Ticket #12: tarantula_less_mem_with_tests2.patch.txt

diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/lib/relevance/tarantula/crawler.rb tarantula.less_mem/lib/relevance/tarantula/crawler.rb
--- tarantula.orig/lib/relevance/tarantula/crawler.rb   2008-05-03 18:49:16.000000000 -0400
+++ tarantula.less_mem/lib/relevance/tarantula/crawler.rb       2008-05-04 00:05:25.000000000 -0400
@@ -20,15 +20,15 @@
     @links_to_crawl = []
     @forms_to_crawl = []
     @referrers = {}
-    @skip_uri_patterns =[
+    @skip_uri_patterns = [
       /^javascript/,
       /^mailto/,
-      /^http/,                                      
+      /^http/,
     ]
     self.transform_url_patterns = [
       [/#.*$/, '']
     ]
-    @reporters = []
+    @reporters = [Relevance::Tarantula::IOReporter.new($stderr)]
     @decoder = HTMLEntities.new
     
   end
@@ -74,9 +74,9 @@
   end  
   
   def save_result(result)
-    return if result.nil?
-    collection = result.success ? successes : failures
-    collection << result
+    reporters.each do |reporter|
+      reporter.report(result)
+    end
   end
   
   def handle_link_results(link, response)
@@ -178,26 +178,21 @@
   end
 
   def generate_reports
-    FileUtils.mkdir_p(report_dir)
+    errors = []
     reporters.each do |reporter|
-      reporter.report(report_dir, self)
-    end
-  end
-  
-  def report_to_console
-    unless (failures).empty?
-      $stderr.puts "****** FAILURES"
-      failures.each do |failure|
-        $stderr.puts "#{failure.code}: #{failure.url}"
+      begin
+        reporter.finish_report
+      rescue RuntimeError => e
+        errors << e
       end
-      raise "#{failures.size} failures"
+    end
+    unless errors.empty?
+      raise errors.map(&:message).join("\n")
     end
   end
   
   def report_results
-    puts "Writing results to #{report_dir}"
     generate_reports
-    report_to_console
   end
   
   def total_links_count
@@ -217,4 +212,4 @@
       print "\r #{links_completed_count} of #{total_links_count} links completed               "
     end
   end  
-end
\ No newline at end of file
+end
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/lib/relevance/tarantula/html_reporter.rb tarantula.less_mem/lib/relevance/tarantula/html_reporter.rb
--- tarantula.orig/lib/relevance/tarantula/html_reporter.rb     2008-05-03 18:49:16.000000000 -0400
+++ tarantula.less_mem/lib/relevance/tarantula/html_reporter.rb 2008-05-04 00:04:11.000000000 -0400
@@ -1,28 +1,33 @@
 class Relevance::Tarantula::HtmlReporter
+  
   include Relevance::Tarantula
-  attr_accessor :basedir, :results 
+  attr_accessor :basedir, :results
   delegate :successes, :failures, :to => :results
-  def self.report(basedir, results)
-    self.new(basedir, results)
-  end
-
-  def initialize(basedir, results)
+  
+  HtmlResultOverview = Struct.new(:code, :url, :description, :method, :referrer, :file_name)
+  
+  def initialize(basedir)
     @basedir = basedir
-    @results = results
-    copy_styles
-    create_index
-    create_detail_reports
+    @results = Struct.new(:successes, :failures).new([], [])
+    FileUtils.mkdir_p(@basedir)
   end
   
-  def template(name)
-    File.read(File.join(File.dirname(__FILE__), name))
+  def report(result)
+    return if result.nil?
+    
+    create_detail_report(result)
+    
+    collection = result.success ? results.successes : results.failures
+    collection << HtmlResultOverview.new(
+      result.code, result.url, result.description, result.method, result.referrer, result.file_name
+    )
   end
   
-  def output(name, body)
-    File.open(File.join(basedir, name), "w") do |file|
-      file.write body
-    end
-  end      
+  def finish_report
+    puts "Writing results to #{basedir}"
+    copy_styles
+    create_index
+  end
   
