nokogiri 1.1.1 → 1.2.0

data/lib/nokogiri/xml/node_set.rb

@@ -41,11 +41,7 @@ module Nokogiri
         last.after datum
       end
 
-      ###
-      # Append +node+ to the NodeSet.
-      def << node
-        push(node)
-      end
+      alias :<< :push
 
       ###
       # Unlink this NodeSet and all Node objects it contains from their
@@ -163,12 +159,8 @@ module Nokogiri
       def wrap(html, &blk)
         each do |j|
           new_parent = Nokogiri.make(html, &blk)
-          j.replace(new_parent)
-          nest = new_parent
-          if nest.child
-            nest = nest.child until nest.child.nil?
-          end
-          j.parent = nest
+          j.parent.add_child(new_parent)
+          new_parent.add_child(j)
         end
         self
       end
@@ -177,21 +169,20 @@ module Nokogiri
         map { |x| x.to_s }.join
       end
 
-      def to_html
-        map { |x| x.to_html }.join('')
+      def to_html *args
+        map { |x| x.to_html(*args) }.join('')
       end
 
-      def to_xml *args
-        map { |x| x.to_xml(*args) }.join('')
+      def to_xhtml *args
+        map { |x| x.to_xhtml(*args) }.join('')
       end
 
-      def size
-        length
+      def to_xml *args
+        map { |x| x.to_xml(*args) }.join('')
       end
 
-      def to_ary
-        to_a
-      end
+      alias :size :length
+      alias :to_ary :to_a
     end
   end
 end

data/lib/nokogiri/xml/processing_instruction.rb

@@ -0,0 +1,6 @@
+module Nokogiri
+  module XML
+    class ProcessingInstruction < Node
+    end
+  end
+end

data/lib/nokogiri/xml/reader.rb

@@ -2,6 +2,11 @@ module Nokogiri
   module XML
     class Reader
       include Enumerable
+      attr_accessor :errors
+
+      def initialize
+        @errors = []
+      end
 
       def attributes
         Hash[*(attribute_nodes.map { |node|

data/lib/nokogiri/xml/sax.rb

@@ -1,5 +1,6 @@
 require 'nokogiri/xml/sax/document'
 require 'nokogiri/xml/sax/parser'
+require 'nokogiri/xml/sax/push_parser'
 
 module Nokogiri
   module XML

data/lib/nokogiri/xml/sax/push_parser.rb

@@ -0,0 +1,47 @@
+module Nokogiri
+  module XML
+    module SAX
+      ###
+      # PushParser can parse a document that is fed to it manually.  It
+      # must be given a SAX::Document object which will be called with
+      # SAX events as the document is being parsed.
+      #
+      # Calling PushParser#<< writes XML to the parser, calling any SAX
+      # callbacks it can.
+      #
+      # PushParser#finish tells the parser that the document is finished
+      # and calls the end_document SAX method.
+      #
+      # Example:
+      #
+      #   parser = PushParser.new(Class.new(XML::SAX::Document) {
+      #     def start_document
+      #       puts "start document called"
+      #     end
+      #   }.new)
+      #   parser << "<div>hello<"
+      #   parser << "/div>"
+      #   parser.finish
+      class PushParser
+        attr_accessor :document
+
+        def initialize(doc = XML::SAX::Document.new, file_name = nil)
+          @document = doc
+          @sax_parser = XML::SAX::Parser.new(doc)
+
+          ## Create our push parser context
+          initialize_native(@sax_parser, file_name)
+        end
+
+        def write chunk, last_chunk = false
+          native_write(chunk, last_chunk)
+        end
+        alias :<< :write
+
+        def finish
+          write '', true
+        end
+      end
+    end
+  end
+end

data/lib/nokogiri/xml/syntax_error.rb

@@ -1,6 +1,6 @@
 module Nokogiri
   module XML
-    class SyntaxError < SyntaxError
+    class SyntaxError < ::Nokogiri::SyntaxError
       def none?
         level == 0
       end
@@ -16,6 +16,8 @@ module Nokogiri
       def fatal?
         level == 3
       end
+
+      alias :to_s :message
     end
   end
 end

