stanford-mods 1.1.5 → 1.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 209b8e1644beeed8c28948b1acdba20b7d63a414
-  data.tar.gz: d3f4b45f2d6f990c3084a67e4eca651a83ecd7e0
+  metadata.gz: 025c7efb3e29b5c56eaf05af7f58eb364e13c8f7
+  data.tar.gz: 4d66ca23ed45283c8374fa50166f694f63c31bb8
 SHA512:
-  metadata.gz: 00b048ecb711ea342bb7071bf5422f06c48f40bf81b7b67590211d9eb0a59774d28991396dfc1c0aa54c7cbe354407183fbd87f28c318517c0a7d08c12bceca4
-  data.tar.gz: a9b70f9cf519fae69b9e557281b156913735d465e83044ab6286caafeac0ca5ce2dde623cbdac58d91522e547041c175c602f5cb3cd39b5b68aed38fcc3ba772
+  metadata.gz: f4c1b8aaf08deb76e6de29153cd492def619aa884b8841e7c68254ea076da2948d91ab81123f81ee565f1cb41f7ba85ee2a4dff4bec1a4b2f10f72cd77775a44
+  data.tar.gz: 8b90601ce966f4ac16cac8dd78c255b692eff2bfdaef8dcd889bdc257acdc594f47925baddd705b2d969cc646c3df4295198edefba1f1bd0c6bfe3011232be6f

data/.hound.yml

@@ -0,0 +1,2 @@
+ruby:
+  config_file: .rubocop.yml

data/.rubocop.yml

@@ -0,0 +1,3 @@
+inherit_from: .rubocop_todo.yml
+
+require: rubocop-rspec

data/.rubocop_todo.yml

@@ -0,0 +1,12 @@
+# This configuration was generated by
+# `rubocop --auto-gen-config`
+# on 2015-11-17 14:05:33 -0800 using RuboCop version 0.35.1.
+# 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: 1
+Lint/UselessAssignment:
+  Exclude:
+    - 'config/mappings_hash.rb'

data/.travis.yml

@@ -1,18 +1,17 @@
 language: ruby
-script: rake rspec
+script: rake
 rvm:
+  - ruby-head
+  - 2.2.3  # spotlight
   - 2.2.0
   - 2.1.5
   - 2.0.0
-  - 1.9.3
-  - ruby-head
-  - jruby-1.7.9-d19 # fails after 1.7.10 for some reason
-#  - jruby-19mode # JRuby in 1.9 mode
-#  - jruby-head  # failing 2014-07
-notifications:
-  email:
-    - ndushay@stanford.edu
-    - laneymcg@stanford.edu
-    - bess@stanford.edu
-#before_install:
-#  gem update --system 1.8.24
+  - 1.9.3  # argo, FRDA
+  # we used to use jruby for merged DOR + MARC records, but no more ...
+  - jruby-head
+matrix:
+  allow_failures:
+    - rvm: jruby-head
+
+notifications: false
+sudo: false

data/Gemfile

@@ -3,9 +3,12 @@ source 'https://rubygems.org'
 # See stanford-mods.gemspec for this gem's dependencies
 gemspec
 
-#gem 'mods', :path => '../mods'
+group :test, :development do
+  gem 'rubocop', require: false
+  gem 'rubocop-rspec', require: false
+end
 
 group :test do
   gem 'coveralls', require: false
+#  gem 'pry-byebug', require: false
 end
-

data/Rakefile

@@ -2,10 +2,6 @@
 require "bundler/gem_tasks"
 require 'bundler'
 
-require 'rspec/core/rake_task'
-require 'yard'
-require 'yard/rake/yardoc_task'
-
 begin
   Bundler.setup(:default, :development)
 rescue Bundler::BundlerError => e
@@ -16,23 +12,31 @@ end
 
 task :default => :ci
 
-desc "run continuous integration suite (tests, coverage, docs)"
-task :ci => [:rspec, :doc]
+desc "run continuous integration suite (tests, coverage, rubocop lint)"
+task :ci => [:rspec, :rubocop]
 
 task :spec => :rspec
 
+require 'rspec/core/rake_task'
 RSpec::Core::RakeTask.new(:rspec) do |spec|
   spec.rspec_opts = ["-c", "--tty", "-f progress", "-r ./spec/spec_helper.rb"]
 end
 
+require 'rubocop/rake_task'
+RuboCop::RakeTask.new(:rubocop) do |task|
+  task.options = ['-l'] # run lint cops only
+end
+
 # Use yard to build docs
+require 'yard'
+require 'yard/rake/yardoc_task'
 begin
   project_root = File.expand_path(File.dirname(__FILE__))
   doc_dest_dir = File.join(project_root, 'doc')
 
