nokogiri 1.0.0-x86-mswin32-60

data/test/css/test_parser.rb

@@ -0,0 +1,224 @@
+require File.expand_path(File.join(File.dirname(__FILE__), '..', "helper"))
+
+module Nokogiri
+  module CSS
+    class TestParser < Nokogiri::TestCase
+      def setup
+        @parser = Nokogiri::CSS::Parser.new
+      end
+
+      def test_find_by_type
+        ast = @parser.parse("a:nth-child(2)").first
+        matches = ast.find_by_type(
+          [:CONDITIONAL_SELECTOR,
+            [:ELEMENT_NAME],
+            [:PSEUDO_CLASS,
+              [:FUNCTION]
+            ]
+          ]
+        )
+        assert_equal(1, matches.length)
+        assert_equal(ast, matches.first)
+      end
+
+      def test_to_type
+        ast = @parser.parse("a:nth-child(2)").first
+        assert_equal(
+          [:CONDITIONAL_SELECTOR,
+            [:ELEMENT_NAME],
+            [:PSEUDO_CLASS,
+              [:FUNCTION]
+            ]
+          ], ast.to_type
+        )
+      end
+
+      def test_to_a
+        asts = @parser.parse("a:nth-child(2)")
+        assert_equal(
+          [:CONDITIONAL_SELECTOR,
+            [:ELEMENT_NAME, ["a"]],
+            [:PSEUDO_CLASS,
+              [:FUNCTION, ["nth-child("], ["2"]]
+            ]
+          ], asts.first.to_a
+        )
+      end
+
+      def test_function_with_arguments
+        assert_xpath  "//*[position() = 2 and self::a]",
+                      @parser.parse("a[2]")
+        assert_xpath  "//*[position() = 2 and self::a]",
+                      @parser.parse("a:nth-child(2)")
+      end
+
+      def test_carrot
+        assert_xpath  "//a[starts-with(@id, 'Boing')]",
+                      @parser.parse("a[id^='Boing']")
+      end
+
+      def test_attributes_with_at
+        ## This is non standard CSS
+        assert_xpath  "//a[@id = 'Boing']",
+                      @parser.parse("a[@id='Boing']")
+      end
+
+      def test_not_equal
+        ## This is non standard CSS
+        assert_xpath  "//a[child::text() != 'Boing']",
+                      @parser.parse("a[text()!='Boing']")
+      end
+
+      def test_function
+        ## This is non standard CSS
+        assert_xpath  "//a[child::text()]",
+                      @parser.parse("a[text()]")
+
+        ## This is non standard CSS
+        assert_xpath  "//child::text()",
+                      @parser.parse("text()")
+
+        ## This is non standard CSS
+        assert_xpath  "//a[contains(child::text(), 'Boing')]",
+                      @parser.parse("a[text()*='Boing']")
+
+        ## This is non standard CSS
+        assert_xpath  "//script//comment()",
+                      @parser.parse("script comment()")
+      end
+
+      def test_nonstandard_nth_selectors
+        ## These are non standard CSS
+        assert_xpath '//a[position() = 99]', @parser.parse('a:eq(99)')
+        assert_xpath '//a[position() = 1]', @parser.parse('a:first') # no parens
+        assert_xpath '//a[position() = last()]', @parser.parse('a:last') # no parens
+        assert_xpath '//a[position() = 99]', @parser.parse('a:nth(99)')
+        assert_xpath '//a[position() = 1]', @parser.parse('a:first()')
+        assert_xpath '//a[position() = last()]', @parser.parse('a:last()')
+        assert_xpath '//a[node()]', @parser.parse('a:parent')
+      end
+
+      def test_standard_nth_selectors
+        assert_xpath '//a[position() = 99]', @parser.parse('a:nth-of-type(99)')
+        assert_xpath '//a[position() = 1]', @parser.parse('a:first-of-type()')
