cocina_display 0.5.0 → 0.7.0

data/lib/cocina_display/subjects/subject_value.rb

@@ -0,0 +1,104 @@
+require_relative "subject"
+require_relative "../contributor"
+require_relative "../title_builder"
+require_relative "../dates/date"
+
+module CocinaDisplay
+  module Subjects
+    # A descriptive value that can be part of a Subject.
+    class SubjectValue
+      attr_reader :cocina
+
+      # The type of the subject value, like "person", "title", or "time".
+      # @see https://github.com/sul-dlss/cocina-models/blob/main/docs/description_types.md#subject-part-types-for-structured-value
+      attr_accessor :type
+
+      # Create a SubjectValue from Cocina structured data.
+      # @param cocina [Hash] The Cocina structured data for the subject.
+      # @return [SubjectValue]
+      def self.from_cocina(cocina)
+        SUBJECT_VALUE_TYPES.fetch(cocina["type"], SubjectValue).new(cocina)
+      end
+
+      # All subject value types that should not be further destructured.
+      # @return [Array<String>]
+      def self.atomic_types
+        SUBJECT_VALUE_TYPES.keys
+      end
+
+      # Initialize a SubjectValue object with Cocina structured data.
+      # @param cocina [Hash] The Cocina structured data for the subject value.
+      def initialize(cocina)
+        @cocina = cocina
+        @type = cocina["type"]
+      end
+
+      # The display string for the subject value.
+      # Subclasses should override this method to provide specific formatting.
+      # @return [String]
+      def display_str
+        cocina["value"]
+      end
+    end
+
+    # A subject value representing a named entity.
+    class NameSubjectValue < SubjectValue
+      attr_reader :name
+
+      # Initialize a NameSubjectValue object with Cocina structured data.
+      # @param cocina [Hash] The Cocina structured data for the subject.
+      def initialize(cocina)
+        super
+        @name = Contributor::Name.new(cocina)
+      end
+
+      # Use the contributor name formatting rules for display.
+      # @return [String] The formatted name string, including life dates
+      # @see CocinaDisplay::Contributor::Name#display_str
+      def display_str
+        @name.display_str(with_date: true)
+      end
+    end
+
+    # A subject value representing an entity with a title.
+    class TitleSubjectValue < SubjectValue
+      # Construct a title string to use for display.
+      # @see CocinaDisplay::TitleBuilder.build
+      # @note Unclear how often structured title subjects occur "in the wild".
+      # @return [String]
+      def display_str
+        TitleBuilder.build([cocina])
+      end
+    end
+
+    # A subject value representing a date and/or time.
+    class TemporalSubjectValue < SubjectValue
+      attr_reader :date
+
+      def initialize(cocina)
+        super
+        @date = Dates::Date.from_cocina(cocina)
+      end
+
+      # @return [String] The formatted date/time string for display
+      def display_str
+        @date.qualified_value
+      end
+    end
+  end
+end
+
+# Map Cocina subject types to specific SubjectValue classes for rendering.
+# @see SubjectValue#type
+SUBJECT_VALUE_TYPES = {
+  "person" => CocinaDisplay::Subjects::NameSubjectValue,
+  "family" => CocinaDisplay::Subjects::NameSubjectValue,
+  "organization" => CocinaDisplay::Subjects::NameSubjectValue,
+  "conference" => CocinaDisplay::Subjects::NameSubjectValue,
+  "event" => CocinaDisplay::Subjects::NameSubjectValue,
+  "name" => CocinaDisplay::Subjects::NameSubjectValue,
+  "title" => CocinaDisplay::Subjects::TitleSubjectValue,
+  "time" => CocinaDisplay::Subjects::TemporalSubjectValue
+  # TODO: special handling for geospatial subjects
+  # "map coordinates", "bounding box coordinates", "point coordinates"
+}.freeze

data/lib/cocina_display/title_builder.rb

@@ -53,7 +53,7 @@ module CocinaDisplay
     def self.sort_title(titles, catalog_links: [])
       part_label = catalog_links.find { |link| link["catalog"] == "folio" }&.fetch("partLabel", nil)
       [new(strategy: :first, add_punctuation: false, only_one_parallel_value: false, part_label: part_label, sortable: true).build(titles)]
-        .flatten.compact.map { |title| title.gsub(/[[:punct:]]*/, "").strip }
+        .flatten.compact.map { |title| title.gsub(/[[:punct:]]*/, "").squeeze(" ").strip }
     end
 
