From 19fbb31464da45ac0935b3db193305f18e245ba9 Mon Sep 17 00:00:00 2001 From: Rick Olson Date: Sun, 4 Feb 2007 20:04:40 +0000 Subject: Add much-needed html-scanner tests. Fixed CDATA parsing bug. [Rick] git-svn-id: http://svn-commit.rubyonrails.org/rails/trunk@6117 5ecf4fe2-1ee6-0310-87b1-e25e094e27de --- .../test/controller/html-scanner/document_test.rb | 104 +++++++++ .../test/controller/html-scanner/node_test.rb | 69 ++++++ .../test/controller/html-scanner/tag_node_test.rb | 239 +++++++++++++++++++++ .../test/controller/html-scanner/text_node_test.rb | 51 +++++ .../test/controller/html-scanner/tokenizer_test.rb | 125 +++++++++++ 5 files changed, 588 insertions(+) create mode 100644 actionpack/test/controller/html-scanner/document_test.rb create mode 100644 actionpack/test/controller/html-scanner/node_test.rb create mode 100644 actionpack/test/controller/html-scanner/tag_node_test.rb create mode 100644 actionpack/test/controller/html-scanner/text_node_test.rb create mode 100644 actionpack/test/controller/html-scanner/tokenizer_test.rb (limited to 'actionpack/test/controller') diff --git a/actionpack/test/controller/html-scanner/document_test.rb b/actionpack/test/controller/html-scanner/document_test.rb new file mode 100644 index 0000000000..a6ba70dde1 --- /dev/null +++ b/actionpack/test/controller/html-scanner/document_test.rb @@ -0,0 +1,104 @@ +require File.dirname(__FILE__) + '/../../abstract_unit' +require 'test/unit' + +class DocumentTest < Test::Unit::TestCase + def test_handle_doctype + doc = nil + assert_nothing_raised do + doc = HTML::Document.new <<-HTML.strip + + + + HTML + end + assert_equal 3, doc.root.children.length + assert_equal %{}, doc.root.children[0].content + assert_match %r{\s+}m, doc.root.children[1].content + assert_equal "html", doc.root.children[2].name + end + + def test_find_img + doc = HTML::Document.new <<-HTML.strip + + +

+ + + HTML + assert doc.find(:tag=>"img", :attributes=>{"src"=>"hello.gif"}) + end + + def test_find_all + doc = HTML::Document.new <<-HTML.strip + + +

+
+

something

+

here is more

+
+ + + HTML + all = doc.find_all :attributes => { :class => "test" } + assert_equal 3, all.length + assert_equal [ "p", "p", "em" ], all.map { |n| n.name } + end + + def test_find_with_text + doc = HTML::Document.new <<-HTML.strip + + +

Some text

+ + + HTML + assert doc.find(:content => "Some text") + assert doc.find(:tag => "p", :child => { :content => "Some text" }) + assert doc.find(:tag => "p", :child => "Some text") + assert doc.find(:tag => "p", :content => "Some text") + end + + def test_parse_xml + assert_nothing_raised { HTML::Document.new("", true, true) } + assert_nothing_raised { HTML::Document.new("something", true, true) } + end + + def test_parse_document + doc = HTML::Document.new(<<-HTML) +
+

blah

+ +
+
+ HTML + assert_not_nil doc.find(:tag => "div", :children => { :count => 1, :only => { :tag => "table" } }) + end + + def test_parse_cdata + doc = HTML::Document.new(<<-HTML) + + + + <![CDATA[<br>]]> + + +

