RubyGems - mechanize - Versions diffs - 2.4 → 2.5 - Mend

mechanize 2.4 → 2.5

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of mechanize might be problematic. Click here for more details.

Files changed (57) hide show

data.tar.gz.sig +0 -0
data/CHANGELOG.rdoc +30 -4
data/EXAMPLES.rdoc +2 -2
data/GUIDE.rdoc +13 -13
data/Manifest.txt +30 -0
data/README.rdoc +19 -2
data/lib/mechanize.rb +30 -7
data/lib/mechanize/chunked_termination_error.rb +7 -0
data/lib/mechanize/form.rb +7 -5
data/lib/mechanize/form/radio_button.rb +14 -0
data/lib/mechanize/http/agent.rb +54 -24
data/lib/mechanize/http/auth_store.rb +6 -0
data/lib/mechanize/http/content_disposition_parser.rb +1 -1
data/lib/mechanize/monkey_patch.rb +3 -2
data/lib/mechanize/page.rb +0 -2
data/lib/mechanize/parser.rb +2 -2
data/lib/mechanize/test_case.rb +70 -435
data/lib/mechanize/test_case/.document +1 -0
data/lib/mechanize/test_case/bad_chunking_servlet.rb +14 -0
data/lib/mechanize/test_case/basic_auth_servlet.rb +24 -0
data/lib/mechanize/test_case/content_type_servlet.rb +8 -0
data/lib/mechanize/test_case/digest_auth_servlet.rb +33 -0
data/lib/mechanize/test_case/file_upload_servlet.rb +20 -0
data/lib/mechanize/test_case/form_servlet.rb +55 -0
data/lib/mechanize/test_case/gzip_servlet.rb +32 -0
data/lib/mechanize/test_case/header_servlet.rb +14 -0
data/lib/mechanize/test_case/http_refresh_servlet.rb +9 -0
data/lib/mechanize/test_case/infinite_redirect_servlet.rb +10 -0
data/lib/mechanize/test_case/infinite_refresh_servlet.rb +10 -0
data/lib/mechanize/test_case/many_cookies_as_string_servlet.rb +37 -0
data/lib/mechanize/test_case/many_cookies_servlet.rb +33 -0
data/lib/mechanize/test_case/modified_since_servlet.rb +21 -0
data/lib/mechanize/test_case/ntlm_servlet.rb +30 -0
data/lib/mechanize/test_case/one_cookie_no_spaces_servlet.rb +11 -0
data/lib/mechanize/test_case/one_cookie_servlet.rb +11 -0
data/lib/mechanize/test_case/quoted_value_cookie_servlet.rb +11 -0
data/lib/mechanize/test_case/redirect_servlet.rb +13 -0
data/lib/mechanize/test_case/referer_servlet.rb +12 -0
data/lib/mechanize/test_case/refresh_with_empty_url.rb +15 -0
data/lib/mechanize/test_case/refresh_without_url.rb +14 -0
data/lib/mechanize/test_case/response_code_servlet.rb +15 -0
data/lib/mechanize/test_case/send_cookies_servlet.rb +19 -0
data/lib/mechanize/test_case/server.rb +36 -0
data/lib/mechanize/test_case/servlets.rb +55 -0
data/lib/mechanize/test_case/verb_servlet.rb +11 -0
data/test/test_mechanize.rb +12 -12
data/test/test_mechanize_file.rb +11 -0
data/test/test_mechanize_file_response.rb +23 -0
data/test/test_mechanize_form.rb +34 -0
data/test/test_mechanize_form_radio_button.rb +19 -2
data/test/test_mechanize_http_agent.rb +104 -26
data/test/test_mechanize_http_auth_store.rb +23 -0
data/test/test_mechanize_http_content_disposition_parser.rb +6 -0
data/test/test_mechanize_page.rb +3 -5
data/test/test_mechanize_page_link.rb +0 -32
metadata +39 -8
metadata.gz.sig +2 -2

data/lib/mechanize/test_case/refresh_with_empty_url.rb ADDED Viewed

@@ -0,0 +1,15 @@
+class RefreshWithEmptyUrl < WEBrick::HTTPServlet::AbstractServlet
+  @@count = 0
+  def do_GET(req, res)
+    address = "#{req.host}:#{req.port}"
+    res.content_type = "text/html"
+    @@count += 1
+    if @@count > 1
+      res['Refresh'] = "0; url=http://#{address}/";
+    else
+      res['Refresh'] = "0; url=";
+    end
+  end
+end

