cocina-models 0.75.0 → 0.78.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (142) hide show
  1. checksums.yaml +4 -4
  2. data/.rubocop.yml +40 -12
  3. data/.rubocop_todo.yml +71 -2
  4. data/README.md +41 -5
  5. data/cocina-models.gemspec +2 -0
  6. data/description_types.yml +167 -38
  7. data/docs/description_types.md +471 -216
  8. data/lib/cocina/generator/generator.rb +7 -12
  9. data/lib/cocina/generator/schema.rb +1 -3
  10. data/lib/cocina/generator/schema_base.rb +0 -8
  11. data/lib/cocina/generator/schema_ref.rb +1 -1
  12. data/lib/cocina/generator/schema_value.rb +14 -4
  13. data/lib/cocina/models/access.rb +4 -4
  14. data/lib/cocina/models/admin_policy.rb +1 -1
  15. data/lib/cocina/models/admin_policy_access_template.rb +7 -7
  16. data/lib/cocina/models/admin_policy_administrative.rb +1 -1
  17. data/lib/cocina/models/admin_policy_with_metadata.rb +3 -3
  18. data/lib/cocina/models/builders/name_title_group_builder.rb +0 -4
  19. data/lib/cocina/models/builders/title_builder.rb +0 -2
  20. data/lib/cocina/models/citation_only_access.rb +2 -2
  21. data/lib/cocina/models/collection_access.rb +4 -4
  22. data/lib/cocina/models/collection_identification.rb +1 -1
  23. data/lib/cocina/models/collection_with_metadata.rb +2 -2
  24. data/lib/cocina/models/contributor.rb +4 -4
  25. data/lib/cocina/models/controlled_digital_lending_access.rb +2 -2
  26. data/lib/cocina/models/dark_access.rb +4 -4
  27. data/lib/cocina/models/description.rb +3 -3
  28. data/lib/cocina/models/descriptive_basic_value.rb +13 -13
  29. data/lib/cocina/models/descriptive_parallel_contributor.rb +5 -5
  30. data/lib/cocina/models/descriptive_parallel_event.rb +3 -3
  31. data/lib/cocina/models/descriptive_value.rb +13 -13
  32. data/lib/cocina/models/descriptive_value_language.rb +6 -6
  33. data/lib/cocina/models/dro.rb +1 -1
  34. data/lib/cocina/models/dro_access.rb +8 -8
  35. data/lib/cocina/models/dro_with_metadata.rb +3 -3
  36. data/lib/cocina/models/embargo.rb +5 -5
  37. data/lib/cocina/models/event.rb +3 -3
  38. data/lib/cocina/models/file.rb +4 -4
  39. data/lib/cocina/models/file_access.rb +4 -4
  40. data/lib/cocina/models/identification.rb +2 -2
  41. data/lib/cocina/models/language.rb +12 -12
  42. data/lib/cocina/models/location_based_access.rb +1 -1
  43. data/lib/cocina/models/location_based_download_access.rb +1 -1
  44. data/lib/cocina/models/mapping/error_notifier.rb +36 -0
  45. data/lib/cocina/models/mapping/from_mods/access.rb +177 -0
  46. data/lib/cocina/models/mapping/from_mods/admin_metadata.rb +217 -0
  47. data/lib/cocina/models/mapping/from_mods/alt_rep_group.rb +26 -0
  48. data/lib/cocina/models/mapping/from_mods/authority.rb +51 -0
  49. data/lib/cocina/models/mapping/from_mods/contributor.rb +161 -0
  50. data/lib/cocina/models/mapping/from_mods/description.rb +98 -0
  51. data/lib/cocina/models/mapping/from_mods/description_builder.rb +61 -0
