cul_hydra 1.0.0

data/lib/cul_hydra/solrizer/terminology_based_solrizer.rb

@@ -0,0 +1,35 @@
+require 'om'
+module Cul::Hydra::Solrizer::TerminologyBasedSolrizer
+# copied from Solrizer::XML::TerminologyBasedSolrizer
+
+  def self.included(klass)
+    klass.send(:extend, ClassMethods)
+  end
+  
+  # Module Methods
+      
+  module ClassMethods
+  # Populate a solr document with solr fields corresponding to the given xml node
+  # Field names are generated using settings from the term in the +doc+'s terminology corresponding to +term_pointer+
+  # @param [Nokogiri::XML::Node] node to solrize
+  # @param [OM::XML::Document] doc document the node came from
+  # @param [Array] term_pointer Array pointing to the term that should be used for solrization settings
+  # @param [Hash] (optional) solr_doc (values hash) to populate
+    def solrize_node(node_value, doc, term_pointer, term, solr_doc = Hash.new, field_mapper = nil, opts = {})
+      return solr_doc unless term.index_as && !term.index_as.empty?
+      generic_field_name_base = OM::XML::Terminology.term_generic_name(*term_pointer)
+      create_and_insert_terms(generic_field_name_base, node_value, term.index_as, solr_doc)
+      if term_pointer.length > 1
+        #hierarchical_field_name_base = OM::XML::Terminology.term_hierarchical_name(*term_pointer)
+        #create_and_insert_terms(hierarchical_field_name_base, node_value, term.index_as, solr_doc)
+      end
+      if term.variant_of and term.variant_of[:field_base]
+        #create_and_insert_terms(term.variant_of[:field_base], node_value, term.index_as, solr_doc)
+        create_and_insert_terms(term.variant_of[:field_base], node_value, term.index_as, solr_doc)
+      end
+      solr_doc
+    end
+
+  end
+
+end # module

data/lib/cul_hydra/solrizer/value_mapper.rb

@@ -0,0 +1,46 @@
+module Cul::Hydra::Solrizer
+# This class uses a yaml map to translate field values for solr
+class ValueMapper
+  def self.load_value_maps(config_path=nil)
+    if config_path.nil?
+      if defined?(Rails.root) && !Rails.root.nil?
+        config_path = File.join(Rails.root, "config", "solr_value_maps.yml")
+      end
+      # Default to using the config file within the gem
+      if !File.exist?(config_path.to_s)
+        config_path = File.join(File.dirname(__FILE__), "..", "..", "..", "config", "solr_value_maps.yml")
+      end
+    end
+    #logger.info("ValueMapper: loading field value maps from #{File.expand_path(config_path)}")
+    YAML::load(File.open(config_path))
+  end
+
+  def self.default_value_maps
+    @@value_maps ||= self.load_value_maps
+  end
+  # Instance methods
+
+  def initialize(value_maps=nil)
+    @value_maps = (value_maps || Cul::Hydra::Solrizer::ValueMapper.default_value_maps).with_indifferent_access
+    @mapped_fields = {}.with_indifferent_access
+  end
+
+  def maps_field?(field_key)
+    @mapped_fields.has_key? field_key
+  end
+
+  def map_field(field_key, map_key)
+    @mapped_fields[field_key] = map_key
+  end
+
+  def map_value(field_key, value_key)
+    map_key = @mapped_fields[field_key]
+    return value_key unless @mapped_fields.has_key? field_key and @value_maps.has_key? map_key
+    if value_key.is_a? Array
+      value_key.collect{ |val| @value_maps[map_key].fetch(val, val) }
+    else
+      @value_maps[map_key].fetch(value_key, value_key)
+    end
+  end
+end
+end

