stanford-mods 2.6.4 → 3.1.0

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: stanford-mods
 version: !ruby/object:Gem::Version
-  version: 2.6.4
+  version: 3.1.0
 platform: ruby
 authors:
 - Naomi Dushay
@@ -9,7 +9,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2022-01-12 00:00:00.000000000 Z
+date: 2022-05-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mods
@@ -17,14 +17,20 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '2.2'
+        version: '3.0'
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: 3.0.1
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '2.2'
+        version: '3.0'
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: 3.0.1
 - !ruby/object:Gem::Dependency
   name: activesupport
   requirement: !ruby/object:Gem::Requirement
@@ -160,21 +166,20 @@ files:
 - LICENSE
 - README.md
 - Rakefile
-- lib/marc_countries.rb
 - lib/stanford-mods.rb
+- lib/stanford-mods/concerns/geo_spatial.rb
+- lib/stanford-mods/concerns/name.rb
+- lib/stanford-mods/concerns/origin_info.rb
+- lib/stanford-mods/concerns/physical_location.rb
+- lib/stanford-mods/concerns/searchworks.rb
+- lib/stanford-mods/concerns/searchworks_subjects.rb
+- lib/stanford-mods/concerns/title.rb
 - lib/stanford-mods/coordinate.rb
 - lib/stanford-mods/date_parsing.rb
-- lib/stanford-mods/geo_spatial.rb
-- lib/stanford-mods/geo_utils.rb
 - lib/stanford-mods/imprint.rb
-- lib/stanford-mods/name.rb
-- lib/stanford-mods/origin_info.rb
-- lib/stanford-mods/physical_location.rb
-- lib/stanford-mods/searchworks.rb
-- lib/stanford-mods/searchworks_languages.rb
-- lib/stanford-mods/searchworks_subjects.rb
+- lib/stanford-mods/record.rb
 - lib/stanford-mods/version.rb
-- spec/date_parsing_spec.rb
+- lib/stanford-mods/vocabularies/searchworks_languages.rb
 - spec/fixtures/searchworks_imprint_data.rb
 - spec/fixtures/searchworks_pub_date_data.rb
 - spec/fixtures/spotlight_pub_date_data.rb
@@ -217,7 +222,6 @@ signing_key:
 specification_version: 4
 summary: Stanford specific wrangling of MODS metadata
 test_files:
-- spec/date_parsing_spec.rb
 - spec/fixtures/searchworks_imprint_data.rb
 - spec/fixtures/searchworks_pub_date_data.rb
 - spec/fixtures/spotlight_pub_date_data.rb

data/lib/marc_countries.rb

@@ -1,387 +0,0 @@
-# encoding: utf-8
-
-# Map 3 letter MARC Country/Location Code to User Friendly Name
-# includes discontinued codes, because they're probably in our data
-# from https://www.loc.gov/marc/countries/countries_code.html 01/20/2016
-MARC_COUNTRIES =
-  {
-    'aa' => 'Albania',
-    'abc' => 'Alberta',
-    'ac' => 'Ashmore and Cartier Islands',
-    'aca' => 'Australian Capital Territory',
-    'ae' => 'Algeria',
-    'af' => 'Afghanistan',
-    'ag' => 'Argentina',
-    'ai' => 'Armenia (Republic)',
-    'air' => 'Armenian S.S.R.',
-    'aj' => 'Azerbaijan',
-    'ajr' => 'Azerbaijan S.S.R.',
-    'aku' => 'Alaska',
-    'alu' => 'Alabama',
-    'am' => 'Anguilla',
-    'an' => 'Andorra',
-    'ao' => 'Angola',
-    'aq' => 'Antigua and Barbuda',
-    'aru' => 'Arkansas',
-    'as' => 'American Samoa',
-    'at' => 'Australia',
-    'au' => 'Austria',
-    'aw' => 'Aruba',
-    'ay' => 'Antarctica',
-    'azu' => 'Arizona',
-    'ba' => 'Bahrain',
