gdor-indexer 0.5.0 → 0.6.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 4724a7e6363dc6d287c4bf1bf6033e000930bf90
-  data.tar.gz: 72fbd1746a114c918ebc9dd838f9674f7fddb8b0
+  metadata.gz: d023031326d6543f222881f79d0a102a6a27f0e9
+  data.tar.gz: 5dd3419e6e38601413f737b3b3011ce3e9baecd9
 SHA512:
-  metadata.gz: 88b7024e0faf8b01d0091b6f53e1683d9f9fc9ba6602d4905d2078e682d7cdb284c95e63d362698ef22709961f313f002ec483dcad43f5f4e224149e6269d339
-  data.tar.gz: 8b9b76607dabc0f1409a66b37efb1fd0e07c8afc29a4c36d65a3f144b60b6b531afb3251512a97cd75a47164ce8b599cdf79a0d12fd795e2f85c8385983fbb21
+  metadata.gz: f182f105f741f8c21bb15c369573b6c3da905b28ef8e0d8920a421dcdacce4beea22891fcbc78dca2c60690c087e800f14f986749b657c218e88fe93247ce072
+  data.tar.gz: 882369a75b3c62d0176746241f60e38204573644855514bc4fc5764ab3fcf97c3758bf7efcdc0fe8bd579102ab355255dfe76ec4730a6c69bffcfdab9b910e76

data/.rspec

@@ -0,0 +1,2 @@
+--color
+--require spec_helper

data/.rubocop_todo.yml

@@ -1,29 +1,11 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2016-01-13 15:12:03 -0800 using RuboCop version 0.34.2.
+# on 2016-07-20 15:53:51 -0700 using RuboCop version 0.41.2.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
 # versions of RuboCop, may require this file to be generated again.
 
-# Offense count: 6
-Lint/AmbiguousRegexpLiteral:
-  Exclude:
-    - 'spec/unit/indexer_spec.rb'
-
-# Offense count: 1
-# Cop supports --auto-correct.
-# Configuration parameters: AlignWith, SupportedStyles, AutoCorrect.
-Lint/EndAlignment:
-  Enabled: false
-
-# Offense count: 4
-Lint/UselessAssignment:
-  Exclude:
-    - 'lib/gdor/indexer.rb'
-    - 'spec/unit/indexer_spec.rb'
-    - 'spec/unit/solr_doc_builder_spec.rb'
-
 # Offense count: 21
 Metrics/AbcSize:
   Max: 82
@@ -31,18 +13,19 @@ Metrics/AbcSize:
 # Offense count: 1
 # Configuration parameters: CountComments.
 Metrics/ClassLength:
-  Max: 243
+  Max: 240
 
 # Offense count: 5
 Metrics/CyclomaticComplexity:
   Max: 9
 
-# Offense count: 318
-# Configuration parameters: AllowURI, URISchemes.
+# Offense count: 314
+# Configuration parameters: AllowHeredoc, AllowURI, URISchemes.
+# URISchemes: http, https
 Metrics/LineLength:
   Max: 258
 
-# Offense count: 15
+# Offense count: 13
 # Configuration parameters: CountComments.
 Metrics/MethodLength:
   Max: 43
@@ -51,6 +34,23 @@ Metrics/MethodLength:
 Metrics/PerceivedComplexity:
   Max: 10
 
+# Offense count: 15
+RSpec/AnyInstance:
+  Exclude:
+    - 'spec/unit/indexer_spec.rb'
+    - 'spec/unit/solr_doc_hash_spec.rb'
+
+# Offense count: 43
+# Configuration parameters: Max.
+RSpec/ExampleLength:
+  Exclude:
+    - 'spec/unit/indexer_spec.rb'
+    - 'spec/unit/mods_fields_spec.rb'
+    - 'spec/unit/mods_pub_fields_spec.rb'
+    - 'spec/unit/mods_subject_fields_spec.rb'
+    - 'spec/unit/public_xml_fields_spec.rb'
+    - 'spec/unit/solr_doc_hash_spec.rb'
+
 # Offense count: 7
 # Configuration parameters: CustomTransform.
 RSpec/FilePath:
@@ -63,22 +63,30 @@ RSpec/FilePath:
     - 'spec/unit/solr_doc_builder_spec.rb'
     - 'spec/unit/solr_doc_hash_spec.rb'
 
-# Offense count: 193
+# Offense count: 163
 RSpec/InstanceVariable:
   Exclude:
     - 'spec/unit/indexer_spec.rb'
     - 'spec/unit/public_xml_fields_spec.rb'
     - 'spec/unit/solr_doc_builder_spec.rb'
 
+# Offense count: 7
+# Configuration parameters: IgnoreSymbolicNames.
+RSpec/VerifiedDoubles:
+  Exclude:
+    - 'spec/unit/indexer_spec.rb'
+
 # Offense count: 7
 # Cop supports --auto-correct.
 # Configuration parameters: EnforcedStyle, SupportedStyles.
+# SupportedStyles: braces, no_braces, context_dependent
 Style/BracesAroundHashParameters:
   Exclude:
     - 'spec/unit/indexer_spec.rb'
 
