tarantula 0.0.5

data/test/relevance/tarantula/form_submission_test.rb

@@ -0,0 +1,71 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+describe "Relevance::Tarantula::FormSubmission" do
+  
+  # TODO: add more from field types to this example form as needed
+  before do
+    @tag = Hpricot(<<END)
+<form action="/session" method="post">
+  <input id="email" name="email" size="30" type="text" />
+  <textarea id="comment" name="comment"value="1" />
+  <input name="commit" type="submit" value="Postit" />
+  <input name="secret" type="hidden" value="secret" />
+  <select id="foo_opened_on_1i" name="foo[opened_on(1i)]">
+    <option value="2003">2003</option>
+    <option value="2004">2004</option>
+  </select> 
+</form>
+END
+    @form = Relevance::Tarantula::Form.new(@tag.at('form'))
+    @fs = Relevance::Tarantula::FormSubmission.new(@form)
+  end
+  
+  it "can mutate text areas" do
+    @fs.stubs(:random_int).returns("42")
+    @fs.mutate_text_areas(@form).should == {"comment" => "42"}
+  end
+  
+  it "can mutate selects" do
+    Hpricot::Elements.any_instance.stubs(:rand).returns(stub(:[] => "2006-stub"))
+    @fs.mutate_selects(@form).should == {"foo[opened_on(1i)]" => "2006-stub"}
+  end
+  
+  it "can mutate inputs" do
+    @fs.stubs(:random_int).returns("43")
+    @fs.mutate_inputs(@form).should == {"commit"=>"43", "secret"=>"43", "email"=>"43"}
+  end
+
+  it "has a signature based on action and fields" do
+    @fs.signature.should == ['/session', [
+      "comment", 
+      "commit", 
+      "email", 
+      "foo[opened_on(1i)]", 
+      "secret"]]
+  end
+  
+  it "has a friendly to_s" do
+    @fs.to_s.should =~ %r{^/session post}
+  end
+  
+  it "can generate a random whole number" do
+    @fs.random_whole_number.should >= 0
+    Fixnum.should === @fs.random_whole_number
+  end
+end
+
+describe "Relevance::Tarantula::FormSubmission for a crummy form" do
+  before do
+    @tag = Hpricot(<<END)
+<form action="/session" method="post">
+  <input value="no_name" />
+</form>
+END
+    @form = Relevance::Tarantula::Form.new(@tag.at('form'))
+    @fs = Relevance::Tarantula::FormSubmission.new(@form)
+  end
+  
+  it "ignores unnamed inputs" do
+    @fs.mutate_inputs(@form).should == {}
+  end
+end

data/test/relevance/tarantula/form_test.rb

@@ -0,0 +1,50 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+describe "Relevance::Tarantula::Form large example" do
+  before do
+    @tag = Hpricot(<<END)
+<form action="/session" method="post">
+  <input name="authenticity_token" type="hidden" value="1be0d07c6e13669a87b8f52a3c7e1d1ffa77708d" />
+  <input id="email" name="email" size="30" type="text" />
+  <input id="password" name="password" size="30" type="password" />
+  <input id="remember_me" name="remember_me" type="checkbox" value="1" />
+  <input name="commit" type="submit" value="Log in" />
+</form>
+END
+    @form = Relevance::Tarantula::Form.new(@tag.at('form'))
+  end
+  
+  it "has an action" do
+    @form.action.should == "/session"
+  end
+  
+  it "has a method" do
+    @form.method.should == "post"
+  end
+  
+end
+
+describe "A Relevance::Tarantula::Form" do
+  it "defaults method to 'get'" do
+    @tag = Hpricot("<form/>")
+    @form = Relevance::Tarantula::Form.new(@tag.at('form'))
+    @form.method.should == 'get'
+  end
+end
+
+describe "A Relevance::Tarantula::Form with a hacked _method" do
+  before do
+    @tag = Hpricot(<<END)
+<form action="/foo">
+  <input name="authenticity_token" type="hidden" value="1be0d07c6e13669a87b8f52a3c7e1d1ffa77708d" />
+  <input id="_method" name="_method" size="30" type="text" value="PUT"/>
+</form>
+END
+    @form = Relevance::Tarantula::Form.new(@tag.at('form'))
+  end
+
+  it "has a method" do
+    @form.method.should == "put"
+  end
+  
+end

