active-fedora 2.3.7 → 2.3.8

data/.rvmrc

@@ -4,7 +4,7 @@
 # development environment upon cd'ing into the directory
 
 ruby_string="ree-1.8.7"
-gemset_name="active_fedora"
+gemset_name="active_fedora_2.x"
 
 # Install rubies when used instead of only displaying a warning and exiting
 rvm_install_on_use_flag=1

data/Gemfile.lock

@@ -1,13 +1,14 @@
 PATH
   remote: .
   specs:
-    active-fedora (2.3.5)
-      activeresource
+    active-fedora (2.3.7)
+      activeresource (< 3.0.0)
       equivalent-xml
       facets
       mediashelf-loggable
       mime-types (>= 1.16)
       multipart-post (= 1.1.2)
+      net-http-persistent
       nokogiri
       om (>= 1.0)
       solr-ruby (>= 0.0.6)
@@ -38,6 +39,7 @@ GEM
     mime-types (1.16)
     mocha (0.9.12)
     multipart-post (1.1.2)
+    net-http-persistent (2.1)
     nokogiri (1.5.0)
     om (1.4.0)
       mediashelf-loggable

data/History.txt

@@ -1,3 +1,10 @@
+2.3.8
+
+HYDRA-670 -- Use persistent http connections
+Don't set log level to error by default
+Benchmark fedora and solr queries
+
+
 2.3.7 
 
 HYDRA-650 -- Fixed relationship inheritance (Rick Johnson)

data/active-fedora.gemspec

@@ -22,10 +22,11 @@ Gem::Specification.new do |s|
   s.add_dependency('nokogiri')
   s.add_dependency('om', '>= 1.0')
   s.add_dependency('solrizer', '>1.0.0')
-  s.add_dependency("activeresource")
+  s.add_dependency("activeresource", "<3.0.0")
   s.add_dependency("mediashelf-loggable")
   s.add_dependency("equivalent-xml")
   s.add_dependency("facets")
+  s.add_dependency("net-http-persistent")
   s.add_development_dependency("yard")
   s.add_development_dependency("RedCloth") # for RDoc formatting
   s.add_development_dependency("rake")

data/lib/active_fedora.rb

@@ -68,7 +68,6 @@ module ActiveFedora #:nodoc:
   # 2. If it does not find a solr.yml and the fedora.yml contains a solr url, it will raise an configuration error 
   # 3. If it does not find a solr.yml and the fedora.yml does not contain a solr url, it will look in: +Rails.root+/config, +current working directory+/config, then the solr.yml shipped with gem
   def self.init( options={} )
-    logger.level = Logger::ERROR if logger.respond_to? :level ###MediaShelf StubLogger doesn't have a level= method
     # Make config_options into a Hash if nil is passed in as the value
     options = {} if options.nil?
 

data/lib/active_fedora/base.rb

@@ -4,6 +4,7 @@ require 'active_fedora/semantic_node'
 require "solrizer"
 require 'nokogiri'
 require "loggable"
+require 'benchmark'
 
 SOLR_DOCUMENT_ID = "id" unless (defined?(SOLR_DOCUMENT_ID) && !SOLR_DOCUMENT_ID.nil?)
 ENABLE_SOLR_UPDATES = true unless defined?(ENABLE_SOLR_UPDATES)
@@ -122,8 +123,12 @@ module ActiveFedora
     # Refreshes the object's info from Fedora
     # Note: Currently just registers any new datastreams that have appeared in fedora
     def refresh
-      inner_object.load_attributes_from_fedora
-      @datastreams = datastreams_in_fedora.merge(datastreams_in_memory)
+      ms = 1000 * Benchmark.realtime do
+        inner_object.load_attributes_from_fedora
+        @datastreams = datastreams_in_fedora.merge(datastreams_in_memory)
+      end
+      logger.debug "refreshing #{pid} took #{ms} ms"
+      @datastreams
     end
 
     #Deletes a Base object, also deletes the info indexed in Solr, and 
@@ -945,11 +950,17 @@ module ActiveFedora
     def update_index
       if defined?( Solrizer::Fedora::Solrizer ) 
         #logger.info("Trying to solrize pid: #{pid}")
-        solrizer = Solrizer::Fedora::Solrizer.new
-        solrizer.solrize( self )
+        ms = 1000 * Benchmark.realtime do
+          solrizer = Solrizer::Fedora::Solrizer.new
+          solrizer.solrize( self )
+        end
+        logger.debug "solrize for #{pid} took #{ms} ms"
       else
         #logger.info("Trying to update solr for pid: #{pid}")
-        SolrService.instance.conn.update(self.to_solr)
+        ms = 1000 * Benchmark.realtime do
+          SolrService.instance.conn.update(self.to_solr)
+        end
+        logger.debug "solr update for #{pid} took #{ms} ms"
       end
     end
 