 # Offense count: 4
 # Configuration parameters: EnforcedStyle, SupportedStyles.
+# SupportedStyles: nested, compact
 Style/ClassAndModuleChildren:
   Exclude:
     - 'lib/gdor/indexer/mods_fields.rb'
@@ -86,54 +94,28 @@ Style/ClassAndModuleChildren:
     - 'lib/gdor/indexer/solr_doc_builder.rb'
     - 'lib/gdor/indexer/solr_doc_hash.rb'
 
-# Offense count: 4
-# Configuration parameters: Exclude.
-Style/Documentation:
-  Exclude:
-    - 'lib/gdor/indexer.rb'
-    - 'lib/gdor/indexer/nokogiri_xml_node_mixin.rb'
-    - 'lib/gdor/indexer/solr_doc_hash.rb'
-    - 'lib/gdor/indexer/version.rb'
-
-# Offense count: 1
-Style/DoubleNegation:
-  Exclude:
-    - 'lib/gdor/indexer/solr_doc_hash.rb'
-
 # Offense count: 1
 # Cop supports --auto-correct.
 # Configuration parameters: EnforcedStyle, SupportedStyles.
+# SupportedStyles: empty_lines, no_empty_lines
 Style/EmptyLinesAroundBlockBody:
   Exclude:
     - 'spec/unit/mods_pub_fields_spec.rb'
 
-# Offense count: 3
-# Configuration parameters: MinBodyLength.
-Style/GuardClause:
-  Exclude:
-    - 'lib/gdor/indexer.rb'
-
-# Offense count: 21
+# Offense count: 22
 # Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles.
+# Configuration parameters: EnforcedStyle, SupportedStyles, IndentationWidth.
+# SupportedStyles: aligned, indented
 Style/MultilineOperationIndentation:
-  Enabled: false
-
-# Offense count: 1
-# Configuration parameters: NamePrefix, NamePrefixBlacklist.
-Style/PredicateName:
   Exclude:
-    - 'lib/gdor/indexer/mods_fields.rb'
+    - 'Rakefile'
+    - 'spec/unit/mods_pub_fields_spec.rb'
 
 # Offense count: 2
 # Cop supports --auto-correct.
 # Configuration parameters: EnforcedStyle, SupportedStyles, AllowInnerSlashes.
+# SupportedStyles: slashes, percent_r, mixed
 Style/RegexpLiteral:
   Exclude:
     - 'spec/unit/indexer_spec.rb'
     - 'spec/unit/mods_subject_fields_spec.rb'
-
-# Offense count: 1
-Style/UnlessElse:
-  Exclude:
-    - 'lib/gdor/indexer.rb'

data/.travis.yml

@@ -0,0 +1,11 @@
+sudo: false
+language: ruby
+cache: bundler
+notifications:
+  email: false
+rvm:
+  - 2.3.1
+env:
+  global:
+    - NOKOGIRI_USE_SYSTEM_LIBRARIES=true
+jdk: oraclejdk8

data/README.md

@@ -1,4 +1,4 @@
-[![Dependency Status](https://gemnasium.com/sul-dlss/gdor-indexer.svg)](https://gemnasium.com/sul-dlss/gdor-indexer) [![Gem Version](https://badge.fury.io/rb/gdor-indexer.svg)](http://badge.fury.io/rb/gdor-indexer)
+[![Build Status](https://travis-ci.org/sul-dlss/gdor-indexer.svg)](https://travis-ci.org/sul-dlss/gdor-indexer) [![Coverage Status](https://coveralls.io/repos/sul-dlss/gdor-indexer/badge.svg?branch=master&service=github)](https://coveralls.io/github/sul-dlss/gdor-indexer?branch=master) [![Dependency Status](https://gemnasium.com/sul-dlss/gdor-indexer.svg)](https://gemnasium.com/sul-dlss/gdor-indexer) [![Gem Version](https://badge.fury.io/rb/gdor-indexer.svg)](http://badge.fury.io/rb/gdor-indexer)
 
 # gdor-indexer
 
@@ -6,55 +6,59 @@ Code to harvest DOR druids via DOR Fetcher service, mods from PURL, and use it t
 
 ## Prerequisites
 
-1. ruby
+1. ruby 2.x+
 2. bundler gem must be installed
 
 ## Install steps for running locally
 
 Add this line to your application's Gemfile:
-
-    gem 'harvestdor-indexer'
+```ruby
+gem 'harvestdor-indexer'
+```
 
 Then execute:
-
-    $ bundle
+```bash
+bundle
+```
 
 ## Configuration
 
-#### Create a collections folder in the config directory:
+### Create a collections folder in the config directory:
 
-    $ cd /path/to/gdor-indexer/config
-    $ mkdir collections
+```bash
+cd /path/to/gdor-indexer/config
+mkdir collections
+```
 
-#### Create a yml config file for your collection(s) to be harvested and indexed.
+### Create a yml config file for your collection(s) to be harvested and indexed.
 