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

@@ -1,7 +1,7 @@
 module Nokogiri
   module XML
     class XPath
-      class SyntaxError < ::SyntaxError
+      class SyntaxError < XML::SyntaxError
       end
     end
   end

data/tasks/test.rb

@@ -0,0 +1,136 @@
+# partial-loads-ok and undef-value-errors necessary to ignore
+# spurious (and eminently ignorable) warnings from the ruby
+# interpreter
+VALGRIND_BASIC_OPTS = "--num-callers=50 --error-limit=no --partial-loads-ok=yes --undef-value-errors=no"
+
+class NokogiriTestTask < Rake::TestTask
+  def initialize *args
+    super
+    %w[ ext lib bin test ].each do |dir|
+      self.libs << dir
+    end
+    self.test_files = FileList['test/**/test_*.rb'] +
+      FileList['test/**/*_test.rb']
+    self.verbose = true
+    self.warning = true
+  end
+end
+
+desc "run test suite under valgrind with basic ruby options"
+NokogiriTestTask.new('test:valgrind').extend(Module.new {
+  def ruby *args
+    cmd = "valgrind #{VALGRIND_BASIC_OPTS} #{RUBY} #{args.join(' ')}"
+    puts cmd
+    system cmd
+  end
+})
+
+desc "run test suite under valgrind with memory-fill ruby options"
+NokogiriTestTask.new('test:valgrind_mem').extend(Module.new {
+  def ruby *args
+    cmd = "valgrind #{VALGRIND_BASIC_OPTS} --freelist-vol=100000000 --malloc-fill=6D --free-fill=66 #{RUBY} #{args.join(' ')}"
+    puts cmd
+    system cmd
+  end
+})
+
+desc "run test suite under valgrind with memory-zero ruby options"
+NokogiriTestTask.new('test:valgrind_mem0').extend(Module.new {
+  def ruby *args
+    cmd = "valgrind #{VALGRIND_BASIC_OPTS} --freelist-vol=100000000 --malloc-fill=00 --free-fill=00 #{RUBY} #{args.join(' ')}"
+    puts cmd
+    system cmd
+  end
+})
+
+desc "run test suite under gdb"
+NokogiriTestTask.new('test:gdb').extend(Module.new {
+  def ruby *args
+    cmd = "gdb --args #{RUBY} #{args.join(' ')}"
+    puts cmd
+    system cmd
+  end
+})
+
+desc "test coverage"
+NokogiriTestTask.new('test:coverage').extend(Module.new {
+  def ruby *args
+    rm_rf "coverage"
+    cmd = "rcov -x Library -I lib:ext:test #{args.join(' ')}"
+    puts cmd
+    system cmd
+  end
+})
+
+namespace :test do
+  desc "run test suite with aggressive GC"
+  task :gc => :build do
+    ENV['NOKOGIRI_GC'] = "true"
+    Rake::Task["test"].invoke
+  end
+
+  desc "find call-seq in the rdoc"
+  task :rdoc => 'docs' do
+    Dir['doc/**/*.html'].each { |docfile|
+      next if docfile =~ /\.src/
+      puts "FAIL: #{docfile}" if File.read(docfile) =~ /call-seq/
+    }
+  end
+
+  desc "Test against multiple versions of libxml2"
+  task :multixml2 do
+    MULTI_XML = File.join(ENV['HOME'], '.multixml2')
+    unless File.exists?(MULTI_XML)
+      %w{ versions install build }.each { |x|
+        FileUtils.mkdir_p(File.join(MULTI_XML, x))
+      }
+      Dir.chdir File.join(MULTI_XML, 'versions') do
+        require 'net/ftp'
+        ftp = Net::FTP.new('xmlsoft.org')
+        ftp.login('anonymous', 'anonymous')
+        ftp.chdir('libxml2')
+        ftp.list('libxml2-2.*.tar.gz').each do |x|
+          file = x[/[^\s]*$/]
+          puts "Downloading #{file}"
+          ftp.getbinaryfile(file)
+        end
+      end
+    end
+
+    # Build any libxml2 versions in $HOME/.multixml2/versions that
+    # haven't been built yet
+    Dir[File.join(MULTI_XML, 'versions','*.tar.gz')].each do |f|
+      filename = File.basename(f, '.tar.gz')
+
+      install_dir = File.join(MULTI_XML, 'install', filename)
+      next if File.exists?(install_dir)
+
+      Dir.chdir File.join(MULTI_XML, 'versions') do
+        system "tar zxvf #{f} -C #{File.join(MULTI_XML, 'build')}"
+      end
+
+      Dir.chdir File.join(MULTI_XML, 'build', filename) do
+        system "./configure --prefix=#{install_dir}"
+        system "make && make install"
+      end
+    end
+
+    test_results = {}
+    Dir[File.join(MULTI_XML, 'install', '*')].each do |xml2_version|
+      extopts = "--with-xml2-include=#{xml2_version}/include/libxml2 --with-xml2-lib=#{xml2_version}/lib --with-xslt-dir=/usr/local"
+      cmd = "#{$0} clean test EXTOPTS='#{extopts}'"
+
+      version = File.basename(xml2_version)
+      result = system(cmd)
+      test_results[version] = {
+        :result => result,
+        :cmd    => cmd
+      }
+    end
+    test_results.sort_by { |k,v| k }.each do |k,v|
+      passed = v[:result]
+      puts "#{k}: #{passed ? 'PASS' : 'FAIL'}"
+      puts "repro: #{v[:cmd]}" unless passed
+    end
+  end
+end