   YARD::Rake::YardocTask.new(:doc) do |yt|
     yt.files = Dir.glob(File.join(project_root, 'lib', '**', '*.rb')) +
-                 [ File.join(project_root, 'README.md') ]
+               [File.join(project_root, 'README.md')]
     yt.options = ['--output-dir', doc_dest_dir, '--readme', 'README.md', '--title', 'Stanford-Mods Documentation']
   end
 rescue LoadError

data/config/mappings_hash.rb

@@ -17,7 +17,7 @@ name_to_xpath = {
 
 # titleInfo
   'title'         => '//mods/titleInfo[not(@type="alternative")]', # .first, extract_title_from_title_info(node)
-  'subtitle'      => '//mods/titleInfo/subTitle' #.first.text
+  'subtitle'      => '//mods/titleInfo/subTitle', #.first.text
   'full_title'    => '//mods/titleInfo', # .first, extract_full_title_from_title_info(node)
   'title_variant' => '//mods/titleInfo[@type="alternative"]', # .first, extract_title_from_title_info(node)
 

data/lib/stanford-mods/name.rb

@@ -0,0 +1,84 @@
+# encoding: UTF-8
+require 'logger'
+require 'mods'
+
+# NON-SearchWorks specific wranglings of MODS <name> metadata as a mixin to the Stanford::Mods::Record object
+module Stanford
+  module Mods
+
+    class Record < ::Mods::Record
+
+      # the first encountered <mods><name> element with marcrelator flavor role of 'Creator' or 'Author'.
+      # if no marcrelator 'Creator' or 'Author', the first name without a role.
+      # if no name without a role, then nil
+      # @return [String] a name in the display_value_w_date form
+      # see Mods::Record.name  in nom_terminology for details on the display_value algorithm
+      def main_author_w_date
+        result = nil
+        first_wo_role = nil
+        @mods_ng_xml.plain_name.each { |n|
+          if n.role.size == 0
+            first_wo_role ||= n
+          end
+          n.role.each { |r|
+            if r.authority.include?('marcrelator') &&
+                  (r.value.include?('Creator') || r.value.include?('Author'))
+              result ||= n.display_value_w_date
+            end
+          }
+        }
+        if !result && first_wo_role
+          result = first_wo_role.display_value_w_date
+        end
+        result
+      end # main_author
+
+      # all names, in display form, except the main_author
+      #  names will be the display_value_w_date form
+      #  see Mods::Record.name  in nom_terminology for details on the display_value algorithm
+      def additional_authors_w_dates
+        results = []
+        @mods_ng_xml.plain_name.each { |n|
+          results << n.display_value_w_date
+        }
+        results.delete(main_author_w_date)
+        results
+      end
+
+      COLLECTOR_ROLE_URI = 'http://id.loc.gov/vocabulary/relators/col'
+
+      # @return Array of Strings, each containing the computed display value of a personal name
+      #   except for the collector role (see mods gem nom_terminology for display value algorithm)
+      # FIXME:  this is broken if there are multiple role codes and some of them are not marcrelator
+      def non_collector_person_authors
+        result = []
+        @mods_ng_xml.personal_name.map do |n|
+          r = n.role
+          unless (r.authority.include?('marcrelator') && r.value.include?('Collector')) ||
+                  r.roleTerm.valueURI.first == COLLECTOR_ROLE_URI
+            result << n.display_value_w_date
+          end
+        end
+        result unless result.empty?
+      end
+
+      # @return Array of Strings, each containing the computed display value of
+      #  a personal name with the role of Collector (see mods gem nom_terminology for display value algorithm)
+      def collectors_w_dates
+        result = []
+        @mods_ng_xml.personal_name.each do |n|
+          unless n.role.size == 0
+            n.role.each { |r|
+              if (r.authority.include?('marcrelator') && r.value.include?('Collector')) ||
+                  r.roleTerm.valueURI.first == COLLECTOR_ROLE_URI
+                result << n.display_value_w_date
+              end
+            }
+          end
+        end
+        result unless result.empty?
+      end
+
+    end # class Record
+  end # Module Mods
+end # Module Stanford