+        assert_xpath '//a[position() = last()]', @parser.parse('a:last-of-type()')
+        assert_xpath '//a[position() = 1]', @parser.parse('a:first-of-type') # no parens
+        assert_xpath '//a[position() = last()]', @parser.parse('a:last-of-type') # no parens
+        assert_xpath '//a[position() = last() - 99]', @parser.parse('a:nth-last-of-type(99)')
+        assert_xpath '//a[position() = last() - 99]', @parser.parse('a:nth-last-of-type(99)')
+      end
+
+      def test_nth_child_selectors
+        assert_xpath '//*[position() = 1 and self::a]', @parser.parse('a:first-child')
+        assert_xpath '//*[position() = last() and self::a]', @parser.parse('a:last-child')
+        assert_xpath '//*[position() = 99 and self::a]', @parser.parse('a:nth-child(99)')
+        assert_xpath '//*[position() = last() - 99 and self::a]', @parser.parse('a:nth-last-child(99)')
+      end
+
+      def test_miscellaneous_selectors
+        assert_xpath '//*[last() = 1 and self::a]',
+          @parser.parse('a:only-child')
+        assert_xpath '//a[last() = 1]', @parser.parse('a:only-of-type')
+        assert_xpath '//a[not(node())]', @parser.parse('a:empty')
+      end
+
+      def test_nth_a_n_plus_b
+        assert_xpath '//a[(position() mod 2) = 0]', @parser.parse('a:nth-of-type(2n)')
+        assert_xpath '//a[(position() >= 1) and (((position()-1) mod 2) = 0)]', @parser.parse('a:nth-of-type(2n+1)')
+        assert_xpath '//a[(position() mod 2) = 0]', @parser.parse('a:nth-of-type(even)')
+        assert_xpath '//a[(position() >= 1) and (((position()-1) mod 2) = 0)]', @parser.parse('a:nth-of-type(odd)')
+        assert_xpath '//a[(position() >= 3) and (((position()-3) mod 4) = 0)]', @parser.parse('a:nth-of-type(4n+3)')
+        assert_xpath '//a[(position() <= 3) and (((position()-3) mod 1) = 0)]', @parser.parse('a:nth-of-type(-1n+3)')
+        assert_xpath '//a[(position() <= 3) and (((position()-3) mod 1) = 0)]', @parser.parse('a:nth-of-type(-n+3)')
+        assert_xpath '//a[(position() >= 3) and (((position()-3) mod 1) = 0)]', @parser.parse('a:nth-of-type(1n+3)')
+        assert_xpath '//a[(position() >= 3) and (((position()-3) mod 1) = 0)]', @parser.parse('a:nth-of-type(n+3)')
+      end
+
+      def test_preceding_selector
+        assert_xpath  "//F[preceding-sibling::E]",
+                      @parser.parse("E ~ F")
+      end
+
+      def test_direct_preceding_selector
+        assert_xpath  "//E/following-sibling::*[1]/self::F",
+                      @parser.parse("E + F")
+      end
+
+      def test_attribute
+        assert_xpath  "//h1[@a = 'Tender Lovemaking']",
+                      @parser.parse("h1[a='Tender Lovemaking']")
+      end
+
+      def test_id
+        assert_xpath "//*[@id = 'foo']", @parser.parse('#foo')
+      end
+
+      def test_pseudo_class_no_ident
+        assert_xpath "//*[1 = 1]", @parser.parse(':link')
+      end
+
+      def test_pseudo_class
+        assert_xpath "//a[1 = 1]", @parser.parse('a:link')
+        assert_xpath "//a[1 = 1]", @parser.parse('a:visited')
+        assert_xpath "//a[1 = 1]", @parser.parse('a:hover')
+        assert_xpath "//a[1 = 1]", @parser.parse('a:active')
+        assert_xpath  "//a[1 = 1 and contains(concat(' ', @class, ' '), ' foo ')]",
+                      @parser.parse('a:active.foo')
+      end
+
+      def test_star
+        assert_xpath "//*", @parser.parse('*')
+        assert_xpath "//*[contains(concat(' ', @class, ' '), ' pastoral ')]",