data/test/css/test_parser.rb

@@ -7,6 +7,10 @@ module Nokogiri
         @parser = Nokogiri::CSS::Parser.new
       end
 
+      def test_extra_single_quote
+        assert_raises(CSS::SyntaxError) { @parser.parse("'") }
+      end
+
       def test_syntax_error_raised
         assert_raises(CSS::SyntaxError) { @parser.parse("a[x=]") }
       end

data/test/css/test_tokenizer.rb

@@ -7,14 +7,29 @@ module Nokogiri
         @scanner = Nokogiri::CSS::Tokenizer.new
       end
 
+      def test_tokenize_bad_single_quote
+        @scanner.scan("'")
+        assert_tokens([["'", "'"]], @scanner)
+      end
+
       def test_not_equal
         @scanner.scan("h1[a!='Tender Lovemaking']")
         assert_tokens([ [:IDENT, 'h1'],
-                        ['[', '['],
+                        [:LSQUARE, '['],
                         [:IDENT, 'a'],
                         [:NOT_EQUAL, '!='],
                         [:STRING, "'Tender Lovemaking'"],
-                        [']', ']'],
+                        [:RSQUARE, ']'],
+        ], @scanner)
+      end
+
+      def test_negation
+        @scanner.scan("p:not(.a)")
+        assert_tokens([ [:IDENT, 'p'],
+                        [:NOT, ':not('],
+                        ['.', '.'],
+                        [:IDENT, 'a'],
+                        [:RPAREN, ')'],
         ], @scanner)
       end
 