data/lib/cul_hydra/solrizer_patch.rb

@@ -0,0 +1,172 @@
+require 'solrizer'
+module Solrizer::DefaultDescriptors
+  def self.date_sortable
+    @date_sortable ||= Solrizer::Descriptor.new(:date, :stored, :indexed, converter: date_sortable_converter)
+  end
+
+  # Produces the field name 'all_text_teim'
+  def self.textable
+    @textable_type ||= TextableDescriptor.new(:text_en, :indexed, :multivalued)
+  end
+
+  # Produces _sim suffix
+  def self.project_facetable
+    @project_facet_type ||= ProjectFacetDescriptor.new(:string, :indexed, :multivalued)
+  end
+
+  def self.project_textable
+    @project_textable_type ||= ProjectTextableDescriptor.new(:text_en, :indexed, :multivalued)
+  end
+
+  ## Produces _sim suffix and a value-mapping converter
+  #def self.marc_code_facetable
+  #  @marc_code_facet_type ||= MarcCodeFacetDescriptor.new(:string, :indexed, :multivalued)
+  #end
+
+  ## Produces _ssm suffix and a value-mapping converter
+  #def self.marc_code_displayable
+  #  @marc_code_type ||= MarcCodeDisplayDescriptor.new(:string, :stored, :indexed, :multivalued)
+  #end
+
+  # Produces all_text_timv fieldname and a value-mapping converter
+  def self.marc_code_textable
+    @marc_code_map_text_type ||= MarcCodeTextableDescriptor.new(:text_en, :indexed, :multivalued)
+  end
+
+  def self.load_value_maps(config_path=nil)
+    if config_path.nil?
+      if defined?(Rails.root) && !Rails.root.nil?
+        config_path = File.join(Rails.root, "config", "solr_value_maps.yml")
+      end
+      # Default to using the config file within the gem
+      if !File.exist?(config_path.to_s)
+        logger.warn("ValueMapper: no field value maps at #{config_path}")
+        config_path = File.join(File.dirname(__FILE__), "..", "..", "..", "config", "solr_value_maps.yml")
+      end
+      if !File.exist?(config_path.to_s)
+        logger.warn("ValueMapper: no field value maps at #{File.expand_path(config_path)}")
+        return {}
+      end
+    end
+    logger.info("ValueMapper: loading field value maps from #{File.expand_path(config_path)}")
+    YAML::load(File.open(config_path))
+  end
+
+  def self.value_maps
+    @@value_maps ||= load_value_maps
+  end
+
+  def self.date_sortable_converter
+    lambda do |type|
+      lambda do |val|
+        begin
+          d = val.length < 11 ? Date.new(*(val.split('-').collect {|s| s.to_i})) : Date.parse(val)
+          iso8601_date(d)
+        rescue ArgumentError
+          nil
+        end
+      end
+    end
+  end
+
+  module Normal
+    SHORT_REPO = "ldpd.short.repo"
+    LONG_REPO  = "ldpd.long.repo"
+    FULL_REPO = "ldpd.full.repo"
+    SHORT_PROJ = "ldpd.short.project"
+    FULL_PROJ = "ldpd.full.project"
+    def normal(value)
+      normal!(value.clone)
+    end
+    def normal!(value)
+      value.gsub!(/\s+/,' ')
+      value.strip!
+      value
+    end
+    def translate_with_default(prefix, value, default)
+      begin
+        # Using method below to handle translations because our YAML keys can contain periods and this doesn't play well with the translation dot-syntax
+        translations = HashWithIndifferentAccess.new(I18n.t(prefix))
+        if translations.has_key?(value)
+          return translations[value]
+        else
+          return default
+        end
+      rescue
+        return default
+      end
+    end
+  end
+
+  class TextableDescriptor < Solrizer::Descriptor
+    include Normal
+    def name_and_converter(field_name, args=nil)
+      super('all_text', args)
+    end
+    def converter(field_type)
+      lambda {|value| normal(value)}
+    end
+  end
+
+  class ProjectTextableDescriptor < Solrizer::Descriptor
+    include Normal
+    def name_and_converter(field_name, args=nil)
+      super('all_text', args)
+    end
+    def converter(field_type)
+      lambda do |value|
+        if value.is_a? String
+          normal!(value)
+          r = [translate_with_default(SHORT_PROJ, value, normal!(value))]
+          r << translate_with_default(FULL_PROJ, value, normal!(value))
+          r.uniq!
+          r.join(' ')
+        else
+          raise "unexpected project_textable #{value.inspect}"
+          value
+        end
+      end
+    end
+  end
+
+  class ProjectFacetDescriptor < Solrizer::Descriptor
+    include Normal
+    def converter(field_type)
+      lambda {|value| translate_with_default(SHORT_PROJ, normal!(value), normal!(value))}
+    end
+  end
+
+  #class MarcCodeFacetDescriptor < Solrizer::Descriptor
+  #  include Normal
+  #  def converter(field_type)
+  #    lambda {|value| translate_with_default(SHORT_REPO, normal!(value), 'Non-Columbia Location')}
+  #  end
+  #end
+
+  #class MarcCodeDisplayDescriptor < Solrizer::Descriptor
+  #  include Normal
+  #  def converter(field_type)
+  #    lambda {|value| translate_with_default(LONG_REPO, normal!(value), 'Non-Columbia Location')}
+  #  end
+  #end
+
+  class MarcCodeTextableDescriptor < Solrizer::Descriptor
+    include Normal
+    def name_and_converter(field_name, args=nil)
+      super('all_text', args)
+    end
+    def converter(field_type)
+      lambda do |value|
+        if value.is_a? String
+          normal!(value)
+          r = [translate_with_default(SHORT_REPO, value, 'Non-Columbia Location')]
+          r << translate_with_default(LONG_REPO, value, 'Non-Columbia Location')
+          r.uniq!
+          r.join(' ')
+        else
+          value
+        end
+      end
+    end
+  end
+end