  52. data/lib/cocina/models/mapping/from_mods/event.rb +543 -0
  53. data/lib/cocina/models/mapping/from_mods/form.rb +381 -0
  54. data/lib/cocina/models/mapping/from_mods/geographic.rb +219 -0
  55. data/lib/cocina/models/mapping/from_mods/hydrus_default_title_builder.rb +28 -0
  56. data/lib/cocina/models/mapping/from_mods/identifier.rb +51 -0
  57. data/lib/cocina/models/mapping/from_mods/identifier_builder.rb +71 -0
  58. data/lib/cocina/models/mapping/from_mods/identifier_type.rb +292 -0
  59. data/lib/cocina/models/mapping/from_mods/language.rb +36 -0
  60. data/lib/cocina/models/mapping/from_mods/language_script.rb +30 -0
  61. data/lib/cocina/models/mapping/from_mods/language_term.rb +106 -0
  62. data/lib/cocina/models/mapping/from_mods/name_builder.rb +307 -0
  63. data/lib/cocina/models/mapping/from_mods/note.rb +162 -0
  64. data/lib/cocina/models/mapping/from_mods/part_builder.rb +147 -0
  65. data/lib/cocina/models/mapping/from_mods/primary.rb +27 -0
  66. data/lib/cocina/models/mapping/from_mods/purl.rb +53 -0
  67. data/lib/cocina/models/mapping/from_mods/related_resource.rb +105 -0
  68. data/lib/cocina/models/mapping/from_mods/subject.rb +413 -0
  69. data/lib/cocina/models/mapping/from_mods/subject_authority_codes.rb +794 -0
  70. data/lib/cocina/models/mapping/from_mods/title.rb +160 -0
  71. data/lib/cocina/models/mapping/from_mods/title_builder.rb +106 -0
  72. data/lib/cocina/models/mapping/from_mods/title_builder_strategy.rb +19 -0
  73. data/lib/cocina/models/mapping/from_mods/value_uri.rb +25 -0
  74. data/lib/cocina/models/mapping/normalizers/base.rb +16 -0
  75. data/lib/cocina/models/mapping/normalizers/mods/geo_extension_normalizer.rb +69 -0
  76. data/lib/cocina/models/mapping/normalizers/mods/name_normalizer.rb +191 -0
  77. data/lib/cocina/models/mapping/normalizers/mods/origin_info_normalizer.rb +157 -0
  78. data/lib/cocina/models/mapping/normalizers/mods/subject_normalizer.rb +296 -0
  79. data/lib/cocina/models/mapping/normalizers/mods/title_normalizer.rb +91 -0
  80. data/lib/cocina/models/mapping/normalizers/mods_normalizer.rb +409 -0
  81. data/lib/cocina/models/mapping/purl.rb +27 -0
  82. data/lib/cocina/models/mapping/to_mods/access.rb +155 -0
  83. data/lib/cocina/models/mapping/to_mods/admin_metadata.rb +129 -0
  84. data/lib/cocina/models/mapping/to_mods/contributor.rb +49 -0
  85. data/lib/cocina/models/mapping/to_mods/description.rb +63 -0
  86. data/lib/cocina/models/mapping/to_mods/event.rb +200 -0
  87. data/lib/cocina/models/mapping/to_mods/form.rb +292 -0
  88. data/lib/cocina/models/mapping/to_mods/geographic.rb +151 -0
  89. data/lib/cocina/models/mapping/to_mods/id_generator.rb +25 -0
  90. data/lib/cocina/models/mapping/to_mods/identifier.rb +57 -0
  91. data/lib/cocina/models/mapping/to_mods/language.rb +82 -0
  92. data/lib/cocina/models/mapping/to_mods/mods_writer.rb +38 -0