-See ```spec/config/walters_integration_spec.yml``` for an example.  Copy that file to ```config/collections``` and change the following settings:
+See `spec/config/walters_integration_spec.yml` for an example.  Copy that file to `config/collections` and change the following settings:
 
 * whitelist
 * dor_fetcher service_url
 * harvestdor log_dir and log_name
 * solr_url
 
-##### whitelist
+#### whitelist
 
-The whitelist is how you specify which objects to index.  The whitelist can be
+The whitelist is how you specify which objects to index.  The whitelist can be:
 
 * an Array of druids inline in the config yml file
 * a filename containing a list of druids (one per line)
 
-If a druid, per the object's identityMetadata at purl page, is for a
+If a druid, per the object's identityMetadata at purl page, is for a:
 
-* collection record:  then we process all the item druids in that collection (as if they were included individually in the whitelist)
+* collection record: then we process all the item druids in that collection (as if they were included individually in the whitelist)
 * non-collection record: then we process the druid as an individual item
 
-#### Run the indexer script
+### Run the indexer script
 
     $ cd /path/to/gdor-indexer
     $ nohup ./bin/indexer -c my_collection &>path/to/nohup.output
 
 ## Running the tests
 
-  ```$ rake```
+`rake`
 
 ## Contributing
 

data/gdor-indexer.gemspec

@@ -18,7 +18,7 @@ Gem::Specification.new do |spec|
   spec.require_paths = ['lib']
 
   spec.add_dependency 'harvestdor-indexer'
-  spec.add_dependency 'stanford-mods', '>= 1.4.0' # for new pub date methods
+  spec.add_dependency 'stanford-mods', '~> 2.2.1'
   spec.add_dependency 'nokogiri'
   spec.add_dependency 'rsolr'
   spec.add_dependency 'activesupport'

data/lib/gdor/indexer.rb

@@ -12,8 +12,8 @@ require 'logger'
 require 'net/smtp'
 require 'set'
 
-# Base class to harvest from DOR via harvestdor gem
 module GDor
+  # Base class to harvest from DOR via harvestdor gem
   class Indexer
     include Hooks
 
@@ -89,12 +89,12 @@ module GDor
         index_with_exception_handling resource
       end
 
-      unless nocommit
+      if nocommit
+        logger.info('Skipping commit per nocommit flag')
+      else
         logger.info('Beginning Commit.')
         solr_client.commit!
         logger.info('Finished Commit.')
-      else
-        logger.info('Skipping commit per nocommit flag')
       end
 
       @total_time = elapsed_time(start_time)
@@ -156,6 +156,7 @@ module GDor
     # Create Solr document for the collection druid suitable for SearchWorks
     #  and write the result to the SearchWorks Solr Index
     # @param [Harvestdor::Indexer::Resource] resource a collection record
+    # @return [Hash]
     def collection_solr_document(resource)
       coll_sdb = GDor::Indexer::SolrDocBuilder.new(resource, logger)
 
@@ -166,7 +167,6 @@ module GDor
         collection_type: 'Digital Collection',
         display_type: coll_display_types_from_items(resource),
         format_main_ssim: 'Archive/Manuscript', # per INDEX-12, add this to all collection records (does not add dups)
-        format: 'Manuscript/Archive', # per INDEX-144, add this to all collection records (does not add dups)
         building_facet: 'Stanford Digital Repository' # INDEX-53 add building_facet = Stanford Digital Repository here for collection
       )
 
@@ -181,7 +181,7 @@ module GDor
 
     # add coll level data to this solr doc and/or cache collection level information
     # @param [Hash] doc_hash representing the Solr document (for an item)
-    # @param [Array<Harvestdor::Indexer::Resource>] collections  the collections the item is a member of
+    # @param [Array<Harvestdor::Indexer::Resource>] collections the collections the item is a member of
     def add_coll_info(doc_hash, collections)
       if collections
         doc_hash[:collection] = []
@@ -213,9 +213,8 @@ module GDor
     # cache the display_type of this (item) object with a collection, so when the collection rec
     # is being indexed, it can get all of the display_types of the members
     def cache_display_type_for_collection(resource, display_type)
-      if display_type && display_type.instance_of?(String)
-        coll_display_types_from_items(resource) << display_type
-      end
+      return unless display_type && display_type.instance_of?(String)
+      coll_display_types_from_items(resource) << display_type
     end
 
     # count the number of records in solr for this collection (and the collection record itself)
@@ -238,8 +237,7 @@ module GDor
     # @return [Array<String>] Array of messages suitable for notificaiton email and/or logs
     def record_count_msgs
       @record_count_msgs ||= begin
-        msgs = []
-        msgs << "Successful count (items + coll record indexed w/o error): #{metrics.success_count}"
+        msgs = ["Successful count (items + coll record indexed w/o error): #{metrics.success_count}"]
 
