri::HTML::Builder.new { |html|
+          html.body {
+            html.p "hello world"
+          }
+        }.doc
+
+        assert node = doc.xpath('//body/p').first
+        assert_equal 'hello world', node.content
+      end
+
+      def test_builder_with_id
+        text = "hello world"
+        doc = Nokogiri::HTML::Builder.new { |html|
+          html.body {
+            html.id_ text
+          }
+        }.doc
+
+        assert node = doc.xpath('//body/id').first
+        assert_equal text, node.content
+      end
+    end
+  end
+end

data/test/html/test_document.rb

@@ -0,0 +1,385 @@
+require "helper"
+
+module Nokogiri
+  module HTML
+    class TestDocument < Nokogiri::TestCase
+      def setup
+        super
+        @html = Nokogiri::HTML.parse(File.read(HTML_FILE))
+      end
+
+      def test_fragment
+        fragment = @html.fragment
+        assert_equal 0, fragment.children.length
+      end
+
+      def test_document_takes_config_block
+        options = nil
+        Nokogiri::HTML(File.read(HTML_FILE), HTML_FILE) do |cfg|
+          options = cfg
+          options.nonet.nowarning.dtdattr
+        end
+        assert options.nonet?
+        assert options.nowarning?
+        assert options.dtdattr?
+      end
+
+      def test_parse_takes_config_block
+        options = nil
+        Nokogiri::HTML.parse(File.read(HTML_FILE), HTML_FILE) do |cfg|
+          options = cfg
+          options.nonet.nowarning.dtdattr
+        end
+        assert options.nonet?
+        assert options.nowarning?
+        assert options.dtdattr?
+      end
+
+      def test_subclass
+        klass = Class.new(Nokogiri::HTML::Document)
+        doc = klass.new
+        assert_instance_of klass, doc
+      end
+
+      def test_subclass_initialize
+        klass = Class.new(Nokogiri::HTML::Document) do
+          attr_accessor :initialized_with
+
+          def initialize(*args)
+            @initialized_with = args
+          end
+        end
+        doc = klass.new("uri", "external_id", 1)
+        assert_equal ["uri", "external_id", 1], doc.initialized_with
+      end
+
+      def test_subclass_dup
+        klass = Class.new(Nokogiri::HTML::Document)
+        doc = klass.new.dup
+        assert_instance_of klass, doc
+      end
+
+      def test_subclass_parse
+        klass = Class.new(Nokogiri::HTML::Document)
+        doc = klass.parse(File.read(HTML_FILE))
+        assert_equal @html.to_s, doc.to_s
+        assert_instance_of klass, doc
+      end
+
+      def test_document_parse_method
+        html = Nokogiri::HTML::Document.parse(File.read(HTML_FILE))
+        assert_equal @html.to_s, html.to_s
+      end
+
+      ###
+      # Nokogiri::HTML returns an empty Document when given a blank string GH#11
+      def test_empty_string_returns_empty_doc
+        doc = Nokogiri::HTML('')
+        assert_instance_of Nokogiri::HTML::Document, doc
+        assert_nil doc.root
+      end
+
+      unless Nokogiri.uses_libxml? && %w[2 6] === LIBXML_VERSION.split('.')[0..1]
+        # FIXME: this is a hack around broken libxml versions
+        def test_to_xhtml_with_indent
+          doc = Nokogiri::HTML('<html><body><a>foo</a></body></html>')
+          doc = Nokogiri::HTML(doc.to_xhtml(:indent => 2))
+          assert_indent 2, doc
+        end
+
+        def test_write_to_xhtml_with_indent
+          io = StringIO.new
+          doc = Nokogiri::HTML('<html><body><a>foo</a></body></html>')
+          doc.write_xhtml_to io, :indent => 5
+          io.rewind
+          doc = Nokogiri::HTML(io.read)
+          assert_indent 5, doc
+        end
+      end
+