  93. data/lib/cocina/models/mapping/to_mods/name_title_group.rb +29 -0
  94. data/lib/cocina/models/mapping/to_mods/name_writer.rb +228 -0
  95. data/lib/cocina/models/mapping/to_mods/note.rb +105 -0
  96. data/lib/cocina/models/mapping/to_mods/part_writer.rb +115 -0
  97. data/lib/cocina/models/mapping/to_mods/related_resource.rb +108 -0
  98. data/lib/cocina/models/mapping/to_mods/role_writer.rb +50 -0
  99. data/lib/cocina/models/mapping/to_mods/subject.rb +486 -0
  100. data/lib/cocina/models/mapping/to_mods/title.rb +260 -0
  101. data/lib/cocina/models/object_metadata.rb +2 -2
  102. data/lib/cocina/models/presentation.rb +2 -2
  103. data/lib/cocina/models/related_resource.rb +9 -9
  104. data/lib/cocina/models/release_tag.rb +4 -4
  105. data/lib/cocina/models/request_admin_policy.rb +1 -1
  106. data/lib/cocina/models/request_administrative.rb +1 -1
  107. data/lib/cocina/models/request_collection.rb +2 -2
  108. data/lib/cocina/models/request_description.rb +3 -3
  109. data/lib/cocina/models/request_dro.rb +4 -4
  110. data/lib/cocina/models/request_file.rb +5 -5
  111. data/lib/cocina/models/request_identification.rb +1 -1
  112. data/lib/cocina/models/sequence.rb +1 -1
  113. data/lib/cocina/models/source.rb +4 -4
  114. data/lib/cocina/models/standard.rb +5 -5
  115. data/lib/cocina/models/stanford_access.rb +2 -2
  116. data/lib/cocina/models/title.rb +13 -13
  117. data/lib/cocina/models/validators/dark_validator.rb +4 -2
  118. data/lib/cocina/models/validators/description_values_validator.rb +77 -0
  119. data/lib/cocina/models/validators/open_api_validator.rb +0 -4
  120. data/lib/cocina/models/validators/validator.rb +2 -1
  121. data/lib/cocina/models/version.rb +1 -1
  122. data/lib/cocina/models/world_access.rb +2 -2
  123. data/lib/cocina/models.rb +4 -0
  124. data/lib/cocina/rspec/factories.rb +205 -0
  125. data/lib/cocina/rspec.rb +2 -0
  126. data/openapi.yml +5 -5
  127. metadata +89 -17
  128. data/docs/_config.yml +0 -1
  129. data/docs/maps/Agent.json +0 -18
  130. data/docs/maps/Collection.json +0 -240
  131. data/docs/maps/DRO.json +0 -316
  132. data/docs/maps/Description.json +0 -17
  133. data/docs/maps/File.json +0 -196
  134. data/docs/maps/Fileset.json +0 -143
  135. data/docs/maps/README.md +0 -7
  136. data/docs/maps/ReleaseTag.json +0 -39
  137. data/docs/maps/Sequence.json +0 -46
  138. data/docs/maps/Title.json +0 -18
  139. data/docs/sampleETD/foxml-export.xml +0 -935
  140. data/docs/sampleETD/foxml.xml +0 -3475
  141. data/docs/sampleETD/xn109qc9773_bibframe.ttl +0 -95
  142. data/docs/sampleETD/xn109qc9773_taco.json +0 -158
@@ -0,0 +1,409 @@
1
+ # frozen_string_literal: true
2
+
3
+ module Cocina
4
+ module Models
5
+ module Mapping
6
+ module Normalizers
7
+ # Normalizes a Fedora MODS document, accounting for differences between Fedora MODS and MODS generated from Cocina.
8
+ # these adjustments have been approved by our metadata authority, Arcadia.