@@ -23,15 +38,14 @@ module Nokogiri
         assert_tokens([ [:IDENT, 'script'],
                         [:S, ' '],
                         [:FUNCTION, 'comment('],
-                        [')', ')'],
+                        [:RPAREN, ')'],
         ], @scanner)
       end
 
       def test_preceding_selector
         @scanner.scan("E ~ F")
         assert_tokens([ [:IDENT, 'E'],
-                        [:TILDE, ' ~'],
-                        [:S, ' '],
+                        [:TILDE, ' ~ '],
                         [:IDENT, 'F'],
         ], @scanner)
       end
@@ -39,19 +53,19 @@ module Nokogiri
       def test_scan_attribute_string
         @scanner.scan("h1[a='Tender Lovemaking']")
         assert_tokens([ [:IDENT, 'h1'],
-                        ['[', '['],
+                        [:LSQUARE, '['],
                         [:IDENT, 'a'],
-                        ['=', '='],
+                        [:EQUAL, '='],
                         [:STRING, "'Tender Lovemaking'"],
-                        [']', ']'],
+                        [:RSQUARE, ']'],
         ], @scanner)
         @scanner.scan('h1[a="Tender Lovemaking"]')
         assert_tokens([ [:IDENT, 'h1'],
-                        ['[', '['],
+                        [:LSQUARE, '['],
                         [:IDENT, 'a'],
-                        ['=', '='],
+                        [:EQUAL, '='],
                         [:STRING, '"Tender Lovemaking"'],
-                        [']', ']'],
+                        [:RSQUARE, ']'],
         ], @scanner)
       end
 
@@ -84,8 +98,7 @@ module Nokogiri
       def test_scan_greater
         @scanner.scan('x > y')
         assert_tokens([ [:IDENT, 'x'],
-                        [:GREATER, ' >'],
-                        [:S, ' '],
+                        [:GREATER, ' > '],
                         [:IDENT, 'y']
         ], @scanner)
       end
@@ -112,7 +125,7 @@ module Nokogiri
                         [':', ':'],
                         [:FUNCTION, 'eq('],
                         [:NUMBER, "0"],
-                        [")", ")"]
+                        [:RPAREN, ')'],
         ], @scanner)
       end
 
@@ -125,7 +138,7 @@ module Nokogiri
                         [:IDENT, 'n'],
                         [:PLUS, '+'],
                         [:NUMBER, '3'],
-                        [")", ")"]
+                        [:RPAREN, ')'],
         ], @scanner)
 
         @scanner.scan('x:nth-child(-1n+3)')
@@ -136,7 +149,7 @@ module Nokogiri
                         [:IDENT, 'n'],
                         [:PLUS, '+'],
                         [:NUMBER, '3'],
-                        [")", ")"]
+                        [:RPAREN, ')'],
         ], @scanner)
 
         @scanner.scan('x:nth-child(-n+3)')
@@ -146,7 +159,7 @@ module Nokogiri
                         [:IDENT, '-n'],
                         [:PLUS, '+'],
                         [:NUMBER, '3'],
-                        [")", ")"]
+                        [:RPAREN, ')'],
         ], @scanner)
       end
 

data/test/css/test_xpath_visitor.rb

@@ -7,6 +7,17 @@ module Nokogiri
         @parser = Nokogiri::CSS::Parser.new
       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)'))
+        assert_xpath("//a[foo(., a, 10)]", @parser.parse('a:foo(a, 10)'))
+      end
+
+      def test_namespace_conversion
+        assert_xpath("//aaron:a", @parser.parse('aaron|a'))
+        assert_xpath("//a", @parser.parse('|a'))
+      end
+
       def test_unknown_psuedo_classes_get_pushed_down
         assert_xpath("//a[aaron(.)]", @parser.parse('a:aaron'))
       end

data/test/helper.rb

@@ -34,6 +34,7 @@ module Nokogiri
         attr_reader :start_elements, :start_document_called
         attr_reader :end_elements, :end_document_called
         attr_reader :data, :comments, :cdata_blocks
+        attr_reader :errors, :warnings
 
         def start_document
           @start_document_called = true
@@ -45,6 +46,16 @@ module Nokogiri
           super
         end
 
+        def error error
+          (@errors ||= []) << error
+          super
+        end
+
+        def warning warning
+          (@warning ||= []) << warning
+          super
+        end
+
         def start_element *args
           (@start_elements ||= []) << args
           super