Changeset 284

Show
Ignore:
Timestamp:
03/21/08 10:22:36 (5 months ago)
Author:
stu
Message:

fixes #5

Files:

Legend:

Unmodified
Added
Removed
Modified
Copied
Moved

  • rubygems/tarantula/trunk/lib/relevance/tarantula/crawler.rb

    r272 r284  
    2222    @referrers = {} 
    2323    @skip_uri_patterns =[ 
     24      /^javascript/, 
    2425      /^mailto/, 
    2526      /^http/,                                       

  • rubygems/tarantula/trunk/lib/relevance/tarantula/html_report_helper.rb

    r273 r284  
    2525                                                                             
    2626  def textmate_url(file, line_no) 
    27     "txmt://open?url=#{File.expand_path(File.join(rails_root,file))}&line_no=#{line_no}" 
     27    "txmt://open?url=file://#{File.expand_path(File.join(rails_root,file))}&line_no=#{line_no}" 
    2828  end 
    2929   

  • rubygems/tarantula/trunk/test/relevance/tarantula/crawler_test.rb

    r272 r284  
    219219  end 
    220220   
     221end 
     222                          
     223describe "Crawler link skipping" do    
     224  before do 
     225    @crawler = Crawler.new 
     226  end 
     227   
    221228  it "skips links that are too long" do 
    222     crawler = Crawler.new 
    223     crawler.should_skip_link?("/foo").should == false 
    224     crawler.max_url_length = 2 
    225     crawler.expects(:log).with("Skipping long url /foo") 
    226     crawler.should_skip_link?("/foo").should == true 
     229    @crawler.should_skip_link?("/foo").should == false 
     230    @crawler.max_url_length = 2 
     231    @crawler.expects(:log).with("Skipping long url /foo") 
     232    @crawler.should_skip_link?("/foo").should == true 
    227233  end 
    228234   
    229235  it "skips outbound links (those that begin with http)" do 
    230     crawler = Crawler.new 
    231     crawler.expects(:log).with("Skipping http-anything") 
    232     crawler.should_skip_link?("http-anything").should == true 
     236    @crawler.expects(:log).with("Skipping http-anything") 
     237    @crawler.should_skip_link?("http-anything").should == true 
     238  end 
     239 
     240  it "skips javascript links (those that begin with javascript)" do 
     241    @crawler.expects(:log).with("Skipping javascript-anything") 
     242    @crawler.should_skip_link?("javascript-anything").should == true 
    233243  end 
    234244 
    235245  it "skips mailto links (those that begin with http)" do 
    236     crawler = Crawler.new 
    237     crawler.expects(:log).with("Skipping mailto-anything") 
    238     crawler.should_skip_link?("mailto-anything").should == true 
     246    @crawler.expects(:log).with("Skipping mailto-anything") 
     247    @crawler.should_skip_link?("mailto-anything").should == true 
    239248  end 
    240249   
    241250  it 'skips blank links' do 
    242     crawler = Crawler.new 
    243     crawler.queue_link(nil) 
    244     crawler.links_to_crawl.should == [] 
    245     crawler.queue_link("") 
    246     crawler.links_to_crawl.should == [] 
     251    @crawler.queue_link(nil) 
     252    @crawler.links_to_crawl.should == [] 
     253    @crawler.queue_link("") 
     254    @crawler.links_to_crawl.should == [] 
    247255  end 
    248256   
    249257  it "logs and skips links that match a pattern" do 
    250     crawler = Crawler.new 
    251     crawler.expects(:log).with("Skipping /the-red-button") 
    252     crawler.skip_uri_patterns << /red-button/ 
    253     crawler.queue_link("/blue-button").should == "/blue-button" 
    254     crawler.queue_link("/the-red-button").should == nil 
     258    @crawler.expects(:log).with("Skipping /the-red-button") 
     259    @crawler.skip_uri_patterns << /red-button/ 
     260    @crawler.queue_link("/blue-button").should == "/blue-button" 
     261    @crawler.queue_link("/the-red-button").should == nil 
    255262  end    
    256263   
    257264  it "logs and skips form submissions that match a pattern" do 
    258     crawler = Crawler.new 
    259     crawler.expects(:log).with("Skipping /reset-password-form") 
    260     crawler.skip_uri_patterns << /reset-password/              
     265    @crawler.expects(:log).with("Skipping /reset-password-form") 
     266    @crawler.skip_uri_patterns << /reset-password/              
    261267    fs = stub_everything(:action => "/reset-password-form") 
    262     crawler.should_skip_form_submission?(fs).should == true 
    263   end 
    264    
     268    @crawler.should_skip_form_submission?(fs).should == true 
     269  end 
    265270end 
    266271