tenderlove-mechanize 0.9.3.20090617085936

data/test/test_hash_api.rb

@@ -0,0 +1,45 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+
+class TestHashApi < Test::Unit::TestCase
+  def setup
+    @agent = WWW::Mechanize.new
+  end
+
+  def test_title
+    page = @agent.get(:url => "http://localhost/file_upload.html")
+    assert_equal('File Upload Form', page.title)
+  end
+
+  def test_page_gets_yielded
+    pages = nil
+    @agent.get(:url => "http://localhost/file_upload.html") { |page|
+      pages = page
+    }
+    assert pages
+    assert_equal('File Upload Form', pages.title)
+  end
+
+  def test_get_with_params
+    page = @agent.get(:url => 'http://localhost/', :params => { :q => 'hello' })
+    assert_equal('http://localhost/?q=hello', page.uri.to_s)
+  end
+
+  def test_get_with_referer
+    request = nil
+    @agent.pre_connect_hooks << lambda { |params|
+      request = params[:request]
+    }
+
+    @agent.get( :url => 'http://localhost/',
+                :referer => URI.parse('http://google.com/')
+              )
+
+    assert request
+    assert_equal 'http://google.com/', request['Referer']
+
+    @agent.get( :url => 'http://localhost/',
+                :params => [],
+                :referer => 'http://tenderlovemaking.com/')
+    assert_equal 'http://tenderlovemaking.com/', request['Referer']
+  end
+end

data/test/test_history.rb

@@ -0,0 +1,142 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+
+class TestHistory < Test::Unit::TestCase
+  def setup
+    @agent    = WWW::Mechanize.new
+    @history  = WWW::Mechanize::History.new
+  end
+
+  def test_push
+    assert_equal(0, @history.length)
+
+    page = @agent.get("http://localhost/tc_bad_links.html")
+    x = @history.push(page)
+    assert_equal(x, @history)
+    assert_equal(1, @history.length)
+    assert(@history.visited?(page))
+    assert(@history.visited?(page.uri))
+    assert(@history.visited?(page.uri.to_s))
+    assert_equal(page, @history.visited_page(page))
+    assert_equal(page, @history.visited_page(page.uri))
+    assert_equal(page, @history.visited_page(page.uri.to_s))
+
+    @history.push(@agent.get("/tc_bad_links.html"))
+    assert_equal(2, @history.length)
+  end
+
+  def test_shift
+    assert_equal(0, @history.length)
+    page = @agent.get("http://localhost/tc_bad_links.html")
+    @history.push(page)
+    assert_equal(1, @history.length)
+
+    @history.push(@agent.get("/tc_bad_links.html"))
+    assert_equal(2, @history.length)
+
+    @history.push(@agent.get("/index.html"))
+    assert_equal(3, @history.length)
+
+    page2 = @history.shift
+    assert_equal(page, page2)
+    assert_equal(2, @history.length)
+
+    @history.shift
+    assert_equal(1, @history.length)
+    assert_equal(false, @history.visited?(page))
+
+    @history.shift
+    assert_equal(0, @history.length)
+
+    assert_nil(@history.shift)
+    assert_equal(0, @history.length)
+  end
+
+  def test_pop
+    assert_equal(0, @history.length)
+    page = @agent.get("http://localhost/tc_bad_links.html")
+    @history.push(page)
+    assert_equal(1, @history.length)
+
+    page2 = @agent.get("/index.html")
+    @history.push(page2)
+    assert_equal(2, @history.length)
+    assert_equal(page2, @history.pop)
+    assert_equal(1, @history.length)
+    assert_equal(true, @history.visited?(page))
+    assert_equal(false, @history.visited?(page2))
+    assert_equal(page, @history.pop)
+    assert_equal(0, @history.length)
+    assert_equal(false, @history.visited?(page))
+    assert_equal(false, @history.visited?(page2))
+    assert_nil(@history.pop)
+  end
+
+  def test_max_size
+    @history  = WWW::Mechanize::History.new(10)
+    1.upto(20) do |i|
+      page = @agent.get('http://localhost/index.html')
+      @history.push page
+      assert_equal(true, @history.visited?(page))
+      if i < 10
+        assert_equal(i, @history.length)
+      else
+        assert_equal(10, @history.length)
+      end
+    end
+
+    @history.clear
+    @history.max_size = 5
+    1.upto(20) do |i|
+      page = @agent.get('http://localhost/index.html')
+      @history.push page
+      assert_equal(true, @history.visited?(page))
+      if i < 5
+        assert_equal(i, @history.length)
+      else
+        assert_equal(5, @history.length)
+      end
+    end
+
+    @history.max_size = 0
+    1.upto(20) do |i|
+      page = @agent.get('http://localhost/index.html')
+      @history.push page
+      assert_equal(false, @history.visited?(page))
+      assert_equal(0, @history.length)
+    end
+  end
+
+  def test_no_slash
+    page = @agent.get('http://localhost')
+
+    node = Struct.new(:href, :inner_text).new('http://localhost/', 'blah')
+    link = WWW::Mechanize::Page::Link.new(node, nil, nil)
+    assert(@agent.visited?(link))
+
+    node = Struct.new(:href, :inner_text).new('http://localhost', 'blah')
+    link = WWW::Mechanize::Page::Link.new(node, nil, nil)
+    assert(@agent.visited?(link))
+  end
+
+  def test_with_slash
+    page = @agent.get('http://localhost/')
+
+    node = Struct.new(:href, :inner_text).new('http://localhost/', 'blah')
+    link = WWW::Mechanize::Page::Link.new(node, nil, nil)
+    assert(@agent.visited?(link))
+
+    node = Struct.new(:href, :inner_text).new('http://localhost', 'blah')
+    link = WWW::Mechanize::Page::Link.new(node, nil, nil)
+    assert(@agent.visited?(link))
+  end
+
+  def test_clear
+    page = nil
+    20.times { @history.push(page = @agent.get('http://localhost/index.html')) }
+    assert_equal(20, @history.length)
+    assert_equal(true, @history.visited?(page))
+    @history.clear
+    assert_equal(0, @history.length)
+    assert_equal(false, @history.visited?(page))
+  end
+end