@@ -1090,17 +1101,23 @@ module ActiveFedora
     
     # Pushes the object and all of its new or dirty datastreams into Fedora
     def update
-      result = Fedora::Repository.instance.save(@inner_object)      
-      datastreams_in_memory.each do |k,ds|
-        if ds.dirty? || ds.new_object? 
-          if ds.class.included_modules.include?(ActiveFedora::MetadataDatastreamHelper) || ds.instance_of?(ActiveFedora::RelsExtDatastream)
-          # if ds.kind_of?(ActiveFedora::MetadataDatastream) || ds.kind_of?(ActiveFedora::NokogiriDatastream) || ds.instance_of?(ActiveFedora::RelsExtDatastream)
-            @metadata_is_dirty = true
-          end
-          result = ds.save
-        end 
+      result = nil
+      ms = 1000 * Benchmark.realtime do
+        result = Fedora::Repository.instance.save(@inner_object)      
+      end
+      logger.debug "instance save for #{pid} took #{ms} ms"
+      ms = 1000 * Benchmark.realtime do
+        datastreams_in_memory.each do |k,ds|
+          if ds.dirty? || ds.new_object? 
+            if ds.class.included_modules.include?(ActiveFedora::MetadataDatastreamHelper) || ds.instance_of?(ActiveFedora::RelsExtDatastream)
+              @metadata_is_dirty = true
+            end
+            result = ds.save
+          end 
+        end
+        refresh
       end
-      refresh
+      logger.debug "datastream save for #{pid} took #{ms} ms"
       return result
     end
 

data/lib/active_fedora/datastream.rb

@@ -21,6 +21,7 @@ module ActiveFedora
     #set this Datastream's content
     def content=(content)
       self.blob = content
+      self.dirty = true
     end
 
     def self.delete(parent_pid, dsid)

data/lib/active_fedora/nokogiri_datastream.rb

@@ -14,7 +14,8 @@ class ActiveFedora::NokogiriDatastream < ActiveFedora::Datastream
   alias_method(:om_term_values, :term_values) unless method_defined?(:om_term_values)
   alias_method(:om_update_values, :update_values) unless method_defined?(:om_update_values)
   
-  attr_accessor :ng_xml, :internal_solr_doc
+  attr_accessor :internal_solr_doc
+  attr_reader :ng_xml
   
   #constructor, calls up to ActiveFedora::Datastream's constructor
   def initialize(attrs=nil)
@@ -43,22 +44,22 @@ class ActiveFedora::NokogiriDatastream < ActiveFedora::Datastream
     Nokogiri::XML::Document.parse("<xml/>")
   end
   
-  # class << self
-  #   from_xml_original = self.instance_method(:from_xml)
-  #   
-  #   define_method(:from_xml, xml, tmpl=self.new) do
-  #     from_xml_original.bind(self).call(xml, tmpl)
-  #     tmpl.send(:dirty=, false)
-  #   end
-  #   
-  #   # def from_xml_custom(xml, tmpl=self.new)
-  #   #   from_xml_original(xml, tmpl)
-  #   #   tmpl.send(:dirty=, false)
-  #   # end
-  #   # 
-  #   # alias_method :from_xml_original, :from_xml 
-  #   # alias_method :from_xml, :from_xml_custom
-  # end
+  def ng_xml=(new_xml)
+    case new_xml 
+    when Nokogiri::XML::Document, Nokogiri::XML::Element, Nokogiri::XML::Node
+      @ng_xml = new_xml
+    when String 
+      @ng_xml = Nokogiri::XML::Document.parse(new_xml)
+    else
+      raise TypeError, "You passed a #{new_xml.class} into the ng_xml of the #{self.dsid} datastream. NokogiriDatastream.ng_xml= only accepts Nokogiri::XML::Document, Nokogiri::XML::Element, Nokogiri::XML::Node, or raw XML (String) as inputs."
+    end
+    self.dirty = true
+  end
+  
+  def content=(content)
+    super
+    self.ng_xml = Nokogiri::XML::Document.parse(content)
+  end
   
   
   def to_xml(xml = self.ng_xml)

data/lib/active_fedora/solr_service.rb

@@ -65,6 +65,13 @@ module ActiveFedora
       return uri.gsub(/(:)/, '\\:')
     end
     
+    # Escapes these characters
+    # + - || ! ( ) { } [ ] ^ " ~ * ? : \
+    # See: http://lucene.apache.org/java/2_4_0/queryparsersyntax.html#Escaping%20Special%20Characters
+    def self.escape_characters_for_query(value)
+      value.gsub(/([\#\+\-\|\{\}\^\"\~\*\:\>\<\?\(\)\[\]\+\!\\])/) {|v| "\\#{v}"}
+    end
+    
   
 end #SolrService
 class SolrNotInitialized < StandardError;end

data/lib/active_fedora/version.rb

@@ -1,3 +1,3 @@
 module ActiveFedora
-    VERSION = "2.3.7"
+    VERSION = "2.3.8"
 end

data/lib/fedora/connection.rb

@@ -1,6 +1,7 @@
 require "base64"
 gem 'multipart-post'
 require 'net/http/post/multipart'
+require 'net/http/persistent'
 require 'cgi'
 require "mime/types"
 require 'net/http'
@@ -46,6 +47,9 @@ module Fedora
   # 409 Conflict
   class ResourceConflict < ClientError; end # :nodoc:
 
