RubyGems - tenderlove-mechanize - Versions diffs - 0.9.3.20090623142847 → 0.9.3.20090911221705 - Mend

tenderlove-mechanize 0.9.3.20090623142847 → 0.9.3.20090911221705

Files changed (165) hide show

data/Manifest.txt +55 -48
data/Rakefile +12 -22
data/lib/mechanize.rb +618 -4
data/lib/mechanize/chain.rb +33 -0
data/lib/mechanize/chain/auth_headers.rb +78 -0
data/lib/mechanize/chain/body_decoding_handler.rb +46 -0
data/lib/mechanize/chain/connection_resolver.rb +76 -0
data/lib/mechanize/chain/custom_headers.rb +21 -0
data/lib/{www/mechanize → mechanize}/chain/handler.rb +1 -1
data/lib/mechanize/chain/header_resolver.rb +51 -0
data/lib/mechanize/chain/parameter_resolver.rb +22 -0
data/lib/{www/mechanize → mechanize}/chain/post_connect_hook.rb +0 -0
data/lib/mechanize/chain/pre_connect_hook.rb +20 -0
data/lib/mechanize/chain/request_resolver.rb +30 -0
data/lib/mechanize/chain/response_body_parser.rb +38 -0
data/lib/mechanize/chain/response_header_handler.rb +48 -0
data/lib/mechanize/chain/response_reader.rb +39 -0
data/lib/mechanize/chain/ssl_resolver.rb +40 -0
data/lib/mechanize/chain/uri_resolver.rb +75 -0
data/lib/mechanize/content_type_error.rb +14 -0
data/lib/mechanize/cookie.rb +70 -0
data/lib/mechanize/cookie_jar.rb +188 -0
data/lib/mechanize/file.rb +71 -0
data/lib/mechanize/file_response.rb +60 -0
data/lib/mechanize/file_saver.rb +37 -0
data/lib/mechanize/form.rb +378 -0
data/lib/mechanize/form/button.rb +9 -0
data/lib/mechanize/form/check_box.rb +11 -0
data/lib/mechanize/form/field.rb +30 -0
data/lib/mechanize/form/file_upload.rb +22 -0
data/lib/mechanize/form/image_button.rb +21 -0
data/lib/mechanize/form/multi_select_list.rb +67 -0
data/lib/mechanize/form/option.rb +49 -0
data/lib/mechanize/form/radio_button.rb +49 -0
data/lib/mechanize/form/select_list.rb +43 -0
data/lib/mechanize/headers.rb +11 -0
data/lib/mechanize/history.rb +65 -0
data/lib/mechanize/inspect.rb +88 -0
data/lib/{www/mechanize → mechanize}/monkey_patch.rb +4 -6
data/lib/mechanize/page.rb +206 -0
data/lib/mechanize/page/base.rb +8 -0
data/lib/mechanize/page/frame.rb +20 -0
data/lib/mechanize/page/image.rb +26 -0
data/lib/mechanize/page/label.rb +20 -0
data/lib/mechanize/page/link.rb +48 -0
data/lib/mechanize/page/meta.rb +50 -0
data/lib/mechanize/pluggable_parsers.rb +101 -0
data/lib/mechanize/redirect_limit_reached_error.rb +16 -0
data/lib/mechanize/redirect_not_get_or_head_error.rb +18 -0
data/lib/mechanize/response_code_error.rb +22 -0
data/lib/mechanize/unsupported_scheme_error.rb +8 -0
data/lib/mechanize/util.rb +67 -0
data/mechanize.gemspec +8 -8
data/test/chain/test_argument_validator.rb +2 -2
data/test/chain/test_auth_headers.rb +2 -2
data/test/chain/test_custom_headers.rb +2 -2
data/test/chain/test_header_resolver.rb +3 -3
data/test/chain/test_parameter_resolver.rb +4 -4
data/test/chain/test_request_resolver.rb +4 -4
data/test/chain/test_response_reader.rb +3 -3
data/test/helper.rb +1 -1
data/test/htdocs/tc_bad_charset.html +9 -0
data/test/htdocs/tc_charset.html +6 -0
data/test/htdocs/test_bad_encoding.html +52 -0
data/test/test_authenticate.rb +3 -3
data/test/test_bad_links.rb +1 -1
data/test/test_blank_form.rb +1 -1