data/lib/cul_hydra/version.rb

@@ -0,0 +1,8 @@
+module Cul
+  module Hydra
+    VERSION = '1.0.0'
+    def self.version
+      VERSION
+    end
+  end
+end

data/lib/cul_hydra/version.rb~

@@ -0,0 +1,8 @@
+module Cul
+  module Hydra
+    VERSION = '0.22.9.10'
+    def self.version
+      VERSION
+    end
+  end
+end

data/lib/tasks/cmodel.rake

@@ -0,0 +1,122 @@
+APP_ROOT = File.expand_path("#{File.dirname(__FILE__)}/../../") unless defined?(APP_ROOT)
+require 'active-fedora'
+
+def logger
+  @logger ||= Logger.new($stdout)
+end
+
+def filename_for_pid(pid)
+  pid.gsub(/\:/,'_') + '.xml'
+end
+
+def pid_for_filename(fname)
+  fname.sub(/\.xml$/,'').sub(/_/,':')
+end
+
+def cmodel_fixture(name)
+  path = File.join(APP_ROOT, 'fixtures','cmodels',name)
+  File.open(path, 'rb')
+end
+
+def each_cmodel
+  path = File.join(APP_ROOT, 'fixtures','cmodels')
+  Dir.new(path).each do |fname|
+    if fname =~ /\.xml$/
+      yield pid_for_filename(fname)
+    end
+  end
+end
+
+def config_subs
+  @subs ||= begin
+    cfile = File.join(APP_ROOT,'config','subs.yml')
+    subs = {}
+    if File.exists? cfile
+      open(cfile) {|blob| subs = YAML::load(blob)[ENV['RAILS_ENV'] || 'test'] }
+    else
+      logger.warn("No subs.yml found; CModels will be loaded without inline substitutions")
+    end
+    subs
+  end
+  @subs
+end
+
+def do_subs(orig)
+  content = orig.clone
+  config_subs.each do |key, val|
+    content.gsub!(/\$#{key.to_s}\$/, val)
+  end
+  content
+end
+
+def connection
+  # no need to go through AF for this except laziness re: finding the YAML
+  @connection ||= (ActiveFedora::Base.fedora_connection[0] ||= ActiveFedora::RubydoraConnection.new(ActiveFedora.config.credentials)).connection
+end
+
+def content_for(pid)
+  fname = filename_for_pid(pid)
+  fcontent = cmodel_fixture(fname).read
+  fcontent = do_subs(fcontent)
+end
+
+def load_content(content, pid)
+  begin
+    connection.ingest(:file=>StringIO.new(content), :pid=>pid)
+  rescue Exception => e
+    puts "possible problem with ingest of #{pid}: #{e.message}"
+    raise e
+  end
+end
+
+def purge(pid)
+  begin
+    connection.purge_object :pid=>pid
+  rescue Exception => e
+    puts "possible problem with purge of #{pid}: #{e.message}"
+  end
+
+end
+
+def reload(pid)
+  fcontent = content_for(pid)
+  purge(pid)
+  load_content(fcontent, pid)
+end
+
+
+namespace :cul_hydra do
+  namespace :cmodel do
+    task :test do #=> :environment do
+      pid = ENV["PID"]
+      puts content_for(pid)
+    end
+
+    task :load do #=> :environment do
+      pid = ENV["PID"]
+      load_content(content_for(pid),pid)
+    end
+
+    task :purge do #=> :environment do
+      pid = ENV["PID"]
+      purge(pid)
+    end
+
+    task :reload do #=> :environment do
+      pid = ENV["PID"]
+      reload(pid)
+    end
+
+    task :reload_all do #=> :environment do
+      pattern = ENV["PATTERN"]
+      pattern = Regexp.compile(pattern) if pattern
+      reload("ldpd:nullbind")
+      each_cmodel do |pid|
+        unless (pattern and not pid =~ pattern)
+          puts "reloading #{pid}"
+          reload(pid)
+        end
+      end
+    end
+  end
+end