this document has <br> for a title

+ + +HTML + + assert_nil doc.find(:tag => "title", :descendant => { :tag => "br" }) + assert doc.find(:tag => "title", :child => "
") + end + + def test_find_empty_tag + doc = HTML::Document.new("
") + assert_nil doc.find(:tag => "div", :attributes => { :id => "map" }, :content => /./) + assert doc.find(:tag => "div", :attributes => { :id => "map" }, :content => /\A\Z/) + assert doc.find(:tag => "div", :attributes => { :id => "map" }, :content => /^$/) + assert doc.find(:tag => "div", :attributes => { :id => "map" }, :content => "") + assert doc.find(:tag => "div", :attributes => { :id => "map" }, :content => nil) + end +end diff --git a/actionpack/test/controller/html-scanner/node_test.rb b/actionpack/test/controller/html-scanner/node_test.rb new file mode 100644 index 0000000000..1cf0a4bb67 --- /dev/null +++ b/actionpack/test/controller/html-scanner/node_test.rb @@ -0,0 +1,69 @@ +require File.dirname(__FILE__) + '/../../abstract_unit' +require 'test/unit' + +class NodeTest < Test::Unit::TestCase + + class MockNode + def initialize(matched, value) + @matched = matched + @value = value + end + + def find(conditions) + @matched && self + end + + def to_s + @value.to_s + end + end + + def setup + @node = HTML::Node.new("parent") + @node.children.concat [MockNode.new(false,1), MockNode.new(true,"two"), MockNode.new(false,:three)] + end + + def test_match + assert !@node.match("foo") + end + + def test_tag + assert !@node.tag? + end + + def test_to_s + assert_equal "1twothree", @node.to_s + end + + def test_find + assert_equal "two", @node.find('blah').to_s + end + + def test_parse_strict + s = "" + assert_raise(RuntimeError) { HTML::Node.parse(nil,0,0,s) } + end + + def test_parse_relaxed + s = "" + node = nil + assert_nothing_raised { node = HTML::Node.parse(nil,0,0,s,false) } + assert node.attributes.has_key?("foo") + assert !node.attributes.has_key?("bar") + end + + def test_to_s_with_boolean_attrs + s = "" + node = HTML::Node.parse(nil,0,0,s) + assert node.attributes.has_key?("foo") + assert node.attributes.has_key?("bar") + assert "", node.to_s + end + + def test_parse_with_unclosed_tag + s = "") + assert_equal "tag", node.name + assert_equal Hash.new, node.attributes + assert_nil node.closing + end + + def test_open_with_attributes + node = tag("") + assert_equal "tag1", node.name + assert_equal "hey_ho", node["foo"] + assert_equal "blah blah", node["x:bar"] + assert_equal "blah blah blah", node["baz"] + end + + def test_self_closing_without_attributes + node = tag("") + assert_equal "tag", node.name + assert_equal Hash.new, node.attributes + assert_equal :self, node.closing + end + + def test_self_closing_with_attributes + node = tag("") + assert_equal "tag", node.name + assert_equal( { "a" => "b" }, node.attributes ) + assert_equal :self, node.closing + end + + def test_closing_without_attributes + node = tag("") + assert_equal "tag", node.name + assert_nil node.attributes + assert_equal :close, node.closing + end + + def test_bracket_op_when_no_attributes + node = tag("") + assert_nil node["foo"] + end + + def test_bracket_op_when_attributes + node = tag("") + assert_equal "b", node["a"] + end + + def test_attributes_with_escaped_quotes + node = tag("") + assert_equal "b\\'c", node["a"] + assert_equal "bob \\\"float\\\"", node["b"] + end + + def test_to_s + node = tag("") + assert_equal %(), node.to_s + end + + def test_tag + assert tag("").tag? + end + + def test_match_tag_as_string + assert tag("").match(:tag => "tag") + assert !tag("").match(:tag => "b") + end + + def test_match_tag_as_regexp + assert tag("").match(:tag => /t.g/) + assert !tag("").match(:tag => /t[bqs]g/) + end + + def test_match_attributes_as_string + t = tag("") + assert t.match(:attributes => {"a" => "something"}) + assert t.match(:attributes => {"b" => "else"}) + end + + def test_match_attributes_as_regexp + t = tag("") + assert t.match(:attributes => {"a" => /^something$/}) + assert t.match(:attributes => {"b" => /e.