data/lib/mechanize/test_case/refresh_without_url.rb ADDED Viewed

@@ -0,0 +1,14 @@
+class RefreshWithoutUrl < WEBrick::HTTPServlet::AbstractServlet
+  @@count = 0
+  def do_GET(req, res)
+    address = "#{req.host}:#{req.port}"
+    res['Content-Type'] = "text/html"
+    @@count += 1
+    if @@count > 1
+      res['Refresh'] = "0; url=http://#{address}/";
+    else
+      res['Refresh'] = "0";
+    end
+  end
+end

data/lib/mechanize/test_case/response_code_servlet.rb ADDED Viewed

@@ -0,0 +1,15 @@
+class ResponseCodeServlet < WEBrick::HTTPServlet::AbstractServlet
+  def do_GET(req, res)
+    res['Content-Type'] = req.query['ct'] || "text/html"
+    if req.query['code']
+      code = req.query['code'].to_i
+      case code
+      when 300, 301, 302, 303, 304, 305, 307
+        res['Location'] = "/index.html"
+      end
+      res.status = code
+    else
+    end
+  end
+end

data/lib/mechanize/test_case/send_cookies_servlet.rb ADDED Viewed

@@ -0,0 +1,19 @@
+class SendCookiesServlet < WEBrick::HTTPServlet::AbstractServlet
+  def do_GET(req, res)
+    res.content_type = 'text/html'
+    cookies = req.cookies.map do |c|
+      "<li><a href=\"#\">#{c.name}:#{c.value}</a>"
+    end.join "\n"
+    res.body = <<-BODY
+<!DOCTYPE html>
+<title>Your cookies</title>
+<ul>
+#{cookies}
+</ul>
+    BODY
+  end
+end

data/lib/mechanize/test_case/server.rb ADDED Viewed

@@ -0,0 +1,36 @@
+require 'webrick'
+require 'mechanize/test_case/servlets'
+server = WEBrick::HTTPServer.new :Port => 8000
+server.mount_proc '/' do |req, res|
+  res.content_type = 'text/html'
+  servlets = MECHANIZE_TEST_CASE_SERVLETS.map do |path, servlet|
+    "<dt>#{servlet}<dd><a href=\"#{path}\">#{path}</a>"
+  end.join "\n"
+  res.body = <<-BODY
+<!DOCTYPE html>
+<title>Mechanize Test Case Servlets</title>
+<p>This server allows you to test various mechanize behavior against other
+HTTP clients.  Some endpoints may require headers be set to have a reasonable
+function, or may respond diffently to POST vs GET requests.  Please see the
+servlet implementation and mechanize tests for further details.
+<p>Here are the servlet endpoints available:
+<dl>
+#{servlets}
+</dl>
+  BODY
+end
+MECHANIZE_TEST_CASE_SERVLETS.each do |path, servlet|
+  server.mount path, servlet
+end
+trap 'INT'  do server.shutdown end
+trap 'TERM' do server.shutdown end
+server.start