data/test/test_history_added.rb

@@ -0,0 +1,16 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+
+class HistoryAddedTest < Test::Unit::TestCase
+  def setup
+    @agent = WWW::Mechanize.new
+  end
+
+  def test_history_added_gets_called
+    onload = 0
+    @agent.history_added = lambda { |page|
+      onload += 1
+    }
+    page = @agent.get('http://localhost/tc_blank_form.html')
+    assert_equal(1, onload)
+  end
+end

data/test/test_html_unscape_forms.rb

@@ -0,0 +1,39 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+
+class TestCheckBoxes < Test::Unit::TestCase
+  def test_field
+    f = WWW::Mechanize::Form::Field.new('a&amp;b', 'a&amp;b')
+    assert_equal('a&b', f.name)
+    assert_equal('a&b', f.value)
+
+    f = WWW::Mechanize::Form::Field.new('a&b', 'a&b')
+    assert_equal('a&b', f.name)
+    assert_equal('a&b', f.value)
+
+    f = WWW::Mechanize::Form::Field.new('a&#38;b', 'a&#38;b')
+    assert_equal('a&b', f.name)
+    assert_equal('a&b', f.value)
+  end
+
+  def test_file_upload
+    f = WWW::Mechanize::Form::FileUpload.new('a&amp;b', 'a&amp;b')
+    assert_equal('a&b', f.name)
+    assert_equal('a&b', f.file_name)
+
+    f = WWW::Mechanize::Form::FileUpload.new('a&b', 'a&b')
+    assert_equal('a&b', f.name)
+    assert_equal('a&b', f.file_name)
+  end
+
+  def test_image_button
+    f = WWW::Mechanize::Form::ImageButton.new('a&amp;b', 'a&amp;b')
+    assert_equal('a&b', f.name)
+    assert_equal('a&b', f.value)
+  end
+
+  def test_radio_button
+    f = WWW::Mechanize::Form::RadioButton.new('a&amp;b', 'a&amp;b', nil, nil)
+    assert_equal('a&b', f.name)
+    assert_equal('a&b', f.value)
+  end
+end