         harvestdor.resources.select(&:collection?).each do |collection|
           solr_count = num_found_in_solr(collection: collection.bare_druid)
@@ -270,12 +268,11 @@ module GDor
       logger.info("Avg complete index time per object (all): #{(@total_time / metrics.total).round(2)} seconds") unless metrics.total == 0
     end
 
+    # @return [String] the composed email body
     def email_report_body
-      body = ''
-
-      body += "\n" + record_count_msgs.join("\n") + "\n"
+      body = "\n" + record_count_msgs.join("\n") + "\n"
 
-      if @druids_failed_to_ix.size > 0
+      unless @druids_failed_to_ix.empty?
         body += "\n"
         body += "records that may have failed to index: \n"
         body += @druids_failed_to_ix.join("\n") + "\n"
@@ -285,7 +282,7 @@ module GDor
       body += "full log is at gdor_indexer/shared/#{config.harvestdor.log_dir}/#{config.harvestdor.log_name} on #{Socket.gethostname}"
       body += "\n"
 
-      body += @validation_messages.join("\n") + "\n"
+      body + @validation_messages.join("\n") + "\n"
     end
 
     # email the results of indexing if we are on one of the harvestdor boxes
@@ -305,12 +302,14 @@ module GDor
       end
     end
 
+    # @param [String] to target email address
+    # @param [Hash] opts options
     def send_email(to, opts = {})
-      opts[:server] ||= 'localhost'
-      opts[:from] ||= 'gryphondor@stanford.edu'
+      opts[:server]     ||= 'localhost'
+      opts[:from]       ||= 'gryphondor@stanford.edu'
       opts[:from_alias] ||= 'gryphondor'
-      opts[:subject] ||= 'default subject'
-      opts[:body] ||= 'default message body'
+      opts[:subject]    ||= 'default subject'
+      opts[:body]       ||= 'default message body'
       mail = Mail.new do
         from opts[:from]
         to to

data/lib/gdor/indexer/mods_fields.rb

@@ -81,14 +81,14 @@ module GDor::Indexer::ModsFields
   # @return [Hash] updated Hash representing the Solr document
   def add_pub_year_tisim(doc_hash)
     pub_date_sort_val = doc_hash[:pub_year_isi]
-    if is_positive_int? pub_date_sort_val
+    if positive_int? pub_date_sort_val
       doc_hash[:pub_year_tisim] = pub_date_sort_val # for date slider
     end
     doc_hash
   end
 
   # @return true if the string parses into an int, and if so, the int is >= 0
-  def is_positive_int?(str)
+  def positive_int?(str)
     str.to_i >= 0
   rescue
     false

data/lib/gdor/indexer/nokogiri_xml_node_mixin.rb

@@ -1,6 +1,6 @@
-# Monkey patch for Nokogiri to cache xpath contexts and make things faster under jRuby
 module Nokogiri
   module XML
+    # Monkey patch for Nokogiri to cache xpath contexts and make things faster under jRuby
     class Node
       @context = nil
 

data/lib/gdor/indexer/solr_doc_hash.rb

@@ -1,33 +1,34 @@
 require 'delegate'
 
 class GDor::Indexer
+  # Hash-like SolrDoc Object class
   class SolrDocHash < SimpleDelegator
     def initialize(hash = {})
       super(hash)
     end
 
-    # looks for non-empty existence of field when exp_val is nil;
+    # @param [String,Regexp] exp_val
+    # when exp_val is nil, looks for non-empty existence of field
     # when exp_val is a String, looks for matching value as a String or as a member of an Array
     # when exp_val is a Regexp, looks for String value that matches, or Array with a String member that matches
-    # @return true if the field is non-trivially present in the hash, false otherwise
+    # @return [Boolean] true if the field is non-trivially present in the hash, false otherwise
     def field_present?(field, exp_val = nil)
-      !!(if self.include?(field) && Array(self[field]).any? { |v| !v.blank? }
-           actual = Array(self[field])
-
-           case exp_val
-           when nil
-             true
-           when Regexp
-             actual.index { |s| exp_val.match(s) }
-           else
-             actual.include? exp_val
-           end
-      end)
+      return false unless include?(field) && Array(self[field]).any? { |v| !v.blank? }
+      case exp_val
+      when nil
+        true
+      when Regexp
+        Array(self[field]).index { |s| exp_val.match(s) }
+      else
+        Array(self[field]).include? exp_val
+      end
     end
 
     # merge in field values from the new hash, with the following guarantees:
     #  values for keys in new_hash will be a non-empty String or flat Array
     #  keys will be removed from hash if all values are nil or empty
+    # @param [Hash] new_hash
+    # @return [GDor::Indexer::SolrDocHash] self
     def combine(new_hash)
       new_hash.select { |_key, value| Array(value).any? { |v| !v.blank? } }.each do |key, new_val|
         if field_present? key
@@ -46,10 +47,10 @@ class GDor::Indexer
       end
 
       compact_blank_fields!
-
       self
     end
 
+    # @return [GDor::Indexer::SolrDocHash] self
     def compact_blank_fields!
