wgibbs-xpath 0.1.4

data/lib/xpath/union.rb

@@ -0,0 +1,27 @@
+module XPath
+  class Union
+    include Enumerable
+
+    attr_reader :expressions
+
+    def initialize(*expressions)
+      @expressions = expressions
+    end
+
+    def each(&block)
+      expressions.each(&block)
+    end
+
+    def to_xpath(predicate=nil)
+      expressions.map { |e| e.to_xpath(predicate) }.join(' | ')
+    end
+
+    def to_xpaths
+      [to_xpath(:exact), to_xpath(:fuzzy)].uniq
+    end
+
+    def method_missing(*args)
+      XPath::Union.new(*expressions.map { |e| e.send(*args) })
+    end
+  end
+end

data/lib/xpath/version.rb

@@ -0,0 +1,3 @@
+module XPath
+  VERSION = '0.1.4'
+end

data/spec/fixtures/form.html

@@ -0,0 +1,87 @@
+<h1>Form</h1>
+
+<p>
+  <a id="awesome-link" data="link-text" href="#link">An awesome link</a>
+  <a id="some-id" data="link-id" href="#id">With id</a>
+  <a title="My title" data="link-title" href="#title">A cool title</a>
+  <a href="#image" data="link-img"><img src="foo.png" alt="Alt link"/></a>
+  <a title="This title is too long" data="link-fuzzy" href="#bar">A link at a time</a>
+  <a title="This title" data="link-exact" href="#foo">A link</a>
+  <a href="#bar" data="link-img-fuzzy"><img src="foo.png" alt="An image that is beautiful"/></a>
+  <a href="#foo" data="link-img-exact"><img src="foo.ong" alt="An image"/></a>
+  <a href="http://www.example.com" data="link-href">Href-ed link</a>
+  <a>Wrong Link</a>
+  <a href="#spacey" data="link-whitespace">    My
+
+      whitespaced
+    link</a>
+  <a href="#has-children" data="link-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 data="btag-with-children">
+    An <em>emphatic</em> button with some children
+  </button>
+
+  <input type="schmoo" value="schmoo button" data="schmoo"/>
+  <label for="text-with-id">Label text <input type="text" id="text-with-id" data="id-text" value="monkey"/></label>
+  <input type="text" id="text-with-placeholder" data="placeholder-text" value="monkey" placeholder="Placeholder text"/>
+  <label for="problem-text-with-id">Label text's got an apostrophe <input type="text" id="problem-text-with-id" data="id-problem-text" value="monkey"/></label>
+</p>
+
+<p>
+  <fieldset id="some-fieldset-id" data="fieldset-id"></fieldset>
+  <fieldset data="fieldset-legend"><legend>Some Legend</legend></fieldset>
+  <fieldset data="fieldset-legend-span"><legend><span>Span Legend</span></legend></fieldset>
+  <fieldset data="fieldset-fuzzy"><legend>Long legend yo</legend></fieldset>
+  <fieldset data="fieldset-exact"><legend>Long legend</legend></fieldset>
+</p>
+
+<p>
+  <select>
+    <optgroup label="Group A" data="optgroup-a"></optgroup>
+    <optgroup label="Group B" data="optgroup-b"></optgroup>
+  </select>
+</p>
+
+<p>
+  <table id="whitespaced-table" data="table-with-whitespace">
+    <tr>
+      <td data="cell-whitespaced">I have
+        <span>nested whitespace</span>
+      </td>
+      <td>I don't</td>
+    </tr>
+  </table>
+</p>

data/spec/fixtures/simple.html

@@ -0,0 +1,32 @@
+<div id="bar" title="barDiv">
+
+</div>
+
+<div title="noId"></div>
+
+<div id="foo" title="fooDiv" data="id">
+  <p id="fooDiv">Blah</p>
+  <p>Bax</p>
+  <p title="monkey">Bax</p>
+  <ul><li>A list</li></ul>
+</div>
+
+<div id="baz" title="bazDiv"></div>
+
+<div id="preference">
+  <p id="is-fuzzy">allamas</p>
+  <p id="is-exact">llama</p>
+</div>
+
+<p id="whitespace">
+A     lot
+
+  of
+      whitespace
+</p>
+
+<div id="moar">
+  <p id="impchay">chimp</p>
+  <div id="elephantay">elephant</div>
+  <p id="amingoflay">flamingo</p>
+</div>