*e/}) + assert t.match(:attributes => {"a" => /me..i/, "b" => /.ls.$/}) + end + + def test_match_attributes_as_number + t = tag("") + assert t.match(:attributes => {"a" => 15}) + assert t.match(:attributes => {"b" => 3.1415}) + assert t.match(:attributes => {"a" => 15, "b" => 3.1415}) + end + + def test_match_attributes_exist + t = tag("") + assert t.match(:attributes => {"a" => true}) + assert t.match(:attributes => {"b" => true}) + assert t.match(:attributes => {"a" => true, "b" => true}) + end + + def test_match_attributes_not_exist + t = tag("") + assert t.match(:attributes => {"c" => false}) + assert t.match(:attributes => {"c" => nil}) + assert t.match(:attributes => {"a" => true, "c" => false}) + end + + def test_match_parent_success + t = tag("", tag("")) + assert t.match(:parent => {:tag => "foo", :attributes => {"k" => /v.l/, "j" => false}}) + end + + def test_match_parent_fail + t = tag("", tag("")) + assert !t.match(:parent => {:tag => /kafka/}) + end + + def test_match_child_success + t = tag("") + tag("", t) + tag("", t) + assert t.match(:child => { :tag => "sib", :attributes => {"v" => /j/}}) + assert t.match(:child => { :attributes => {"a" => "kelly"}}) + end + + def test_match_child_fail + t = tag("") + tag("", t) + tag("", t) + assert !t.match(:child => { :tag => "sib", :attributes => {"v" => /r/}}) + assert !t.match(:child => { :attributes => {"v" => false}}) + end + + def test_match_ancestor_success + t = tag("", tag("", tag(""))) + assert t.match(:ancestor => {:tag => "parent", :attributes => {"a" => /ll/}}) + assert t.match(:ancestor => {:attributes => {"m" => "vaughn"}}) + end + + def test_match_ancestor_fail + t = tag("", tag("", tag(""))) + assert !t.match(:ancestor => {:tag => /^parent/, :attributes => {"v" => /m/}}) + assert !t.match(:ancestor => {:attributes => {"v" => false}}) + end + + def test_match_descendant_success + tag("", tag("", t = tag(""))) + assert t.match(:descendant => {:tag => "child", :attributes => {"a" => /ll/}}) + assert t.match(:descendant => {:attributes => {"m" => "vaughn"}}) + end + + def test_match_descendant_fail + tag("", tag("", t = tag(""))) + assert !t.match(:descendant => {:tag => /^child/, :attributes => {"v" => /m/}}) + assert !t.match(:descendant => {:attributes => {"v" => false}}) + end + + def test_match_child_count + t = tag("") + tag("hello", t) + tag("", t) + tag("", t) + assert t.match(:children => { :count => 2 }) + assert t.match(:children => { :count => 2..4 }) + assert t.match(:children => { :less_than => 4 }) + assert t.match(:children => { :greater_than => 1 }) + assert !t.match(:children => { :count => 3 }) + end + + def test_conditions_as_strings + t = tag("") + assert t.match("tag" => "tag") + assert t.match("attributes" => { "x:k" => "something" }) + assert !t.match("tag" => "gat") + assert !t.match("attributes" => { "x:j" => "something" }) + end + + def test_attributes_as_symbols + t = tag("") + assert t.match(:attributes => { :v => /oh/ }) + assert t.match(:attributes => { :a => /ll/ }) + end + + def test_match_sibling + t = tag("") + tag("hello", t) + tag("", t) + tag("world", t) + m = tag("", t) + tag("", t) + + assert m.match(:sibling => {:tag => "span", :attributes => {:a => true}}) + assert m.match(:sibling => {:tag => "span", :attributes => {:m => true}}) + assert !m.match(:sibling => {:tag => "span", :attributes => {:k => true}}) + end + + def test_match_sibling_before + t = tag("") + tag("hello", t) + tag("", t) + tag("world", t) + m = tag("", t) + tag("", t) + + assert m.match(:before => {:tag => "span", :attributes => {:m => true}}) + assert !m.match(:before => {:tag => "span", :attributes => {:a => true}}) + assert !m.match(:before => {:tag => "span", :attributes => {:k => true}}) + end + + def test_match_sibling_after + t = tag("") + tag("hello", t) + tag("", t) + tag("world", t) + m = tag("", t) + tag("", t) + + assert m.match(:after => {:tag => "span", :attributes => {:a => true}}) + assert !m.match(:after => {:tag => "span", :attributes => {:m => true}}) + assert !m.match(:after => {:tag => "span", :attributes => {:k => true}}) + end + + def test_to_s + t = tag("") + tag("hello", t) + tag("