+      def test_swap_should_not_exist
+        assert_raises(NoMethodError) {
+          @html.swap
+        }
+      end
+
+      def test_namespace_should_not_exist
+        assert_raises(NoMethodError) {
+          @html.namespace
+        }
+      end
+
+#      def test_meta_encoding
+#        assert_equal 'UTF-8', @html.meta_encoding
+#      end
+#
+#      def test_meta_encoding=
+#        @html.meta_encoding = 'EUC-JP'
+#        assert_equal 'EUC-JP', @html.meta_encoding
+#      end
+
+      def test_meta_encoding_without_head
+        html = Nokogiri::HTML('<html><body>foo</body></html>')
+        assert_nil html.meta_encoding
+
+        html.meta_encoding = 'EUC-JP'
+        assert_nil html.meta_encoding
+      end
+
+      def test_root_node_parent_is_document
+        parent = @html.root.parent
+        assert_equal @html, parent
+        assert_instance_of Nokogiri::HTML::Document, parent
+      end
+
+      def test_parse_handles_nil_gracefully
+        assert_nothing_raised do
+          @doc = Nokogiri::HTML::Document.parse(nil)
+        end
+        assert_instance_of Nokogiri::HTML::Document, @doc
+      end
+
+      def test_parse_empty_document
+        doc = Nokogiri::HTML("\n")
+        assert_equal 0, doc.css('a').length
+        assert_equal 0, doc.xpath('//a').length
+        assert_equal 0, doc.search('//a').length
+      end
+
+      def test_HTML_function
+        html = Nokogiri::HTML(File.read(HTML_FILE))
+        assert html.html?
+      end
+
+      def test_parse_io
+        assert doc = File.open(HTML_FILE, 'rb') { |f|
+          Document.read_io(f, nil, 'UTF-8',
+                           XML::ParseOptions::NOERROR | XML::ParseOptions::NOWARNING
+                          )
+        }
+      end
+
+      def test_to_xhtml
+        assert_match 'XHTML', @html.to_xhtml
+        assert_match 'XHTML', @html.to_xhtml(:encoding => 'UTF-8')
+        assert_match 'UTF-8', @html.to_xhtml(:encoding => 'UTF-8')
+      end
+
+      def test_no_xml_header
+        html = Nokogiri::HTML(<<-eohtml)
+        <html>
+        </html>
+        eohtml
+        assert html.to_html.length > 0, 'html length is too short'
+        assert_no_match(/^<\?xml/, html.to_html)
+      end
+
+#      def test_document_has_error
+#        html = Nokogiri::HTML(<<-eohtml)
+#        <html>
+#          <body>
+#            <div awesome="asdf>
+#              <p>inside div tag</p>
+#            </div>
+#            <p>outside div tag</p>
+#          </body>
+#        </html>
+#        eohtml
+#        assert html.errors.length > 0
+#      end
+#
+#      def test_relative_css
+#        html = Nokogiri::HTML(<<-eohtml)
+#        <html>
+#          <body>
+#            <div>
+#              <p>inside div tag</p>
+#            </div>
+#            <p>outside div tag</p>
+#          </body>
+#        </html>
+#        eohtml
+#        set = html.search('div').search('p')
+#        assert_equal(1, set.length)
+#        assert_equal('inside div tag', set.first.inner_text)
+#      end
+
+      def test_multi_css
+        html = Nokogiri::HTML(<<-eohtml)
+        <html>
+          <body>
+            <div>
+              <p>p tag</p>
+              <a>a tag</a>
+            </div>
+          </body>
+        </html>
+        eohtml
+        set = html.css('p, a')
+        assert_equal(2, set.length)
+        assert_equal ['a tag', 'p tag'].sort, set.map { |x| x.content }.sort
+      end
+
+#      def test_inner_text
+#        html = Nokogiri::HTML(<<-eohtml)
+#        <html>
+#          <body>
+#            <div>
+#              <p>
+#                Hello world!