data/lib/mechanize/test_case/servlets.rb ADDED Viewed

@@ -0,0 +1,55 @@
+require 'mechanize/test_case/bad_chunking_servlet'
+require 'mechanize/test_case/basic_auth_servlet'
+require 'mechanize/test_case/content_type_servlet'
+require 'mechanize/test_case/digest_auth_servlet'
+require 'mechanize/test_case/file_upload_servlet'
+require 'mechanize/test_case/form_servlet'
+require 'mechanize/test_case/gzip_servlet'
+require 'mechanize/test_case/header_servlet'
+require 'mechanize/test_case/http_refresh_servlet'
+require 'mechanize/test_case/infinite_redirect_servlet'
+require 'mechanize/test_case/infinite_refresh_servlet'
+require 'mechanize/test_case/many_cookies_as_string_servlet'
+require 'mechanize/test_case/many_cookies_servlet'
+require 'mechanize/test_case/modified_since_servlet'
+require 'mechanize/test_case/ntlm_servlet'
+require 'mechanize/test_case/one_cookie_no_spaces_servlet'
+require 'mechanize/test_case/one_cookie_servlet'
+require 'mechanize/test_case/quoted_value_cookie_servlet'
+require 'mechanize/test_case/redirect_servlet'
+require 'mechanize/test_case/referer_servlet'
+require 'mechanize/test_case/refresh_with_empty_url'
+require 'mechanize/test_case/refresh_without_url'
+require 'mechanize/test_case/response_code_servlet'
+require 'mechanize/test_case/send_cookies_servlet'
+require 'mechanize/test_case/verb_servlet'
+MECHANIZE_TEST_CASE_SERVLETS = {
+  '/bad_chunking'           => BadChunkingServlet,
+  '/basic_auth'             => BasicAuthServlet,
+  '/content_type_test'      => ContentTypeServlet,
+  '/digest_auth'            => DigestAuthServlet,
+  '/file_upload'            => FileUploadServlet,
+  '/form post'              => FormServlet,
+  '/form_post'              => FormServlet,
+  '/gzip'                   => GzipServlet,
+  '/http_headers'           => HeaderServlet,
+  '/http_refresh'           => HttpRefreshServlet,
+  '/if_modified_since'      => ModifiedSinceServlet,
+  '/infinite_redirect'      => InfiniteRedirectServlet,
+  '/infinite_refresh'       => InfiniteRefreshServlet,
+  '/many_cookies'           => ManyCookiesServlet,
+  '/many_cookies_as_string' => ManyCookiesAsStringServlet,
+  '/ntlm'                   => NTLMServlet,
+  '/one_cookie'             => OneCookieServlet,
+  '/one_cookie_no_space'    => OneCookieNoSpacesServlet,
+  '/quoted_value_cookie'    => QuotedValueCookieServlet,
+  '/redirect'               => RedirectServlet,
+  '/referer'                => RefererServlet,
+  '/refresh_with_empty_url' => RefreshWithEmptyUrl,
+  '/refresh_without_url'    => RefreshWithoutUrl,
+  '/response_code'          => ResponseCodeServlet,
+  '/send_cookies'           => SendCookiesServlet,
+  '/verb'                   => VerbServlet,
+}

data/lib/mechanize/test_case/verb_servlet.rb ADDED Viewed

@@ -0,0 +1,11 @@
+class VerbServlet < WEBrick::HTTPServlet::AbstractServlet
+  %w[HEAD GET POST PUT DELETE].each do |verb|
+    eval <<-METHOD
+      def do_#{verb}(req, res)
+        res.header['X-Request-Method'] = #{verb.dump}
+        res.body = #{verb.dump}
+      end
+    METHOD
+  end
+end

data/test/test_mechanize.rb CHANGED Viewed

@@ -434,15 +434,15 @@ but not <a href="/" rel="me nofollow">this</a>!
     @mech.follow_meta_refresh = true
     @mech.follow_meta_refresh_self = true
-    page = @mech.get('http://localhost/refresh_with_empty_url')
+    page = @mech.get('http://example/refresh_with_empty_url')
     assert_equal(3, @mech.history.length)
-    assert_equal('http://localhost/refresh_with_empty_url',
+    assert_equal('http://example/refresh_with_empty_url',
                  @mech.history[0].uri.to_s)
-    assert_equal('http://localhost/refresh_with_empty_url',
+    assert_equal('http://example/refresh_with_empty_url',
                  @mech.history[1].uri.to_s)
-    assert_equal('http://localhost/index.html', page.uri.to_s)
-    assert_equal('http://localhost/index.html', @mech.history.last.uri.to_s)
+    assert_equal('http://example/', page.uri.to_s)
+    assert_equal('http://example/', @mech.history.last.uri.to_s)
   end
   def test_get_follow_meta_refresh_in_body
@@ -460,15 +460,15 @@ but not <a href="/" rel="me nofollow">this</a>!
     @mech.follow_meta_refresh = true
     @mech.follow_meta_refresh_self = true
-    page = @mech.get('http://localhost/refresh_without_url')
+    page = @mech.get('http://example/refresh_without_url')
     assert_equal(3, @mech.history.length)
-    assert_equal('http://localhost/refresh_without_url',
+    assert_equal('http://example/refresh_without_url',
                  @mech.history[0].uri.to_s)
-    assert_equal('http://localhost/refresh_without_url',
+    assert_equal('http://example/refresh_without_url',
                  @mech.history[1].uri.to_s)
-    assert_equal('http://localhost/index.html', page.uri.to_s)
-    assert_equal('http://localhost/index.html', @mech.history.last.uri.to_s)
+    assert_equal('http://example/', page.uri.to_s)
+    assert_equal('http://example/', @mech.history.last.uri.to_s)
   end
   def test_get_follow_meta_refresh_referer_not_sent
@@ -566,9 +566,9 @@ but not <a href="/" rel="me nofollow">this</a>!
       requests << request
     }
-    page = @mech.get('http://localhost/http_refresh?refresh_time=0')
+    page = @mech.get('http://example/http_refresh?refresh_time=0')
-    assert_equal('http://localhost/index.html', page.uri.to_s)
+    assert_equal('http://example/', page.uri.to_s)
     assert_equal(2, @mech.history.length)
     assert_nil requests.last['referer']
   end