data/lib/tasks/cul_hydra_dev.rake

@@ -0,0 +1,54 @@
+APP_ROOT = File.expand_path("#{File.dirname(__FILE__)}/../../") unless defined?(APP_ROOT)
+
+require 'jettywrapper'
+JETTY_ZIP_BASENAME = '7.x-stable'
+Jettywrapper.url = "https://github.com/projecthydra/hydra-jetty/archive/#{JETTY_ZIP_BASENAME}.zip"
+
+namespace :cul_hydra do
+
+  begin
+    # This code is in a begin/rescue block so that the Rakefile is usable
+    # in an environment where RSpec is unavailable (i.e. production).
+
+    require 'rspec/core/rake_task'
+
+    RSpec::Core::RakeTask.new(:rspec) do |spec|
+      spec.pattern = FileList['spec/**/*_spec.rb']
+      spec.pattern += FileList['spec/*_spec.rb']
+      spec.rspec_opts = ['--backtrace'] if ENV['CI']
+    end
+
+    RSpec::Core::RakeTask.new(:rcov) do |spec|
+      spec.pattern = FileList['spec/**/*_spec.rb']
+      spec.pattern += FileList['spec/*_spec.rb']
+      spec.rcov = true
+    end
+
+  rescue LoadError => e
+    puts "[Warning] Exception creating rspec rake tasks.  This message can be ignored in environments that intentionally do not pull in the RSpec gem (i.e. production)."
+    puts e
+  end
+
+  desc "CI build"
+  task :ci do
+    ENV['environment'] = "test"
+    #Rake::Task["active_fedora:configure_jetty"].invoke
+    jetty_params = Jettywrapper.load_config
+    error = Jettywrapper.wrap(jetty_params) do
+      Rake::Task["cul_hydra:cmodel:reload_all"].invoke
+      Rake::Task['cul_hydra:coverage'].invoke
+    end
+    raise "test failures: #{error}" if error
+  end
+
+  desc "Execute specs with coverage"
+  task :coverage do
+    # Put spec opts in a file named .rspec in root
+    ruby_engine = defined?(RUBY_ENGINE) ? RUBY_ENGINE : "ruby"
+    ENV['COVERAGE'] = 'true' unless ruby_engine == 'jruby'
+
+   # Rake::Task["active_fedora:fixtures"].invoke
+    Rake::Task["cul_hydra:rspec"].invoke
+  end
+
+end