+  # 415 Unsupported Media Type
+  class UnsupportedMediaType < ClientError; end # :nodoc:
+
   # 5xx Server Error
   class ServerError < ConnectionError; end # :nodoc:
 
@@ -60,6 +64,23 @@ module Fedora
   # This class is used by ActiveResource::Base to interface with REST
   # services.
   class Connection
+
+    CLASSES = [
+      Net::HTTP::Delete,
+      Net::HTTP::Get,
+      Net::HTTP::Head,
+      Net::HTTP::Post,
+      Net::HTTP::Put
+    ].freeze
+
+    MIME_TYPES = {
+      :binary => "application/octet-stream",
+      :json   => "application/json",
+      :xml    => "text/xml",
+      :none   => "text/plain"
+    }.freeze
+
+
     attr_reader :site, :surrogate
     attr_accessor :format
 
@@ -78,84 +99,102 @@ module Fedora
       @surrogate=surrogate
     end
 
+    ##
+    # Perform an HTTP Delete, Head, Get, Post, or Put.
+
+    CLASSES.each do |clazz|
+      verb = clazz.to_s.split("::").last.downcase
+
+      define_method verb do |*args|
+        path = args[0]
+        params = args[1] || {}
+
+        response_for clazz, path, params
+      end
+    end
+
     # Set URI for remote service.
     def site=(site)
       @site = site.is_a?(URI) ? site : URI.parse(site)
     end
 
-    # Execute a GET request.
-    # Used to get (find) resources.
-    def get(path, headers = {})
-      format.decode(request(:get, path, build_request_headers(headers)).body)
-    end
-
-    # Execute a DELETE request (see HTTP protocol documentation if unfamiliar).
-    # Used to delete resources.
-    def delete(path, headers = {})
-      request(:delete, path, build_request_headers(headers))
-    end
 
+    private
 
+    # Makes request to remote service.
+    def response_for(clazz, path, params)
+      logger.debug "#{clazz} #{path}"
+      request = clazz.new path
+      request.body = params[:body]
 
-    def raw_get(path, headers = {})
-      request(:get, path, build_request_headers(headers))
-    end
-    def post(path, body='', headers={})
-      do_method(:post, path, body, headers)
-    end
-    def put( path, body='', headers={})
-      do_method(:put, path, body, headers)
+      handle_request request, params[:upload], params[:type], params[:headers] || {}
     end
 
-    private
-    def do_method(method, path, body = '', headers = {})
-      meth_map={:put=>Net::HTTP::Put::Multipart, :post=>Net::HTTP::Post::Multipart}
-      raise "undefined method: #{method}" unless meth_map.has_key? method
-      headers = build_request_headers(headers) 
-      if body.respond_to?(:read)
-        if body.respond_to?(:original_filename?)
-          filename = File.basename(body.original_filename)
-          io = UploadIO.new(body, mime_type,filename)
-        elsif body.path
-          filename = File.basename(body.path)
-        else
-          filename="NOFILE"
-        end
-        mime_types = MIME::Types.of(filename)
-        mime_type ||= mime_types.empty? ? "application/octet-stream" : mime_types.first.content_type
-
-        io = nil
-        if body.is_a?(File)
-          io = UploadIO.new(body.path,mime_type)
-        else
-          io =UploadIO.new(body, mime_type, filename)
-        end
 
-        req = meth_map[method].new(path, {:file=>io}, headers)
-        multipart_request(req)
+    def handle_request request, upload, type, headers
+      handle_uploads request, upload, type
+      handle_headers request, upload, type, headers
+      result = http.request self.site, request
+      handle_response(result)
+    end
+
+    ##
+    # Handle chunked uploads.
+    #
+    # +request+: A Net::HTTP request Object.
+    # +upload+: A Hash with the following keys:
+    #   +:file+: The file to be HTTP chunked uploaded.
+    #   +:headers+: A Hash containing additional HTTP headers.
+    # +:type+: A Symbol with the mime_type.
+
+    def handle_uploads request, upload, type
+      return unless upload
+      io = nil
+      if upload[:file].is_a?(File)
+         io = File.open upload[:file].path
       else
-        request(method, path, body.to_s, headers)
+        io = upload[:file]
       end
+
+      request.body_stream = io
     end
-    def multipart_request(req)
-      result = nil
-      result = http.start do |conn|
-        conn.read_timeout=60600 #these can take a while
-        conn.request(req)
+
+
+    def handle_headers request, upload, type, headers
+      request.basic_auth(self.site.user, self.site.password) if self.site.user
+
+      request.add_field "Accept", mime_type(type)
+      request.add_field "Content-Type", mime_type(type) if requires_content_type? request
+
+      headers.merge! chunked_headers upload
+      headers.each do |header, value|
+        request[header] = value
       end
-      handle_response(result)
+    end
+
+    ##
+    # Setting of chunked upload headers.
+    #
+    # +upload+: A Hash with the following keys:
+    #   +:file+: The file to be HTTP chunked uploaded.
+
+    def chunked_headers upload
+      return {} unless upload
 
