krikri 0.1.3 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: bff72a4104f298f6090646716eb511e2178e274a
-  data.tar.gz: 1ef9b3b5a53d1eb17d83adc78943c8ac3ba05946
+  metadata.gz: 88f2fda2ab3d73eefca94acf0d09535360c2f8a2
+  data.tar.gz: 7e7728069c141de05865fcc778eff47478a32307
 SHA512:
-  metadata.gz: 3d72e9587388409a9a0cb565664c81330d740f9c311dfb41e7b61bec8b98e44440510ae9cd5f9fc3f45b90412f0e0836aacb2728ff277705f3fb776f7a54067a
-  data.tar.gz: 35006b6621f1235649446b1a9b0e207ae02a341d9be705b33ee5b73f7d4329593391b231c91cdb7295519567972ed01c545c79524c3611a832c0d4e0a984ee83
+  metadata.gz: 86e64af30ab18c16a0873aa9ae5394a8f8e60cc10b7b7a58025128fb89bfb93f24dc9c8868083d1609ef3741a03aeeb77c69bb597d834f44a162100e853945b9
+  data.tar.gz: d752ec0976968d5ae7a7a0534e63e7785fa38cbede0a2f8bfecb0e065bb33642bf0a03203411f569fa336051bdf36ecfcd6c3d508c57dff3074280cf5f11955d

data/Rakefile

@@ -25,7 +25,7 @@ require 'jettywrapper'
 import 'lib/tasks/jetty.rake'
 
 desc "Run all specs in spec directory (excluding plugin specs) in an engine_cart-generated app"
-task :ci => ['jetty:clean', 'engine_cart:generate'] do
+task :ci => ['jetty:clean', 'engine_cart:clean', 'engine_cart:generate'] do
   Rake::Task['jetty:config'].invoke
 
   Jettywrapper.wrap(quiet: true,

data/app/views/krikri/records/show.html.erb

@@ -5,6 +5,9 @@
 
   <h2>QA Record: <%= @document.id.html_safe %></h2>
 
+  <%= button_to 'Random record', random_record_url,
+    :class => "btn btn-primary", :method => :get %>
+
   <div class="row">
     <div class="col-md-6 col-sm-6">
       <h3>Enriched Record</h3>

data/lib/krikri.rb

@@ -1,5 +1,5 @@
 ##
-# Require 'blacklight' before 'krikri/engine' to ensure that view order is 
+# Require 'blacklight' before 'krikri/engine' to ensure that view order is
 # configured properly.
 require 'rails'
 require 'devise'
@@ -7,10 +7,10 @@ require 'blacklight'
 require "krikri/engine"
 
 module Krikri
-  autoload :XmlParser,      'krikri/parsers/xml_parser'
-  # autoload Krikri::OaiDcParser
-  autoload :OaiDcParser,    'krikri/parsers/oai_dc_parser'
-  autoload :JsonParser,     'krikri/parsers/json_parser'
-  autoload :ModsParser,     'krikri/parsers/mods_parser'
-  autoload :QdcParser,      'krikri/parsers/qdc_parser'
+  autoload :XmlParser,        'krikri/parsers/xml_parser'
+  autoload :OaiDcParser,      'krikri/parsers/oai_dc_parser'
+  autoload :JsonParser,       'krikri/parsers/json_parser'
+  autoload :ModsParser,       'krikri/parsers/mods_parser'
+  autoload :QdcParser,        'krikri/parsers/qdc_parser'
+  autoload :OaiParserHeaders, 'krikri/parsers/oai_parser_headers'
 end

data/lib/krikri/harvester.rb

@@ -36,6 +36,16 @@ module Krikri
 
     attr_accessor :uri, :name
 
+    included do
+      class << self
+        ##
+        # @see SoftwareAgent#queue_name
+        def queue_name
+          :harvest
+        end
+      end
+    end
+
     ##
     # Accepts options for a generic harvester:
     #   uri: a URI for the harvest endpoint or provider