data/test/relevance/tarantula/html_document_handler_test.rb

@@ -0,0 +1,43 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+describe "Relevance::Tarantula::HtmlDocumentHandler" do
+  
+  before do
+    @handler = Relevance::Tarantula::HtmlDocumentHandler.new(nil)
+  end
+  
+  it "does not write HTML Scanner warnings to the console" do
+    bad_html = "<html><div></form></html>"    
+    err = Recording.stderr do
+      @handler.handle(Relevance::Tarantula::Result.new(:response => stub(:html? => true, :body => bad_html)))
+    end
+    err.should == ""
+  end
+  
+  it "ignores non-html" do
+    @handler.expects(:queue_link).never
+    @handler.handle(Relevance::Tarantula::Result.new(:response => stub(:html? => false, :body => '<a href="/foo">foo</a>')))
+  end
+  
+  it "queues anchor tags" do
+    @handler.expects(:queue_link).with {|*args| args[0]['href'] == "/foo" && args[1] == nil}
+    @handler.handle(Relevance::Tarantula::Result.new(:response => stub(:html? => true, :body => '<a href="/foo">foo</a>')))
+  end
+
+  it "queues link tags" do
+    @handler.expects(:queue_link).with {|*args| args[0]['href'] == "/bar" && args[1] == nil}
+    @handler.handle(Relevance::Tarantula::Result.new(:response => stub(:html? => true, :body => '<link href="/bar">bar</a>')))
+  end
+  
+  it "queues forms" do
+    @handler.expects(:queue_form).with{|tag,referrer| Hpricot::Elem === tag}
+    @handler.handle(Relevance::Tarantula::Result.new(:url => "/page-url", :response => stub(:html? => true, :body => '<form>stuff</form>')))
+  end
+  
+  it "infers form action from page url if form is not explicit" do
+    @handler.expects(:queue_form).with{|tag,referrer| tag['action'].should == '/page-url'; true }
+    @handler.handle(Relevance::Tarantula::Result.new(:url => "/page-url", :response => stub(:html? => true, :body => '<form>stuff</form>')))
+  end
+  
+end
+

data/test/relevance/tarantula/html_report_helper_test.rb

@@ -0,0 +1,47 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+module HtmlReportHelperSpec
+  
+  # Is there an idiom for this?
+  def self.included(base)
+    base.before do
+      @reporter = Object.new
+      @reporter.extend Relevance::Tarantula::HtmlReportHelper
+    end                                                                   
+  end 
+end
+
+describe 'Relevance::Tarantula::HtmlReportHelper#wrap_in_line_number_table' do
+  include HtmlReportHelperSpec
+  it "can wrap text in a line number table" do
+    html = @reporter.wrap_in_line_number_table("Line 1\nLine 2")
+    html.should == <<-END.strip
+<table class="grid tablesorter"><thead><tr><th class="sort asc"><span>Line #</span><span class="sort"><em>&#8613;</em></span></th><th class="sort left"><span>Line</span><span class="sort"><em>&#8613;</em></span></th></tr></thead><tr><td>1</td><td>Line 1</td></tr><tr><td>2</td><td>Line 2</td></tr></table>    
+END
+  end  
+end
+
+describe 'Relevance::Tarantula::HtmlReportHelper#wrap_stack_trace_line' do
+  include HtmlReportHelperSpec
+  it "can wrap stack trace line in links" do                       
+    line = %{/action_controller/filters.rb:697:in `call_filters'}
+    @reporter.stubs(:textmate_url).returns("ide_url")
+    html = @reporter.wrap_stack_trace_line(line)
+    html.should == "<a href='ide_url'>/action_controller/filters.rb:697</a>:in `call_filters'"
+  end  
+  
+  it "converts html entities for non-stack trace lines" do
+    line = %{<a href="foo">escape me</a>}
+    html = @reporter.wrap_stack_trace_line(line)
+    html.should == %{&lt;a href=&quot;foo&quot;&gt;escape me&lt;/a&gt;}
+  end
+
+end
+
+describe 'Relevance::Tarantula::HtmlReportHelper IDE help' do
+  include HtmlReportHelperSpec
+  it "can create a textmate url" do
+    @reporter.stubs(:rails_root).returns("STUB_RAILS_ROOT")
+    @reporter.textmate_url("/etc/somewhere", 100).should =~ %r{txmt://open\?url=.*/STUB_RAILS_ROOT/etc/somewhere&line_no=100}
+  end
+end