+      chunked_headers = {
+        "Content-Type"      => mime_type(:binary),
+        "Transfer-Encoding" => "chunked"
+      }.merge upload[:headers] || {}
     end
 
-    # Makes request to remote service.
-    def request(method, path, *arguments)
-      result = http.send(method, path, *arguments)
-      handle_response(result)
+    def requires_content_type? request
+      [Net::HTTP::Post, Net::HTTP::Put].include? request.class
     end
 
     # Handles response and error codes from remote service.
     def handle_response(response)
       message = "Error from Fedora: #{response.body}"
+      logger.debug "Response: #{response.code}"
       case response.code.to_i
       when 301,302
         raise(Redirection.new(response))
@@ -173,6 +212,8 @@ module Fedora
         raise(MethodNotAllowed.new(response, message))
       when 409
         raise(ResourceConflict.new(response, message))
+      when 415
+        raise UnsupportedMediaType.new(response, message)
       when 422
         raise(ResourceInvalid.new(response, message))
       when 423...500
@@ -184,35 +225,30 @@ module Fedora
       end
     end
 
+    def mime_type type
+      if type.kind_of? String
+        type
+      else
+        MIME_TYPES[type] || MIME_TYPES[:xml]
+      end
+    end
+
     # Creates new Net::HTTP instance for communication with
     # remote service and resources.
     def http
-      http = Net::HTTP.new(@site.host, @site.port)
+      return @http if @http
+      @http = Net::HTTP::Persistent.new#(@site)
       if(@site.is_a?(URI::HTTPS))
-        http.use_ssl = true
-        http.verify_mode = OpenSSL::SSL::VERIFY_NONE
+        @http.use_ssl = true
+        @http.verify_mode = OpenSSL::SSL::VERIFY_NONE
 
         if (defined?(SSL_CLIENT_CERT_FILE) && !SSL_CLIENT_CERT_FILE.nil? && defined?(SSL_CLIENT_KEY_FILE) && !SSL_CLIENT_KEY_FILE.nil? && defined?(SSL_CLIENT_KEY_PASS) && !SSL_CLIENT_KEY_PASS.nil?)
-          http.cert = OpenSSL::X509::Certificate.new( File.read(SSL_CLIENT_CERT_FILE) )
-          http.key = OpenSSL::PKey::RSA.new( File.read(SSL_CLIENT_KEY_FILE), SSL_CLIENT_KEY_PASS )
+          @http.cert = OpenSSL::X509::Certificate.new( File.read(SSL_CLIENT_CERT_FILE) )
+          @http.key = OpenSSL::PKey::RSA.new( File.read(SSL_CLIENT_KEY_FILE), SSL_CLIENT_KEY_PASS )
         end
       end
-      http
+      @http
     end
 
-    def default_header
-      @default_header ||= { 'Content-Type' => format.mime_type }
-    end
-
-    # Builds headers for request to remote service.
-    def build_request_headers(headers)
-      headers.merge!({"From"=>surrogate}) if @surrogate
-      authorization_header.update(default_header).update(headers)
-    end
-
-    # Sets authorization header; authentication information is pulled from credentials provided with site URI.
-    def authorization_header
-      (@site.user || @site.password ? { 'Authorization' => 'Basic ' + ["#{@site.user}:#{ @site.password}"].pack('m').delete("\r\n") } : {})
-    end
   end
 end

data/lib/fedora/repository.rb

@@ -59,7 +59,7 @@ module Fedora
     
     # Fetch the raw content of either a fedora object or datastream
     def fetch_content(object_uri)
-      response = connection.raw_get("#{url_for(object_uri)}?format=xml")
+      response = connection.get("#{url_for(object_uri)}?format=xml")
       StringResponse.new(response.body, response.content_type)
     end
     
@@ -95,7 +95,7 @@ module Fedora
       params[:sessionToken] = options[:sessionToken] if options[:sessionToken]
       includes = fields.inject("") { |s, f| s += "&#{f}=true"; s }
       
-      convert_xml(connection.get("#{fedora_url.path}/objects?#{params.to_fedora_query}#{includes}"))
+      convert_xml(XmlFormat.decode(connection.get("#{fedora_url.path}/objects?#{params.to_fedora_query}#{includes}").body))
     end
     