data/spec/fixtures/stuff.html

@@ -0,0 +1,43 @@
+<h1>This is a test</h1>
+
+<p id="first">
+  Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod
+  tempor incididunt ut <a href="/with_simple_html" title="awesome title" class="simple">labore</a>
+  et dolore magna aliqua. Ut enim ad minim veniam,
+  quis nostrud exercitation <a href="/foo" id="foo">ullamco</a> laboris nisi
+  ut aliquip ex ea commodo consequat.
+  <a href="/with_simple_html"><img src="http://www.foobar.sun/dummy_image.jpg" width="20" height="20" alt="awesome image" /></a>
+</p>
+
+<p id="second">
+  Duis aute irure dolor in reprehenderit in voluptate velit esse cillum
+  dolore eu fugiat <a href="/redirect" id="red">Redirect</a> pariatur. Excepteur sint occaecat cupidatat non proident,
+  sunt in culpa qui officia
+  text   with
+  whitespace
+  id est laborum.
+</p>
+
+<p>
+  <input type="text" id="test_field" value="monkey"/>
+  <textarea>banana</textarea>
+  <a href="/redirect_back">BackToMyself</a>
+  <a title="twas a fine link" href="/redirect">A link came first</a>
+  <a title="a fine link" href="/with_simple_html">A link</a>
+  <a title="a fine link with data method" data-method="delete" href="/delete">A link with data-method</a>
+  <a>No Href</a>
+  <a href="">Blank Href</a>
+  <a href="#">Blank Anchor</a>
+  <a href="#anchor">Anchor</a>
+  <a href="/with_simple_html#anchor">Anchor on different page</a>
+  <a href="/with_html#anchor">Anchor on same page</a>
+  <input type="text" value="" id="test_field">
+  <input type="text" checked="checked" id="checked_field">
+  <a href="/redirect"><img src="http://www.foobar.sun/dummy_image.jpg" width="20" height="20" alt="very fine image" /></a>
+  <a href="/with_simple_html"><img src="http://www.foobar.sun/dummy_image.jpg" width="20" height="20" alt="fine image" /></a>
+</p>
+
+<div id="hidden" style="display: none;">
+  <div id="hidden_via_ancestor">Inside element with hidden ancestor</div>
+  <a href="/with_simple_html" title="awesome title" class="simple">hidden link</a>
+</div>

data/spec/html_spec.rb

@@ -0,0 +1,187 @@
+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(*args)
+    all(*args).first
+  end
+
+  def all(*args)
+    XPath::HTML.send(subject, *args).to_xpaths.map do |xpath|
+      doc.xpath(xpath)
+    end.flatten.uniq.map { |node| node[:data] }
+  end
+
+  describe '#link' do
+    subject { :link }
+
+    it("finds links by id")                                { get('some-id').should == 'link-id' }
+    it("finds links by content")                           { get('An awesome link').should == 'link-text' }
+    it("finds links by content regardless of whitespace")  { get('My whitespaced link').should == 'link-whitespace' }
+    it("finds links with child tags by content")           { get('An emphatic link').should == 'link-children' }
+    it("finds links by the content of theur child tags")   { get('emphatic').should == 'link-children' }
+    it("finds links by approximate content")               { get('awesome').should == 'link-text' }
+    it("prefers exact matches of content")                 { all('A link').should == ['link-exact', 'link-fuzzy'] }
+    it("finds links by title")                             { get('My title').should == 'link-title' }
+    it("finds links by approximate title")                 { get('title').should == 'link-title' }
+    it("prefers exact matches of title")                   { all('This title').should == ['link-exact', 'link-fuzzy'] }
+    it("finds links by image's alt attribute")             { get('Alt link').should == 'link-img' }
+    it("finds links by image's approximate alt attribute") { get('Alt').should == 'link-img' }
+    it("prefers exact matches of image's alt attribute")   { all('An image').should == ['link-img-exact', 'link-img-fuzzy'] }
+    it("does not find links without href attriutes")       { get('Wrong Link').should be_nil }
+    it("finds links with an href")                         { get("Href-ed link", :href => 'http://www.example.com').should == 'link-href' }
+    it("does not find links with an incorrect href")       { get("Href-ed link", :href => 'http://www.somewhere.com').should be_nil }
+  end
+
+  describe '#button' do
+    subject { :button }
+
+    context "with submit type" do
+      it("finds buttons by id")                { get('submit-with-id').should == 'id-submit' }
+      it("finds buttons by value")             { get('submit-with-value').should == 'value-submit' }
+      it("finds buttons by approximate value") { get('mit-with-val').should == 'value-submit' }
+      it("prefers buttons with exact value")   { all('exact value submit').should == ['exact-value-submit', 'not-exact-value-submit'] }
+    end
+
+    context "with button type" do
+      it("finds buttons by id")                { get('button-with-id').should == 'id-button' }
+      it("finds buttons by value")             { get('button-with-value').should == 'value-button' }
+      it("finds buttons by approximate value") { get('ton-with-val').should == 'value-button' }
+      it("prefers buttons with exact value")   { all('exact value button').should == ['exact-value-button', 'not-exact-value-button'] }
+    end
+
+    context "with image type" do
+      it("finds buttons by id")                { get('imgbut-with-id').should == 'id-imgbut' }
+      it("finds buttons by value")             { get('imgbut-with-value').should == 'value-imgbut' }
+      it("finds buttons by approximate value") { get('gbut-with-val').should == 'value-imgbut' }
+      it("finds buttons by alt attribute")     { get('imgbut-with-alt').should == 'alt-imgbut' }
+      it("prefers buttons with exact value")   { all('exact value imgbut').should == ['exact-value-imgbut', 'not-exact-value-imgbut'] }
+    end
+
+    context "with button tag" do
+      it("finds buttons by id")                       { get('btag-with-id').should == 'id-btag' }
+      it("finds buttons by value")                    { get('btag-with-value').should == 'value-btag' }
+      it("finds buttons by approximate value")        { get('tag-with-val').should == 'value-btag' }
+      it("finds prefers buttons with exact value")    { all('exact value btag').should == ['exact-value-btag', 'not-exact-value-btag'] }
+      it("finds buttons by text")                     { get('btag-with-text').should == 'text-btag' }
+      it("finds buttons by text ignoring whitespace") { get('My whitespaced button').should == 'btag-with-whitespace' }
+      it("finds buttons by approximate text ")        { get('tag-with-tex').should == 'text-btag' }
+      it("finds buttons with child tags by text")     { get('An emphatic button').should == 'btag-with-children' }
+      it("finds buttons by text of their children")   { get('emphatic').should == 'btag-with-children' }
+      it("prefers buttons with exact text")           { all('exact text btag').should == ['exact-text-btag', 'not-exact-text-btag'] }
+    end
+
+    context "with unkown type" do
+      it("does not find the button") { get('schmoo button').should be_nil }
+    end
+  end
+
+  describe '#fieldset' do
+    subject { :fieldset }
+
+    it("finds fieldsets by id")                  { get('some-fieldset-id').should == 'fieldset-id' }
+    it("finds fieldsets by legend")              { get('Some Legend').should == 'fieldset-legend' }
+    it("finds fieldsets by legend child tags")   { get('Span Legend').should == 'fieldset-legend-span' }
+    it("accepts approximate legends")            { get('Legend').should == 'fieldset-legend' }
+    it("prefers exact legend")                   { all('Long legend').should == ['fieldset-exact', 'fieldset-fuzzy'] }
+  end
+
+  describe '#field' do
+    subject { :field }
+
+    context "by id" do
+      it("finds inputs with text type")     {}
+      it("finds inputs with password type") {}
+      it("finds inputs with custom type")   {}
+      it("finds textareas")                 {}
+      it("finds select boxes")              {}
+      it("does not find submit buttons")    {}
+      it("does not find image buttons")     {}
+      it("does not find hidden fields")     {}
+    end
+
+    context "by name" do
+      it("finds inputs with text type")     {}
+      it("finds inputs with password type") {}
+      it("finds inputs with custom type")   {}
+      it("finds textareas")                 {}
+      it("finds select boxes")              {}
+      it("does not find submit buttons")    {}
+      it("does not find image buttons")     {}
+      it("does not find hidden fields")     {}
+    end
+
+    context "by referenced label" do
+      it("finds inputs with text type")     {}
+      it("finds inputs with password type") {}
+      it("finds inputs with custom type")   {}
+      it("finds textareas")                 {}
+      it("finds select boxes")              {}
+      it("does not find submit buttons")    {}
+      it("does not find image buttons")     {}
+      it("does not find hidden fields")     {}
+    end
+
+    context "by parent label" do
+      it("finds inputs with text type")                    { get('Label text').should == 'id-text' }
+      it("finds inputs where label has problem chars")     { get("Label text's got an apostrophe").should == 'id-problem-text' }
+      it("finds inputs with password type")                {}
+      it("finds inputs with custom type")                  {}
+      it("finds textareas")                                {}
+      it("finds select boxes")                             {}
+      it("does not find submit buttons")                   {}
+      it("does not find image buttons")                    {}
+      it("does not find hidden fields")                    {}
+    end
+
+    context "with :with option" do
+      it("finds inputs that match option")          {}
+      it("omits inputs that don't match option")    {}
+      it("finds textareas that match option")       {}
+      it("omits textareas that don't match option") {}
+    end
+
+    context "with :checked option" do
+      context "when true" do
+        it("finds checked fields")   {}
+        it("omits unchecked fields") {}
+      end
+      context "when false" do
+        it("finds unchecked fields") {}
+        it("omits checked fields")   {}
+      end
+      context "when ommitted" do
+        it("finds unchecked fields") {}
+        it("finds checked fields")   {}
+      end
+    end
+  end
+
+  describe '#fillable_field' do
+    subject{ :fillable_field }
+    context "by parent label" do
+      it("finds inputs with text type")                    { get('Label text').should == 'id-text' }
+      it("finds inputs where label has problem chars")     { get("Label text's got an apostrophe").should == 'id-problem-text' }
+    end
+    context "by placeholder" do
+      it("finds inputs with placeholder text")              { get('Placeholder text').should == 'placeholder-text' }
+    end
+
+  end
+
+  describe "#optgroup" do
+    subject { :optgroup }
+
+    it("finds optgroups by label") { get('Group A').should == 'optgroup-a' }
+  end
+
+  describe "#table" do
+    subject {:table}
+
+    it("finds cell content regardless of whitespace") {get('whitespaced-table', :rows => [["I have nested whitespace", "I don't"]]).should == 'table-with-whitespace'}
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1,4 @@
+require 'rubygems'
+require 'bundler/setup'
+
+require 'xpath'

data/spec/union_spec.rb

@@ -0,0 +1,63 @@
+require 'spec_helper'
+
+describe XPath::Union do
+  let(:template) { File.read(File.expand_path('fixtures/simple.html', File.dirname(__FILE__))) }
+  let(:doc) { Nokogiri::HTML(template) }
+
+  describe '#expressions' do
+    it "should return the expressions" do
+      @expr1 = XPath.generate { |x| x.descendant(:p) } 
+      @expr2 = XPath.generate { |x| x.descendant(:div) } 
+      @collection = XPath::Union.new(@expr1, @expr2)
+      @collection.expressions.should == [@expr1, @expr2]
+    end
+  end
+
+  describe '#each' do
+    it "should iterate through the expressions" do
+      @expr1 = XPath.generate { |x| x.descendant(:p) } 
+      @expr2 = XPath.generate { |x| x.descendant(:div) } 
+      @collection = XPath::Union.new(@expr1, @expr2)
+      exprs = []
+      @collection.each { |expr| exprs << expr }
+      exprs.should == [@expr1, @expr2]
+    end
+  end
+
+  describe '#map' do
+    it "should map the expressions" do
+      @expr1 = XPath.generate { |x| x.descendant(:p) } 
+      @expr2 = XPath.generate { |x| x.descendant(:div) } 
+      @collection = XPath::Union.new(@expr1, @expr2)
+      @collection.map { |expr| expr.class }.should == [XPath::Expression::Descendant, XPath::Expression::Descendant]
+    end
+  end
+
+  describe '#to_xpath' do
+    it "should create a valid xpath expression" do
+      @expr1 = XPath.generate { |x| x.descendant(:p) } 
+      @expr2 = XPath.generate { |x| x.descendant(:div).where(x.attr(:id) == 'foo') } 
+      @collection = XPath::Union.new(@expr1, @expr2)
+      @results = doc.xpath(@collection.to_xpath)
+      @results[0][:title].should == 'fooDiv'
+      @results[1].text.should == 'Blah'
+      @results[2].text.should == 'Bax'
+    end
+  end
+
+  describe '#where, #apply and others' do
+    it "should be delegated to the individual expressions" do
+      @expr1 = XPath.generate { |x| x.descendant(:p) } 
+      @expr2 = XPath.generate { |x| x.descendant(:div) } 
+      @collection = XPath::Union.new(@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
+      @results = doc.xpath(@xpath1)
+      @results[0][:title].should == 'fooDiv'
+      @results = doc.xpath(@xpath2)
+      @results[0][:id].should == 'fooDiv'
+    end
+  end
+
+end
+