       keys.reject { |key| field_present? key }.each do |key|
         delete key
@@ -57,11 +58,13 @@ class GDor::Indexer
       self
     end
 
+    # @return [String]
     def druid
       self[:druid]
     end
 
     # validate fields that should be in hash for any item object in SearchWorks Solr
+    # @param [Object] config Configuration object
     # @return [Array<String>] Array of messages suitable for notificaiton email and/or logs
     def validate_item(config)
       result = validate_gdor_fields(config)
@@ -75,7 +78,7 @@ class GDor::Indexer
     end
 
     # validate fields that should be in hash for any collection object in SearchWorks Solr
-    # @return [Array<String>] Array of messages suitable for notificaiton email and/or logs
+    # @see #validate_item for param and return
     def validate_collection(config)
       result = validate_gdor_fields(config)
       result << "#{druid} missing collection_type 'Digital Collection'\n" unless field_present?(:collection_type, 'Digital Collection')
@@ -84,7 +87,7 @@ class GDor::Indexer
     end
 
     # validate fields that should be in hash for every gryphonDOR object in SearchWorks Solr
-    # @return [Array<String>] Array of messages suitable for notificaiton email and/or logs
+    # @see #validate_item for param and return
     def validate_gdor_fields(config)
       result = []
       result << "#{druid} missing druid field\n" unless field_present?(:druid, druid)
@@ -96,7 +99,7 @@ class GDor::Indexer
     end
 
     # validate fields that should be in doc hash for every unmerged gryphonDOR object in SearchWorks Solr
-    # @return [Array<String>] array of Strings indicating absence of required fields
+    # @see #validate_item for param and return
     def validate_mods(_config)
       result = []
       result << "#{druid} missing modsxml\n" unless field_present?(:modsxml)

data/lib/gdor/indexer/version.rb

@@ -1,5 +1,5 @@
 module GDor
   class Indexer
-    VERSION = '0.5.0'
+    VERSION = '0.6.0'.freeze
   end
 end

data/spec/unit/indexer_spec.rb

@@ -1,9 +1,8 @@
-require 'spec_helper'
+require 'yaml'
 
 describe GDor::Indexer do
   before(:all) do
     @config_yml_path = File.join(File.dirname(__FILE__), '..', 'config', 'walters_integration_spec.yml')
-    require 'yaml'
     @yaml = YAML.load_file(@config_yml_path)
     @ns_decl = "xmlns='#{Mods::MODS_NS}'"
     @fake_druid = 'oo000oo0000'
@@ -82,7 +81,7 @@ describe GDor::Indexer do
 
       @indexer.harvest_and_index
     end
-    it 'indexs each resource' do
+    it 'indexes each resource' do
       allow(@indexer).to receive(:harvestdor).and_return(Class.new do
         def initialize(*items)
           @items = items
@@ -109,13 +108,13 @@ describe GDor::Indexer do
       @indexer.harvest_and_index
     end
     it 'does not commit if nocommit is set' do
-      expect(@indexer.solr_client).to_not receive(:commit!)
+      expect(@indexer.solr_client).not_to receive(:commit!)
       @indexer.harvest_and_index(true)
     end
   end
 
   describe '#index' do
-    it 'indexs collections as collections' do
+    it 'indexes collections as collections' do
       expect(@indexer).to receive(:collection_solr_document).with(collection)
       @indexer.index collection
     end
@@ -196,17 +195,16 @@ describe GDor::Indexer do
   end # item_solr_document
 
   context '#collection_solr_document' do
+    let(:doc_hash) { GDor::Indexer::SolrDocHash.new }
     it 'calls validate_collection' do
-      doc_hash = GDor::Indexer::SolrDocHash.new
       allow_any_instance_of(GDor::Indexer::SolrDocBuilder).to receive(:doc_hash).and_return(doc_hash) # speed up the test
       expect(doc_hash).to receive(:validate_collection).and_return([])
-      doc_hash = @indexer.collection_solr_document collection
+      @indexer.collection_solr_document collection
     end
     it 'calls GDor::Indexer::SolrDocBuilder.validate_mods' do
-      doc_hash = GDor::Indexer::SolrDocHash.new
       allow_any_instance_of(GDor::Indexer::SolrDocBuilder).to receive(:doc_hash).and_return(doc_hash) # speed up the test
       expect(doc_hash).to receive(:validate_mods).and_return([])
-      doc_hash = @indexer.collection_solr_document collection
+      @indexer.collection_solr_document collection
     end
     it 'populates druid and access_facet fields' do
       doc_hash = @indexer.collection_solr_document collection
@@ -222,26 +220,25 @@ describe GDor::Indexer do
       doc_hash = @indexer.collection_solr_document collection
       expect(doc_hash).to include collection_type: 'Digital Collection'
     end
+
     context 'add format_main_ssim Archive/Manuscript' do
       it 'no other values' do
         allow_any_instance_of(GDor::Indexer::SolrDocBuilder).to receive(:doc_hash).and_return(GDor::Indexer::SolrDocHash.new)
-
         doc_hash = @indexer.collection_solr_document collection
         expect(doc_hash).to include format_main_ssim: 'Archive/Manuscript'
       end