     # @param strategy [Symbol] ":first" selects a single title value based on precedence of
@@ -116,6 +116,7 @@ module CocinaDisplay
     end
 
     def extract_title(cocina_title)
+      return if cocina_title.blank?
       title_values = if cocina_title["value"]
         cocina_title["value"]
       elsif cocina_title["structuredValue"].present?

data/lib/cocina_display/utils.rb

@@ -19,11 +19,12 @@ module CocinaDisplay
       end.delete_suffix(delimiter)
     end
 
-    # Recursively flatten structured and parallel values in Cocina metadata.
+    # Recursively flatten structured, parallel, and grouped values in Cocina metadata.
     # Returns a list of hashes representing the "leaf" nodes with +value+ key.
     # @return [Array<Hash>] List of node hashes with "value" present
     # @param cocina [Hash] The Cocina structured data to flatten
     # @param output [Array] Used for recursion, should be empty on first call
+    # @param atomic_types [Array<String>] Types considered atomic; will not be flattened
     # @example simple value
     #  cocina = { "value" => "John Doe", "type" => "name" }
     #  Utils.flatten_nested_values(cocina)
@@ -36,14 +37,38 @@ module CocinaDisplay
     #  cocina = { "parallelValue" => [{"value" => "foo" }, { "structuredValue" => [{"value" => "bar"},  {"value" => "baz"}] }] }
     #  Utils.flatten_nested_values(cocina)
     #  #=> [{"value" => "foo"}, {"value" => "foo"}, {"value" => "baz"}]
-    def self.flatten_nested_values(cocina, output = [])
+    def self.flatten_nested_values(cocina, output = [], atomic_types: [])
       return [cocina] if cocina["value"].present?
-      return cocina.flat_map { |node| flatten_nested_values(node, output) } if cocina.is_a?(Array)
+      return [cocina] if atomic_types.include?(cocina["type"])
+      return cocina.flat_map { |node| flatten_nested_values(node, output, atomic_types: atomic_types) } if cocina.is_a?(Array)
 
-      nested_values = Array(cocina["structuredValue"]) + Array(cocina["parallelValue"])
+      nested_values = Array(cocina["structuredValue"]) + Array(cocina["parallelValue"]) + Array(cocina["groupedValue"])
       return output unless nested_values.any?
 
-      nested_values.flat_map { |node| flatten_nested_values(node, output) }
+      nested_values.flat_map { |node| flatten_nested_values(node, output, atomic_types: atomic_types) }
+    end
+
+    # Recursively remove empty values from a hash, including nested hashes and arrays.
+    # @param hash [Hash] The hash to process
+    # @param output [Hash] Used for recursion, should be empty on first call
+    # @return [Hash] The hash with empty values removed
+    # @example
+    #  hash = { "name" => "", "age" => nil, "address => { "city" => "Anytown", "state" => [] } }
+    #  #  Utils.remove_empty_values(hash)
+    #  #=> { "address" => { "city" => "Anytown" } }
+    def self.deep_compact_blank(hash, output = {})
+      hash.each do |key, value|
+        if value.is_a?(Hash)
+          nested = deep_compact_blank(value)
+          output[key] = nested unless nested.empty?
+        elsif value.is_a?(Array)
+          compacted_array = value.map { |v| deep_compact_blank(v) }.reject(&:blank?)
+          output[key] = compacted_array unless compacted_array.empty?
+        elsif value.present?
+          output[key] = value
+        end
+      end
+      output
     end
   end
 end