     # Retrieve an object from fedora and load it as an instance of the given model/class
@@ -103,10 +103,14 @@ module Fedora
     # @param pid of the Fedora object to retrieve and deserialize
     # @param klazz the Model whose deserialize method the object's FOXML will be passed into
     def find_model(pid, klazz)
-      obj = self.find_objects("pid=#{pid}").first
-      if obj.nil?
-        raise ActiveFedora::ObjectNotFoundError, "The repository does not have an object with pid #{pid}.  The repository URL is #{self.base_url}"
+      obj = nil
+      ms = 1000 * Benchmark.realtime do
+        obj = self.find_objects("pid=#{pid}").first
+        if obj.nil?
+          raise ActiveFedora::ObjectNotFoundError, "The repository does not have an object with pid #{pid}.  The repository URL is #{self.base_url}"
+        end
       end
+      logger.debug "loading #{pid} took #{ms} ms"
       doc = Nokogiri::XML::Document.parse(obj.object_xml)     
       klazz.deserialize(doc)
     end
@@ -132,7 +136,7 @@ module Fedora
       case object
         when Fedora::FedoraObject
           pid = (object.pid ? object : 'new')
-          response = connection.post("#{url_for(pid)}?" + object.attributes.to_fedora_query, object.blob)
+          response = connection.post("#{url_for(pid)}?" + object.attributes.to_fedora_query, :body=>object.blob)
           if response.code == '201'
             object.pid = extract_pid(response) 
             object.new_object = false
@@ -142,10 +146,14 @@ module Fedora
           end
         when Fedora::Datastream
           raise ArgumentError, "Missing dsID attribute" if object.dsid.nil?
-          extra_headers = {}
-          extra_headers['Content-Type'] = object.attributes[:mimeType] if object.attributes[:mimeType]
-          response = connection.post("#{url_for(object)}?" + object.attributes.to_fedora_query, 
-            object.blob, extra_headers)
+          headers = {}
+          headers[:type] = object.attributes[:mimeType] if object.attributes[:mimeType]
+          if object.blob.respond_to? :read
+            headers[:upload] = {:file=>object.blob}
+          else
+            headers[:body]=object.blob
+          end
+          response = connection.post("#{url_for(object)}?" + object.attributes.to_fedora_query, headers)
           if response.code == '201'
             object.new_object = false
             true
@@ -170,7 +178,13 @@ module Fedora
         response.code == '200' || '307'
       when Fedora::Datastream
         raise ArgumentError, "Missing dsID attribute" if object.dsid.nil?
-        response = connection.put("#{url_for(object)}?" + object.attributes.to_fedora_query, object.blob)
+        headers = {}
+        if object.blob.respond_to? :read
+          headers[:upload] = {:file=>object.blob}
+        else
+          headers[:body]=object.blob
+        end
+        response = connection.put("#{url_for(object)}?" + object.attributes.to_fedora_query, headers)
         response.code == '200' || '201'
         return response.code
       else
@@ -225,7 +239,7 @@ module Fedora
         content_to_ingest = content_to_ingest.read
       end
         
-      connection.post(url,content_to_ingest)
+      connection.post(url, :body=>content_to_ingest)
     end
 
     # Fetch the given object using custom method.  This is used to fetch other aspects of a fedora object,
@@ -245,11 +259,11 @@ module Fedora
       end
       
       extra_params.delete(:format) if method == :export
-      connection.raw_get("#{url_for(object)}#{path}?#{extra_params.to_fedora_query}").body
+      connection.get("#{url_for(object)}#{path}?#{extra_params.to_fedora_query}").body
     end
     
     def describe_repository
-      result_body = connection.raw_get("#{fedora_url.path}/describe?xml=true").body
+      result_body = connection.get("#{fedora_url.path}/describe?xml=true").body
       XmlSimple.xml_in(result_body)
     end
     

data/spec/integration/repository_spec.rb

@@ -140,6 +140,8 @@ end
 describe Fedora::Repository, "find_objects" do
   before(:all) do
     Fedora::Repository.register(ActiveFedora.fedora_config[:url])
+    @mock_response = mock("Response")
+    @mock_response.stubs(:body).returns(sample_response_xml)
   end
   
   def fields_to_s(fields)
@@ -156,31 +158,33 @@ describe Fedora::Repository, "find_objects" do
   
   it "should include all fields by default" do
     when_find_objects('label~Image*') { |conn|
-      conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s(Fedora::ALL_FIELDS))
+      conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s(Fedora::ALL_FIELDS)).returns(@mock_response)
     }
   end
   
   it "should include all fields when :include => :all " do
     when_find_objects('label~Image*', :include => :all) { |conn|
-      conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s(Fedora::ALL_FIELDS))
+      conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s(Fedora::ALL_FIELDS)).returns(@mock_response)
     }
   end
   
   it "should fetch results with limit" do
     when_find_objects('label~Image*', :limit => 10) { |conn|
-      conn.expects(:get).with('/fedora/objects?maxResults=10&query=label%7EImage%2A&resultFormat=xml' + fields_to_s(Fedora::ALL_FIELDS))
+      conn.expects(:get).with('/fedora/objects?maxResults=10&query=label%7EImage%2A&resultFormat=xml' + fields_to_s(Fedora::ALL_FIELDS)).returns(@mock_response)
     }
   end
   
   it "should fetch results with some fields but not :pid" do
     when_find_objects('label~Image*', :select => [:label, :mDate]) { |conn|
-      conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s([:pid, :label, :mDate]))
+      conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s([:pid, :label, :mDate])).returns(@mock_response)
+
     }
   end
   
   it "should fetch results with some fields with :pid" do
     when_find_objects('label~Image*', :select => [:pid, :label, :mDate]) { |conn|
-      conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s([:pid, :label, :mDate]))
+      conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s([:pid, :label, :mDate])).returns(@mock_response)
+
     }
   end
   