-    'bb' => 'Barbados',
-    'bcc' => 'British Columbia',
-    'bd' => 'Burundi',
-    'be' => 'Belgium',
-    'bf' => 'Bahamas',
-    'bg' => 'Bangladesh',
-    'bh' => 'Belize',
-    'bi' => 'British Indian Ocean Territory',
-    'bl' => 'Brazil',
-    'bm' => 'Bermuda Islands',
-    'bn' => 'Bosnia and Herzegovina',
-    'bo' => 'Bolivia',
-    'bp' => 'Solomon Islands',
-    'br' => 'Burma',
-    'bs' => 'Botswana',
-    'bt' => 'Bhutan',
-    'bu' => 'Bulgaria',
-    'bv' => 'Bouvet Island',
-    'bw' => 'Belarus',
-    'bwr' => 'Byelorussian S.S.R.',
-    'bx' => 'Brunei',
-    'ca' => 'Caribbean Netherlands',
-    'cau' => 'California',
-    'cb' => 'Cambodia',
-    'cc' => 'China',
-    'cd' => 'Chad',
-    'ce' => 'Sri Lanka',
-    'cf' => 'Congo (Brazzaville)',
-    'cg' => 'Congo (Democratic Republic)',
-    'ch' => 'China (Republic : 1949 )',
-    'ci' => 'Croatia',
-    'cj' => 'Cayman Islands',
-    'ck' => 'Colombia',
-    'cl' => 'Chile',
-    'cm' => 'Cameroon',
-    'cn' => 'Canada',
-    'co' => 'Curaçao',
-    'cou' => 'Colorado',
-    'cp' => 'Canton and Enderbury Islands',
-    'cq' => 'Comoros',
-    'cr' => 'Costa Rica',
-    'cs' => 'Czechoslovakia',
-    'ctu' => 'Connecticut',
-    'cu' => 'Cuba',
-    'cv' => 'Cabo Verde',
-    'cw' => 'Cook Islands',
-    'cx' => 'Central African Republic',
-    'cy' => 'Cyprus',
-    'cz' => 'Canal Zone',
-    'dcu' => 'District of Columbia',
-    'deu' => 'Delaware',
-    'dk' => 'Denmark',
-    'dm' => 'Benin',
-    'dq' => 'Dominica',
-    'dr' => 'Dominican Republic',
-    'ea' => 'Eritrea',
-    'ec' => 'Ecuador',
-    'eg' => 'Equatorial Guinea',
-    'em' => 'TimorLeste',
-    'enk' => 'England',
-    'er' => 'Estonia',
-    'err' => 'Estonia',
-    'es' => 'El Salvador',
-    'et' => 'Ethiopia',
-    'fa' => 'Faroe Islands',
-    'fg' => 'French Guiana',
-    'fi' => 'Finland',
-    'fj' => 'Fiji',
-    'fk' => 'Falkland Islands',
-    'flu' => 'Florida',
-    'fm' => 'Micronesia (Federated States)',
-    'fp' => 'French Polynesia',
-    'fr' => 'France',
-    'fs' => 'Terres australes et antarctiques françaises',
-    'ft' => 'Djibouti',
-    'gau' => 'Georgia',
-    'gb' => 'Kiribati',
-    'gd' => 'Grenada',
-    'ge' => 'Germany (East)',
-    'gh' => 'Ghana',
-    'gi' => 'Gibraltar',
-    'gl' => 'Greenland',
-    'gm' => 'Gambia',
-    'gn' => 'Gilbert and Ellice Islands',
-    'go' => 'Gabon',
-    'gp' => 'Guadeloupe',
-    'gr' => 'Greece',
-    'gs' => 'Georgia (Republic)',
-    'gsr' => 'Georgian S.S.R.',
-    'gt' => 'Guatemala',
-    'gu' => 'Guam',
-    'gv' => 'Guinea',
-    'gw' => 'Germany',
-    'gy' => 'Guyana',
-    'gz' => 'Gaza Strip',
-    'hiu' => 'Hawaii',
-    'hk' => 'Hong Kong',
-    'hm' => 'Heard and McDonald Islands',
-    'ho' => 'Honduras',
-    'ht' => 'Haiti',
-    'hu' => 'Hungary',
-    'iau' => 'Iowa',
-    'ic' => 'Iceland',
-    'idu' => 'Idaho',
-    'ie' => 'Ireland',
-    'ii' => 'India',
-    'ilu' => 'Illinois',
-    'inu' => 'Indiana',
-    'io' => 'Indonesia',
-    'iq' => 'Iraq',
-    'ir' => 'Iran',
-    'is' => 'Israel',
-    'it' => 'Italy',
-    'iu' => 'IsraelSyria Demilitarized Zones',
-    'iv' => "Côte d'Ivoire",
-    'iw' => 'IsraelJordan Demilitarized Zones',