   def copy_styles
     # not using cp_r because it picks up .svn crap
@@ -44,19 +49,26 @@
     template = ERB.new(template("index.html.erb"))
     output("index.html", template.result(binding))
   end
-
-  def create_detail_reports
-    template = ERB.new(template("detail.html.erb"))
-    results.successes.each do |result|
-      output(result.file_name, template.result(result.send(:binding)))
-    end
-    results.failures.each do |result|
-      output(result.file_name, template.result(result.send(:binding)))
+  
+  def template(name)
+    File.read(File.join(File.dirname(__FILE__), name))
+  end
+  
+  def output(name, body)
+    File.open(File.join(basedir, name), "w") do |file|
+      file.write body
     end
+  end      
+  
+  def create_detail_report(result)
+    template = ERB.new(template("detail.html.erb"))
+    output(result.file_name, template.result(result.send(:binding)))
   end 
-
+  
   # CSS class for HTML status codes
   def class_for_code(code)
     "r#{Integer(code)/100}" 
   end
+  
+  
 end
\ No newline at end of file
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/lib/relevance/tarantula/io_reporter.rb tarantula.less_mem/lib/relevance/tarantula/io_reporter.rb
--- tarantula.orig/lib/relevance/tarantula/io_reporter.rb       1969-12-31 19:00:00.000000000 -0500
+++ tarantula.less_mem/lib/relevance/tarantula/io_reporter.rb   2008-05-04 00:04:11.000000000 -0400
@@ -0,0 +1,34 @@
+class Relevance::Tarantula::IOReporter
+  
+  include Relevance::Tarantula
+  attr_accessor :io, :results
+  delegate :successes, :failures, :to => :results
+  
+  IOResultOverview = Struct.new(:code, :url)
+  
+  def initialize(io)
+    @io = io
+    @results = Struct.new(:successes, :failures).new([], [])
+  end
+  
+  def report(result)
+    return if result.nil?
+    
+    unless result.success # collection = result.success ? results.successes : results.failures
+      results.failures << IOResultOverview.new(
+        result.code, result.url
+      )
+    end
+  end
+  
+  def finish_report
+    unless (failures).empty?
+      io.puts "****** FAILURES"
+      failures.each do |failure|
+        io.puts "#{failure.code}: #{failure.url}"
+      end
+      raise "#{failures.size} failures"
+    end
+  end
+  
+end
\ No newline at end of file
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/lib/relevance/tarantula/rails_integration_proxy.rb tarantula.less_mem/lib/relevance/tarantula/rails_integration_proxy.rb
--- tarantula.orig/lib/relevance/tarantula/rails_integration_proxy.rb   2008-05-03 18:49:16.000000000 -0400
+++ tarantula.less_mem/lib/relevance/tarantula/rails_integration_proxy.rb       2008-05-04 00:04:11.000000000 -0400
@@ -18,7 +18,7 @@
       [/\?\d+$/, ''],                               # strip trailing numbers for assets
       [/^http:\/\/#{integration_test.host}/, '']    # strip full path down to relative
     ]
-    t.reporters << Relevance::Tarantula::HtmlReporter
+    t.reporters << Relevance::Tarantula::HtmlReporter.new(t.report_dir)
     t
   end
 
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/lib/relevance/tarantula/result.rb tarantula.less_mem/lib/relevance/tarantula/result.rb
--- tarantula.orig/lib/relevance/tarantula/result.rb    2008-05-03 18:49:16.000000000 -0400
+++ tarantula.less_mem/lib/relevance/tarantula/result.rb        2008-05-04 00:04:11.000000000 -0400
@@ -30,7 +30,7 @@
     "#{DEFAULT_LOCALHOST}#{url}"
   end
   ALLOW_NNN_FOR = /^allow_(\d\d\d)_for$/
-  class <<self
+  class << self
     attr_accessor :next_number
     def handle(result)
       retval = result.dup
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/lib/relevance/tarantula.rb tarantula.less_mem/lib/relevance/tarantula.rb
--- tarantula.orig/lib/relevance/tarantula.rb   2008-05-03 18:49:16.000000000 -0400
+++ tarantula.less_mem/lib/relevance/tarantula.rb       2008-05-04 00:04:11.000000000 -0400
@@ -47,6 +47,7 @@
 
 require 'relevance/tarantula/html_reporter'
 require 'relevance/tarantula/html_report_helper'
+require 'relevance/tarantula/io_reporter'
 require 'relevance/tarantula/recording'
 require 'relevance/tarantula/response'
 require 'relevance/tarantula/result'
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/test/relevance/core_extensions/test_case_test.rb tarantula.less_mem/test/relevance/core_extensions/test_case_test.rb
--- tarantula.orig/test/relevance/core_extensions/test_case_test.rb     2008-05-03 18:49:15.000000000 -0400
+++ tarantula.less_mem/test/relevance/core_extensions/test_case_test.rb 2008-05-04 00:04:11.000000000 -0400
@@ -4,6 +4,7 @@
 describe "TestCase extensions" do
   it "can create the crawler" do 
     RailsIntegrationProxy.stubs(:rails_root).returns("STUB_RAILS_ROOT")
+    Crawler.any_instance.stubs(:rails_root).returns("STUB_RAILS_ROOT")
     tarantula_crawler(stub_everything)
   end
   