data/test/test_checkboxes.rb +1 -1
data/test/test_content_type.rb +2 -2
data/test/test_cookie_class.rb +12 -12
data/test/test_cookie_jar.rb +13 -13
data/test/test_cookies.rb +1 -1
data/test/test_encoded_links.rb +1 -1
data/test/test_errors.rb +2 -2
data/test/test_follow_meta.rb +3 -3
data/test/test_form_action.rb +1 -1
data/test/test_form_as_hash.rb +1 -1
data/test/test_form_button.rb +2 -2
data/test/test_form_no_inputname.rb +1 -1
data/test/test_forms.rb +1 -1
data/test/test_frames.rb +1 -1
data/test/test_get_headers.rb +1 -1
data/test/test_gzipping.rb +2 -2
data/test/test_hash_api.rb +1 -1
data/test/test_history.rb +7 -7
data/test/test_history_added.rb +1 -1
data/test/test_html_unscape_forms.rb +7 -7
data/test/test_if_modified_since.rb +1 -1
data/test/test_keep_alive.rb +1 -1
data/test/test_links.rb +2 -2
data/test/test_mech.rb +2 -2
data/test/test_mechanize_file.rb +7 -7
data/test/test_meta.rb +2 -2
data/test/test_multi_select.rb +1 -1
data/test/test_no_attributes.rb +1 -1
data/test/test_option.rb +1 -1
data/test/test_page.rb +3 -3
data/test/test_pluggable_parser.rb +14 -14
data/test/test_post_form.rb +1 -1
data/test/test_pretty_print.rb +2 -2
data/test/test_radiobutton.rb +1 -1
data/test/test_redirect_limit_reached.rb +1 -3
data/test/test_redirect_verb_handling.rb +1 -3
data/test/test_referer.rb +1 -1
data/test/test_relative_links.rb +1 -1
data/test/test_request.rb +1 -1
data/test/test_response_code.rb +3 -3
data/test/test_save_file.rb +3 -3
data/test/test_scheme.rb +3 -3
data/test/test_select.rb +2 -2
data/test/test_select_all.rb +1 -1
data/test/test_select_none.rb +1 -1
data/test/test_select_noopts.rb +1 -1
data/test/test_set_fields.rb +1 -1
data/test/test_ssl_server.rb +1 -1
data/test/test_subclass.rb +1 -1
data/test/test_textarea.rb +1 -1
data/test/test_upload.rb +1 -1
data/test/test_verbs.rb +1 -1
metadata +61 -56
data/lib/www/mechanize.rb +0 -619
data/lib/www/mechanize/chain.rb +0 -34
data/lib/www/mechanize/chain/auth_headers.rb +0 -80
data/lib/www/mechanize/chain/body_decoding_handler.rb +0 -48
data/lib/www/mechanize/chain/connection_resolver.rb +0 -78
data/lib/www/mechanize/chain/custom_headers.rb +0 -23
data/lib/www/mechanize/chain/header_resolver.rb +0 -53
data/lib/www/mechanize/chain/parameter_resolver.rb +0 -24
data/lib/www/mechanize/chain/pre_connect_hook.rb +0 -22
data/lib/www/mechanize/chain/request_resolver.rb +0 -32
data/lib/www/mechanize/chain/response_body_parser.rb +0 -40
data/lib/www/mechanize/chain/response_header_handler.rb +0 -50
data/lib/www/mechanize/chain/response_reader.rb +0 -41
data/lib/www/mechanize/chain/ssl_resolver.rb +0 -42
data/lib/www/mechanize/chain/uri_resolver.rb +0 -77
data/lib/www/mechanize/content_type_error.rb +0 -16
data/lib/www/mechanize/cookie.rb +0 -72
data/lib/www/mechanize/cookie_jar.rb +0 -191
data/lib/www/mechanize/file.rb +0 -73
data/lib/www/mechanize/file_response.rb +0 -62
data/lib/www/mechanize/file_saver.rb +0 -39
data/lib/www/mechanize/form.rb +0 -360
data/lib/www/mechanize/form/button.rb +0 -8
data/lib/www/mechanize/form/check_box.rb +0 -13
data/lib/www/mechanize/form/field.rb +0 -28
data/lib/www/mechanize/form/file_upload.rb +0 -24
