From 196334c5a1a978f918ab435be90dc8b6673bc506 Mon Sep 17 00:00:00 2001 From: Willem van Bergen Date: Fri, 1 Jan 2010 14:01:43 +0100 Subject: [PATCH] Added two SAX-based backends for XmlMini, using both LibXML and Nokogiri. --- .../lib/active_support/xml_mini/libxmlsax.rb | 74 +++++++ .../lib/active_support/xml_mini/nokogirisax.rb | 73 +++++++ activesupport/test/core_ext/hash_ext_test.rb | 8 +- .../test/xml_mini/libxmlsax_engine_test.rb | 181 ++++++++++++++++ .../test/xml_mini/nokogirisax_engine_test.rb | 217 ++++++++++++++++++++ 5 files changed, 550 insertions(+), 3 deletions(-) create mode 100644 activesupport/lib/active_support/xml_mini/libxmlsax.rb create mode 100644 activesupport/lib/active_support/xml_mini/nokogirisax.rb create mode 100644 activesupport/test/xml_mini/libxmlsax_engine_test.rb create mode 100644 activesupport/test/xml_mini/nokogirisax_engine_test.rb diff --git a/activesupport/lib/active_support/xml_mini/libxmlsax.rb b/activesupport/lib/active_support/xml_mini/libxmlsax.rb new file mode 100644 index 0000000..a88607d --- /dev/null +++ b/activesupport/lib/active_support/xml_mini/libxmlsax.rb @@ -0,0 +1,74 @@ +require 'libxml' + +# = XmlMini LibXML implementation using a SAX-based parser +module ActiveSupport + module XmlMini_LibXMLSAX + extend self + + # Class that will build the hash while the XML document + # is being parsed using SAX events. + class HashBuilder + + include LibXML::XML::SaxParser::Callbacks + + CONTENT_KEY = '__content__'.freeze + HASH_SIZE_KEY = '__hash_size__'.freeze + + attr_reader :hash + + def current_hash + @hash_stack.last + end + + def on_start_document + @hash = { CONTENT_KEY => '' } + @hash_stack = [@hash] + end + + def on_end_document + @hash = @hash_stack.pop + @hash.delete(CONTENT_KEY) + end + + def on_start_element(name, attrs = {}) + new_hash = { CONTENT_KEY => '' }.merge(attrs) + new_hash[HASH_SIZE_KEY] = new_hash.size + 1 + + case current_hash[name] + when Array then current_hash[name] << new_hash + when Hash then current_hash[name] = [current_hash[name], new_hash] + when nil then current_hash[name] = new_hash + end + + @hash_stack.push(new_hash) + end + + def on_end_element(name) + if current_hash.length > current_hash.delete(HASH_SIZE_KEY) && current_hash[CONTENT_KEY].blank? || current_hash[CONTENT_KEY] == '' + current_hash.delete(CONTENT_KEY) + end + @hash_stack.pop + end + + def on_characters(string) + current_hash[CONTENT_KEY] << string + end + + alias_method :on_cdata_block, :on_characters + end + + attr_accessor :document_class + self.document_class = HashBuilder + + def parse(string) + return {} if string.blank? + LibXML::XML::Error.set_handler(&LibXML::XML::Error::QUIET_HANDLER) + parser = LibXML::XML::SaxParser.string(string) + document = self.document_class.new + + parser.callbacks = document + parser.parse + document.hash + end + end +end \ No newline at end of file diff --git a/activesupport/lib/active_support/xml_mini/nokogirisax.rb b/activesupport/lib/active_support/xml_mini/nokogirisax.rb new file mode 100644 index 0000000..f13c9bf --- /dev/null +++ b/activesupport/lib/active_support/xml_mini/nokogirisax.rb @@ -0,0 +1,73 @@ +require 'nokogiri' + +# = XmlMini Nokogiri implementation using a SAX-based parser +module ActiveSupport + module XmlMini_NokogiriSAX + extend self + + # Class that will build the hash while the XML document + # is being parsed using SAX events. + class HashBuilder < Nokogiri::XML::SAX::Document + + CONTENT_KEY = '__content__'.freeze + HASH_SIZE_KEY = '__hash_size__'.freeze + + attr_reader :hash + + def current_hash + @hash_stack.last + end + + def start_document + @hash = {} + @hash_stack = [@hash] + end + + def end_document + raise "Parse stack not empty!" if @hash_stack.size > 1 + end + + def error(error_message) + raise error_message + end + + def start_element(name, attrs = []) + new_hash = { CONTENT_KEY => '' } + new_hash[attrs.shift] = attrs.shift while attrs.length > 0 + new_hash[HASH_SIZE_KEY] = new_hash.size + 1 + + case current_hash[name] + when Array then current_hash[name] << new_hash + when Hash then current_hash[name] = [current_hash[name], new_hash] + when nil then current_hash[name] = new_hash + end + + @hash_stack.push(new_hash) + end + + def end_element(name) + if current_hash.length > current_hash.delete(HASH_SIZE_KEY) && current_hash[CONTENT_KEY].blank? || current_hash[CONTENT_KEY] == '' + current_hash.delete(CONTENT_KEY) + end + @hash_stack.pop + end + + def characters(string) + current_hash[CONTENT_KEY] << string + end + + alias_method :cdata_block, :characters + end + + attr_accessor :document_class + self.document_class = HashBuilder + + def parse(string) + return {} if string.blank? + document = self.document_class.new + parser = Nokogiri::XML::SAX::Parser.new(document) + parser.parse(string) + document.hash + end + end +end \ No newline at end of file diff --git a/activesupport/test/core_ext/hash_ext_test.rb b/activesupport/test/core_ext/hash_ext_test.rb index 736c1c1..44308c3 100644 --- a/activesupport/test/core_ext/hash_ext_test.rb +++ b/activesupport/test/core_ext/hash_ext_test.rb @@ -982,9 +982,11 @@ class QueryTest < Test::Unit::TestCase def test_expansion_count_is_limited expected = { - 'ActiveSupport::XmlMini_REXML' => 'RuntimeError', - 'ActiveSupport::XmlMini_Nokogiri' => 'Nokogiri::XML::SyntaxError', - 'ActiveSupport::XmlMini_LibXML' => 'LibXML::XML::Error', + 'ActiveSupport::XmlMini_REXML' => 'RuntimeError', + 'ActiveSupport::XmlMini_Nokogiri' => 'Nokogiri::XML::SyntaxError', + 'ActiveSupport::XmlMini_NokogiriSAX' => 'RuntimeError', + 'ActiveSupport::XmlMini_LibXML' => 'LibXML::XML::Error', + 'ActiveSupport::XmlMini_LibXMLSAX' => 'LibXML::XML::Error', }[ActiveSupport::XmlMini.backend.name].constantize assert_raise expected do diff --git a/activesupport/test/xml_mini/libxmlsax_engine_test.rb b/activesupport/test/xml_mini/libxmlsax_engine_test.rb new file mode 100644 index 0000000..51543e8 --- /dev/null +++ b/activesupport/test/xml_mini/libxmlsax_engine_test.rb @@ -0,0 +1,181 @@ +require 'abstract_unit' +require 'active_support/xml_mini' +require 'active_support/core_ext/hash/conversions' + +begin + require 'libxml' +rescue LoadError + # Skip libxml tests +else + +class LibXMLSAXEngineTest < Test::Unit::TestCase + include ActiveSupport + + def setup + @default_backend = XmlMini.backend + XmlMini.backend = 'LibXMLSAX' + end + + def teardown + XmlMini.backend = @default_backend + end + + def test_exception_thrown_on_expansion_attack + assert_raise LibXML::XML::Error do + attack_xml = <<-EOT + + + + + + + + + ]> + + &a; + + EOT + + Hash.from_xml(attack_xml) + end + end + + def test_setting_libxml_as_backend + XmlMini.backend = 'LibXMLSAX' + assert_equal XmlMini_LibXMLSAX, XmlMini.backend + end + + def test_blank_returns_empty_hash + assert_equal({}, XmlMini.parse(nil)) + assert_equal({}, XmlMini.parse('')) + end + + def test_array_type_makes_an_array + assert_equal_rexml(<<-eoxml) + + + a post + another post + + + eoxml + end + + def test_one_node_document_as_hash + assert_equal_rexml(<<-eoxml) + + eoxml + end + + def test_one_node_with_attributes_document_as_hash + assert_equal_rexml(<<-eoxml) + + eoxml + end + + def test_products_node_with_book_node_as_hash + assert_equal_rexml(<<-eoxml) + + + + eoxml + end + + def test_products_node_with_two_book_nodes_as_hash + assert_equal_rexml(<<-eoxml) + + + + + eoxml + end + + def test_single_node_with_content_as_hash + assert_equal_rexml(<<-eoxml) + + hello world + + eoxml + end + + def test_children_with_children + assert_equal_rexml(<<-eoxml) + + + + + + eoxml + end + + def test_children_with_text + assert_equal_rexml(<<-eoxml) + + + hello everyone + + + eoxml + end + + def test_children_with_non_adjacent_text + assert_equal_rexml(<<-eoxml) + + good + + hello everyone + + morning + + eoxml + end + + def test_children_with_simple_cdata + assert_equal_rexml(<<-eoxml) + + + + + + eoxml + end + + def test_children_with_multiple_cdata + assert_equal_rexml(<<-eoxml) + + + + + + eoxml + end + + def test_children_with_text_and_cdata + assert_equal_rexml(<<-eoxml) + + + hello + morning + + + eoxml + end + + def test_children_with_blank_text + assert_equal_rexml(<<-eoxml) + + + + eoxml + end + + private + def assert_equal_rexml(xml) + hash = XmlMini.with_backend('REXML') { XmlMini.parse(xml) } + assert_equal(hash, XmlMini.parse(xml)) + end +end + +end diff --git a/activesupport/test/xml_mini/nokogirisax_engine_test.rb b/activesupport/test/xml_mini/nokogirisax_engine_test.rb new file mode 100644 index 0000000..1149d0f --- /dev/null +++ b/activesupport/test/xml_mini/nokogirisax_engine_test.rb @@ -0,0 +1,217 @@ +require 'abstract_unit' +require 'active_support/xml_mini' +require 'active_support/core_ext/hash/conversions' + +begin + require 'nokogiri' +rescue LoadError + # Skip nokogiri tests +else + +class NokogiriSAXEngineTest < Test::Unit::TestCase + include ActiveSupport + + def setup + @default_backend = XmlMini.backend + XmlMini.backend = 'NokogiriSAX' + end + + def teardown + XmlMini.backend = @default_backend + end + + def test_file_from_xml + hash = Hash.from_xml(<<-eoxml) + + + + + eoxml + assert hash.has_key?('blog') + assert hash['blog'].has_key?('logo') + + file = hash['blog']['logo'] + assert_equal 'logo.png', file.original_filename + assert_equal 'image/png', file.content_type + end + + def test_exception_thrown_on_expansion_attack + assert_raise RuntimeError do + attack_xml = <<-EOT + + + + + + + + + ]> + + &a; + + EOT + + Hash.from_xml(attack_xml) + end + end + + def test_setting_nokogiri_as_backend + XmlMini.backend = 'Nokogiri' + assert_equal XmlMini_Nokogiri, XmlMini.backend + end + + def test_blank_returns_empty_hash + assert_equal({}, XmlMini.parse(nil)) + assert_equal({}, XmlMini.parse('')) + end + + def test_array_type_makes_an_array + assert_equal_rexml(<<-eoxml) + + + a post + another post + + + eoxml + end + + def test_one_node_document_as_hash + assert_equal_rexml(<<-eoxml) + + eoxml + end + + def test_one_node_with_attributes_document_as_hash + assert_equal_rexml(<<-eoxml) + + eoxml + end + + def test_products_node_with_book_node_as_hash + assert_equal_rexml(<<-eoxml) + + + + eoxml + end + + def test_products_node_with_two_book_nodes_as_hash + assert_equal_rexml(<<-eoxml) + + + + + eoxml + end + + def test_single_node_with_content_as_hash + assert_equal_rexml(<<-eoxml) + + hello world + + eoxml + end + + def test_children_with_children + assert_equal_rexml(<<-eoxml) + + + + + + eoxml + end + + def test_children_with_text + assert_equal_rexml(<<-eoxml) + + + hello everyone + + + eoxml + end + + def test_children_with_non_adjacent_text + assert_equal_rexml(<<-eoxml) + + good + + hello everyone + + morning + + eoxml + end + + def test_parse_from_io + io = StringIO.new(<<-eoxml) + + good + + hello everyone + + morning + + eoxml + XmlMini.parse(io) + end + + def test_children_with_simple_cdata + assert_equal_rexml(<<-eoxml) + + + + + + eoxml + end + + def test_children_with_multiple_cdata + assert_equal_rexml(<<-eoxml) + + + + + + eoxml + end + + def test_children_with_text_and_cdata + assert_equal_rexml(<<-eoxml) + + + hello + morning + + + eoxml + end + + def test_children_with_blank_text + assert_equal_rexml(<<-eoxml) + + + + eoxml + end + + def test_children_with_blank_text_and_attribute + assert_equal_rexml(<<-eoxml) + + + + eoxml + end + + private + def assert_equal_rexml(xml) + hash = XmlMini.with_backend('REXML') { XmlMini.parse(xml) } + assert_equal(hash, XmlMini.parse(xml)) + end +end + +end -- 1.6.5.1