+                      @parser.parse('*.pastoral')
+      end
+
+      def test_class
+        assert_xpath  "//*[contains(concat(' ', @class, ' '), ' a ') and contains(concat(' ', @class, ' '), ' b ')]",
+                      @parser.parse('.a.b')
+        assert_xpath  "//*[contains(concat(' ', @class, ' '), ' awesome ')]",
+                      @parser.parse('.awesome')
+        assert_xpath  "//foo[contains(concat(' ', @class, ' '), ' awesome ')]",
+                      @parser.parse('foo.awesome')
+        assert_xpath  "//foo//*[contains(concat(' ', @class, ' '), ' awesome ')]",
+                      @parser.parse('foo .awesome')
+      end
+
+      def test_ident
+        assert_xpath '//x', @parser.parse('x')
+      end
+
+      def test_parse_space
+        assert_xpath '//x//y', @parser.parse('x y')
+      end
+
+      def test_parse_descendant
+        assert_xpath '//x/y', @parser.parse('x > y')
+      end
+
+      def test_parse_slash
+        ## This is non standard CSS
+        assert_xpath '//x/y', @parser.parse('x/y')
+      end
+
+      def test_parse_doubleslash
+        ## This is non standard CSS
+        assert_xpath '//x//y', @parser.parse('x//y')
+      end
+
+      def test_multi_path
+        assert_xpath ['//x/y', '//y/z'], @parser.parse('x > y, y > z')
+        assert_xpath ['//x/y', '//y/z'], @parser.parse('x > y,y > z')
+        ###
+        # TODO: should we make this work?
+        # assert_xpath ['//x/y', '//y/z'], @parser.parse('x > y | y > z')
+      end
+
+      def assert_xpath expecteds, asts
+        expecteds = [expecteds].flatten
+        expecteds.zip(asts).each do |expected, actual|
+          assert_equal expected, actual.to_xpath
+        end
+      end
+    end
+  end
+end

data/test/css/test_tokenizer.rb

@@ -0,0 +1,162 @@
+require File.expand_path(File.join(File.dirname(__FILE__), '..', "helper"))
+
+module Nokogiri
+  module CSS
+    class TestTokenizer < Nokogiri::TestCase
+      def setup
+        @scanner = Nokogiri::CSS::Tokenizer.new
+      end
+
+      def test_not_equal
+        @scanner.scan("h1[a!='Tender Lovemaking']")
+        assert_tokens([ [:IDENT, 'h1'],
+                        ['[', '['],
+                        [:IDENT, 'a'],
+                        [:NOT_EQUAL, '!='],
+                        [:STRING, "'Tender Lovemaking'"],
+                        [']', ']'],
+        ], @scanner)
+      end
+
+      def test_function
+        @scanner.scan("script comment()")
+        assert_tokens([ [:IDENT, 'script'],
+                        [:S, ' '],
+                        [:FUNCTION, 'comment('],
+                        [')', ')'],
+        ], @scanner)
+      end
+
+      def test_preceding_selector
+        @scanner.scan("E ~ F")
+        assert_tokens([ [:IDENT, 'E'],
+                        [:TILDE, ' ~'],
+                        [:S, ' '],
+                        [:IDENT, 'F'],
+        ], @scanner)
+      end
+
+      def test_scan_attribute_string
+        @scanner.scan("h1[a='Tender Lovemaking']")
+        assert_tokens([ [:IDENT, 'h1'],
+                        ['[', '['],
+                        [:IDENT, 'a'],
+                        ['=', '='],
+                        [:STRING, "'Tender Lovemaking'"],
+                        [']', ']'],
+        ], @scanner)
+        @scanner.scan('h1[a="Tender Lovemaking"]')
+        assert_tokens([ [:IDENT, 'h1'],
+                        ['[', '['],
+                        [:IDENT, 'a'],
+                        ['=', '='],
+                        [:STRING, '"Tender Lovemaking"'],
+                        [']', ']'],
+        ], @scanner)
+      end
+
+      def test_scan_id
+        @scanner.scan('#foo')
+        assert_tokens([ [:HASH, '#foo'] ], @scanner)
+      end
+
+      def test_scan_pseudo
+        @scanner.scan('a:visited')
+        assert_tokens([ [:IDENT, 'a'],
+                        [':', ':'],
+                        [:IDENT, 'visited']
+        ], @scanner)
+      end
+
+      def test_scan_star
+        @scanner.scan('*')
+        assert_tokens([ ['*', '*'], ], @scanner)
+      end
+
+      def test_scan_class
+        @scanner.scan('x.awesome')
+        assert_tokens([ [:IDENT, 'x'],
+                        ['.', '.'],
+                        [:IDENT, 'awesome'],
+        ], @scanner)
+      end
+
+      def test_scan_greater
+        @scanner.scan('x > y')
+        assert_tokens([ [:IDENT, 'x'],
+                        [:GREATER, ' >'],
+                        [:S, ' '],
+                        [:IDENT, 'y']
+        ], @scanner)
+      end
+
+      def test_scan_slash
+        @scanner.scan('x/y')
+        assert_tokens([ [:IDENT, 'x'],
+                        [:SLASH, '/'],
+                        [:IDENT, 'y']
+        ], @scanner)
+      end
+
+      def test_scan_doubleslash
+        @scanner.scan('x//y')
+        assert_tokens([ [:IDENT, 'x'],
+                        [:DOUBLESLASH, '//'],
+                        [:IDENT, 'y']
+        ], @scanner)
+      end
+
+      def test_scan_function_selector
+        @scanner.scan('x:eq(0)')
+        assert_tokens([ [:IDENT, 'x'],
+                        [':', ':'],
+                        [:FUNCTION, 'eq('],
+                        [:NUMBER, "0"],
+                        [")", ")"]
+        ], @scanner)
+      end
+
+      def test_scan_an_plus_b
+        @scanner.scan('x:nth-child(5n+3)')
+        assert_tokens([ [:IDENT, 'x'],
+                        [':', ':'],
+                        [:FUNCTION, 'nth-child('],
+                        [:NUMBER, '5'],
+                        [:IDENT, 'n'],
+                        [:PLUS, '+'],
+                        [:NUMBER, '3'],
+                        [")", ")"]
+        ], @scanner)
+
+        @scanner.scan('x:nth-child(-1n+3)')
+        assert_tokens([ [:IDENT, 'x'],
+                        [':', ':'],
+                        [:FUNCTION, 'nth-child('],
+                        [:NUMBER, '-1'],
+                        [:IDENT, 'n'],
+                        [:PLUS, '+'],
+                        [:NUMBER, '3'],
+                        [")", ")"]
+        ], @scanner)
+
+        @scanner.scan('x:nth-child(-n+3)')
+        assert_tokens([ [:IDENT, 'x'],
+                        [':', ':'],
+                        [:FUNCTION, 'nth-child('],
+                        [:IDENT, '-n'],
+                        [:PLUS, '+'],
+                        [:NUMBER, '3'],
+                        [")", ")"]
+        ], @scanner)
+      end
+
+      def assert_tokens(tokens, scanner)
+        toks = []
+        while tok = @scanner.next_token
+          toks << tok
+        end
+        assert_equal(tokens, toks)
+      end
+    end
+  end
+end

data/test/css/test_xpath_visitor.rb

@@ -0,0 +1,54 @@
+require File.expand_path(File.join(File.dirname(__FILE__), '..', "helper"))
+
+module Nokogiri
+  module CSS
+    class TestXPathVisitor < Nokogiri::TestCase
+      def setup
+        @parser = Nokogiri::CSS::Parser.new
+      end
+
+      def test_class_selectors
+        assert_xpath  "//*[contains(concat(' ', @class, ' '), ' red ')]",
+                      @parser.parse(".red")
+      end
+
+      def test_pipe
+        assert_xpath  "//a[@id = 'Boing' or starts-with(@id, concat('Boing', '-'))]",
+                      @parser.parse("a[id|='Boing']")
+      end
+
+      def test_custom_functions
+        visitor = Class.new(XPathVisitor) do
+          attr_accessor :awesome
+          def visit_function_aaron node
+            @awesome = true
+            'aaron() = 1'
+          end
+        end.new
+        ast = @parser.parse('a:aaron()').first
+        assert_equal 'a[aaron() = 1]', visitor.accept(ast)
+        assert visitor.awesome
+      end
+
+      def test_custom_psuedo_classes
+        visitor = Class.new(XPathVisitor) do
+          attr_accessor :awesome
+          def visit_pseudo_class_aaron node
+            @awesome = true
+            'aaron() = 1'
+          end
+        end.new
+        ast = @parser.parse('a:aaron').first
+        assert_equal 'a[aaron() = 1]', visitor.accept(ast)
+        assert visitor.awesome
+      end
+
+      def assert_xpath expecteds, asts
+        expecteds = [expecteds].flatten
+        expecteds.zip(asts).each do |expected, actual|
+          assert_equal expected, actual.to_xpath
+        end
+      end
+    end
+  end
+end

data/test/files/staff.xml

@@ -0,0 +1,59 @@
+<?xml version="1.0"?><?TEST-STYLE PIDATA?>
+<!DOCTYPE staff SYSTEM "staff.dtd" [
+   <!ENTITY ent1 "es">
+   <!ENTITY ent2 "1900 Dallas Road">
+   <!ENTITY ent3 "Texas">
+   <!ENTITY ent4 "<entElement domestic='Yes'>Element data</entElement><?PItarget PIdata?>">
+   <!ENTITY ent5 PUBLIC "entityURI" "entityFile" NDATA notation1>
+   <!ENTITY ent1 "This entity should be discarded">
+   <!ELEMENT br EMPTY>
+   <!ATTLIST br width CDATA "0">
+   <!NOTATION notation1 PUBLIC "notation1File">
+   <!NOTATION notation2 SYSTEM "notation2File">
+]>
+<!-- This is comment number 1.-->
+<staff>
+ <employee>
+  <employeeId>EMP0001</employeeId>
+  <name>Margaret Martin</name>
+  <position>Accountant</position>           
+  <salary>56,000</salary>
+  <gender>Female</gender>
+  <address domestic="Yes">1230 North Ave. Dallas, Texas 98551</address>
+ </employee>
+ <employee>
+  <employeeId>EMP0002</employeeId>
+  <name>Martha Raynolds<![CDATA[This is a CDATASection with EntityReference number 2 &ent2;]]>
+<![CDATA[This is an adjacent CDATASection with a reference to a tab &tab;]]></name>
+  <position>Secretary</position>
+  <salary>35,000</salary>
+  <gender>Female</gender>
+  <address domestic="Yes" street="Yes">&ent2; Dallas, &ent3;
+ 98554</address>
+ </employee>
+ <employee>
+  <employeeId>EMP0003</employeeId>
+  <name>Roger
+ Jones</name>
+  <position>Department Manager</position>
+  <salary>100,000</salary>
+  <gender>&ent4;</gender>
+  <address domestic="Yes" street="No">PO Box 27 Irving, texas 98553</address>
+ </employee>
+ <employee>
+  <employeeId>EMP0004</employeeId>
+  <name>Jeny Oconnor</name>
+  <position>Personnel Director</position>
+  <salary>95,000</salary>
+  <gender>Female</gender>
+  <address domestic="Yes" street="Y&ent1;">27 South Road. Dallas, Texas 98556</address>
+ </employee>
+ <employee>
+  <employeeId>EMP0005</employeeId>
+  <name>Robert Myers</name>
+  <position>Computer Specialist</position>
+  <salary>90,000</salary>
+  <gender>male</gender>
+  <address street="Yes">1821 Nordic. Road, Irving Texas 98558</address>
+ </employee>
+ </staff>