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/test/relevance/tarantula/crawler_test.rb tarantula.less_mem/test/relevance/tarantula/crawler_test.rb
--- tarantula.orig/test/relevance/tarantula/crawler_test.rb     2008-05-03 18:49:16.000000000 -0400
+++ tarantula.less_mem/test/relevance/tarantula/crawler_test.rb 2008-05-04 00:06:59.000000000 -0400
@@ -98,12 +98,9 @@
 end
 
 describe 'Relevance::Tarantula::Crawler#report_results' do
-  it "delegates to generate_reports then report_to_console" do
+  it "delegates to generate_reports" do
     crawler = Crawler.new
-    crawler.expects(:report_dir)
-    crawler.expects(:puts)
     crawler.expects(:generate_reports)
-    crawler.expects(:report_to_console)
     crawler.report_results
   end
 end
@@ -187,21 +184,14 @@
     crawler.do_crawl
   end
   
-  it "reports errors to stderr and then raises" do
-    crawler = Crawler.new
-    crawler.failures << stub(:code => "404", :url => "/uh-oh")
-    $stderr.expects(:puts).with("****** FAILURES")
-    $stderr.expects(:puts).with("404: /uh-oh")
-    lambda {crawler.report_to_console}.should.raise RuntimeError
-  end
-  
   it "asks each reporter to write its report in report_dir" do
     crawler = Crawler.new
-    crawler.failures << stub(:code => "404", :url => "/uh-oh")
     crawler.stubs(:report_dir).returns(test_output_dir)
     reporter = stub_everything
     reporter.expects(:report)
+    reporter.expects(:finish_report)
     crawler.reporters = [reporter]
+    crawler.save_result stub(:code => "404", :url => "/uh-oh")
     crawler.generate_reports
   end
   
@@ -286,4 +276,4 @@
     crawler = Crawler.new
     lambda{crawler.foo}.should.raise(NoMethodError)
   end
-end
\ No newline at end of file
+end
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/test/relevance/tarantula/html_reporter_test.rb tarantula.less_mem/test/relevance/tarantula/html_reporter_test.rb
--- tarantula.orig/test/relevance/tarantula/html_reporter_test.rb       2008-05-03 18:49:16.000000000 -0400
+++ tarantula.less_mem/test/relevance/tarantula/html_reporter_test.rb   2008-05-04 00:04:11.000000000 -0400
@@ -6,7 +6,7 @@
     FileUtils.rm_rf(test_output_dir)
     FileUtils.mkdir_p(test_output_dir)
     Relevance::Tarantula::Result.next_number = 0
-    @results = (1..10).map do |index|
+    @success_results = (1..10).map do |index|
       Relevance::Tarantula::Result.new(
         :success => true, 
         :method => "get", 
@@ -23,6 +23,23 @@
         :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)}", 
+        :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, "1.html")
@@ -31,8 +48,10 @@
   
   it "creates a report based on tarantula results" do    
     Relevance::Tarantula::Result.any_instance.stubs(:rails_root).returns("STUB_ROOT")
-    results = stub(:successes => @results, :failures => @results)
-    Relevance::Tarantula::HtmlReporter.report(test_output_dir, results)
+    # results = stub(:successes => @results, :failures => @results)
+    reporter = Relevance::Tarantula::HtmlReporter.new(test_output_dir)
+    (@success_results + @fail_results).each {|r| reporter.report(r)}
+    reporter.finish_report
     File.should.exist @index
     File.should.exist @detail
   end
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/test/relevance/tarantula/io_reporter_test.rb tarantula.less_mem/test/relevance/tarantula/io_reporter_test.rb
--- tarantula.orig/test/relevance/tarantula/io_reporter_test.rb 1969-12-31 19:00:00.000000000 -0500
+++ tarantula.less_mem/test/relevance/tarantula/io_reporter_test.rb     2008-05-04 00:04:11.000000000 -0400
@@ -0,0 +1,12 @@
+require File.join(File.dirname(__FILE__), "..", "..", "test_helper.rb")
+include Relevance::Tarantula
+
+describe 'Relevance::Tarantula::IOReporter' do
+  it "reports errors to stderr and then raises" do
+    reporter = 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}.should.raise RuntimeError
+  end
+end
diff -Naru --exclude=.svn --exclude='*~' tarantula.orig/test/relevance/tarantula/rails_integration_proxy_test.rb tarantula.less_mem/test/relevance/tarantula/rails_integration_proxy_test.rb
--- tarantula.orig/test/relevance/tarantula/rails_integration_proxy_test.rb     2008-05-03 18:49:16.000000000 -0400
+++ tarantula.less_mem/test/relevance/tarantula/rails_integration_proxy_test.rb 2008-05-04 00:04:11.000000000 -0400
@@ -4,6 +4,7 @@
 describe "Relevance::Tarantula::RailsIntegrationProxy rails_integration_test" do
   before {
     Crawler.any_instance.stubs(:crawl)
+    Crawler.any_instance.stubs(:rails_root).returns("STUB_RAILS_ROOT")
     RailsIntegrationProxy.stubs(:rails_root).returns("STUB_RAILS_ROOT")
   }