data/test/test_if_modified_since.rb

@@ -0,0 +1,20 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+
+class TestIfModifiedSince < Test::Unit::TestCase
+  def setup
+    @agent = WWW::Mechanize.new
+  end
+
+  def test_get_twice
+    assert_equal(0, @agent.history.length)
+    page = @agent.get('http://localhost/if_modified_since')
+    assert_match(/You did not send/, page.body)
+
+    assert_equal(1, @agent.history.length)
+    page2 = @agent.get('http://localhost/if_modified_since')
+
+    assert_equal(2, @agent.history.length)
+    assert_equal(page.object_id, page2.object_id)
+    assert_match(/You did not send/, page.body)
+  end
+end

data/test/test_keep_alive.rb

@@ -0,0 +1,31 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+
+class TestKeepAlive < Test::Unit::TestCase
+  def setup
+    @agent = WWW::Mechanize.new
+  end
+
+  def test_keep_alive
+    page = @agent.get('http://localhost/http_headers')
+    headers = {}
+    page.body.split(/[\r\n]+/).each do |header|
+      headers.[]=(*header.chomp.split(/\|/))
+    end
+    assert(headers.has_key?('connection'))
+    assert_equal('keep-alive', headers['connection'])
+    assert(headers.has_key?('keep-alive'))
+    assert_equal('300', headers['keep-alive'])
+  end
+
+  def test_close_connection
+    @agent.keep_alive = false
+    page = @agent.get('http://localhost/http_headers')
+    headers = {}
+    page.body.split(/[\r\n]+/).each do |header|
+      headers.[]=(*header.chomp.split(/\|/))
+    end
+    assert(headers.has_key?('connection'))
+    assert_equal('close', headers['connection'])
+    assert(!headers.has_key?('keep-alive'))
+  end
+end