data/lib/www/mechanize/form/image_button.rb +0 -23
data/lib/www/mechanize/form/multi_select_list.rb +0 -69
data/lib/www/mechanize/form/option.rb +0 -51
data/lib/www/mechanize/form/radio_button.rb +0 -38
data/lib/www/mechanize/form/select_list.rb +0 -45
data/lib/www/mechanize/headers.rb +0 -12
data/lib/www/mechanize/history.rb +0 -67
data/lib/www/mechanize/inspect.rb +0 -90
data/lib/www/mechanize/page.rb +0 -181
data/lib/www/mechanize/page/base.rb +0 -10
data/lib/www/mechanize/page/frame.rb +0 -22
data/lib/www/mechanize/page/link.rb +0 -50
data/lib/www/mechanize/page/meta.rb +0 -51
data/lib/www/mechanize/pluggable_parsers.rb +0 -103
data/lib/www/mechanize/redirect_limit_reached_error.rb +0 -18
data/lib/www/mechanize/redirect_not_get_or_head_error.rb +0 -20
data/lib/www/mechanize/response_code_error.rb +0 -25
data/lib/www/mechanize/unsupported_scheme_error.rb +0 -10
data/lib/www/mechanize/util.rb +0 -76

data/lib/mechanize/file_response.rb ADDED Viewed

@@ -0,0 +1,60 @@
+class Mechanize
+  ###
+  # Fake response for dealing with file:/// requests
+  class FileResponse
+    def initialize(file_path)
+      @file_path = file_path
+    end
+    def read_body
+      if ::File.exists?(@file_path)
+        if directory?
+          yield dir_body
+        else
+          yield ::File.read(@file_path)
+        end
+      else
+        yield ''
+      end
+    end
+    def code
+      ::File.exists?(@file_path) ? 200 : 400
+    end
+    def content_length
+      return dir_body.length if directory?
+      ::File.exists?(@file_path) ? ::File.stat(@file_path).size : 0
+    end
+    def each_header; end
+    def [](key)
+      return nil unless key.downcase == 'content-type'
+      return 'text/html' if directory?
+      return 'text/html' if ['.html', '.xhtml'].any? { |extn|
+        @file_path =~ /#{extn}$/
+      }
+      nil
+    end
+    def each
+    end
+    def get_fields(key)
+      []
+    end
+    private
+    def dir_body
+      '<html><body>' +
+        Dir[::File.join(@file_path, '*')].map { |f|
+        "<a href=\"file://#{f}\">#{::File.basename(f)}</a>"
+      }.join("\n") + '</body></html>'
+    end
+    def directory?
+      ::File.directory?(@file_path)
+    end
+  end
+end

data/lib/mechanize/file_saver.rb ADDED Viewed

@@ -0,0 +1,37 @@
+class Mechanize
+  # = Synopsis
+  # This is a pluggable parser that automatically saves every file
+  # it encounters.  It saves the files as a tree, reflecting the
+  # host and file path.
+  #
+  # == Example to save all PDF's
+  #  require 'rubygems'
+  #  require 'mechanize'
+  #
+  #  agent = Mechanize.new
+  #  agent.pluggable_parser.pdf = Mechanize::FileSaver
+  #  agent.get('http://example.com/foo.pdf')
+  #
+  class FileSaver < File
+    attr_reader :filename
+    def initialize(uri=nil, response=nil, body=nil, code=nil)
+      super(uri, response, body, code)
+      path = uri.path.empty? ? 'index.html' : uri.path.gsub(/^[\/]*/, '')
+      path += 'index.html' if path =~ /\/$/
+      split_path = path.split(/\//)
+      filename = split_path.length > 0 ? split_path.pop : 'index.html'
+      joined_path = split_path.join(::File::SEPARATOR)
+      path = if joined_path.empty?