+#              </p>
+#            </div>
+#          </body>
+#        </html>
+#        eohtml
+#        node = html.xpath('//div').first
+#        assert_equal('Hello world!', node.inner_text.strip)
+#      end
+#
+#      def test_find_by_xpath
+#        found = @html.xpath('//div/a')
+#        assert_equal 3, found.length
+#      end
+#
+#      def test_find_by_css
+#        found = @html.css('div > a')
+#        assert_equal 3, found.length
+#      end
+#
+#      def test_find_by_css_with_square_brackets
+#        found = @html.css("div[@id='header'] > h1")
+#        found = @html.css("div[@id='header'] h1") # this blows up on commit 6fa0f6d329d9dbf1cc21c0ac72f7e627bb4c05fc
+#        assert_equal 1, found.length
+#      end
+#
+#      def test_find_with_function
+#        found = @html.css("div:awesome() h1", Class.new {
+#          def awesome divs
+#            [divs.first]
+#          end
+#        }.new)
+#      end
+#
+#      def test_dup_shallow
+#        found = @html.search('//div/a').first
+#        dup = found.dup(0)
+#        assert dup
+#        assert_equal '', dup.content
+#      end
+#
+#      def test_search_can_handle_xpath_and_css
+#        found = @html.search('//div/a', 'div > p')
+#        length = @html.xpath('//div/a').length +
+#          @html.css('div > p').length
+#        assert_equal length, found.length
+#      end
+
+      def test_dup_document
+        assert dup = @html.dup
+        assert_not_equal dup, @html
+        assert @html.html?
+        assert_instance_of Nokogiri::HTML::Document, dup
+        assert dup.html?, 'duplicate should be html'
+        assert_equal @html.to_s, dup.to_s
+      end
+
+      def test_dup_document_shallow
+        assert dup = @html.dup(0)
+        assert_not_equal dup, @html
+      end
+
+#      def test_dup
+#        found = @html.search('//div/a').first
+#        dup = found.dup
+#        assert dup
+#        assert_equal found.content, dup.content
+#        assert_equal found.document, dup.document
+#      end
+#
+#      def test_inner_html
+#        html = Nokogiri::HTML(<<-eohtml)
+#        <html>
+#          <body>
+#            <div>
+#              <p>
+#                Hello world!
+#              </p>
+#            </div>
+#          </body>
+#        </html>
+#        eohtml
+#        node = html.xpath('//div').first
+#        assert_equal('<p>Helloworld!</p>', node.inner_html.gsub(/\s/, ''))
+#      end
+#
+#      def test_fragment_contains_text_node
+#        fragment = Nokogiri::HTML.fragment('fooo')
+#        assert_equal 1, fragment.children.length
+#        assert_equal 'fooo', fragment.inner_text
+#      end
+#
+#      def test_fragment_includes_two_tags
+#        assert_equal 2, Nokogiri::HTML.fragment("<br/><hr/>").children.length
+#      end
+#
+#      def test_relative_css_finder
+#        doc = Nokogiri::HTML(<<-eohtml)
+#          <html>
+#            <body>
+#              <div class="red">
+#                <p>
+#                  inside red
+#                </p>
+#              </div>
+#              <div class="green">
+#                <p>
+#                  inside green
+#                </p>
+#              </div>
+#            </body>
+#          </html>
+#        eohtml
+#        red_divs = doc.css('div.red')
+#        assert_equal 1, red_divs.length
+#        p_tags = red_divs.first.css('p')
+#        assert_equal 1, p_tags.length
+#        assert_equal 'inside red', p_tags.first.text.strip
+#      end
+#
+#      def test_find_classes
+#        doc = Nokogiri::HTML(<<-eohtml)
+#          <html>
+#            <body>
+#              <p class="red">RED</p>
+#              <p class="awesome red">RED</p>
+#              <p class="notred">GREEN</p>
+#              <p class="green notred">GREEN</p>
+#            </body>
+#          </html>
+#        eohtml
+#        list = doc.css('.red')
+#        assert_equal 2, list.length
+#        assert_equal %w{ RED RED }, list.map { |x| x.text }
+#      end
+
+      def test_parse_can_take_io
+        html = nil
+        File.open(HTML_FILE, 'rb') { |f|
+          html = Nokogiri::HTML(f)
+        }
+        assert html.html?
+      end
+
+      def test_html?
+        assert !@html.xml?
+        assert @html.html?
+      end
+
+      def test_serialize
+        assert @html.serialize
+        assert @html.to_html
+      end
+    end
+  end
+end
+