-    'iy' => 'IraqSaudi Arabia Neutral Zone',
-    'ja' => 'Japan',
-    'ji' => 'Johnston Atoll',
-    'jm' => 'Jamaica',
-    'jn' => 'Jan Mayen',
-    'jo' => 'Jordan',
-    'ke' => 'Kenya',
-    'kg' => 'Kyrgyzstan',
-    'kgr' => 'Kirghiz S.S.R.',
-    'kn' => 'Korea (North)',
-    'ko' => 'Korea (South)',
-    'ksu' => 'Kansas',
-    'ku' => 'Kuwait',
-    'kv' => 'Kosovo',
-    'kyu' => 'Kentucky',
-    'kz' => 'Kazakhstan',
-    'kzr' => 'Kazakh S.S.R.',
-    'lau' => 'Louisiana',
-    'lb' => 'Liberia',
-    'le' => 'Lebanon',
-    'lh' => 'Liechtenstein',
-    'li' => 'Lithuania',
-    'lir' => 'Lithuania',
-    'ln' => 'Central and Southern Line Islands',
-    'lo' => 'Lesotho',
-    'ls' => 'Laos',
-    'lu' => 'Luxembourg',
-    'lv' => 'Latvia',
-    'lvr' => 'Latvia',
-    'ly' => 'Libya',
-    'mau' => 'Massachusetts',
-    'mbc' => 'Manitoba',
-    'mc' => 'Monaco',
-    'mdu' => 'Maryland',
-    'meu' => 'Maine',
-    'mf' => 'Mauritius',
-    'mg' => 'Madagascar',
-    'mh' => 'Macao',
-    'miu' => 'Michigan',
-    'mj' => 'Montserrat',
-    'mk' => 'Oman',
-    'ml' => 'Mali',
-    'mm' => 'Malta',
-    'mnu' => 'Minnesota',
-    'mo' => 'Montenegro',
-    'mou' => 'Missouri',
-    'mp' => 'Mongolia',
-    'mq' => 'Martinique',
-    'mr' => 'Morocco',
-    'msu' => 'Mississippi',
-    'mtu' => 'Montana',
-    'mu' => 'Mauritania',
-    'mv' => 'Moldova',
-    'mvr' => 'Moldavian S.S.R.',
-    'mw' => 'Malawi',
-    'mx' => 'Mexico',
-    'my' => 'Malaysia',
-    'mz' => 'Mozambique',
-    'na' => 'Netherlands Antilles',
-    'nbu' => 'Nebraska',
-    'ncu' => 'North Carolina',
-    'ndu' => 'North Dakota',
-    'ne' => 'Netherlands',
-    'nfc' => 'Newfoundland and Labrador',
-    'ng' => 'Niger',
-    'nhu' => 'New Hampshire',
-    'nik' => 'Northern Ireland',
-    'nju' => 'New Jersey',
-    'nkc' => 'New Brunswick',
-    'nl' => 'New Caledonia',
-    'nm' => 'Northern Mariana Islands',
-    'nmu' => 'New Mexico',
-    'nn' => 'Vanuatu',
-    'no' => 'Norway',
-    'np' => 'Nepal',
-    'nq' => 'Nicaragua',
-    'nr' => 'Nigeria',
-    'nsc' => 'Nova Scotia',
-    'ntc' => 'Northwest Territories',
-    'nu' => 'Nauru',
-    'nuc' => 'Nunavut',
-    'nvu' => 'Nevada',
-    'nw' => 'Northern Mariana Islands',
-    'nx' => 'Norfolk Island',
-    'nyu' => 'New York (State)',
-    'nz' => 'New Zealand',
-    'ohu' => 'Ohio',
-    'oku' => 'Oklahoma',
-    'onc' => 'Ontario',
-    'oru' => 'Oregon',
-    'ot' => 'Mayotte',
-    'pau' => 'Pennsylvania',
-    'pc' => 'Pitcairn Island',
-    'pe' => 'Peru',
-    'pf' => 'Paracel Islands',
-    'pg' => 'GuineaBissau',
-    'ph' => 'Philippines',
-    'pic' => 'Prince Edward Island',
-    'pk' => 'Pakistan',
-    'pl' => 'Poland',
-    'pn' => 'Panama',
-    'po' => 'Portugal',
-    'pp' => 'Papua New Guinea',
-    'pr' => 'Puerto Rico',
-    'pt' => 'Portuguese Timor',
-    'pw' => 'Palau',
-    'py' => 'Paraguay',
-    'qa' => 'Qatar',
-    'qea' => 'Queensland',
-    'quc' => 'Québec (Province)',
-    'rb' => 'Serbia',
-    're' => 'Réunion',
-    'rh' => 'Zimbabwe',
-    'riu' => 'Rhode Island',
-    'rm' => 'Romania',
-    'ru' => 'Russia (Federation)',
-    'rur' => 'Russian S.F.S.R.',