+               uri.host
+             else
+               "#{uri.host}#{::File::SEPARATOR}#{joined_path}"
+             end
+      @filename = "#{path}#{::File::SEPARATOR}#{filename}"
+      FileUtils.mkdir_p(path)
+      save_as(@filename)
+    end
+  end
+end

data/lib/mechanize/form.rb ADDED Viewed

@@ -0,0 +1,378 @@
+require 'mechanize/form/field'
+require 'mechanize/form/file_upload'
+require 'mechanize/form/button'
+require 'mechanize/form/image_button'
+require 'mechanize/form/radio_button'
+require 'mechanize/form/check_box'
+require 'mechanize/form/multi_select_list'
+require 'mechanize/form/select_list'
+require 'mechanize/form/option'
+class Mechanize
+  # =Synopsis
+  # This class encapsulates a form parsed out of an HTML page.  Each type
+  # of input fields available in a form can be accessed through this object.
+  # See GlobalForm for more methods.
+  #
+  # ==Example
+  # Find a form and print out its fields
+  #  form = page.forms.first # => Mechanize::Form
+  #  form.fields.each { |f| puts f.name }
+  # Set the input field 'name' to "Aaron"
+  #  form['name'] = 'Aaron'
+  #  puts form['name']
+  class Form
+    attr_accessor :method, :action, :name
+    attr_reader :fields, :buttons, :file_uploads, :radiobuttons, :checkboxes
+    attr_accessor :enctype
+    alias :elements :fields
+    attr_reader :form_node
+    attr_reader :page
+    def initialize(node, mech=nil, page=nil)
+      @enctype = node['enctype'] || 'application/x-www-form-urlencoded'
+      @form_node        = node
+      @action           = Util.html_unescape(node['action'])
+      @method           = (node['method'] || 'GET').upcase
+      @name             = node['name']
+      @clicked_buttons  = []
+      @page             = page
+      @mech             = mech
+      parse
+    end
+    # Returns whether or not the form contains a field with +field_name+
+    def has_field?(field_name)
+      ! fields.find { |f| f.name.eql? field_name }.nil?
+    end
+    alias :has_key? :has_field?
+    def has_value?(value)
+      ! fields.find { |f| f.value.eql? value }.nil?
+    end
+    def keys; fields.map { |f| f.name }; end
+    def values; fields.map { |f| f.value }; end
+    def submits  ; @submits   ||= buttons.select { |f| f.class == Submit   }; end
+    def resets   ; @resets    ||= buttons.select { |f| f.class == Reset    }; end
+    def texts    ; @texts     ||=  fields.select { |f| f.class == Text     }; end
+    def hiddens  ; @hiddens   ||=  fields.select { |f| f.class == Hidden   }; end
+    def textareas; @textareas ||=  fields.select { |f| f.class == Textarea }; end
+    def submit_button?(button_name) !!  submits.find{|f| f.name == button_name}; end
+    def reset_button?(button_name)  !!   resets.find{|f| f.name == button_name}; end
+    def text_field?(field_name)     !!    texts.find{|f| f.name == field_name}; end
+    def hidden_field?(field_name)   !!  hiddens.find{|f| f.name == field_name}; end
+    def textarea_field?(field_name) !!textareas.find{|f| f.name == field_name}; end
+    # Add a field with +field_name+ and +value+
+    def add_field!(field_name, value = nil)
+      fields << Field.new(field_name, value)
+    end
+    # This method sets multiple fields on the form.  It takes a list of field
+    # name, value pairs.  If there is more than one field found with the
+    # same name, this method will set the first one found.  If you want to
+    # set the value of a duplicate field, use a value which is a Hash with
+    # the key as the index in to the form.  The index
+    # is zero based.  For example, to set the second field named 'foo', you
+    # could do the following:
+    #  form.set_fields( :foo => { 1 => 'bar' } )
+    def set_fields(fields = {})
+      fields.each do |k,v|
+        case v
+        when Hash
+          v.each do |index, value|
+            self.fields_with(:name => k.to_s).[](index).value = value
+          end
+        else
+          value = nil
+          index = 0
+          [v].flatten.each do |val|
+            index = val.to_i unless value.nil?
+            value = val if value.nil?