data/test/relevance/tarantula/html_reporter_test.rb

@@ -0,0 +1,82 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+describe "Relevance::Tarantula::HtmlReporter file output" do
+
+  before do
+    FileUtils.rm_rf(test_output_dir)
+    FileUtils.mkdir_p(test_output_dir)
+    @test_name = "test_user_pages"
+    Relevance::Tarantula::Result.next_number = 0
+    @success_results = (1..10).map do |index|
+      Relevance::Tarantula::Result.new(
+        :success => true, 
+        :method => "get", 
+        :url => "/widgets/#{index}", 
+        :response => stub(:code => 200, :body => "<h1>header</h1>\n<p>text</p>"), 
+        :referrer => "/random/#{rand(100)}", 
+        :test_name => @test_name,
+        :log => <<-END,
+Made-up stack trace:
+/some_module/some_class.rb:697:in `bad_method'
+/some_module/other_class.rb:12345677:in `long_method'
+this link should be <a href="#">escaped</a>
+blah blah blah
+        END
+        :data => "{:param1 => :value, :param2 => :another_value}"
+      )
+    end
+    @fail_results = (1..10).map do |index|
+      Relevance::Tarantula::Result.new(
+        :success => false, 
+        :method => "get", 
+        :url => "/widgets/#{index}", 
+        :response => stub(:code => 500, :body => "<h1>header</h1>\n<p>text</p>"), 
+        :referrer => "/random/#{rand(100)}", 
+        :test_name => @test_name,
+        :log => <<-END,
+Made-up stack trace:
+/some_module/some_class.rb:697:in `bad_method'
+/some_module/other_class.rb:12345677:in `long_method'
+this link should be <a href="#">escaped</a>
+blah blah blah
+        END
+        :data => "{:param1 => :value, :param2 => :another_value}"
+      )
+    end
+    @index = File.join(test_output_dir, "index.html")
+    FileUtils.rm_f @index
+    @detail = File.join(test_output_dir, @test_name,"1.html")
+    FileUtils.rm_f @detail
+  end
+  
+  it "creates a final report based on tarantula results" do
+    Relevance::Tarantula::Result.any_instance.stubs(:rails_root).returns("STUB_ROOT")        
+    reporter = Relevance::Tarantula::HtmlReporter.new(test_output_dir)
+    stub_puts_and_print(reporter)
+    (@success_results + @fail_results).each {|r| reporter.report(r)}
+    reporter.finish_report(@test_name)
+    File.should.exist @index
+  end 
+  
+  it "creates a final report with links to detailed reports in subdirs" do
+    Relevance::Tarantula::Result.any_instance.stubs(:rails_root).returns("STUB_ROOT")
+    reporter = Relevance::Tarantula::HtmlReporter.new(test_output_dir)
+    stub_puts_and_print(reporter)
+    (@success_results + @fail_results).each {|r| reporter.report(r)}
+    reporter.finish_report(@test_name)
+    links = Hpricot(File.read(@index)).search('.left a')
+    links.each do |link|
+      link['href'].should.match(/#{@test_name}\/\d+\.html/)
+    end
+  end
+
+  it "creates detailed reports based on tarantula results" do
+    Relevance::Tarantula::Result.any_instance.stubs(:rails_root).returns("STUB_ROOT")        
+    reporter = Relevance::Tarantula::HtmlReporter.new(test_output_dir)
+    stub_puts_and_print(reporter)
+    (@success_results + @fail_results).each {|r| reporter.report(r)}
+    reporter.finish_report(@test_name)    
+    File.should.exist @detail
+  end
+
+end

data/test/relevance/tarantula/invalid_html_handler_test.rb

@@ -0,0 +1,33 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+describe "Relevance::Tarantula::InvalidHtmlHandler" do
+  before do
+    @handler = Relevance::Tarantula::InvalidHtmlHandler.new
+  end
+    
+  it "does not write HTML Scanner warnings to the console" do
+    bad_html = "<html><div></form></html>"    
+    err = Recording.stderr do
+      @handler.handle(Relevance::Tarantula::Result.new(:response => stub(:html? => true, :body => bad_html)))
+    end
+    err.should == ""
+  end
+  
+  it "rejects unclosed html" do
+    response = stub(:html? => true, :body => '<html><div></html>', :code => 200)
+    result = @handler.handle(Relevance::Tarantula::Result.new(:response => response))
+    result.success.should == false
+    result.description.should == "Bad HTML (Scanner)"
+  end
+
+  it "loves the good html" do
+    response = stub(:html? => true, :body => '<html><div></div></html>', :code => 200)
+    @handler.handle(Relevance::Tarantula::Result.new(:response => response)).should == nil
+  end
+
+  it "ignores non html" do
+    response = stub(:html? => false, :body => '<html><div></html>', :code => 200)
+    @handler.handle(Relevance::Tarantula::Result.new(:response => response)).should == nil
+  end
+end
+