data/lib/krikri/job.rb

@@ -21,13 +21,14 @@ module Krikri
     end
 
     ##
-    # @abstract run the job's task. Implement the actual task
-    #   against the agent passed in.
-    # @param agent  the agent to run the task
+    # Run the job's task. Receieves a `Krikri::SoftwareAgent` or other object
+    # responding to `#run`.
+    #
+    # @param agent [#run] the agent to run the task
     # @param activity_uri  the URI of the activity responsible for
     #   generating the resources. Set this to (e.g.) prov:wasGeneratedBy
-    def self.run(_, _)
-      raise NotImplementedError
+    def self.run(agent, activity_uri = nil)
+      agent.run(activity_uri)
     end
   end
 end

data/lib/krikri/mapper.rb

@@ -62,13 +62,27 @@ module Krikri
         begin
           Registry.get(name).process_record(rec)
         rescue => e
-          Rails.logger.error("Error processing mapping for #{rec.rdf_subject}" \
-                             "\n#{e.message}\n#{e.backtrace}")
+          desc = mapping_exception_desc(rec)
+          bt = e.backtrace.join("\n")
+          Rails.logger.error("Error processing mapping.\n" \
+                             "#{desc}\n#{e.message}\n#{bt}")
           nil
         end
       end
     end
 
+    ##
+    # Return a string that describes the object encountered by the exception
+    # handler in #map
+    def mapping_exception_desc(rec)
+      if defined? rec.content
+        "content:\n#{rec.content || '[no content]'}"
+      else
+        "object:\n#{rec.inspect}"
+      end
+    end
+    private_class_method :mapping_exception_desc
+
     ##
     # An application-wide registry of defined mappings
     Registry = Class.new(Krikri::Registry)
@@ -82,6 +96,10 @@ module Krikri
 
       attr_reader :name, :generator_uri
 
+      def self.queue_name
+        :mapping
+      end
+
       def initialize(opts = {})
         @name = opts.fetch(:name).to_sym
         @generator_uri = RDF::URI(opts.fetch(:generator_uri))

data/lib/krikri/mapping_dsl/parser_methods.rb

@@ -41,6 +41,20 @@ module Krikri::MappingDSL
       end
     end
 
+    ##
+    # Gives access to a delayed call for the `#header` of a parsed record.
+    #
+    # @return [Proc] a proc that, when called, returns the #rdf_subject of the
+    #   OriginalRecord associated with the parsed record passed as its argument.
+    # @todo consider a more generalized approach
+    def header
+      lambda do |parsed|
+        raise "#{parsed} does not have a `header`" unless
+          parsed.respond_to? :header
+        parsed.header
+      end
+    end
+
     ##
     # This class acts as a proxy for a parsed record's nodes, wrapped in the
     # class passed as the second argument. All methods available on the wrapper

data/lib/krikri/parser.rb

@@ -133,13 +133,38 @@ module Krikri
     class ValueArray
       include Enumerable
 
-      delegate :<<, :[], :[]=, :each, :empty?, :map, :to_a, :to_ary,
+      delegate :[], :each, :empty?, :map, :to_a, :to_ary,
       :to => :@array
 
       def initialize(array = [])
         @array = array
       end
 
+      ##
+      # @see Array#[]=
+      # @raise [InvalidParserValueError] when the value is not a Parser::Value
+      def []=(index, value)
+        raise InvalidParserValueError unless value.is_a? Value
+        @array[index] = value
+        self
+      end
+
+      ##
+      # @see Array#<<
+      # @raise [InvalidParserValueError] when the value is not a Parser::Value
+      def <<(value)
+        raise InvalidParserValueError unless value.is_a? Value
+        @array << value
+        value
+      end
+
+      ##
+      # @see Array#concat
+      # @return [ValueArray]
+      def concat(*args, &block)
+        self.class.new(@array.concat(*args, &block))
+      end
+
       ##
       # @return [Array] literal values from the objects in this array.
       # @see Parser::Value#value
@@ -161,6 +186,18 @@ module Krikri
         result
       end
 