+          end
+          self.fields_with(:name => k.to_s).[](index).value = value
+        end
+      end
+    end
+    # Fetch the value of the first input field with the name passed in
+    # ==Example
+    # Fetch the value set in the input field 'name'
+    #  puts form['name']
+    def [](field_name)
+      f = field(field_name)
+      f && f.value
+    end
+    # Set the value of the first input field with the name passed in
+    # ==Example
+    # Set the value in the input field 'name' to "Aaron"
+    #  form['name'] = 'Aaron'
+    def []=(field_name, value)
+      f = field(field_name)
+      if f.nil?
+        add_field!(field_name, value)
+      else
+        f.value = value
+      end
+    end
+    # Treat form fields like accessors.
+    def method_missing(id,*args)
+      method = id.to_s.gsub(/=$/, '')
+      if field(method)
+        return field(method).value if args.empty?
+        return field(method).value = args[0]
+      end
+      super
+    end
+    # Submit this form with the button passed in
+    def submit button=nil, headers = {}
+      @mech.submit(self, button, headers)
+    end
+    # Submit form using +button+. Defaults
+    # to the first button.
+    def click_button(button = buttons.first)
+      submit(button)
+    end
+    # This method is sub-method of build_query.
+    # It converts charset of query value of fields into expected one.
+    def proc_query(field)
+      return unless field.query_value
+      field.query_value.map{|(name, val)|
+        [from_native_charset(name), from_native_charset(val.to_s)]
+      }
+    end
+    private :proc_query
+    def from_native_charset(str, enc=nil)
+      if page
+        enc ||= page.encoding
+        Util.from_native_charset(str,enc) rescue str
+      else
+        str
+      end
+    end
+    private :from_native_charset
+    # This method builds an array of arrays that represent the query
+    # parameters to be used with this form.  The return value can then
+    # be used to create a query string for this form.
+    def build_query(buttons = [])
+      query = []
+      fields().each do |f|
+        qval = proc_query(f)
+        query.push(*qval)
+      end
+      checkboxes().each do |f|
+        if f.checked
+          qval = proc_query(f)
+          query.push(*qval)
+        end
+      end
+      radio_groups = {}
+      radiobuttons().each do |f|
+        fname = from_native_charset(f.name)
+        radio_groups[fname] ||= []
+        radio_groups[fname] << f
+      end
+      # take one radio button from each group
+      radio_groups.each_value do |g|
+        checked = g.select {|f| f.checked}
+        if checked.size == 1
+          f = checked.first
+          qval = proc_query(f)
+          query.push(*qval)
+        elsif checked.size > 1
+          raise "multiple radiobuttons are checked in the same group!"
+        end
+      end
+      @clicked_buttons.each { |b|
+        qval = proc_query(b)
+        query.push(*qval)
+      }
+      query
+    end
+    # This method adds a button to the query.  If the form needs to be
+    # submitted with multiple buttons, pass each button to this method.
+    def add_button_to_query(button)
+      @clicked_buttons << button
+    end
+    # This method calculates the request data to be sent back to the server
+    # for this form, depending on if this is a regular post, get, or a
+    # multi-part post,
+    def request_data
+      query_params = build_query()
+      case @enctype.downcase
+      when /^multipart\/form-data/
+        boundary = rand_string(20)
+        @enctype = "multipart/form-data; boundary=#{boundary}"
+        params = []
+        query_params.each { |k,v| params << param_to_multipart(k, v) unless k.nil? }
+        @file_uploads.each { |f| params << file_to_multipart(f) }
+        params.collect { |p| "--#{boundary}\r\n#{p}" }.join('') +
+          "--#{boundary}--\r\n"
+      else
+        Mechanize::Util.build_query_string(query_params)
+      end
+    end
+    # Removes all fields with name +field_name+.
+    def delete_field!(field_name)
+      @fields.delete_if{ |f| f.name == field_name}
+    end
+    { :field        => :fields,
+      :button       => :buttons,
+      :file_upload  => :file_uploads,
+      :radiobutton  => :radiobuttons,
+      :checkbox     => :checkboxes,
+    }.each do |singular,plural|
+      eval(<<-eomethod)
+          def #{plural}_with criteria = {}
+            criteria = {:name => criteria} if String === criteria
+            f = #{plural}.find_all do |thing|
+              criteria.all? { |k,v| v === thing.send(k) }
+            end
+            yield f if block_given?