data/test/test_mechanize_file.rb CHANGED Viewed

@@ -16,7 +16,18 @@ class TestMechanizeFile < Mechanize::TestCase
       Dir.chdir dir do
         page.save 'test.html'
+        assert File.exist? 'test.html'
         assert_equal '0123456789', File.read('test.html')
+        page.save 'test.html'
+        assert File.exist? 'test.html.1'
+        assert_equal '0123456789', File.read('test.html.1')
+        page.save 'test.html'
+        assert File.exist? 'test.html.2'
+        assert_equal '0123456789', File.read('test.html.2')
       end
     end
   end

data/test/test_mechanize_file_response.rb ADDED Viewed

@@ -0,0 +1,23 @@
+require 'mechanize/test_case'
+class TestMechanizeFileResponse < Mechanize::TestCase
+  def test_content_type
+    Tempfile.open %w[pi .nothtml] do |tempfile|
+      res = Mechanize::FileResponse.new tempfile.path
+      assert_equal nil, res['content-type']
+    end
+    Tempfile.open %w[pi .xhtml] do |tempfile|
+      res = Mechanize::FileResponse.new tempfile.path
+      assert_equal 'text/html', res['content-type']
+    end
+    Tempfile.open %w[pi .html] do |tempfile|
+      res = Mechanize::FileResponse.new tempfile.path
+      assert_equal 'text/html', res['Content-Type']
+    end
+  end
+end

data/test/test_mechanize_form.rb CHANGED Viewed

@@ -65,6 +65,40 @@ class TestMechanizeForm < Mechanize::TestCase
     assert query.all? { |x| x[1] == '' }
   end
+  def test_build_query_radio_button_duplicate
+    html = Nokogiri::HTML <<-HTML
+<form>
+  <input type=radio name=name value=a checked=true>
+  <input type=radio name=name value=a checked=true>
+</form>
+    HTML
+    form = Mechanize::Form.new html.at('form'), @mech, @page
+    query = form.build_query
+    assert_equal [%w[name a]], query
+  end
+  def test_build_query_radio_button_multiple_checked
+    html = Nokogiri::HTML <<-HTML
+<form>
+  <input type=radio name=name value=a checked=true>
+  <input type=radio name=name value=b checked=true>
+</form>
+    HTML
+    form = Mechanize::Form.new html.at('form'), @mech, @page
+    e = assert_raises Mechanize::Error do
+      form.build_query
+    end
+    assert_equal 'radiobuttons "a, b" are checked in the "name" group, ' \
+                 'only one is allowed',
+                 e.message
+  end
   def test_method_missing_get
     page = html_page <<-BODY
 <form>