", t) + assert_equal %(hello
), t.to_s + end + + private + + def tag(content, parent=nil) + node = HTML::Node.parse(parent,0,0,content) + parent.children << node if parent + node + end +end diff --git a/actionpack/test/controller/html-scanner/text_node_test.rb b/actionpack/test/controller/html-scanner/text_node_test.rb new file mode 100644 index 0000000000..9853701fda --- /dev/null +++ b/actionpack/test/controller/html-scanner/text_node_test.rb @@ -0,0 +1,51 @@ +require File.dirname(__FILE__) + '/../../abstract_unit' +require 'test/unit' + +class TextNodeTest < Test::Unit::TestCase + def setup + @node = HTML::Text.new(nil, 0, 0, "hello, howdy, aloha, annyeong") + end + + def test_to_s + assert_equal "hello, howdy, aloha, annyeong", @node.to_s + end + + def test_find_string + assert_equal @node, @node.find("hello, howdy, aloha, annyeong") + assert_equal false, @node.find("bogus") + end + + def test_find_regexp + assert_equal @node, @node.find(/an+y/) + assert_nil @node.find(/b/) + end + + def test_find_hash + assert_equal @node, @node.find(:content => /howdy/) + assert_nil @node.find(:content => /^howdy$/) + assert_equal false, @node.find(:content => "howdy") + end + + def test_find_other + assert_nil @node.find(:hello) + end + + def test_match_string + assert @node.match("hello, howdy, aloha, annyeong") + assert_equal false, @node.match("bogus") + end + + def test_match_regexp + assert_not_nil @node, @node.match(/an+y/) + assert_nil @node.match(/b/) + end + + def test_match_hash + assert_not_nil @node, @node.match(:content => "howdy") + assert_nil @node.match(:content => /^howdy$/) + end + + def test_match_other + assert_nil @node.match(:hello) + end +end \ No newline at end of file diff --git a/actionpack/test/controller/html-scanner/tokenizer_test.rb b/actionpack/test/controller/html-scanner/tokenizer_test.rb new file mode 100644 index 0000000000..437136b9a7 --- /dev/null +++ b/actionpack/test/controller/html-scanner/tokenizer_test.rb @@ -0,0 +1,125 @@ +require File.dirname(__FILE__) + '/../../abstract_unit' +require 'test/unit' + +class TokenizerTest < Test::Unit::TestCase + + def test_blank + tokenize "" + assert_end + end + + def test_space + tokenize " " + assert_next " " + assert_end + end + + def test_tag_simple_open + tokenize "" + assert_next "" + assert_end + end + + def test_tag_simple_self_closing + tokenize "" + assert_next "" + assert_end + end + + def test_tag_simple_closing + tokenize "" + assert_next "" + end + + def test_tag_with_single_quoted_attribute + tokenize %{x} + assert_next %{} + end + + def test_tag_with_single_quoted_attribute_with_escape + tokenize %{x} + assert_next %{} + end + + def test_tag_with_double_quoted_attribute + tokenize %{x} + assert_next %{} + end + + def test_tag_with_double_quoted_attribute_with_escape + tokenize %{x} + assert_next %{} + end + + def test_tag_with_unquoted_attribute + tokenize %{x} + assert_next %{} + end + + def test_tag_with_lt_char_in_attribute + tokenize %{x} + assert_next %{} + end + + def test_tag_with_gt_char_in_attribute + tokenize %{x} + assert_next %{} + end + + def test_doctype_tag + tokenize %{\n } + assert_next %{} + assert_next %{\n } + assert_next %{} + end + + def test_cdata_tag + tokenize %{]]>} + assert_next %{]]>} + assert_end + end + + def test_less_than_with_space + tokenize %{original < hello > world} + assert_next %{original } + assert_next %{< hello > world} + end + + def test_less_than_without_matching_greater_than + tokenize %{hello foo\nbar} + assert_next %{hello } + assert_next %{} + assert_next %{foo} + assert_next %{} + assert_next %{\nbar} + assert_next %{} + assert_end + end + + def test_unterminated_comment + tokenize %{hello