data/lib/cocina_display/version.rb

@@ -2,5 +2,5 @@
 
 # :nodoc:
 module CocinaDisplay
-  VERSION = "0.5.0" # :nodoc:
+  VERSION = "0.7.0" # :nodoc:
 end

data/lib/cocina_display/vocabularies/marc_country_codes.rb

@@ -0,0 +1,393 @@
+# frozen_string_literal: true
+
+module CocinaDisplay
+  module Vocabularies
+    # Map of MARC country codes to country names.
+    # @see http://www.loc.gov/marc/countries/countries_code.html
+    # @note Ported from stanford-mods gem.
+    MARC_COUNTRY = {
+      "aa" => "Albania",
+      "abc" => "Alberta",
+      "ac" => "Ashmore and Cartier Islands", # discontinued
+      "aca" => "Australian Capital Territory",
+      "ae" => "Algeria",
+      "af" => "Afghanistan",
+      "ag" => "Argentina",
+      "ai" => "Armenia (Republic)",
+      "air" => "Armenian S.S.R.", # discontinued
+      "aj" => "Azerbaijan",
+      "ajr" => "Azerbaijan S.S.R.", # discontinued
+      "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.", # discontinued
+      "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", # discontinued
+      "co" => "Curaçao",
+      "cou" => "Colorado",
+      "cp" => "Canton and Enderbury Islands", # discontinued
+      "cq" => "Comoros",
+      "cr" => "Costa Rica",
+      "cs" => "Czechoslovakia", # discontinued
+      "ctu" => "Connecticut",
+      "cu" => "Cuba",
+      "cv" => "Cabo Verde",
+      "cw" => "Cook Islands",
+      "cx" => "Central African Republic",
+      "cy" => "Cyprus",
+      "cz" => "Canal Zone", # discontinued
+      "dcu" => "District of Columbia",
+      "deu" => "Delaware",
+      "dk" => "Denmark",
+      "dm" => "Benin",
+      "dq" => "Dominica",
+      "dr" => "Dominican Republic",
+      "ea" => "Eritrea",
+      "ec" => "Ecuador",
+      "eg" => "Equatorial Guinea",
+      "em" => "Timor-Leste",
+      "enk" => "England",
+      "er" => "Estonia",
+      "err" => "Estonia", # discontinued
+      "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)", # discontinued
+      "gg" => "Guernsey",
+      "gh" => "Ghana",
+      "gi" => "Gibraltar",
+      "gl" => "Greenland",
+      "gm" => "Gambia",
+      "gn" => "Gilbert and Ellice Islands", # discontinued
+      "go" => "Gabon",
+      "gp" => "Guadeloupe",
+      "gr" => "Greece",
+      "gs" => "Georgia (Republic)",
+      "gsr" => "Georgian S.S.R.", # discontinued
+      "gt" => "Guatemala",
+      "gu" => "Guam",
+      "gv" => "Guinea",
+      "gw" => "Germany",
+      "gy" => "Guyana",
+      "gz" => "Gaza Strip",
+      "hiu" => "Hawaii",
+      "hk" => "Hong Kong", # discontinued
+      "hm" => "Heard and McDonald Islands",
+      "ho" => "Honduras",
+      "ht" => "Haiti",
+      "hu" => "Hungary",
+      "iau" => "Iowa",
+      "ic" => "Iceland",
+      "idu" => "Idaho",
+      "ie" => "Ireland",
+      "ii" => "India",
+      "ilu" => "Illinois",
+      "im" => "Isle of Man",
+      "inu" => "Indiana",
+      "io" => "Indonesia",
+      "iq" => "Iraq",
+      "ir" => "Iran",
+      "is" => "Israel",
+      "it" => "Italy",
+      "iu" => "Israel-Syria Demilitarized Zones", # discontinued
+      "iv" => "Côte d'Ivoire",
+      "iw" => "Israel-Jordan Demilitarized Zones", # discontinued
+      "iy" => "Iraq-Saudi Arabia Neutral Zone",
+      "ja" => "Japan",
+      "je" => "Jersey",
+      "ji" => "Johnston Atoll",
+      "jm" => "Jamaica",
+      "jn" => "Jan Mayen", # discontinued
+      "jo" => "Jordan",
+      "ke" => "Kenya",
+      "kg" => "Kyrgyzstan",
+      "kgr" => "Kirghiz S.S.R.", # discontinued
+      "kn" => "Korea (North)",
+      "ko" => "Korea (South)",
+      "ksu" => "Kansas",
+      "ku" => "Kuwait",
+      "kv" => "Kosovo",
+      "kyu" => "Kentucky",
+      "kz" => "Kazakhstan",
+      "kzr" => "Kazakh S.S.R.", # discontinued
+      "lau" => "Louisiana",
+      "lb" => "Liberia",
+      "le" => "Lebanon",
+      "lh" => "Liechtenstein",
+      "li" => "Lithuania",
+      "lir" => "Lithuania", # discontinued
+      "ln" => "Central and Southern Line Islands", # discontinued
+      "lo" => "Lesotho",
+      "ls" => "Laos",
+      "lu" => "Luxembourg",
+      "lv" => "Latvia",
+      "lvr" => "Latvia", # discontinued
+      "ly" => "Libya",
+      "mau" => "Massachusetts",
+      "mbc" => "Manitoba",