data/test/test_mechanize_form_radio_button.rb CHANGED Viewed

@@ -12,7 +12,8 @@ class TestMechanizeFormRadioButton < Mechanize::TestCase
   <input type="radio" name="color" value="brown">
   <input type="radio" name="color" value="green">
   <input type="radio" name="color" value="red">
-  <input type="radio" name="color" value="yellow">
+  <input type="radio" name="color" value="yellow" id="a">
+  <input type="radio" name="color" value="yellow" id="b">
   <input type="submit" value="Submit">
 </form>
@@ -24,7 +25,7 @@ class TestMechanizeFormRadioButton < Mechanize::TestCase
     @brown  = @form.radiobutton_with :value => 'brown'
     @green  = @form.radiobutton_with :value => 'green'
     @red    = @form.radiobutton_with :value => 'red'
-    @yellow = @form.radiobutton_with :value => 'yellow'
+    @yellow = @form.radiobutton_with :id    => 'a'
   end
   def test_check
@@ -58,6 +59,22 @@ class TestMechanizeFormRadioButton < Mechanize::TestCase
     refute @blue.checked?
   end
+  def test_equals2
+    refute_equal @yellow, @red
+    other_yellow = @form.radiobutton_with :id => 'b'
+    assert_equal @yellow, other_yellow
+  end
+  def test_hash
+    refute_equal @yellow.hash, @red.hash
+    other_yellow = @form.radiobutton_with :id => 'b'
+    assert_equal @yellow.hash, other_yellow.hash
+  end
   def test_label
     assert_equal 'Blue', @blue.label.text
   end