@@ -203,7 +207,7 @@ describe Fedora::Repository, "find_objects" do
   it "should convert xml response with 0 objectFields into array of FedoraObject" do
     objects = when_find_objects('label~Image*', :select => [:pid, :label, :mDate]) { |conn|
       conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s([:pid, :label, :mDate])).
-        returns(Fedora::XmlFormat.decode(sample_response_xml))
+        returns(@mock_response)
     }
     
     objects.session_token.should == 'aToken'
@@ -213,7 +217,7 @@ describe Fedora::Repository, "find_objects" do
   it "should return FedoraObjects with new_object set to false" do
     objects = when_find_objects('label~Image*', :select => [:pid, :label, :mDate]) { |conn|
       conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s([:pid, :label, :mDate])).
-        returns(Fedora::XmlFormat.decode(sample_response_xml))
+        returns(@mock_response)
     }
     objects.each do |obj|
       obj.should_not be_new_object
@@ -224,7 +228,7 @@ describe Fedora::Repository, "find_objects" do
   it "should convert xml response with single objectFields into array of FedoraObject" do
     objects = when_find_objects('label~Image*', :select => [:pid, :label, :mDate]) { |conn|
       conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s([:pid, :label, :mDate])).
-        returns(Fedora::XmlFormat.decode(sample_response_xml(1)))
+        returns(@mock_response)
     }
     
     objects.session_token.should == 'aToken'
@@ -234,9 +238,10 @@ describe Fedora::Repository, "find_objects" do
   end
   
   it "should convert xml response 2 objectFields into array of FedoraObject" do
+    @mock_response.stubs(:body).returns(sample_response_xml(2))
     objects = when_find_objects('label~Image*', :select => [:pid, :label, :mDate]) { |conn|
       conn.expects(:get).with('/fedora/objects?query=label%7EImage%2A&resultFormat=xml' + fields_to_s([:pid, :label, :mDate])).
-        returns(Fedora::XmlFormat.decode(sample_response_xml(2)))
+        returns(@mock_response)
     }
     
     objects.session_token.should == 'aToken'

data/spec/spec_helper.rb

@@ -14,6 +14,8 @@ end
 ENV["RAILS_ENV"] ||= 'test'
 RAILS_ENV = ENV["RAILS_ENV"]
 
+logger.level = Logger::WARN if logger.respond_to? :level ###MediaShelf StubLogger doesn't have a level= method
+
 $:.unshift(File.dirname(__FILE__) + '/../lib')
 Dir[File.join(File.dirname(__FILE__)+'/../lib/')+'**/*.rb'].each{|x| require x unless x.match(/railtie.rb$/)}
 $VERBOSE=nil

data/spec/unit/connection_spec.rb

@@ -2,24 +2,24 @@ require File.join( File.dirname(__FILE__),  "../spec_helper" )
 
 require 'active_resource'
 require 'fedora/repository'
-
-describe Fedora::Connection do
-  it "should be creatable w/ a surrogate id" do
-    c = Fedora::Connection.new('http://127.0.0.1/fedora', 'fubar', 'bob')
-    c.site.to_s.should == "http://127.0.0.1/fedora"
-    c.format.should == 'fubar'
-    c.surrogate.should == 'bob'
-  end
-  it "should set a from header if surrogate defined." do
-    c = Fedora::Connection.new('http://127.0.0.1/fedora', ActiveResource::Formats[:xml], 'bob')
-    h = Hash.new
-    r= c.send(:build_request_headers,h)
-    r['From'].should == 'bob'
-  end
-  it "should not set a from header if surrogate undefined." do
-    c = Fedora::Connection.new('http://127.0.0.1/fedora' )
-    h = Hash.new
-    r= c.send(:build_request_headers,h)
-    r['From'].should be_nil
-  end
-end
+### Removed by Justin 2011-9-23
+# describe Fedora::Connection do
+#   it "should be creatable w/ a surrogate id" do
+#     c = Fedora::Connection.new('http://127.0.0.1/fedora', 'fubar', 'bob')
+#     c.site.to_s.should == "http://127.0.0.1/fedora"
+#     c.format.should == 'fubar'
+#     c.surrogate.should == 'bob'
+#   end
+#   it "should set a from header if surrogate defined." do
+#     c = Fedora::Connection.new('http://127.0.0.1/fedora', ActiveResource::Formats[:xml], 'bob')
+#     h = Hash.new
+#     r= c.send(:build_request_headers,h)
+#     r['From'].should == 'bob'
+#   end
+#   it "should not set a from header if surrogate undefined." do
+#     c = Fedora::Connection.new('http://127.0.0.1/fedora' )
+#     h = Hash.new
+#     r= c.send(:build_request_headers,h)
+#     r['From'].should be_nil
+#   end
+# end