9
+ class ModsNormalizer # rubocop:disable Metrics/ClassLength
10
+ include Cocina::Models::Mapping::Normalizers::Base
11
+
12
+ MODS_NS = Cocina::Models::Mapping::FromMods::Description::DESC_METADATA_NS
13
+ XLINK_NS = Cocina::Models::Mapping::FromMods::Description::XLINK_NS
14
+
15
+ # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
16
+ # @param [String] druid
17
+ # @param [String] label
18
+ # @return [Nokogiri::Document] normalized MODS
19
+ def self.normalize(mods_ng_xml:, druid:, label:)
20
+ new(mods_ng_xml: mods_ng_xml, druid: druid, label: label).normalize
21
+ end
22
+
23
+ # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
24
+ # @param [String] druid
25
+ # @return [Nokogiri::Document] normalized MODS
26
+ def self.normalize_purl(mods_ng_xml:, druid:)
27
+ new(mods_ng_xml: mods_ng_xml, druid: druid).normalize_purl
28
+ end
29
+
30
+ # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
31
+ # @param [String] druid
32
+ # @param [String] label
33
+ # @return [Nokogiri::Document] normalized MODS
34
+ def self.normalize_purl_and_missing_title(mods_ng_xml:, druid:, label:)
35
+ new(mods_ng_xml: mods_ng_xml, druid: druid, label: label).normalize_purl_and_missing_title
36
+ end
37
+
38
+ # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
39
+ # @return [Nokogiri::Document] normalized MODS
40
+ def self.normalize_identifier_type(mods_ng_xml:)
41
+ new(mods_ng_xml: mods_ng_xml, druid: nil).normalize_identifier_type
42
+ end
43
+
44
+ def initialize(mods_ng_xml:, druid:, label: nil)
45
+ @ng_xml = mods_ng_xml.root ? mods_ng_xml.dup : blank_ng_xml
46
+ @ng_xml.encoding = 'UTF-8'
47
+ @druid = druid
48
+ @label = label
49
+ end
50
+
51
+ def normalize
52
+ normalize_default_namespace
53
+ normalize_xsi
54
+ normalize_version
55
+ normalize_empty_attributes
56
+ normalize_authority_uris # must be called before OriginInfoNormalizer
57
+ @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::OriginInfoNormalizer.normalize(mods_ng_xml: ng_xml)
58
+ @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::SubjectNormalizer.normalize(mods_ng_xml: ng_xml)
59
+ @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::NameNormalizer.normalize(mods_ng_xml: ng_xml)
60
+ normalize_related_item_other_type
61
+ normalize_unmatched_altrepgroup
62
+ normalize_unmatched_nametitlegroup
63
+ normalize_xml_space
64
+ normalize_language_term_type
65
+ normalize_access_condition
66
+ normalize_identifier_type_attr
67
+ normalize_location_physical_location
68
+ normalize_purl_location
69
+ normalize_empty_notes
70
+ @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::TitleNormalizer.normalize(mods_ng_xml: ng_xml, label: label)
71
+ @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::GeoExtensionNormalizer.normalize(mods_ng_xml: ng_xml, druid: druid)
72
+ normalize_empty_type_of_resource # Must be after normalize_empty_attributes
73
+ normalize_notes
74
+ normalize_abstracts
75
+ normalize_usage_primary
76
+ normalize_related_item_attributes
77
+ # This should be last-ish.
78
+ normalize_empty_related_items
79
+ remove_empty_elements(ng_xml.root) # this must be last
80
+ ng_xml
81
+ end
82
+
83
+ def normalize_purl
84
+ normalize_purl_location
85
+ ng_xml
86
+ end
87
+
88
+ def normalize_purl_and_missing_title
89
+ normalize_purl_location
90
+ @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::TitleNormalizer.normalize_missing_title(mods_ng_xml: ng_xml, label: label)
91
+ ng_xml
92
+ end
93
+
94
+ def normalize_identifier_type
95
+ normalize_identifier_type_attr
96
+ ng_xml
97
+ end
98
+
99
+ private
100
+
101
+ attr_reader :ng_xml, :druid, :label
102
+
103
+ # remove all empty elements that have no attributes and no children, recursively
104
+ def remove_empty_elements(start_node)
105
+ return unless start_node
106
+
107
+ # remove node if there are no element children, there is no text value and there are no attributes
108
+ if start_node.elements.size.zero? &&
109
+ start_node.text.blank? &&
110
+ start_node.attributes.size.zero? &&
111
+ start_node.name != 'etal'
112
+ parent = start_node.parent
113
+ start_node.remove
114
+ remove_empty_elements(parent) # need to call again after child has been deleted
115
+ else
116
+ start_node.element_children.each { |e| remove_empty_elements(e) }
117
+ end
118
+ end
119
+
120
+ def normalize_default_namespace
121
+ xml = ng_xml.to_s
122
+
123
+ unless xml.include?('xmlns="http://www.loc.gov/mods/v3"')
124
+ xml.sub!('mods:mods', 'mods:mods xmlns="http://www.loc.gov/mods/v3"')
125
+ xml.gsub!('mods:', '')
126
+ end
127
+
128
+ regenerate_ng_xml(xml)
129
+ end
130
+
131
+ def normalize_xsi
132
+ return if ng_xml.namespaces.include?('xmlns:xsi')
133
+
134
+ xml = ng_xml.to_s
135
+ xml.sub!('<mods ', '<mods xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" ')
136
+
137
+ regenerate_ng_xml(xml)
138
+ end
139
+
140
+ def normalize_version
141
+ # Only normalize version when version isn't mapped.