data/test/test_mechanize_http_agent.rb CHANGED Viewed

@@ -218,6 +218,14 @@ class TestMechanizeHttpAgent < Mechanize::TestCase
     @agent.fetch URI 'http://example/index.html'
   end
+  def test_fetch_ignore_bad_chunking
+    @agent.ignore_bad_chunking = true
+    file = @agent.fetch 'http://example/bad_chunking'
+    assert_equal '0123456789', file.content
+  end
   def test_fetch_post_connect_hook
     response = nil
     @agent.post_connect_hooks << lambda { |_, _, res, _| response = res }
@@ -227,6 +235,14 @@ class TestMechanizeHttpAgent < Mechanize::TestCase
     assert response
   end
+  def test_fetch_redirect_header
+    page = @agent.fetch('http://example/redirect', :get,
+                        'X-Location' => '/http_headers',
+                        'Range' => 'bytes=0-99999')
+    assert_match 'range|bytes=0-999', page.body
+  end
   def test_fetch_server_error
     e = assert_raises Mechanize::ResponseCodeError do
       @mech.get 'http://localhost/response_code?code=500'
@@ -874,13 +890,49 @@ class TestMechanizeHttpAgent < Mechanize::TestCase
     assert_match %r%error handling content-encoding gzip:%, e.message
     assert_match %r%Zlib%, e.message
-    assert_match %r%unable to gunzip response, trying raw inflate%, log.string
-    assert_match %r%unable to gunzip response:%, log.string
+    assert_match %r%unable to gunzip response: unexpected end of file%,
+                 log.string
+    assert_match %r%unable to inflate response: buffer error%,
+                 log.string
+    assert body_io.closed?
+  end
+  def test_response_content_encoding_gzip_checksum_corrupt_crc
+    log = StringIO.new
+    logger = Logger.new log
+    @agent.context.log = logger
+    @res.instance_variable_set :@header, 'content-encoding' => %w[gzip]
+    body_io = StringIO.new \
+      "\037\213\b\0002\002\225M\000\003+H,*\001\000\306p\017J\004\000\000\000"
+    body = @agent.response_content_encoding @res, body_io
+    assert_equal 'part', body.read
+    assert body_io.closed?
+    assert_match %r%invalid compressed data -- crc error%, log.string
+  end
+  def test_response_content_encoding_gzip_checksum_corrupt_length
+    log = StringIO.new
+    logger = Logger.new log
+    @agent.context.log = logger
+    @res.instance_variable_set :@header, 'content-encoding' => %w[gzip]
+    body_io = StringIO.new \
+      "\037\213\b\0002\002\225M\000\003+H,*\001\000\306p\017I\005\000\000\000"
+    body = @agent.response_content_encoding @res, body_io
     assert body_io.closed?
+    assert_match %r%invalid compressed data -- length error%, log.string
   end
-  def test_response_content_encoding_gzip_corrupt_checksum
+  def test_response_content_encoding_gzip_checksum_truncated
     log = StringIO.new
     logger = Logger.new log
     @agent.context.log = logger
@@ -889,14 +941,11 @@ class TestMechanizeHttpAgent < Mechanize::TestCase
     body_io = StringIO.new \
       "\037\213\b\0002\002\225M\000\003+H,*\001\000\306p\017I\004\000\000"
-    e = assert_raises Mechanize::Error do
-      @agent.response_content_encoding @res, body_io
-    end
+    body = @agent.response_content_encoding @res, body_io
-    assert_match %r%error handling content-encoding gzip:%, e.message
-    assert_match %r%Zlib%, e.message
+    assert body_io.closed?
-    assert_match %r%unable to gunzip response, trying raw inflate%, log.string
+    assert_match %r%unable to gunzip response: footer is not found%, log.string
   end
   def test_response_content_encoding_gzip_empty
@@ -926,6 +975,18 @@ class TestMechanizeHttpAgent < Mechanize::TestCase
     assert_equal Encoding::BINARY, content.encoding if have_encoding?
   end
+  def test_response_content_encoding_gzip_no_footer
+    @res.instance_variable_set :@header, 'content-encoding' => %w[gzip]
+    body_io = StringIO.new \
+      "\037\213\b\0002\002\225M\000\003+H,*\001\000"
+    body = @agent.response_content_encoding @res, body_io
+    assert_equal 'part', body.read
+    assert body_io.closed?
+  end
   def test_response_content_encoding_none
     @res.instance_variable_set :@header, 'content-encoding' => %w[none]
@@ -1168,6 +1229,21 @@ class TestMechanizeHttpAgent < Mechanize::TestCase
     assert_equal Encoding::BINARY, body.encoding if body.respond_to? :encoding
   end
+  def test_response_read_chunked_no_trailer
+    @res['Transfer-Encoding'] = 'chunked'
+    def @res.content_length() end
+    def @res.read_body
+      yield 'a' * 10
+      raise EOFError
+    end
+    e = assert_raises Mechanize::ChunkedTerminationError do
+      @agent.response_read @res, @req, @uri
+    end
+    assert_equal 'aaaaaaaaaa', e.body_io.read
+  end
   def test_response_read_content_length_head
     req = Net::HTTP::Head.new '/'
@@ -1239,21 +1315,6 @@ class TestMechanizeHttpAgent < Mechanize::TestCase
     end
   end
-  def test_file_response_content_type
-    Tempfile.open ['pi','.nothtml'] do |tempfile|
-      res = Mechanize::FileResponse.new tempfile.path
-      assert_equal nil, res['content-type']
-    end
-    Tempfile.open ['pi','.xhtml'] do |tempfile|
-      res = Mechanize::FileResponse.new tempfile.path
-      assert_equal 'text/html', res['content-type']
-    end
-    Tempfile.open ['pi','.html'] do |tempfile|
-      res = Mechanize::FileResponse.new tempfile.path
-      assert_equal 'text/html', res['Content-Type']
-    end
-  end
   def test_response_read_large
     @agent.max_file_buffer = 10240
@@ -1310,20 +1371,37 @@ class TestMechanizeHttpAgent < Mechanize::TestCase
     page = fake_page
     page = @agent.response_redirect({ 'Location' => '/index.html' }, :get,
-                                    page, 0, referer)
+                                    page, 0, {}, referer)
     assert_equal URI('http://fake.example/index.html'), page.uri
     assert_equal 'http://example/referer', requests.first['Referer']
   end
+  def test_response_redirect_header
+    @agent.redirect_ok = true
+    referer = page 'http://example/referer'
+    headers = {
+      'Range' => 'bytes=0-9999',
+    }
+    page = fake_page
+    page = @agent.response_redirect({ 'Location' => '/http_headers' }, :get,
+                                    page, 0, headers, referer)
+    assert_equal URI('http://fake.example/http_headers'), page.uri
+    assert_match 'range|bytes=0-999', page.body
+  end
   def test_response_redirect_malformed
     @agent.redirect_ok = true
     referer = page 'http://example/referer'
     page = fake_page
     page = @agent.response_redirect({ 'Location' => '/index.html?q=あ' }, :get,
-                                    page, 0, referer)
+                                    page, 0, {}, referer)
     assert_equal URI('http://fake.example/index.html?q=%E3%81%82'), page.uri