data/test/test_links.rb

@@ -0,0 +1,120 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+
+class LinksMechTest < Test::Unit::TestCase
+  def setup
+    @agent = WWW::Mechanize.new
+  end
+
+  def test_unsupported_link_types
+    page = @agent.get("http://google.com/tc_links.html")
+    link = page.link_with(:text => 'javascript link')
+    assert_raise(WWW::Mechanize::UnsupportedSchemeError) {
+      link.click
+    }
+
+    @agent.scheme_handlers['javascript'] = lambda { |my_link, my_page|
+      URI.parse('http://localhost/tc_links.html')
+    }
+    assert_nothing_raised {
+      link.click
+    }
+  end
+
+  def test_link_with_no_path
+    page = @agent.get("http://localhost/relative/tc_relative_links.html")
+    page = page.link_with(:text => 'just the query string').click
+    assert_equal('http://localhost/relative/tc_relative_links.html?a=b', page.uri.to_s)
+  end
+
+  def test_base
+    page = @agent.get("http://google.com/tc_base_link.html")
+    page = page.links.first.click
+    assert @agent.visited?("http://localhost/index.html")
+  end
+
+  def test_find_meta
+    page = @agent.get("http://localhost/find_link.html")
+    assert_equal(3, page.meta.length)
+    assert_equal(%w{
+      http://www.drphil.com/
+      http://www.upcase.com/
+      http://tenderlovemaking.com/ }.sort,
+      page.meta.map { |x| x.href.downcase }.sort)
+  end
+
+  def test_find_link
+    page = @agent.get("http://localhost/find_link.html")
+    assert_equal(18, page.links.length)
+  end
+
+  def test_alt_text
+    page = @agent.get("http://localhost/alt_text.html")
+    assert_equal(5, page.links.length)
+    assert_equal(1, page.meta.length)
+
+    assert_equal('', page.meta.first.text)
+    assert_equal('alt text', page.link_with(:href => 'alt_text.html').text)
+    assert_equal('', page.link_with(:href => 'no_alt_text.html').text)
+    assert_equal('no image', page.link_with(:href => 'no_image.html').text)
+    assert_equal('', page.link_with(:href => 'no_text.html').text)
+    assert_equal('', page.link_with(:href => 'nil_alt_text.html').text)
+  end
+
+  def test_click_link
+    @agent.user_agent_alias = 'Mac Safari'
+    page = @agent.get("http://localhost/frame_test.html")
+    link = page.link_with(:text => "Form Test")
+    assert_not_nil(link)
+    assert_equal('Form Test', link.text)
+    page = @agent.click(link)
+    assert_equal("http://localhost/form_test.html",
+      @agent.history.last.uri.to_s)
+  end
+
+  def test_click_method
+    page = @agent.get("http://localhost/frame_test.html")
+    link = page.link_with(:text => "Form Test")
+    assert_not_nil(link)
+    assert_equal('Form Test', link.text)
+    page = link.click
+    assert_equal("http://localhost/form_test.html",
+      @agent.history.last.uri.to_s)
+  end
+
+  def test_find_bold_link
+    page = @agent.get("http://localhost/tc_links.html")
+    link = page.links_with(:text => /Bold Dude/)
+    assert_equal(1, link.length)
+    assert_equal('Bold Dude', link.first.text)
+
+    link = page.links_with(:text => 'Aaron James Patterson')
+    assert_equal(1, link.length)
+    assert_equal('Aaron James Patterson', link.first.text)
+
+    link = page.links_with(:text => 'Aaron Patterson')
+    assert_equal(1, link.length)
+    assert_equal('Aaron Patterson', link.first.text)
+
+    link = page.links_with(:text => 'Ruby Rocks!')
+    assert_equal(1, link.length)
+    assert_equal('Ruby Rocks!', link.first.text)
+  end
+
+  def test_link_with_encoded_space
+    page = @agent.get("http://localhost/tc_links.html")
+    link = page.link_with(:text => 'encoded space')
+    page = @agent.click link
+  end
+
+  def test_link_with_space
+    page = @agent.get("http://localhost/tc_links.html")
+    link = page.link_with(:text => 'not encoded space')
+    page = @agent.click link
+  end
+
+  def test_link_with_unusual_characters
+    page = @agent.get("http://localhost/tc_links.html")
+    link = page.link_with(:text => 'unusual characters')
+    assert_nothing_raised { @agent.click link }
+  end
+end