+      ##
+      # Accesses the union of multiple specified fields.
+      #
+      # @return [ValueArray] an array containing the nodes available in the
+      # given fields.
+      def fields(*args)
+        results = args.map do |f|
+          field(*Array(f))
+        end
+        self.class.new(results.flatten)
+      end
+
       ##
       # Retrieves the first element of a ValueArray. Uses an optional argument
       # to specify how many items to return. By design, it behaves similarly
@@ -172,10 +209,18 @@ module Krikri
         self.class.new([@array.first].compact)
       end
 
+      ##
+      # @see Array#concat
+      # @return [ValueArray]
+      def flatten(*args, &block)
+        self.class.new(@array.flatten(*args, &block))
+      end
+
       ##
       # Wraps the result of Array#select in a ValueArray
       #
       # @see Array#select
+      # @return [ValueArray]
       def select(*args, &block)
         self.class.new(@array.select(*args, &block))
       end
@@ -184,6 +229,7 @@ module Krikri
       # Wraps the result of Array#reject in a ValueArray
       #
       # @see Array#reject
+      # @return [ValueArray]
       def reject(*args, &block)
         self.class.new(@array.reject(*args, &block))
       end
@@ -216,6 +262,10 @@ module Krikri
       def get_field(name)
         self.class.new(flat_map { |val| val[name] })
       end
+
+      public
+
+      class InvalidParserValueError < TypeError; end
     end
   end
 end

data/lib/krikri/parsers/oai_dc_parser.rb

@@ -4,6 +4,8 @@ module Krikri
   # metadata path as harvested from OAI-PMH.
   # @see Krikri::XmlParser
   class OaiDcParser < XmlParser
+    include Krikri::OaiParserHeaders
+
     def initialize(record, root_path = '//oai_dc:dc')
       super
     end

data/lib/krikri/parsers/oai_parser_headers.rb

@@ -0,0 +1,20 @@
+module Krikri
+  ##
+  # Concern for Krikri::XmlParsers with oai-style headers
+  # @example
+  #    class MyOaiParser < Krikri::XmlParser
+  #      include Krikri::OaiParserHeaders
+  #    end
+  module OaiParserHeaders
+    extend ActiveSupport::Concern
+
+    ##
+    # @return [Krikri::Parser::ValueArray] a ValueArray containing the
+    #   header node as a `Value` of this parser class
+    def header
+      header_node = Nokogiri::XML(record.to_s).at_xpath('//xmlns:header')
+      Krikri::Parser::ValueArray
+        .new([self.class::Value.new(header_node, root.namespaces)])
+    end
+  end
+end

data/lib/krikri/random_record_generator.rb

@@ -0,0 +1,19 @@
+module Krikri
+  # Gets random records from the search index
+  class RandomRecordGenerator
+
+    def initialize
+      @solr_repo = Blacklight::SolrRepository.new(Blacklight::Configuration.new)
+    end
+
+    # @return Krikri::SearchIndexDocument
+    def record
+      solr_params = { :id => "*:*",
+                      :sort => "random_#{rand(9999)} desc",
+                      :rows => 1 }
+      query_result = @solr_repo.search(solr_params)
+      solr_response = Blacklight::SolrResponse.new(query_result, solr_params)
+      Krikri::SearchIndexDocument.new(solr_response.docs.first)
+    end
+  end
+end

data/lib/krikri/search_results_helper_behavior.rb

@@ -65,7 +65,7 @@ module Krikri
         return error_msg(e.message)
       end
 
-      return error_msg('Original record not found.') unless 
+      return error_msg('Original record not found.') unless
         original_record.present?
       prettify_string(original_record.to_s, original_record.content_type)
     end
@@ -97,5 +97,9 @@ module Krikri
     def error_msg(message = '')
       "There was a problem getting the record.\n\n#{message}"
     end
+
+    def random_record_url
+      url_for_document(Krikri::RandomRecordGenerator.new.record)
+    end
   end
 end

data/lib/krikri/software_agent.rb

@@ -13,7 +13,7 @@ module Krikri
     # This is the name of the most-derived class upon which this is invoked.
     # @see Krikri::Activity
     def agent_name
-      self.class.to_s
+      self.class.agent_name
     end
 