       it 'other values present' do
         allow_any_instance_of(GDor::Indexer::SolrDocBuilder).to receive(:doc_hash).and_return(GDor::Indexer::SolrDocHash.new({ format_main_ssim: %w(Image Video) }))
-
         doc_hash = @indexer.collection_solr_document collection
         expect(doc_hash).to include format_main_ssim: ['Image', 'Video', 'Archive/Manuscript']
       end
       it 'already has values Archive/Manuscript' do
         allow_any_instance_of(GDor::Indexer::SolrDocBuilder).to receive(:doc_hash).and_return(GDor::Indexer::SolrDocHash.new({ format_main_ssim: 'Archive/Manuscript' }))
-
         doc_hash = @indexer.collection_solr_document collection
         expect(doc_hash).to include format_main_ssim: ['Archive/Manuscript']
       end
     end
+
     it 'populates building_facet field with Stanford Digital Repository' do
       doc_hash = @indexer.collection_solr_document collection
       expect(doc_hash).to include building_facet: 'Stanford Digital Repository'
@@ -252,9 +249,9 @@ describe GDor::Indexer do
     before(:each) do
       @coll_druids_array = [collection]
     end
+    let(:doc_hash) { GDor::Indexer::SolrDocHash.new({}) }
 
     it 'adds no collection field values to doc_hash if there are none' do
-      doc_hash = GDor::Indexer::SolrDocHash.new({})
       @indexer.add_coll_info(doc_hash, nil)
       expect(doc_hash[:collection]).to be_nil
       expect(doc_hash[:collection_with_title]).to be_nil
@@ -263,7 +260,6 @@ describe GDor::Indexer do
 
     context 'collection field' do
       it 'is added field to doc hash' do
-        doc_hash = GDor::Indexer::SolrDocHash.new({})
         @indexer.add_coll_info(doc_hash, @coll_druids_array)
         expect(doc_hash[:collection]).to match_array [@coll_druid_from_test_config]
       end
@@ -286,7 +282,6 @@ describe GDor::Indexer do
       it 'adds two values to doc_hash when object belongs to two collections' do
         coll_druid1 = 'oo111oo2222'
         coll_druid2 = 'oo333oo4444'
-        doc_hash = GDor::Indexer::SolrDocHash.new({})
         @indexer.add_coll_info(doc_hash, [double(druid: coll_druid1, bare_druid: coll_druid1, public_xml: @ng_pub_xml, identity_md_obj_label: 'foo'), double(druid: coll_druid2, bare_druid: coll_druid2, public_xml: @ng_pub_xml, identity_md_obj_label: 'bar')])
         expect(doc_hash[:collection_with_title]).to match_array ["#{coll_druid1}-|-foo", "#{coll_druid2}-|-bar"]
       end
@@ -347,25 +342,25 @@ describe GDor::Indexer do
     end
 
     it 'email body includes coll id' do
-      expect(subject).to match /testcoll indexed coll record is: ww121ss5000/
+      expect(subject).to match(/testcoll indexed coll record is: ww121ss5000/)
     end
 
     it 'email body includes coll title' do
-      expect(subject).to match /coll title: testcoll title/
+      expect(subject).to match(/coll title: testcoll title/)
     end
 
     it 'email body includes failed to index druids' do
       @indexer.instance_variable_set(:@druids_failed_to_ix, %w(a b))
-      expect(subject).to match /records that may have failed to index: \na\nb\n\n/
+      expect(subject).to match(/records that may have failed to index: \na\nb\n\n/)
     end
 
     it 'email body include validation messages' do
       @indexer.instance_variable_set(:@validation_messages, ['this is a validation message'])
-      expect(subject).to match /this is a validation message/
+      expect(subject).to match(/this is a validation message/)
     end
 
     it 'email includes reference to full log' do
-      expect(subject).to match /full log is at gdor_indexer\/shared\/spec\/test_logs\/testcoll\.log/
+      expect(subject).to match(/full log is at gdor_indexer\/shared\/spec\/test_logs\/testcoll\.log/)
     end
   end
 
@@ -378,7 +373,7 @@ describe GDor::Indexer do
 
     it 'has an appropriate subject' do
       expect(@indexer).to receive(:send_email) do |_to, opts|
-        expect(opts[:subject]).to match /is finished/
+        expect(opts[:subject]).to match(/is finished/)
       end
 
       @indexer.email_results

data/spec/unit/mods_fields_spec.rb

@@ -1,5 +1,3 @@
-require 'spec_helper'
-
 describe GDor::Indexer::ModsFields do
   let(:fake_druid) { 'oo000oo0000' }
   let(:ns_decl) { "xmlns='#{Mods::MODS_NS}'" }

data/spec/unit/mods_pub_fields_spec.rb

@@ -1,5 +1,3 @@
-require 'spec_helper'
-
 describe GDor::Indexer::ModsFields do
   let(:fake_druid) { 'oo000oo0000' }
   let(:ns_decl) { "xmlns='#{Mods::MODS_NS}'" }

data/spec/unit/mods_subject_fields_spec.rb

@@ -1,5 +1,3 @@
-require 'spec_helper'
-
 describe GDor::Indexer::ModsFields do
   let(:fake_druid) { 'oo000oo0000' }
   let(:ns_decl) { "xmlns='#{Mods::MODS_NS}'" }

data/spec/unit/public_xml_fields_spec.rb

@@ -1,5 +1,3 @@
-require 'spec_helper'
-
 describe GDor::Indexer::PublicXmlFields do
   before(:all) do
     @fake_druid = 'oo000oo0000'

data/spec/unit/solr_doc_builder_spec.rb

@@ -1,13 +1,10 @@
-require 'spec_helper'