-    'rw' => 'Rwanda',
-    'ry' => 'Ryukyu Islands, Southern',
-    'sa' => 'South Africa',
-    'sb' => 'Svalbard',
-    'sc' => 'SaintBarthélemy',
-    'scu' => 'South Carolina',
-    'sd' => 'South Sudan',
-    'sdu' => 'South Dakota',
-    'se' => 'Seychelles',
-    'sf' => 'Sao Tome and Principe',
-    'sg' => 'Senegal',
-    'sh' => 'Spanish North Africa',
-    'si' => 'Singapore',
-    'sj' => 'Sudan',
-    'sk' => 'Sikkim',
-    'sl' => 'Sierra Leone',
-    'sm' => 'San Marino',
-    'sn' => 'Sint Maarten',
-    'snc' => 'Saskatchewan',
-    'so' => 'Somalia',
-    'sp' => 'Spain',
-    'sq' => 'Swaziland',
-    'sr' => 'Surinam',
-    'ss' => 'Western Sahara',
-    'st' => 'SaintMartin',
-    'stk' => 'Scotland',
-    'su' => 'Saudi Arabia',
-    'sv' => 'Swan Islands',
-    'sw' => 'Sweden',
-    'sx' => 'Namibia',
-    'sy' => 'Syria',
-    'sz' => 'Switzerland',
-    'ta' => 'Tajikistan',
-    'tar' => 'Tajik S.S.R.',
-    'tc' => 'Turks and Caicos Islands',
-    'tg' => 'Togo',
-    'th' => 'Thailand',
-    'ti' => 'Tunisia',
-    'tk' => 'Turkmenistan',
-    'tkr' => 'Turkmen S.S.R.',
-    'tl' => 'Tokelau',
-    'tma' => 'Tasmania',
-    'tnu' => 'Tennessee',
-    'to' => 'Tonga',
-    'tr' => 'Trinidad and Tobago',
-    'ts' => 'United Arab Emirates',
-    'tt' => 'Trust Territory of the Pacific Islands',
-    'tu' => 'Turkey',
-    'tv' => 'Tuvalu',
-    'txu' => 'Texas',
-    'tz' => 'Tanzania',
-    'ua' => 'Egypt',
-    'uc' => 'United States Misc. Caribbean Islands',
-    'ug' => 'Uganda',
-    'ui' => 'United Kingdom Misc. Islands',
-    'uik' => 'United Kingdom Misc. Islands',
-    'uk' => 'United Kingdom',
-    'un' => 'Ukraine',
-    'unr' => 'Ukraine',
-    'up' => 'United States Misc. Pacific Islands',
-    'ur' => 'Soviet Union',
-    'us' => 'United States',
-    'utu' => 'Utah',
-    'uv' => 'Burkina Faso',
-    'uy' => 'Uruguay',
-    'uz' => 'Uzbekistan',
-    'uzr' => 'Uzbek S.S.R.',
-    'vau' => 'Virginia',
-    'vb' => 'British Virgin Islands',
-    'vc' => 'Vatican City',
-    've' => 'Venezuela',
-    'vi' => 'Virgin Islands of the United States',
-    'vm' => 'Vietnam',
-    'vn' => 'Vietnam, North',
-    'vp' => 'Various places',
-    'vra' => 'Victoria',
-    'vs' => 'Vietnam, South',
-    'vtu' => 'Vermont',
-    'wau' => 'Washington (State)',
-    'wb' => 'West Berlin',
-    'wea' => 'Western Australia',
-    'wf' => 'Wallis and Futuna',
-    'wiu' => 'Wisconsin',
-    'wj' => 'West Bank of the Jordan River',
-    'wk' => 'Wake Island',
-    'wlk' => 'Wales',
-    'ws' => 'Samoa',
-    'wvu' => 'West Virginia',
-    'wyu' => 'Wyoming',
-    'xa' => 'Christmas Island (Indian Ocean)',
-    'xb' => 'Cocos (Keeling) Islands',
-    'xc' => 'Maldives',
-    'xd' => 'Saint KittsNevis',
-    'xe' => 'Marshall Islands',
-    'xf' => 'Midway Islands',
-    'xga' => 'Coral Sea Islands Territory',
-    'xh' => 'Niue',
-    'xi' => 'Saint KittsNevisAnguilla',
-    'xj' => 'Saint Helena',
-    'xk' => 'Saint Lucia',
-    'xl' => 'Saint Pierre and Miquelon',
-    'xm' => 'Saint Vincent and the Grenadines',
-    'xn' => 'Macedonia',
-    'xna' => 'New South Wales',
-    'xo' => 'Slovakia',
-    'xoa' => 'Northern Territory',