data/spec/unit/datastream_spec.rb

@@ -60,6 +60,17 @@ describe ActiveFedora::Datastream do
     end
   end
   
+  describe '.content=' do
+    it "should update the content and ng_xml, marking the datastream as dirty" do
+      sample_xml = "<foo><xmlelement/></foo>"
+      @test_datastream.should_not be_dirty
+      @test_datastream.blob.should_not be_equivalent_to(sample_xml)
+      @test_datastream.content = sample_xml
+      @test_datastream.should be_dirty
+      @test_datastream.blob.should be_equivalent_to(sample_xml)
+    end
+  end
+  
   describe ".dirty?" do
     it "should return the value of the @dirty attribute" do
       @test_datastream.dirty.should equal(@test_datastream.dirty?)

data/spec/unit/nokogiri_datastream_spec.rb

@@ -10,7 +10,7 @@ describe ActiveFedora::NokogiriDatastream do
                       :empty_field => {:values => {}}
                       } 
     @sample_xml = XmlSimple.xml_in("<fields><coverage>coverage1</coverage><coverage>coverage2</coverage><creation_date>fake-date</creation_date><mydate>fake-date</mydate><publisher>publisher1</publisher></fields>")
-
+    @sample_raw_xml = "<foo><xmlelement/></foo>"
     @solr_doc = {"id"=>"hydrange_article1","name_role_roleTerm_t"=>["creator","submitter","teacher"],"name_0_role_t"=>"\r\ncreator\r\nsubmitter\r\n","name_1_role_t"=>"\r\n teacher \r\n","name_0_role_0_roleTerm_t"=>"creator","name_0_role_1_roleTerm_t"=>"submitter","name_1_role_0_roleTerm_t"=>["teacher"]}
   end
   
@@ -35,8 +35,8 @@ describe ActiveFedora::NokogiriDatastream do
       test_ds1.ng_xml.to_xml.should == "<?xml version=\"1.0\"?>\n<xml>\n  <foo/>\n</xml>\n"
     end
     it "should initialize from #xml_template if no xml is provided" do
-      ActiveFedora::NokogiriDatastream.expects(:xml_template).returns("fake template")
-      ActiveFedora::NokogiriDatastream.new.ng_xml.should == "fake template"
+      ActiveFedora::NokogiriDatastream.expects(:xml_template).returns("<fake template/>")
+      ActiveFedora::NokogiriDatastream.new.ng_xml.should be_equivalent_to("<fake template/>")
     end
   end
   
@@ -234,6 +234,32 @@ describe ActiveFedora::NokogiriDatastream do
     end
   end
   
+  describe '.content=' do
+    it "should update the content and ng_xml, marking the datastream as dirty" do
+      @test_ds.should_not be_dirty
+      @test_ds.blob.should_not be_equivalent_to(@sample_raw_xml)
+      @test_ds.ng_xml.to_xml.should_not be_equivalent_to(@sample_raw_xml)
+      @test_ds.content = @sample_raw_xml
+      @test_ds.should be_dirty
+      @test_ds.blob.should be_equivalent_to(@sample_raw_xml)
+      @test_ds.ng_xml.to_xml.should be_equivalent_to(@sample_raw_xml)
+    end
+  end
+  
+  describe 'ng_xml=' do
+    it "should parse raw xml for you" do
+      @test_ds.ng_xml.to_xml.should_not be_equivalent_to(@sample_raw_xml)
+      @test_ds.ng_xml = @sample_raw_xml
+      @test_ds.ng_xml.class.should == Nokogiri::XML::Document
+      @test_ds.ng_xml.to_xml.should be_equivalent_to(@sample_raw_xml)
+    end
+    it "should mark the datastream as dirty" do
+      @test_ds.should_not be_dirty
+      @test_ds.ng_xml = @sample_raw_xml
+      @test_ds.should be_dirty
+    end
+  end
+  
   describe '.to_xml' do
     it "should provide .to_xml" do
       @test_ds.should respond_to(:to_xml)

data/spec/unit/repository_spec.rb

@@ -99,14 +99,14 @@ describe Fedora::Repository do
     it "should post the provided xml or file to fedora" do
       foxml = fixture("test_12.foxml.xml").read
       connection = Fedora::Repository.instance.send(:connection)
-      connection.expects(:post).with("/fedora/objects/new",foxml)
+      connection.expects(:post).with("/fedora/objects/new",:body=>foxml)
       Fedora::Repository.instance.ingest(foxml)
     end
     it "should accept a file as its input" do
       foxml_file = fixture("test_12.foxml.xml")
       foxml = fixture("test_12.foxml.xml").read
       connection = Fedora::Repository.instance.send(:connection)
-      connection.expects(:post).with("/fedora/objects/new",foxml)
+      connection.expects(:post).with("/fedora/objects/new",:body=>foxml)
       Fedora::Repository.instance.ingest(foxml_file)
     end
   end