-
 describe GDor::Indexer::SolrDocBuilder do
   before(:all) do
-    @fake_druid = 'oo000oo0000'
     @ns_decl = "xmlns='#{Mods::MODS_NS}'"
     @mods_xml = "<mods #{@ns_decl}><note>SolrDocBuilder test</note></mods>"
-    @ng_mods_xml = Nokogiri::XML(@mods_xml)
   end
 
+  let(:fake_druid) { 'oo000oo0000' }
   let :logger do
     lgr = Logger.new(StringIO.new)
     lgr.level = Logger::WARN
@@ -15,7 +12,7 @@ describe GDor::Indexer::SolrDocBuilder do
   end
 
   def sdb_for_data(mods, pub_xml)
-    resource = Harvestdor::Indexer::Resource.new(double, @fake_druid)
+    resource = Harvestdor::Indexer::Resource.new(double, fake_druid)
     allow(resource).to receive(:mods).and_return(Nokogiri::XML(mods))
     allow(resource).to receive(:public_xml).and_return(Nokogiri::XML(pub_xml))
     i = Harvestdor::Indexer.new
@@ -24,49 +21,40 @@ describe GDor::Indexer::SolrDocBuilder do
     GDor::Indexer::SolrDocBuilder.new(resource, logger)
   end
 
-  # NOTE:
-  # "Doubles, stubs, and message expectations are all cleaned out after each example."
-  # per https://www.relishapp.com/rspec/rspec-mocks/docs/scope
-
   context 'doc_hash' do
-    before(:all) do
-      cmd_xml = "<contentMetadata type='image' objectId='#{@fake_druid}'></contentMetadata>"
-      @pub_xml = "<publicObject id='druid#{@fake_druid}'>#{cmd_xml}</publicObject>"
-    end
-    let :doc_hash do
-      sdb_for_data(@mods_xml, @pub_xml).doc_hash
-    end
-    before(:each) do
-      @doc_hash = doc_hash
+    let(:doc_hash) do
+      cmd_xml = "<contentMetadata type='image' objectId='#{fake_druid}'></contentMetadata>"
+      pub_xml = "<publicObject id='druid#{fake_druid}'>#{cmd_xml}</publicObject>"
+      sdb_for_data(@mods_xml, pub_xml).doc_hash
     end
+
     it 'id field should be set to druid' do
-      expect(@doc_hash[:id]).to eq(@fake_druid)
+      expect(doc_hash[:id]).to eq(fake_druid)
     end
     it 'does not have the gdor fields set in indexer.rb' do
-      expect(@doc_hash).to_not have_key(:druid)
-      expect(@doc_hash).to_not have_key(:access_facet)
-      expect(@doc_hash).to_not have_key(:url_fulltext)
-      expect(@doc_hash).to_not have_key(:display_type)
-      expect(@doc_hash).to_not have_key(:file_id)
+      expect(doc_hash).not_to have_key(:druid)
+      expect(doc_hash).not_to have_key(:access_facet)
+      expect(doc_hash).not_to have_key(:url_fulltext)
+      expect(doc_hash).not_to have_key(:display_type)
+      expect(doc_hash).not_to have_key(:file_id)
     end
     it 'has the full MODS in the modsxml field' do
-      expect(@doc_hash[:modsxml]).to be_equivalent_to @mods_xml
+      expect(doc_hash[:modsxml]).to be_equivalent_to @mods_xml
     end
   end # doc hash
 
   context '#catkey' do
-    before(:all) do
-      @identity_md_start = "<publicObject><identityMetadata objectId='#{@fake_druid}'>"
-      @identity_md_end = '</identityMetadata></publicObject>'
-      @empty_id_md = "#{@identity_md_start}#{@identity_md_end}"
-      @barcode_id_md = "#{@identity_md_start}<otherId name=\"barcode\">666</otherId>#{@identity_md_end}"
-    end
+    let(:identity_md_start) { "<publicObject><identityMetadata objectId='#{fake_druid}'>" }
+    let(:identity_md_end) { '</identityMetadata></publicObject>' }
+    let(:empty_id_md) { "#{identity_md_start}#{identity_md_end}" }
+    let(:barcode_id_md) { "#{identity_md_start}<otherId name=\"barcode\">666</otherId>#{identity_md_end}" }
+
     it 'is nil if there is no indication of catkey in identityMetadata' do
-      sdb = sdb_for_data(@mods_xml, @empty_id_md)
+      sdb = sdb_for_data(@mods_xml, empty_id_md)
       expect(sdb.catkey).to be_nil
     end
     it 'takes a catkey in identityMetadata/otherId with name attribute of catkey' do
-      pub_xml = "#{@identity_md_start}<otherId name=\"catkey\">12345</otherId>#{@identity_md_end}"
+      pub_xml = "#{identity_md_start}<otherId name=\"catkey\">12345</otherId>#{identity_md_end}"
       sdb = sdb_for_data(@mods_xml, pub_xml)
       expect(sdb.catkey).to eq('12345')
     end
@@ -74,18 +62,18 @@ describe GDor::Indexer::SolrDocBuilder do
       m = "<mods #{@ns_decl}><recordInfo>
         <recordIdentifier source=\"SIRSI\">a6780453</recordIdentifier>
       </recordInfo></mods>"
-      sdb = sdb_for_data(@mods_xml, @empty_id_md)
+      sdb = sdb_for_data(m, empty_id_md)
       expect(sdb.catkey).to be_nil
     end
     it 'logs an error when there is identityMetadata/otherId with name attribute of barcode but there is no catkey in mods' do
-      sdb = sdb_for_data(@mods_xml, @barcode_id_md)
-      expect(logger).to receive(:error).with(/#{@fake_druid} has barcode .* in identityMetadata but no SIRSI catkey in mods/)
+      sdb = sdb_for_data(@mods_xml, barcode_id_md)
+      expect(logger).to receive(:error).with(/#{fake_druid} has barcode .* in identityMetadata but no SIRSI catkey in mods/)
       sdb.catkey
     end
 
     context 'catkey from mods' do