     ##
@@ -45,18 +45,27 @@ module Krikri
       end
 
       ##
-      # @see SoftwareAgent#agent_name
+      # @return a string representation of this SoftwareAgent class
       def agent_name
         to_s
       end
 
+      ##
+      # @return the name of the default queue for jobs invoking this
+      #   SoftwareAgent
+      def queue_name
+        agent_name.downcase
+      end
+
       ##
       # Enqueue a job.
       #
-      # Example:
+      # @example
+      #   MyAgent.enqueue(:name => my_job)
       #
+      # @example
       #   Krikri::Harvesters::OAIHarvester.enqueue(
-      #     Krikri::HarvestJob,
+      #     :harvest,
       #     opts = {
       #       uri: 'http://vcoai.lib.harvard.edu/vcoai/vc',
       #       oai: { set: 'dag', metadata_prefix: 'mods' }
@@ -71,22 +80,31 @@ module Krikri
       # This depends on Redis and Marmotta being available and properly
       # configured (if necessary) in the Rails app.
       #
+      # @param queue_name [#to_s] the Resque queue name
+      # @param opts [Hash] a hash of options that will be used to initialize
+      #   the agent (an instance of this class).
+      #
+      # @return [Boolean]
+      #
       # @see https://github.com/resque/resque/tree/1-x-stable
-      # @see Krikri::HarvestJob
+      # @see Krikri::Job
       # @see Krikri::SoftwareAgent#agent_name
       # @see Krikri::Harvester::expected_opts
-      # @return [Boolean]
-      def enqueue(job_class, opts = {})
-        fail "#{job_class} has no #perform method" unless
-          job_class.respond_to?(:perform)
-        fail 'opts is not a hash' unless opts.is_a?(Hash)
+      def enqueue(*args)
+        queue = args.shift unless args.first.is_a? Hash
+        queue ||= queue_name
+        opts = args.shift || {}
+        fail ArgumentError, "unexpected arguments #{args}" unless args.empty?
+        fail ArgumentError, 'opts is not a hash' unless opts.is_a?(Hash)
+
         activity = Krikri::Activity.create do |a|
           a.agent = agent_name
           a.opts = JSON.generate(opts)
         end
+
         log :info, "created activity #{activity.id}"
-        Resque.enqueue(job_class, activity.id)
-        log :info, "enqueued #{job_class}"
+        Resque.enqueue_to(queue, Krikri::Job, activity.id)
+        log :info, "enqueued to #{queue}"
         true
       end
 

data/lib/krikri/version.rb

@@ -1,3 +1,3 @@
 module Krikri
-  VERSION = "0.1.3"
+  VERSION = "0.2.0"
 end

data/lib/tasks/jetty.rake

@@ -4,7 +4,10 @@ require 'jettywrapper'
 namespace :jetty do
   Jettywrapper.url = 'https://github.com/dpla/marmotta-jetty/archive/3.3.0-solr-4.9.0.zip'
 
-  MARMOTTA_HOME = ENV['MARMOTTA_HOME'] || File.expand_path(File.join(Jettywrapper.app_root, 'jetty', 'marmotta'))
+  DEFAULT_MARMOTTA = File.expand_path(File.join(Jettywrapper.app_root,
+                                                'jetty',
+                                                'marmotta'))
+  MARMOTTA_HOME = ENV['MARMOTTA_HOME'] || DEFAULT_MARMOTTA
 
   desc 'Configure solr schema'
   task :config do
@@ -13,9 +16,10 @@ namespace :jetty do
        'jetty/solr/development-core/conf/solrconfig.xml')
   end
 
-  desc 'Remove the jetty and marmotta directories and recreate them'
-  task :clean do
-    FileUtils.rm_rf(MARMOTTA_HOME)
-    Jettywrapper.clean
+  desc 'Empty the Marmotta home directory used by Jettywrapper'
+  task :clean_marmotta_home do
+    FileUtils.rm_rf(MARMOTTA_HOME) unless MARMOTTA_HOME == DEFAULT_MARMOTTA
   end
+
+  task :clean => [:clean_marmotta_home]
 end