142
+ match = /MODS version (\d\.\d)/.match(ng_xml.at('//mods:recordInfo/mods:recordOrigin', mods: MODS_NS)&.content)
143
+
144
+ if !match
145
+ ng_xml.root['version'] = '3.7'
146
+ ng_xml.root['xsi:schemaLocation'] = 'http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-3-7.xsd'
147
+ elsif match && match[1] != ng_xml.root['version']
148
+ ng_xml.root['version'] = match[1]
149
+ ng_xml.root['xsi:schemaLocation'] = "http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-#{match[1].sub('.', '-')}.xsd"
150
+ end
151
+ end
152
+
153
+ def normalize_authority_uris
154
+ Cocina::Models::Mapping::FromMods::Authority::NORMALIZE_AUTHORITY_URIS.each do |authority_uri|
155
+ ng_xml.xpath("//mods:*[@authorityURI='#{authority_uri}']", mods: MODS_NS).each do |node|
156
+ node[:authorityURI] = "#{authority_uri}/"
157
+ end
158
+ end
159
+ end
160
+
161
+ def normalize_purl_location
162
+ normalize_purl_for(ng_xml.root, purl: Cocina::Models::Mapping::Purl.for(druid: druid))
163
+ ng_xml.xpath('/mods:mods/mods:relatedItem', mods: MODS_NS).each { |related_item_node| normalize_purl_for(related_item_node) }
164
+ end
165
+
166
+ def normalize_purl_for(base_node, purl: nil)
167
+ purl_nodes(base_node).each do |purl_node|
168
+ purl_node.content = Cocina::Models::Mapping::FromMods::Purl.purl_value(purl_node)
169
+ end
170
+
171
+ # If there is a purl, add a url node if there is not already one.