data/test/test_mech.rb

@@ -0,0 +1,268 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+
+class TestMechMethods < Test::Unit::TestCase
+  def setup
+    @agent = WWW::Mechanize.new
+  end
+
+  def test_get_with_tilde
+    page = @agent.get('http://localhost/?foo=~2')
+    assert_equal('http://localhost/?foo=~2', page.uri.to_s)
+  end
+
+  def test_parser_can_be_set
+    @agent.html_parser = {}
+    assert_raises(NoMethodError) {
+      @agent.get('http://localhost/?foo=~2').links
+    }
+  end
+
+  def test_submit_takes_arbirary_headers
+    page = @agent.get('http://localhost:2000/form_no_action.html')
+    assert form = page.forms.first
+    form.action = '/http_headers'
+    page = @agent.submit(form, nil, { 'foo' => 'bar' })
+    headers = Hash[*(
+      page.body.split("\n").map { |x| x.split('|') }.flatten
+    )]
+    assert_equal 'bar', headers['foo']
+  end
+
+  def test_get_with_params
+    page = @agent.get('http://localhost/', { :q => 'hello' })
+    assert_equal('http://localhost/?q=hello', page.uri.to_s)
+  end
+
+  def test_get_with_upper_http
+    page = @agent.get('HTTP://localhost/', { :q => 'hello' })
+    assert_equal('HTTP://localhost/?q=hello', page.uri.to_s)
+  end
+
+  def test_get_no_referer
+    requests = []
+    @agent.pre_connect_hooks << lambda { |params|
+      requests << params[:request]
+    }
+
+    @agent.get('http://localhost/')
+    @agent.get('http://localhost/')
+    assert_equal(2, requests.length)
+    requests.each do |request|
+      assert_nil request['referer']
+    end
+  end
+
+  def test_with_anchor
+    page = @agent.get('http://localhost/?foo=bar&#34;')
+    assert_equal('http://localhost/?foo=bar%22', page.uri.to_s)
+  end
+
+  def test_post_connect_hook_gets_called
+    response = nil
+    @agent.post_connect_hooks << lambda { |params|
+      response = params[:response]
+    }
+
+    @agent.get('http://localhost/')
+    assert(response)
+  end
+
+  def test_get_with_referer
+    request = nil
+    @agent.pre_connect_hooks << lambda { |params|
+      request = params[:request]
+    }
+
+    @agent.get('http://localhost/', URI.parse('http://google.com/'))
+    assert_equal 'http://google.com/', request['Referer']
+
+    @agent.get('http://localhost/', [], 'http://tenderlovemaking.com/')
+    assert_equal 'http://tenderlovemaking.com/', request['Referer']
+  end
+  
+  def test_get_with_file_referer
+    assert_nothing_raised do
+      @agent.get('http://localhost', [], WWW::Mechanize::File.new(URI.parse('http://tenderlovemaking.com/crossdomain.xml')))
+    end
+  end
+
+  def test_weird_url
+    assert_nothing_raised {
+      @agent.get('http://localhost/?action=bing&bang=boom=1|a=|b=|c=')
+    }
+    assert_nothing_raised {
+      @agent.get('http://localhost/?a=b&#038;b=c&#038;c=d')
+    }
+    assert_nothing_raised {
+      @agent.get("http://localhost/?a=#{[0xd6].pack('U')}")
+    }
+  end
+
+  unless RUBY_VERSION >= '1.9.0'
+    def test_kcode_url
+      $KCODE = 'u'
+      page = @agent.get("http://localhost/?a=#{[0xd6].pack('U')}")
+      assert_not_nil(page)
+      assert_equal('http://localhost/?a=%D6', page.uri.to_s)
+      $KCODE = 'NONE'
+    end
+  end
+
+  def test_history
+    0.upto(25) do |i|
+      assert_equal(i, @agent.history.size)
+      page = @agent.get("http://localhost/")
+    end
+    page = @agent.get("http://localhost/form_test.html")
+
+    assert_equal("http://localhost/form_test.html",
+      @agent.history.last.uri.to_s)
+    assert_equal("http://localhost/",
+      @agent.history[-2].uri.to_s)
+    assert_equal("http://localhost/",
+      @agent.history[-2].uri.to_s)
+
+    assert_equal(true, @agent.visited?("http://localhost/"))
+    assert_equal(true, @agent.visited?("/form_test.html"))
+    assert_equal(false, @agent.visited?("http://google.com/"))
+    assert_equal(true, @agent.visited?(page.links.first))
+
+  end
+
+  def test_visited
+    @agent.get("http://localhost/content_type_test?ct=application/pdf")
+    assert_equal(true,
+      @agent.visited?("http://localhost/content_type_test?ct=application/pdf"))
+    assert_equal(false,
+      @agent.visited?("http://localhost/content_type_test"))
+    assert_equal(false,
+      @agent.visited?("http://localhost/content_type_test?ct=text/html"))
+  end
+
+  def test_visited_after_redirect
+    @agent.get("http://localhost/response_code?code=302")
+    assert_equal("http://localhost/index.html",
+      @agent.current_page.uri.to_s)
+    assert_equal(true,
+                 @agent.visited?('http://localhost/response_code?code=302'))
+  end
+
+  def test_max_history
+    @agent.max_history = 10
+    0.upto(10) do |i|
+      assert_equal(i, @agent.history.size)
+      page = @agent.get("http://localhost/")
+    end
+    
+    0.upto(10) do |i|
+      assert_equal(10, @agent.history.size)
+      page = @agent.get("http://localhost/")
+    end
+  end
+
+  def test_max_history_order
+    @agent.max_history = 2
+    assert_equal(0, @agent.history.length)
+
+    @agent.get('http://localhost/form_test.html')
+    assert_equal(1, @agent.history.length)
+
+    @agent.get('http://localhost/empty_form.html')
+    assert_equal(2, @agent.history.length)
+
+    @agent.get('http://localhost/tc_checkboxes.html')
+    assert_equal(2, @agent.history.length)
+    assert_equal('http://localhost/empty_form.html', @agent.history[0].uri.to_s)
+    assert_equal('http://localhost/tc_checkboxes.html',
+                 @agent.history[1].uri.to_s)
+  end
+
+  def test_back_button
+    0.upto(5) do |i|
+      assert_equal(i, @agent.history.size)
+      page = @agent.get("http://localhost/")
+    end
+    page = @agent.get("http://localhost/form_test.html")
+
+    assert_equal("http://localhost/form_test.html",
+      @agent.history.last.uri.to_s)
+    assert_equal("http://localhost/",
+      @agent.history[-2].uri.to_s)
+
+    assert_equal(7, @agent.history.size)
+    @agent.back
+    assert_equal(6, @agent.history.size)
+    assert_equal("http://localhost/",
+      @agent.history.last.uri.to_s)
+  end
+
+  def test_google
+    page = @agent.get("http://localhost/google.html")
+    search = page.forms.find { |f| f.name == "f" }
+    assert_not_nil(search)
+    assert_not_nil(search.field_with(:name => 'q'))
+    assert_not_nil(search.field_with(:name => 'hl'))
+    assert_not_nil(search.fields.find { |f| f.name == 'ie' })
+  end
+
+  def test_click
+    @agent.user_agent_alias = 'Mac Safari'
+    page = @agent.get("http://localhost/frame_test.html")
+    link = page.link_with(:text => "Form Test")
+    assert_not_nil(link)
+    page = @agent.click(link)
+    assert_equal("http://localhost/form_test.html",
+      @agent.history.last.uri.to_s)
+  end
+
+  def test_click_hpricot
+    page = @agent.get("http://localhost/frame_test.html")
+
+    link = (page/"//a[@class='bar']").first
+    assert_not_nil(link)
+    page = @agent.click(link)
+    assert_equal("http://localhost/form_test.html",
+      @agent.history.last.uri.to_s)
+  end
+
+  def test_click_hpricot_frame
+    page = @agent.get("http://localhost/frame_test.html")
+
+    link = (page/"//frame[@name='frame2']").first
+    assert_not_nil(link)
+    page = @agent.click(link)
+    assert_equal("http://localhost/form_test.html",
+      @agent.history.last.uri.to_s)
+  end
+
+  def test_new_find
+    page = @agent.get("http://localhost/frame_test.html")
+    assert_equal(3, page.frames.size)
+
+    find_orig = page.frames.find_all { |f| f.name == 'frame1' }
+    find1 = page.frames_with(:name => 'frame1')
+
+    find_orig.zip(find1).each { |a,b|
+      assert_equal(a, b)
+    }
+  end
+
+  def test_get_file
+    page = @agent.get("http://localhost/frame_test.html")
+    content_length = page.header['Content-Length']
+    page_as_string = @agent.get_file("http://localhost/frame_test.html")
+    assert_equal(content_length.to_i, page_as_string.length.to_i)
+  end
+
+  def test_transact
+    page = @agent.get("http://localhost/frame_test.html")
+    assert_equal(1, @agent.history.length)
+    @agent.transact { |a|
+      5.times {
+        @agent.get("http://localhost/frame_test.html")
+      }
+      assert_equal(6, @agent.history.length)
+    }
+    assert_equal(1, @agent.history.length)
+  end
+end