data/spec/unit/solr_service_spec.rb

@@ -84,4 +84,11 @@ describe ActiveFedora::SolrService do
     end
   end
   
+  describe "escape_characters_for_query" do
+    it "should escape the Solr special characters" do
+      test_val = '# + - || ! ( ) { } [ ] ^ " ~ * ? : \\'
+      expected_result = '\\# \\+ \\- \\|\\| \\! \\( \\) \\{ \\} \\[ \\] \\^ \\" \\~ \\* \\? \\: \\\\'
+      ActiveFedora::SolrService.escape_characters_for_query(test_val).should == expected_result
+    end
+  end
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: active-fedora
 version: !ruby/object:Gem::Version 
-  hash: 13
+  hash: 19
   prerelease: 
   segments: 
   - 2
   - 3
-  - 7
-  version: 2.3.7
+  - 8
+  version: 2.3.8
 platform: ruby
 authors: 
 - Matt Zumwalt
@@ -16,7 +16,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-09-02 00:00:00 -05:00
+date: 2011-09-23 00:00:00 -05:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -133,12 +133,14 @@ dependencies:
   requirement: &id008 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
-    - - ">="
+    - - <
       - !ruby/object:Gem::Version 
-        hash: 3
+        hash: 7
         segments: 
+        - 3
         - 0
-        version: "0"
+        - 0
+        version: 3.0.0
   type: :runtime
   version_requirements: *id008
 - !ruby/object:Gem::Dependency 
@@ -184,7 +186,7 @@ dependencies:
   type: :runtime
   version_requirements: *id011
 - !ruby/object:Gem::Dependency 
-  name: yard
+  name: net-http-persistent
   prerelease: false
   requirement: &id012 !ruby/object:Gem::Requirement 
     none: false
@@ -195,10 +197,10 @@ dependencies:
         segments: 
         - 0
         version: "0"
-  type: :development
+  type: :runtime
   version_requirements: *id012
 - !ruby/object:Gem::Dependency 
-  name: RedCloth
+  name: yard
   prerelease: false
   requirement: &id013 !ruby/object:Gem::Requirement 
     none: false
@@ -212,7 +214,7 @@ dependencies:
   type: :development
   version_requirements: *id013
 - !ruby/object:Gem::Dependency 
-  name: rake
+  name: RedCloth
   prerelease: false
   requirement: &id014 !ruby/object:Gem::Requirement 
     none: false
@@ -226,7 +228,7 @@ dependencies:
   type: :development
   version_requirements: *id014
 - !ruby/object:Gem::Dependency 
-  name: rcov
+  name: rake
   prerelease: false
   requirement: &id015 !ruby/object:Gem::Requirement 
     none: false
@@ -240,7 +242,7 @@ dependencies:
   type: :development
   version_requirements: *id015
 - !ruby/object:Gem::Dependency 
-  name: solrizer-fedora
+  name: rcov
   prerelease: false
   requirement: &id016 !ruby/object:Gem::Requirement 
     none: false
@@ -254,7 +256,7 @@ dependencies:
   type: :development
   version_requirements: *id016
 - !ruby/object:Gem::Dependency 
-  name: jettywrapper
+  name: solrizer-fedora
   prerelease: false
   requirement: &id017 !ruby/object:Gem::Requirement 
     none: false
@@ -268,9 +270,23 @@ dependencies:
   type: :development
   version_requirements: *id017
 - !ruby/object:Gem::Dependency 
-  name: rspec
+  name: jettywrapper
   prerelease: false
   requirement: &id018 !ruby/object:Gem::Requirement 
+    none: false
+    requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        hash: 3
+        segments: 
+        - 0
+        version: "0"
+  type: :development
+  version_requirements: *id018
+- !ruby/object:Gem::Dependency 
+  name: rspec
+  prerelease: false
+  requirement: &id019 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - <
@@ -282,11 +298,11 @@ dependencies:
         - 0
         version: 2.0.0
   type: :development
-  version_requirements: *id018
+  version_requirements: *id019
 - !ruby/object:Gem::Dependency 
   name: mocha
   prerelease: false
-  requirement: &id019 !ruby/object:Gem::Requirement 
+  requirement: &id020 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -298,11 +314,11 @@ dependencies:
         - 8
         version: 0.9.8
   type: :development
-  version_requirements: *id019
+  version_requirements: *id020
 - !ruby/object:Gem::Dependency 
   name: ruby-debug
   prerelease: false
-  requirement: &id020 !ruby/object:Gem::Requirement 
+  requirement: &id021 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -312,7 +328,7 @@ dependencies:
         - 0
         version: "0"
   type: :development
-  version_requirements: *id020
+  version_requirements: *id021
 description: ActiveFedora provides for creating and managing objects in the Fedora Repository Architecture.
 email: 
 - matt.zumwalt@yourmediashelf.com