data/test/relevance/tarantula/io_reporter_test.rb

@@ -0,0 +1,11 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+describe 'Relevance::Tarantula::IOReporter' do
+  it "reports errors to stderr and then raises" do
+    reporter = Relevance::Tarantula::IOReporter.new($stderr)
+    reporter.report stub(:code => "404", :url => "/uh-oh", :success => false)
+    $stderr.expects(:puts).with("****** FAILURES")
+    $stderr.expects(:puts).with("404: /uh-oh")
+    lambda {reporter.finish_report("test_user_pages")}.should.raise RuntimeError
+  end
+end

data/test/relevance/tarantula/link_test.rb

@@ -0,0 +1,61 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+describe "Relevance::Tarantula::Link" do
+  include ActionView::Helpers::UrlHelper
+  
+  it "parses anchor tags" do
+    link = Relevance::Tarantula::Link.new(Hpricot('<a href="/foo">foo</a>').at('a'))
+    assert_equal "/foo", link.href
+    assert_equal :get, link.method
+  end
+
+  it "parses anchor tags with POST 'method'" do
+    link = Relevance::Tarantula::Link.new(Hpricot(%Q{<a href="/foo" onclick="#{method_javascript_function(:post)}">foo</a>}).at('a'))
+    assert_equal "/foo", link.href
+    assert_equal :post, link.method
+  end
+
+  it "parses anchor tags with PUT 'method'" do
+    link = Relevance::Tarantula::Link.new(Hpricot(%Q{<a href="/foo" onclick="#{method_javascript_function(:put)}">foo</a>}).at('a'))
+    assert_equal "/foo", link.href
+    assert_equal :put, link.method
+  end
+
+  it "parses anchor tags with DELETE 'method'" do
+    link = Relevance::Tarantula::Link.new(Hpricot(%Q{<a href="/foo" onclick="#{method_javascript_function(:delete)}">foo</a>}).at('a'))
+    assert_equal "/foo", link.href
+    assert_equal :delete, link.method
+  end
+
+  it "parses link tags with text" do
+    link = Relevance::Tarantula::Link.new(Hpricot('<link href="/bar">bar</a>').at('link'))
+    assert_equal "/bar", link.href
+    assert_equal :get, link.method
+  end
+  
+  it "parses link tags without text" do
+    link = Relevance::Tarantula::Link.new(Hpricot('<link href="/bar" />').at('link'))
+    assert_equal "/bar", link.href
+    assert_equal :get, link.method
+  end
+  
+  # method_javascript_function needs this method
+  def protect_against_forgery?
+    false
+  end
+  
+end
+
+describe "possible conflict when user has an AR model named Link" do
+  it "does not taint Object with Relevance::Tarantula" do
+    Object.ancestors.should.not.include Relevance::Tarantula
+  end
+  
+  it "doesnt break with a Link model" do
+    lambda {
+      class Link < ActiveRecord::Base
+      end
+    }.should.not.raise
+  end
+  
+end

data/test/relevance/tarantula/log_grabber_test.rb

@@ -0,0 +1,26 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+describe 'Relevance::Tarantula::LogGrabber' do
+  before do
+    @grabber = Relevance::Tarantula::LogGrabber.new(log_file)
+    FileUtils.mkdir_p(test_output_dir)
+  end
+
+  def log_file
+    File.join(File.join(test_output_dir, "example.log"))
+  end
+
+  it "can clear the log file" do
+    File.open(log_file, "w") {|f| f.print "sample log"}
+    File.size(log_file).should == 10
+    @grabber.clear!
+    File.size(log_file).should == 0
+  end
+
+  it "can grab the log file" do
+    File.open(log_file, "w") {|f| f.print "sample log"}
+    @grabber.grab!.should == "sample log"
+    File.size(log_file).should == 0
+  end
+
+end