-    'xp' => 'Spratly Island',
-    'xr' => 'Czech Republic',
-    'xra' => 'South Australia',
-    'xs' => 'South Georgia and the South Sandwich Islands',
-    'xv' => 'Slovenia',
-    # 'xx' => 'No place, unknown, or undetermined',
-    'xxc' => 'Canada',
-    'xxk' => 'United Kingdom',
-    'xxr' => 'Soviet Union',
-    'xxu' => 'United States',
-    'ye' => 'Yemen',
-    'ykc' => 'Yukon Territory',
-    'ys' => "Yemen (People's Democratic Republic)",
-    'yu' => 'Serbia and Montenegro',
-    'za' => 'Zambia'
-  }.freeze

data/lib/stanford-mods/geo_utils.rb

@@ -1,28 +0,0 @@
-# encoding: utf-8
-module Stanford
-  module Mods
-    # Abstract geo methods usable to several classes
-    module GeoUtils
-      # @param [String] val Coordinates value
-      # @return [String] cleaned value (strips parens and period), or the original value
-      def cleaner_coordinate(val)
-        matches = val.match(/^\(?([^)]+)\)?\.?$/)
-        matches ? matches[1] : val
-      end
-
-      # @param [String] point coordinate point in degrees notation
-      # @return [Float] converted value in decimal notation
-      def coord_to_decimal(point)
-        regex = /(?<dir>[NESW])\s*(?<deg>\d+)[°⁰º](?:(?<min>\d+)[ʹ'])?(?:(?<sec>\d+)[ʺ"])?/
-        match = regex.match(point)
-        return Float::INFINITY unless match
-
-        dec = match['deg'].to_i
-        dec += match['min'].to_f / 60
-        dec += match['sec'].to_f / 60 / 60
-        dec = -1 * dec if match['dir'] == 'W' || match['dir'] == 'S'
-        dec
-      end
-    end
-  end
-end

data/lib/stanford-mods/name.rb

@@ -1,80 +0,0 @@
-# encoding: UTF-8
-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|
-          first_wo_role ||= n if n.role.empty?
-          n.role.each { |r|
-            if r.authority.include?('marcrelator') &&
-                 r.value.any? { |v| v.match(/creator/i) || v.match?(/author/i) }
-              result ||= n.display_value_w_date
-            end
-          }
-        }
-        result = first_wo_role.display_value_w_date if !result && first_wo_role
-        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
-
-      # @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|
-          next if n.role.size.zero?
-
-          n.role.each { |r|
-            result << n.display_value_w_date unless includes_marc_relator_collector_role?(r)
-          }
-        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|
-          next if n.role.size.zero?
-
-          n.role.each { |r|
-            result << n.display_value_w_date if includes_marc_relator_collector_role?(r)
-          }
-        end
-        result unless result.empty?
-      end
-
-      COLLECTOR_ROLE_URI = 'http://id.loc.gov/vocabulary/relators/col'.freeze
-
-      # @param Nokogiri::XML::Node role_node the role node from a parent name node
-      # @return true if there is a MARC relator collector role assigned
-      def includes_marc_relator_collector_role?(role_node)
-        (role_node.authority.include?('marcrelator') && role_node.value.include?('Collector')) ||
-        role_node.roleTerm.valueURI.first == COLLECTOR_ROLE_URI
-      end
-    end # class Record
-  end # Module Mods
-end # Module Stanford