+      "mc" => "Monaco",
+      "mdu" => "Maryland",
+      "meu" => "Maine",
+      "mf" => "Mauritius",
+      "mg" => "Madagascar",
+      "mh" => "Macao", # discontinued
+      "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.", # discontinued
+      "mw" => "Malawi",
+      "mx" => "Mexico",
+      "my" => "Malaysia",
+      "mz" => "Mozambique",
+      "na" => "Netherlands Antilles", # discontinued
+      "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", # discontinued
+      "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" => "Guinea-Bissau",
+      "ph" => "Philippines",
+      "pic" => "Prince Edward Island",
+      "pk" => "Pakistan",
+      "pl" => "Poland",
+      "pn" => "Panama",
+      "po" => "Portugal",
+      "pp" => "Papua New Guinea",
+      "pr" => "Puerto Rico",
+      "pt" => "Portuguese Timor", # discontinued
+      "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.", # discontinued
+      "rw" => "Rwanda",
+      "ry" => "Ryukyu Islands, Southern", # discontinued
+      "sa" => "South Africa",
+      "sb" => "Svalbard", # discontinued
+      "sc" => "Saint-Barthélemy",
+      "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", # discontinued
+      "sl" => "Sierra Leone",
+      "sm" => "San Marino",
+      "sn" => "Sint Maarten",
+      "snc" => "Saskatchewan",
+      "so" => "Somalia",
+      "sp" => "Spain",
+      "sq" => "Eswatini",
+      "sr" => "Surinam",
+      "ss" => "Western Sahara",
+      "st" => "Saint-Martin",
+      "stk" => "Scotland",
+      "su" => "Saudi Arabia",
+      "sv" => "Swan Islands", # discontinued
+      "sw" => "Sweden",
+      "sx" => "Namibia",
+      "sy" => "Syria",
+      "sz" => "Switzerland",
+      "ta" => "Tajikistan",
+      "tar" => "Tajik S.S.R.", # discontinued
+      "tc" => "Turks and Caicos Islands",
+      "tg" => "Togo",
+      "th" => "Thailand",
+      "ti" => "Tunisia",
+      "tk" => "Turkmenistan",
+      "tkr" => "Turkmen S.S.R.", # discontinued
+      "tl" => "Tokelau",
+      "tma" => "Tasmania",
+      "tnu" => "Tennessee",
+      "to" => "Tonga",
+      "tr" => "Trinidad and Tobago",
+      "ts" => "United Arab Emirates",
+      "tt" => "Trust Territory of the Pacific Islands", # discontinued
+      "tu" => "Turkey",
+      "tv" => "Tuvalu",
+      "txu" => "Texas",
+      "tz" => "Tanzania",
+      "ua" => "Egypt",
+      "uc" => "United States Misc. Caribbean Islands",
+      "ug" => "Uganda",
+      "ui" => "United Kingdom Misc. Islands", # discontinued
+      "uik" => "United Kingdom Misc. Islands",
+      "uk" => "United Kingdom", # discontinued
+      "un" => "Ukraine",
+      "unr" => "Ukraine", # discontinued
+      "up" => "United States Misc. Pacific Islands",
+      "ur" => "Soviet Union", # discontinued
+      "us" => "United States", # discontinued
+      "utu" => "Utah",
+      "uv" => "Burkina Faso",
+      "uy" => "Uruguay",
+      "uz" => "Uzbekistan",
+      "uzr" => "Uzbek S.S.R.", # discontinued
+      "vau" => "Virginia",
+      "vb" => "British Virgin Islands",
+      "vc" => "Vatican City",
+      "ve" => "Venezuela",
+      "vi" => "Virgin Islands of the United States",
+      "vm" => "Vietnam",
+      "vn" => "Vietnam, North", # discontinued
+      "vp" => "Various places",
+      "vra" => "Victoria",
+      "vs" => "Vietnam, South", # discontinued
+      "vtu" => "Vermont",
+      "wau" => "Washington (State)",
+      "wb" => "West Berlin", # discontinued
+      "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 Kitts-Nevis",
+      "xe" => "Marshall Islands",
+      "xf" => "Midway Islands",
+      "xga" => "Coral Sea Islands Territory",
+      "xh" => "Niue",
+      "xi" => "Saint Kitts-Nevis-Anguilla", # discontinued
+      "xj" => "Saint Helena",
+      "xk" => "Saint Lucia",
+      "xl" => "Saint Pierre and Miquelon",
+      "xm" => "Saint Vincent and the Grenadines",
+      "xn" => "North 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", # discontinued
+      "xxu" => "United States",
+      "ye" => "Yemen",
+      "ykc" => "Yukon Territory",
+      "ys" => "Yemen (People's Democratic Republic)", # discontinued
+      "yu" => "Serbia and Montenegro", # discontinued
+      "za" => "Zambia"
+    }.freeze
+  end
+end