       it 'looks for catkey in mods if identityMetadata/otherId with name attribute of barcode is found' do
-        sdb = sdb_for_data(@mods_xml, @barcode_id_md)
+        sdb = sdb_for_data(@mods_xml, barcode_id_md)
         smr = sdb.smods_rec
         expect(smr).to receive(:record_info).and_call_original # this is as close as I can figure to @smods_rec.record_info.recordIdentifier
         sdb.catkey
@@ -94,29 +82,28 @@ describe GDor::Indexer::SolrDocBuilder do
         m = "<mods #{@ns_decl}><recordInfo>
           <descriptionStandard>dacs</descriptionStandard>
         </recordInfo></mods>"
-        sdb = sdb_for_data(m, @barcode_id_md)
-
+        sdb = sdb_for_data(m, barcode_id_md)
         expect(sdb.catkey).to be_nil
       end
       it 'populated when source attribute is SIRSI' do
         m = "<mods #{@ns_decl}><recordInfo>
           <recordIdentifier source=\"SIRSI\">a6780453</recordIdentifier>
         </recordInfo></mods>"
-        sdb = sdb_for_data(m, @barcode_id_md)
+        sdb = sdb_for_data(m, barcode_id_md)
         expect(sdb.catkey).not_to be_nil
       end
       it 'not populated when source attribute is not SIRSI' do
         m = "<mods #{@ns_decl}><recordInfo>
           <recordIdentifier source=\"FOO\">a6780453</recordIdentifier>
         </recordInfo></mods>"
-        sdb = sdb_for_data(m, @barcode_id_md)
+        sdb = sdb_for_data(m, barcode_id_md)
         expect(sdb.catkey).to be_nil
       end
       it 'removes the a at the beginning of the catkey' do
         m = "<mods #{@ns_decl}><recordInfo>
           <recordIdentifier source=\"SIRSI\">a6780453</recordIdentifier>
         </recordInfo></mods>"
-        sdb = sdb_for_data(m, @barcode_id_md)
+        sdb = sdb_for_data(m, barcode_id_md)
         expect(sdb.catkey).to eq('6780453')
       end
     end

data/spec/unit/solr_doc_hash_spec.rb

@@ -1,5 +1,3 @@
-require 'spec_helper'
-
 describe GDor::Indexer::SolrDocHash do
   context '#field_present?' do
     context 'actual field value is boolean true' do
@@ -132,6 +130,7 @@ describe GDor::Indexer::SolrDocHash do
         expect(subject.combine(foo: {})).to eq({})
       end
     end # orig has no key
+
     context 'orig value is nil' do
       subject do
         described_class.new(foo: nil)
@@ -155,6 +154,7 @@ describe GDor::Indexer::SolrDocHash do
         expect(subject.combine(foo: {})).to eq({})
       end
     end # orig value is nil
+
     context 'orig value is empty String' do
       subject do
         described_class.new(foo: '')
@@ -178,6 +178,7 @@ describe GDor::Indexer::SolrDocHash do
         expect(subject.combine(foo: {})).to eq({})
       end
     end # orig value is empty String
+
     context 'orig value is non-empty String' do
       subject do
         described_class.new(foo: 'a')
@@ -201,6 +202,7 @@ describe GDor::Indexer::SolrDocHash do
         expect(subject.combine(foo: :bar)).to eq(foo: ['a', :bar])
       end
     end # orig value is String
+
     context 'orig value is empty Array' do
       subject do
         described_class.new(foo: [])
@@ -224,6 +226,7 @@ describe GDor::Indexer::SolrDocHash do
         expect(subject.combine(foo: {})).to eq({})
       end
     end # orig value is empty Array
+
     context 'orig value is non-empty Array' do
       subject do
         described_class.new(foo: %w(a b))
@@ -309,7 +312,6 @@ describe GDor::Indexer::SolrDocHash do
     end
     it 'has a value if collection_type is missing' do
       hash = described_class.new(format_main_ssim: 'Archive/Manuscript')
-
       expect(hash.validate_collection(mock_config).first).to match(/collection_type/)
     end
     it "has a value if collection_type is not 'Digital Collection'" do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: gdor-indexer
 version: !ruby/object:Gem::Version
-  version: 0.5.0
+  version: 0.6.0
 platform: ruby
 authors:
 - Naomi Dushay
@@ -10,7 +10,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-01-19 00:00:00.000000000 Z
+date: 2016-08-06 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: harvestdor-indexer
@@ -30,16 +30,16 @@ dependencies:
   name: stanford-mods
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.4.0
+        version: 2.2.1
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.4.0
+        version: 2.2.1
 - !ruby/object:Gem::Dependency
   name: nokogiri
   requirement: !ruby/object:Gem::Requirement
@@ -316,8 +316,10 @@ extra_rdoc_files: []
 files:
 - ".gitignore"
 - ".hound.yml"
+- ".rspec"
 - ".rubocop.yml"
 - ".rubocop_todo.yml"
+- ".travis.yml"
 - ".yardopts"
 - Gemfile
 - LICENSE.txt
@@ -363,7 +365,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.4.6
+rubygems_version: 2.5.1
 signing_key: 
 specification_version: 4
 summary: PURL doc =>  Solr hash logic
@@ -378,4 +380,3 @@ test_files:
 - spec/unit/solr_doc_builder_spec.rb
 - spec/unit/solr_doc_hash_spec.rb
 - spec/vcr_cassettes/no_coll_druid_in_druid_array_call.yml
-has_rdoc: