xpath 0.1.0 → 0.1.1

data/lib/xpath.rb

@@ -1,3 +1,5 @@
+require 'nokogiri'
+
 module XPath
   autoload :Expression, 'xpath/expression'
   autoload :Union, 'xpath/union'
@@ -45,6 +47,17 @@ module XPath
     Expression::Variable.new(name)
   end
 
+  def string
+    Expression::StringFunction.new(current)
+  end
+
+  def css(selector)
+    paths = Nokogiri::CSS.xpath_for(selector).map do |selector|
+      Expression::CSS.new(current, Expression::Literal.new(selector))
+    end
+    Union.new(*paths)
+  end
+
   def varstring(name)
     var(name).string_literal
   end

data/lib/xpath/expression.rb

@@ -25,9 +25,6 @@ module XPath
       def initialize(left, expressions)
         @left = wrap_xpath(left)
         @expressions = expressions.map { |e| wrap_xpath(e) }
-        if @expressions.empty?
-          raise ArgumentError, "must specify at least one expression"
-        end
       end
     end
 
@@ -45,8 +42,10 @@ module XPath
       def to_xpath(predicate=nil)
         if @expressions.length == 1
           "#{@left.to_xpath(predicate)}/#{@expressions.first.to_xpath(predicate)}"
-        else
+        elsif @expressions.length > 1
           "#{@left.to_xpath(predicate)}/*[#{@expressions.map { |e| "self::#{e.to_xpath(predicate)}" }.join(" | ")}]"
+        else
+          "#{@left.to_xpath(predicate)}/*"
         end
       end
     end
@@ -55,8 +54,10 @@ module XPath
       def to_xpath(predicate=nil)
         if @expressions.length == 1
           "#{@left.to_xpath(predicate)}//#{@expressions.first.to_xpath(predicate)}"
-        else
+        elsif @expressions.length > 1
           "#{@left.to_xpath(predicate)}//*[#{@expressions.map { |e| "self::#{e.to_xpath(predicate)}" }.join(" | ")}]"
+        else
+          "#{@left.to_xpath(predicate)}//*"
         end
       end
     end
@@ -65,8 +66,10 @@ module XPath
       def to_xpath(predicate=nil)
         if @expressions.length == 1
           "#{@left.to_xpath(predicate)}/following-sibling::*[1]/self::#{@expressions.first.to_xpath(predicate)}"
-        else
+        elsif @expressions.length > 1
           "#{@left.to_xpath(predicate)}/following-sibling::*[1]/self::*[#{@expressions.map { |e| "self::#{e.to_xpath(predicate)}" }.join(" | ")}]"
+        else
+          "#{@left.to_xpath(predicate)}/following-sibling::*[1]/self::*"
         end
       end
     end
@@ -211,6 +214,12 @@ module XPath
       end
     end
 
+    class CSS < Binary
+      def to_xpath(predicate=nil)
+        "#{@left.to_xpath}#{@right.to_xpath}"
+      end
+    end
+
     def current
       self
     end
@@ -237,10 +246,6 @@ module XPath
       Expression::Is.new(current, expression)
     end
 
-    def string
-      Expression::StringFunction.new(current)
-    end
-
     def or(expression)
       Expression::Or.new(current, expression)
     end

data/lib/xpath/html.rb

@@ -3,13 +3,9 @@ module XPath
     include XPath
     extend self
 
-    def from_css(css)
-      XPath::Union.new(*Nokogiri::CSS.xpath_for(css).map { |selector| ::XPath::Expression::Literal.new(:".#{selector}") }.flatten)
-    end
-
     def link(locator)
       link = descendant(:a)[attr(:href)]
-      link[attr(:id).equals(locator) | text.is(locator) | attr(:title).is(locator) | descendant(:img)[attr(:alt).is(locator)]]
+      link[attr(:id).equals(locator) | string.n.is(locator) | attr(:title).is(locator) | descendant(:img)[attr(:alt).is(locator)]]
     end
 
     def content(locator)
@@ -18,7 +14,7 @@ module XPath
 
     def button(locator)
       button = descendant(:input)[attr(:type).one_of('submit', 'image', 'button')][attr(:id).equals(locator) | attr(:value).is(locator)]
-      button += descendant(:button)[attr(:id).equals(locator) | attr(:value).is(locator) | text.is(locator)]
+      button += descendant(:button)[attr(:id).equals(locator) | attr(:value).is(locator) | string.n.is(locator)]
       button += descendant(:input)[attr(:type).equals('image')][attr(:alt).is(locator)]
     end
 
@@ -86,7 +82,7 @@ module XPath
     end
 
     def table_rows(rows)
-      row_conditions = descendant(:tr)[table_row(rows.first)] 
+      row_conditions = descendant(:tr)[table_row(rows.first)]
       rows.drop(1).each do |row|
         row_conditions = row_conditions.next_sibling(:tr)[table_row(row)]
       end
@@ -94,21 +90,13 @@ module XPath
     end
 
     def table_row(cells)
-      cell_conditions = child(:td, :th)[text.equals(cells.first)] 
+      cell_conditions = child(:td, :th)[text.equals(cells.first)]
       cells.drop(1).each do |cell|
         cell_conditions = cell_conditions.next_sibling(:td, :th)[text.equals(cell)]
       end
       cell_conditions
     end
 
-    def wrap(path)
-      if path.respond_to?(:to_xpaths)
-        path.to_xpaths
-      else
-        [path.to_s].flatten
-      end
-    end
-
   protected
 
     def locate_field(xpath, locator)

data/lib/xpath/version.rb

@@ -1,3 +1,3 @@
 module XPath
-  VERSION = '0.1.0'
+  VERSION = '0.1.1'
 end

data/spec/fixtures/form.html

@@ -1,5 +1,62 @@
 <h1>Form</h1>
 
+<p>
+  <a id="awesome-link" href="#link">An awesome link</a>
+  <a id="some-id" href="#id">With id</a>
+  <a title="My title" href="#title">A cool title</a>
+  <a href="#image"><img src="foo.png" alt="Alt link"/></a>
+  <a title="This title is too long" href="#bar">A link at a time</a>
+  <a title="This title" href="#foo">A link</a>
+  <a href="#bar"><img src="foo.png" alt="An image that is beautiful"/></a>
+  <a href="#foo"><img src="foo.ong" alt="An image"/></a>
+  <a>Wrong Link</a>
+  <a href="#spacey">    My
+
+      whitespaced
+    link</a>
+  <a href="#has-children">
+    An <em>emphatic</em> link with some children
+  </a>
+</p>
+
+<p>
+  <input type="submit" id="submit-with-id" data="id-submit" value="Has ID"/>
+  <input type="submit" value="submit-with-value" data="value-submit"/>
+  <input type="submit" value="not exact value submit" data="not-exact-value-submit"/>
+  <input type="submit" value="exact value submit" data="exact-value-submit"/>
+
+  <input type="button" id="button-with-id" data="id-button" value="Has ID"/>
+  <input type="button" value="button-with-value" data="value-button"/>
+  <input type="button" value="not exact value button" data="not-exact-value-button"/>
+  <input type="button" value="exact value button" data="exact-value-button"/>
+
+  <input type="image" id="imgbut-with-id" data="id-imgbut" value="Has ID"/>
+  <input type="image" value="imgbut-with-value" data="value-imgbut"/>
+  <input type="image" alt="imgbut-with-alt" data="alt-imgbut"/>
+  <input type="image" value="not exact value imgbut" data="not-exact-value-imgbut"/>
+  <input type="image" value="exact value imgbut" data="exact-value-imgbut"/>
+
+  <button id="btag-with-id" data="id-btag" value="Has ID"/>
+  <button value="btag-with-value" data="value-btag"/>
+  <button value="not exact value btag" data="not-exact-value-btag"/>
+  <button value="exact value btag" data="exact-value-btag"/>
+
+  <button data="text-btag">btag-with-text</button>
+  <button data="not-exact-text-btag">not exact text btag</button>
+  <button data="exact-text-btag">exact text btag</button>
+
+  <button data="btag-with-whitespace">    My
+
+      whitespaced
+    button</button>
+  <button href="btag-with-children">
+    An <em>emphatic</em> button with some children
+  </button>
+
+  <input type="schmoo" value="schmoo button" data="schmoo"/>
+</p>
+
+
 <form action="/form" method="post">
 
   <p>
@@ -32,7 +89,7 @@
     <label for="form_last_name">Last Name</label>
     <input type="text" name="form[last_name]" value="Smith" id="form_last_name"/>
   </p>
-  
+
   <p>
     <label for="form_name_explanation">Explanation of Name</label>
     <textarea name="form[name_explanation]" id="form_name_explanation"></textarea>
@@ -47,7 +104,7 @@
     <label for="form_schmooo">Schmooo</label>
     <input type="schmooo" name="form[schmooo]" value="This is Schmooo!" id="form_schmooo"/>
   </p>
-  
+
   <p>
     <label>Street<br/>
       <input type="text" name="form[street]" value="Sesame street 66"/>
@@ -74,7 +131,7 @@
     <label for="form_image">Image</label>
     <input type="file" name="form[image]" id="form_image"/>
   </p>
-  
+
   <p>
     <input type="hidden" name="form[token]" value="12345" id="form_token"/>
   </p>
@@ -196,7 +253,7 @@
     <label for="form_document">Document</label>
     <input type="file" name="form[document]" id="form_document"/>
   </p>
- 
+
   <p>
     <input type="submit" value="Upload"/>
   <p>
@@ -229,7 +286,7 @@
     <label for="html5_color">Html5 Color</label>
     <input type="color" name="form[html5_color]" value="#FFF" id="html5_color"/>
   </p>
-  
+
   <p>
     <input type="submit" name="form[html5_submit]" value="html5_submit"/>
   </p>

data/spec/fixtures/simple.html

@@ -24,3 +24,9 @@ A     lot
   of
       whitespace
 </p>
+
+<div id="moar">
+  <p id="impchay">chimp</p>
+  <p id="amingoflay">flamingo</p>
+  <div id="elephantay">elephant</div>
+</div>

data/spec/html_spec.rb

@@ -0,0 +1,195 @@
+require 'spec_helper'
+require 'nokogiri'
+
+describe XPath::HTML do
+  let(:template) { 'form' }
+  let(:template_path) { File.read(File.expand_path("fixtures/#{template}.html", File.dirname(__FILE__))) }
+  let(:doc) { Nokogiri::HTML(template_path) }
+
+  def get(xpath)
+    all(xpath).first
+  end
+
+  def all(xpath)
+    xpath.to_xpaths.map do |xpath|
+      doc.xpath(xpath)
+    end.flatten
+  end
+
+  describe '#link' do
+    it "finds links by id" do
+      get(XPath::HTML.link('some-id'))[:href].should == '#id'
+    end
+
+    it "finds links by content" do
+      get(XPath::HTML.link('An awesome link'))[:href].should == '#link'
+    end
+
+    it "finds links by content without caring about whitespace" do
+      get(XPath::HTML.link('My whitespaced link'))[:href].should == '#spacey'
+    end
+
+    it "finds links with child tags by content" do
+      get(XPath::HTML.link('An emphatic link'))[:href].should == '#has-children'
+      get(XPath::HTML.link('emphatic'))[:href].should == '#has-children'
+    end
+
+    it "finds links by approximate content" do
+      get(XPath::HTML.link('awesome'))[:href].should == '#link'
+    end
+
+    it "prefers exact matches of content" do
+      result = all(XPath::HTML.link('A link'))
+      result[0][:href].should == '#foo'
+      result[1][:href].should == '#bar'
+    end
+
+    it "finds links by title" do
+      get(XPath::HTML.link('My title'))[:href].should == '#title'
+    end
+
+    it "finds links by approximate title" do
+      get(XPath::HTML.link('title'))[:href].should == '#title'
+    end
+
+    it "prefers exact matches of title" do
+      result = all(XPath::HTML.link('This title'))
+      result[0][:href].should == '#foo'
+      result[1][:href].should == '#bar'
+    end
+
+    it "finds links by an image's alt attribute" do
+      get(XPath::HTML.link('Alt link'))[:href].should == '#image'
+    end
+
+    it "finds links by an image's approximate alt attribute" do
+      get(XPath::HTML.link('Alt'))[:href].should == '#image'
+    end
+
+    it "prefers exact matches of image's alt attribute" do
+      result = all(XPath::HTML.link('An image'))
+      result[0][:href].should == '#foo'
+      result[1][:href].should == '#bar'
+    end
+
+    it "does not find links without href attriutes" do
+      get(XPath::HTML.link('Wrong Link')).should be_nil
+    end
+  end
+
+  describe '#button' do
+    context "with submit type" do
+      it "finds buttons by id" do
+        get(XPath::HTML.button('submit-with-id'))[:data].should == 'id-submit'
+      end
+
+      it "finds buttons by value" do
+        get(XPath::HTML.button('submit-with-value'))[:data].should == 'value-submit'
+      end
+
+      it "finds buttons by approximate value " do
+        get(XPath::HTML.button('mit-with-val'))[:data].should == 'value-submit'
+      end
+
+      it "finds prefers buttons with exact value " do
+        results = all(XPath::HTML.button('exact value submit'))
+        results[0][:data].should == 'exact-value-submit'
+        results[1][:data].should == 'not-exact-value-submit'
+      end
+    end
+
+    context "with button type" do
+      it "finds buttons by id" do
+        get(XPath::HTML.button('button-with-id'))[:data].should == 'id-button'
+      end
+
+      it "finds buttons by value" do
+        get(XPath::HTML.button('button-with-value'))[:data].should == 'value-button'
+      end
+
+      it "finds buttons by approximate value " do
+        get(XPath::HTML.button('ton-with-val'))[:data].should == 'value-button'
+      end
+
+      it "finds prefers buttons with exact value " do
+        results = all(XPath::HTML.button('exact value button'))
+        results[0][:data].should == 'exact-value-button'
+        results[1][:data].should == 'not-exact-value-button'
+      end
+    end
+
+    context "with image type" do
+      it "finds buttons by id" do
+        get(XPath::HTML.button('imgbut-with-id'))[:data].should == 'id-imgbut'
+      end
+
+      it "finds buttons by value" do
+        get(XPath::HTML.button('imgbut-with-value'))[:data].should == 'value-imgbut'
+      end
+
+      it "finds buttons by approximate value " do
+        get(XPath::HTML.button('gbut-with-val'))[:data].should == 'value-imgbut'
+      end
+
+      it "finds buttons by alt attribute" do
+        get(XPath::HTML.button('imgbut-with-alt'))[:data].should == 'alt-imgbut'
+      end
+
+      it "prefers buttons with exact value " do
+        results = all(XPath::HTML.button('exact value imgbut'))
+        results[0][:data].should == 'exact-value-imgbut'
+        results[1][:data].should == 'not-exact-value-imgbut'
+      end
+    end
+
+    context "with button tag" do
+      it "finds buttons by id" do
+        get(XPath::HTML.button('btag-with-id'))[:data].should == 'id-btag'
+      end
+
+      it "finds buttons by value" do
+        get(XPath::HTML.button('btag-with-value'))[:data].should == 'value-btag'
+      end
+
+      it "finds buttons by approximate value " do
+        get(XPath::HTML.button('tag-with-val'))[:data].should == 'value-btag'
+      end
+
+      it "finds prefers buttons with exact value " do
+        results = all(XPath::HTML.button('exact value btag'))
+        results[0][:data].should == 'exact-value-btag'
+        results[1][:data].should == 'not-exact-value-btag'
+      end
+
+      it "finds buttons by text" do
+        get(XPath::HTML.button('btag-with-text'))[:data].should == 'text-btag'
+      end
+
+      it "finds buttons by text ignoring whitespace" do
+        get(XPath::HTML.button('My whitespaced button'))[:data].should == 'btag-with-whitespace'
+      end
+
+      it "finds buttons by approximate text " do
+        get(XPath::HTML.button('tag-with-tex'))[:data].should == 'text-btag'
+      end
+
+      it "finds buttons with child tags by text" do
+        get(XPath::HTML.button('An emphatic button'))[:href].should == 'btag-with-children'
+        get(XPath::HTML.button('emphatic'))[:href].should == 'btag-with-children'
+      end
+
+      it "prefers buttons with exact text" do
+        results = all(XPath::HTML.button('exact text btag'))
+        results[0][:data].should == 'exact-text-btag'
+        results[1][:data].should == 'not-exact-text-btag'
+      end
+    end
+
+    context "with unkown type" do
+      it "does not find the button" do
+        get(XPath::HTML.button('schmoo button')).should be_nil
+      end
+    end
+
+  end
+end

data/spec/xpath_spec.rb

@@ -43,6 +43,12 @@ describe XPath do
       @results[0].text.should == 'Blah'
       @results[3].text.should == 'A list'
     end
+
+    it "should find all nodes when no arguments given" do
+      @results = xpath { |x| x.descendant[x.attr(:id) == 'foo'].descendant }
+      @results[0].text.should == 'Blah'
+      @results[4].text.should == 'A list'
+    end
   end
 
   describe '#child' do
@@ -62,6 +68,12 @@ describe XPath do
       @results[0].text.should == 'Blah'
       @results[3].text.should == 'A list'
     end
+
+    it "should find all nodes when no arguments given" do
+      @results = xpath { |x| x.descendant[x.attr(:id) == 'foo'].child }
+      @results[0].text.should == 'Blah'
+      @results[3].text.should == 'A list'
+    end
   end
 
   describe '#next_sibling' do
@@ -70,6 +82,7 @@ describe XPath do
       xpath { |x| x.descendant(:p)[x.attr(:id) == 'fooDiv'].next_sibling(:ul, :p) }.first.text.should == 'Bax'
       xpath { |x| x.descendant(:p)[x.attr(:title) == 'monkey'].next_sibling(:ul, :p) }.first.text.should == 'A list'
       xpath { |x| x.descendant(:p)[x.attr(:id) == 'fooDiv'].next_sibling(:ul, :li) }.first.should be_nil
+      xpath { |x| x.descendant(:p)[x.attr(:id) == 'fooDiv'].next_sibling }.first.text.should == 'Bax'
     end
   end
 
@@ -221,7 +234,7 @@ describe XPath do
       @results[0][:title].should == "barDiv"
       @results[1][:title].should == "fooDiv"
     end
-    
+
     it "should be closed" do
       @results = xpath do |x|
         foo_div = x.anywhere(:div).where(x.attr(:id) == 'foo')
@@ -231,6 +244,27 @@ describe XPath do
     end
   end
 
+  describe '#css' do
+    it "should find nodes by the given CSS selector" do
+      @results = xpath { |x| x.css('#preference p') }
+      @results[0].text.should == 'allamas'
+      @results[1].text.should == 'llama'
+    end
+
+    it "should respect previous expression" do
+      @results = xpath { |x| x.descendant[x.attr(:id) == 'moar'].css('p') }
+      @results[0].text.should == 'chimp'
+      @results[1].text.should == 'flamingo'
+    end
+
+    it "should allow comma separated selectors" do
+      @results = xpath { |x| x.descendant[x.attr(:id) == 'moar'].css('div, p') }
+      @results[0].text.should == 'chimp'
+      @results[1].text.should == 'flamingo'
+      @results[2].text.should == 'elephant'
+    end
+  end
+
   describe '#name' do
     it "should match the node's name" do
       xpath { |x| x.descendant(:*).where(x.name == 'ul') }.first.text.should == "A list"
@@ -250,7 +284,11 @@ describe XPath do
 
     it "should raise an argument error if the interpolation key is not given" do
       @xpath = XPath.generate { |x| x.descendant(:*).where(x.attr(:id) == x.var(:id).string_literal) }
-      lambda { @xpath.apply.to_xpath }.should raise_error(ArgumentError)
+      if defined?(KeyError)
+        lambda { @xpath.apply.to_xpath }.should raise_error(KeyError)
+      else
+        lambda { @xpath.apply.to_xpath }.should raise_error(ArgumentError)
+      end
     end
   end
 
@@ -274,8 +312,8 @@ describe XPath do
 
   describe '#union' do
     it "should create a union expression" do
-      @expr1 = XPath.generate { |x| x.descendant(:p) } 
-      @expr2 = XPath.generate { |x| x.descendant(:div) } 
+      @expr1 = XPath.generate { |x| x.descendant(:p) }
+      @expr2 = XPath.generate { |x| x.descendant(:div) }
       @collection = @expr1.union(@expr2)
       @xpath1 = @collection.where(XPath.attr(:id) == 'foo').to_xpath
       @xpath2 = @collection.where(XPath.attr(:id) == XPath.varstring(:id)).apply(:id => 'fooDiv').to_xpath
@@ -286,8 +324,8 @@ describe XPath do
     end
 
     it "should be aliased as +" do
-      @expr1 = XPath.generate { |x| x.descendant(:p) } 
-      @expr2 = XPath.generate { |x| x.descendant(:div) } 
+      @expr1 = XPath.generate { |x| x.descendant(:p) }
+      @expr2 = XPath.generate { |x| x.descendant(:div) }
       @collection = @expr1 + @expr2
       @xpath1 = @collection.where(XPath.attr(:id) == 'foo').to_xpath
       @xpath2 = @collection.where(XPath.attr(:id) == XPath.varstring(:id)).apply(:id => 'fooDiv').to_xpath

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: xpath
 version: !ruby/object:Gem::Version 
-  hash: 27
+  hash: 25
   prerelease: false
   segments: 
   - 0
   - 1
-  - 0
-  version: 0.1.0
+  - 1
+  version: 0.1.1
 platform: ruby
 authors: 
 - Jonas Nicklas
@@ -15,39 +15,38 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-08-15 00:00:00 +02:00
+date: 2010-10-06 00:00:00 +02:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
-  name: rspec
+  name: nokogiri
   prerelease: false
   requirement: &id001 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
-    - - ">="
+    - - ~>
       - !ruby/object:Gem::Version 
-        hash: 13
+        hash: 9
         segments: 
         - 1
-        - 2
-        - 9
-        version: 1.2.9
-  type: :development
+        - 3
+        version: "1.3"
+  type: :runtime
   version_requirements: *id001
 - !ruby/object:Gem::Dependency 
-  name: nokogiri
+  name: rspec
   prerelease: false
   requirement: &id002 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
-        hash: 29
+        hash: 13
         segments: 
         - 1
-        - 3
-        - 3
-        version: 1.3.3
+        - 2
+        - 9
+        version: 1.2.9
   type: :development
   version_requirements: *id002
 - !ruby/object:Gem::Dependency 
@@ -84,6 +83,7 @@ files:
 - spec/fixtures/form.html
 - spec/fixtures/simple.html
 - spec/fixtures/stuff.html
+- spec/html_spec.rb
 - spec/spec_helper.rb
 - spec/union_spec.rb
 - spec/xpath_spec.rb