+            f
+          end
+          def #{singular}_with criteria = {}
+            f = #{plural}_with(criteria).first
+            yield f if block_given?
+            f
+          end
+          alias :#{singular} :#{singular}_with
+        eomethod
+    end
+    private
+    def parse
+      @fields       = []
+      @buttons      = []
+      @file_uploads = []
+      @radiobuttons = []
+      @checkboxes   = []
+      # Find all input tags
+      form_node.search('input').each do |node|
+        type = (node['type'] || 'text').downcase
+        name = node['name']
+        next if name.nil? && !(type == 'submit' || type =='button')
+        case type
+        when 'radio'
+          @radiobuttons << RadioButton.new(node['name'], node['value'], !!node['checked'], self, node)
+        when 'checkbox'
+          @checkboxes << CheckBox.new(node['name'], node['value'], !!node['checked'], self, node)
+        when 'file'
+          @file_uploads << FileUpload.new(node['name'], nil)
+        when 'submit'
+          @buttons << Submit.new(node['name'], node['value'])
+        when 'button'
+          @buttons << Button.new(node['name'], node['value'])
+        when 'reset'
+          @buttons << Reset.new(node['name'], node['value'])
+        when 'image'
+          @buttons << ImageButton.new(node['name'], node['value'])
+        when 'hidden'
+          @fields << Hidden.new(node['name'], node['value'] || '')
+        when 'text'
+          @fields << Text.new(node['name'], node['value'] || '')
+        when 'textarea'
+          @fields << Textarea.new(node['name'], node['value'] || '')
+        else
+          @fields << Field.new(node['name'], node['value'] || '')
+        end
+      end
+      # Find all textarea tags
+      form_node.search('textarea').each do |node|
+        next if node['name'].nil?
+        @fields << Field.new(node['name'], node.inner_text)
+      end
+      # Find all select tags
+      form_node.search('select').each do |node|
+        next if node['name'].nil?
+        if node.has_attribute? 'multiple'
+          @fields << MultiSelectList.new(node['name'], node)
+        else
+          @fields << SelectList.new(node['name'], node)
+        end
+      end
+      # Find all submit button tags
+      # FIXME: what can I do with the reset buttons?
+      form_node.search('button').each do |node|
+        type = (node['type'] || 'submit').downcase
+        next if type == 'reset'
+        @buttons << Button.new(node['name'], node['value'])
+      end
+    end
+    def rand_string(len = 10)
+      chars = ("a".."z").to_a + ("A".."Z").to_a
+      string = ""
+      1.upto(len) { |i| string << chars[rand(chars.size-1)] }
+      string
+    end
+    def mime_value_quote(str)
+      str.gsub(/(["\r\\])/){|s| '\\' + s}
+    end
+    def param_to_multipart(name, value)
+      return "Content-Disposition: form-data; name=\"" +
+        "#{mime_value_quote(name)}\"\r\n" +
+        "\r\n#{value}\r\n"
+    end
+    def file_to_multipart(file)
+      file_name = file.file_name ? ::File.basename(file.file_name) : ''
+      body =  "Content-Disposition: form-data; name=\"" +
+        "#{mime_value_quote(file.name)}\"; " +
+        "filename=\"#{mime_value_quote(file_name)}\"\r\n" +
+        "Content-Transfer-Encoding: binary\r\n"
+      if file.file_data.nil? and ! file.file_name.nil?
+        file.file_data = ::File.open(file.file_name, "rb") { |f| f.read }
+        file.mime_type = WEBrick::HTTPUtils.mime_type(file.file_name,
+                                                      WEBrick::HTTPUtils::DefaultMimeTypes)
+      end
+      if file.mime_type != nil
+        body << "Content-Type: #{file.mime_type}\r\n"
+      end
+      body <<
+        if file.file_data.respond_to? :read
+          "\r\n#{file.file_data.read}\r\n"
+        else
+          "\r\n#{file.file_data}\r\n"
+        end
+      body
+    end
+  end
+end