data/lib/stanford-mods/physical_location.rb

@@ -0,0 +1,83 @@
+require 'logger'
+require 'mods'
+
+# Parsing MODS //location/physicalLocation for series, box, and folder for Special Collections
+# This is not used by Searchworks, otherwise it would have been in the searchworks.rb file
+module Stanford
+  module Mods
+    class Record < ::Mods::Record
+      # return box number (note: single valued and might be something like 35A)
+      #   data in location/physicalLocation or in relatedItem/location/physicalLocation
+      #   so use _location to get the data from either one of them
+      # TODO:  should it be hierarchical series/box/folder?
+      def box
+        #   _location.physicalLocation should find top level and relatedItem
+        box_num = @mods_ng_xml._location.physicalLocation.map do |node|
+          val = node.text
+          # note that this will also find Flatbox or Flat-box
+          match_data = val.match(/Box ?:? ?([^,|(Folder)]+)/i)
+          match_data[1].strip if match_data.present?
+        end.compact
+
+        # There should only be one box
+        box_num.first
+      end
+
+      # returns folder number (note: single valued)
+      #   data in location/physicalLocation or in relatedItem/location/physicalLocation
+      #   so use _location to get the data from either one of them
+      # TODO:  should it be hierarchical series/box/folder?
+      def folder
+        #   _location.physicalLocation should find top level and relatedItem
+        folder_num = @mods_ng_xml._location.physicalLocation.map do |node|
+          val = node.text
+
+          match_data = if val =~ /\|/
+                         # we assume the data is pipe-delimited, and may contain commas within values
+                         val.match(/Folder ?:? ?([^|]+)/)
+                       else
+                         # the data should be comma-delimited, and may not contain commas within values
+                         val.match(/Folder ?:? ?([^,]+)/)
+                       end
+
+          match_data[1].strip if match_data.present?
+        end.compact
+
+        # There should be one folder
+        folder_num.first
+      end
+
+      # return entire contents of physicalLocation (note: single valued)
+      #   but only if it has series, accession, box or folder data
+      #   data in location/physicalLocation or in relatedItem/location/physicalLocation
+      #   so use _location to get the data from either one of them
+      # TODO:  should it be hierarchical series/box/folder?
+      def location
+        #   _location.physicalLocation should find top level and relatedItem
+        loc = @mods_ng_xml._location.physicalLocation.map do |node|
+          node.text if node.text.match(/.*(Series)|(Accession)|(Folder)|(Box).*/i)
+        end.compact
+
+        # There should only be one location
+        loc.first
+      end
+
+      # return series/accession 'number' (note: single valued)
+      #   data in location/physicalLocation or in relatedItem/location/physicalLocation
+      #   so use _location to get the data from either one of them
+      # TODO: should it be hierarchical series/box/folder?
+      def series
+        #   _location.physicalLocation should find top level and relatedItem
+        series_num = @mods_ng_xml._location.physicalLocation.map do |node|
+          val = node.text
+          # feigenbaum uses 'Accession'
+          match_data = val.match(/(?:(?:Series)|(?:Accession)):? ([^,|]+)/i)
+          match_data[1].strip if match_data.present?
+        end.compact
+
+        # There should be only one series
+        series_num.first
+      end
+    end # class Record
+  end # Module Mods
+end # Module Stanford

data/lib/stanford-mods/searchworks.rb

@@ -26,7 +26,7 @@ module Stanford
                     result << SEARCHWORKS_LANGUAGES[v.strip]
                   end
                 end
-              rescue => e
+              rescue
                 # TODO:  this should be written to a logger
                 p "Couldn't find english name for #{ct.text}"
               end
@@ -422,7 +422,6 @@ module Stanford
         end
         dates = pub_dates
         if dates
-          year = []
           pruned_dates = []
           dates.each do |f_date|
             #remove ? and []

data/lib/stanford-mods/version.rb

@@ -1,6 +1,6 @@
 module Stanford
   module Mods
     # this is the Ruby Gem version
-    VERSION = "1.1.5"
+    VERSION = "1.2.0"
   end
 end

data/lib/stanford-mods.rb

@@ -1,6 +1,8 @@
 require 'stanford-mods/version'
 require 'mods'
+require 'stanford-mods/name'
 require 'stanford-mods/searchworks'
+require 'stanford-mods/physical_location'
 
 # Stanford specific wranglings of MODS metadata as an extension of the Mods::Record object
 module Stanford
@@ -8,43 +10,6 @@ module Stanford
 
     class Record < ::Mods::Record
 
-      # the first encountered <mods><name> element with marcrelator flavor role of 'Creator' or 'Author'.
-      # if no marcrelator 'Creator' or 'Author', the first name without a role.
-      # if no name without a role, then nil
-      # @return [String] a name in the display_value_w_date form
-      # see Mods::Record.name  in nom_terminology for details on the display_value algorithm
-      def main_author_w_date
-        result = nil
-        first_wo_role = nil
-        @mods_ng_xml.plain_name.each { |n|
-          if n.role.size == 0
-            first_wo_role ||= n
-          end
-          n.role.each { |r|
-            if r.authority.include?('marcrelator') &&
-                  (r.value.include?('Creator') || r.value.include?('Author'))
-              result ||= n.display_value_w_date
-            end
-          }
-        }
-        if !result && first_wo_role
-          result = first_wo_role.display_value_w_date
-        end
-        result
-      end # main_author
-
-      # all names, in display form, except the main_author
-      #  names will be the display_value_w_date form
-      #  see Mods::Record.name  in nom_terminology for details on the display_value algorithm
-      def additional_authors_w_dates
-        results = []
-        @mods_ng_xml.plain_name.each { |n|
-          results << n.display_value_w_date
-        }
-        results.delete(main_author_w_date)
-        results
-      end
-
     end # Record class
   end # Mods module
 end # Stanford module