data/test/relevance/tarantula/rails_integration_proxy_test.rb

@@ -0,0 +1,94 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+
+describe "Relevance::Tarantula::RailsIntegrationProxy rails_integration_test" do
+  before {
+    Relevance::Tarantula::Crawler.any_instance.stubs(:crawl)
+    Relevance::Tarantula::Crawler.any_instance.stubs(:rails_root).returns("STUB_RAILS_ROOT")
+    Relevance::Tarantula::RailsIntegrationProxy.stubs(:rails_root).returns("STUB_RAILS_ROOT")
+    Relevance::Tarantula::RailsIntegrationProxy.stubs(:new).returns(stub(:integration_test => stub(:method_name => @test_name)))
+    @test_name = "test_user_pages"
+  }
+
+  it "strips leading hostname from link urls" do    
+    crawler = Relevance::Tarantula::RailsIntegrationProxy.rails_integration_test(stub(:host => "foo.com"))
+    crawler.transform_url("http://foo.com/path").should == "/path"
+    crawler.transform_url("http://bar.com/path").should == "http://bar.com/path"
+  end
+  
+  it "allows override of max_url_length" do
+    crawler = Relevance::Tarantula::RailsIntegrationProxy.rails_integration_test(stub(:host => "foo.com"), 
+                                             :max_url_length => 16)
+    crawler.max_url_length.should == 16
+  end
+
+  it "has some useful defaults" do
+    crawler = Relevance::Tarantula::RailsIntegrationProxy.rails_integration_test(stub(:host => "foo.com")) 
+    crawler.log_grabber.should.not.be nil
+  end
+end
+
+
+describe "Relevance::Tarantula::RailsIntegrationProxy" do
+  %w{get post}.each do |http_method|
+    it "can #{http_method}" do
+      @rip = Relevance::Tarantula::RailsIntegrationProxy.new(stub)
+      @response = stub({:code => :foo})
+      @rip.integration_test = stub_everything(:response => @response)
+      @rip.send(http_method, "/url").should.be @response
+    end
+  end
+  
+  it "adds a response accessor to its delegate rails integration test" do
+    o = Object.new
+    Relevance::Tarantula::RailsIntegrationProxy.new(o)
+    o.methods(false).sort.should == %w{response response=}
+  end
+
+end
+
+describe "Relevance::Tarantula::RailsIntegrationProxy patching" do
+  before do
+    @rip = Relevance::Tarantula::RailsIntegrationProxy.new(stub)
+    @rip.stubs(:rails_root).returns("faux_rails_root")
+    @response = stub_everything({:code => "404", :headers => {}})
+    File.stubs(:exist?).returns(true)
+  end
+  
+  it "patches in Relevance::CoreExtensions::Response" do
+    @rip = Relevance::Tarantula::RailsIntegrationProxy.new(stub)
+    @rip.stubs(:rails_root).returns("faux_rails_root")
+    @response = stub_everything({:code => "404", :headers => {}, :content_type => "text/html"})
+    @response.meta.ancestors.should.not.include Relevance::CoreExtensions::Response
+    @rip.patch_response("/url", @response)
+    @response.meta.ancestors.should.include Relevance::CoreExtensions::Response
+    @response.html?.should == true
+  end
+  
+  it "ignores 404s for known static binary types" do
+    File.expects(:extension).returns("pdf")
+    @rip.expects(:log).with("Skipping /url (for now)")
+    @rip.patch_response("/url", @response)
+  end
+  
+  it "replaces 404s with 200s, pulling content from public, for known text types" do
+    File.expects(:extension).returns("html")
+    @rip.expects(:static_content_file).with("/url").returns("File body")
+    @rip.patch_response("/url", @response)
+    @response.headers.should == {"type" => "text/html"}
+  end
+  
+  it "logs and skips types we haven't dealt with yet" do
+    File.expects(:extension).returns("whizzy")
+    @rip.expects(:log).with("Skipping unknown type /url")
+    @rip.patch_response("/url", @response)
+  end
+  
+  it "can find static content relative to rails root" do
+    @rip.static_content_path("foo").should == File.expand_path("faux_rails_root/public/foo")
+  end
+  
+  it "can read static content relative to rails root" do
+    File.expects(:read).with(@rip.static_content_path("foo"))
+    @rip.static_content_file("foo")
+  end
+end