172
+ if purl && purl_nodes(base_node).to_a.none? { |purl_node| purl_node.content == purl }
173
+ new_location = Nokogiri::XML::Node.new('location', Nokogiri::XML(nil))
174
+ new_url = Nokogiri::XML::Node.new('url', Nokogiri::XML(nil))
175
+ new_url.content = purl
176
+ new_location << new_url
177
+ base_node << new_location
178
+ end
179
+
180
+ purl_nodes(base_node).each do |purl_node|
181
+ next if purl_node == Cocina::Models::Mapping::FromMods::Purl.primary_purl_node(base_node, purl)
182
+
183
+ # Move into own relatedItem
184
+ new_related_item = Nokogiri::XML::Node.new('relatedItem', Nokogiri::XML(nil))
185
+ location_node = purl_node.parent
186
+ location_node.remove
187
+ new_related_item << location_node
188
+ base_node << new_related_item
189
+ purl_node[:usage] = 'primary display'
190
+ end
191
+
192
+ primary_url_node = primary_url_node_for(base_node, purl)
193
+ base_node.xpath('mods:location/mods:url', mods: MODS_NS).each do |url_node|
194
+ if url_node == primary_url_node
195
+ url_node[:usage] = 'primary display'
196
+ elsif url_node[:usage] == 'primary display'
197
+ url_node.delete('usage')
198
+ end
199
+ end
200
+ end
201
+
202
+ def purl_nodes(base_node)
203
+ base_node.xpath('mods:location/mods:url', mods: MODS_NS).select { |url_node| ::Cocina::Models::Mapping::Purl.purl?(url_node.text) }
204
+ end
205
+
206
+ def primary_url_node_for(base_node, purl)
207
+ primary_purl_nodes, primary_url_nodes = base_node.xpath('mods:location/mods:url[@usage="primary display"]', mods: MODS_NS)
208
+ .partition { |url_node| ::Cocina::Models::Mapping::Purl.purl?(url_node.text) }
209
+ all_purl_nodes = base_node.xpath('mods:location/mods:url', mods: MODS_NS)
210
+ .select { |url_node| ::Cocina::Models::Mapping::Purl.purl?(url_node.text) }
211
+
212
+ this_purl_node = purl ? all_purl_nodes.find { |purl_node| purl_node.content == purl } : nil
213
+
214
+ primary_purl_nodes.first || primary_url_nodes.first || this_purl_node || all_purl_nodes.first
215
+ end
216
+
217
+ def normalize_related_item_other_type
218
+ ng_xml.xpath('//mods:relatedItem[@type and @otherType]', mods: MODS_NS).each do |related_node|
219
+ related_node.delete('otherType')
220
+ related_node.delete('otherTypeURI')
221
+ related_node.delete('otherTypeAuth')
222
+ end
223
+ end
224
+
225
+ def normalize_empty_notes
226
+ ng_xml.xpath('//mods:note[not(text()) and not(@xlink:href)]', mods: MODS_NS, xlink: XLINK_NS).each(&:remove)
227
+ end
228
+
229
+ def normalize_empty_type_of_resource
230
+ ng_xml.xpath('//mods:typeOfResource[not(text())][not(@*)]', mods: MODS_NS).each(&:remove)
231
+ end
232
+
233
+ def normalize_unmatched_altrepgroup
234
+ normalize_unmatched_altrepgroup_for(ng_xml.root)
235
+ ng_xml.xpath('//mods:relatedItem', mods: MODS_NS).each { |related_item_node| normalize_unmatched_altrepgroup_for(related_item_node) }
236
+ end
237
+
238
+ def normalize_unmatched_altrepgroup_for(base_node)
239
+ ids = {}
240
+ base_node.xpath('mods:*[@altRepGroup]', mods: MODS_NS).each do |node|
241
+ id = [node['altRepGroup'], node.name]
242
+ ids[id] ||= []
243
+ ids[id] << node
244
+ end
245
+
246
+ ids.each_value do |nodes|
247
+ next unless nodes.size == 1
248
+
249
+ nodes.first.delete('altRepGroup')
250
+ end
251
+ end
252
+
253
+ def normalize_unmatched_nametitlegroup
254
+ normalize_unmatched_nametitlegroup_for(ng_xml.root)
255
+ ng_xml.xpath('//mods:relatedItem', mods: MODS_NS).each { |related_item_node| normalize_unmatched_nametitlegroup_for(related_item_node) }
256
+ end
257
+
258
+ def normalize_unmatched_nametitlegroup_for(base_node)
259
+ ids = {}
260
+ base_node.xpath('mods:name[@nameTitleGroup] | mods:titleInfo[@nameTitleGroup]', mods: MODS_NS).each do |node|
261
+ id = node['nameTitleGroup']
262
+ ids[id] ||= []
263
+ ids[id] << node
264
+ end
265
+
266
+ ids.each_value do |nodes|
267
+ next unless nodes.size == 1
268
+
269
+ nodes.first.delete('nameTitleGroup')
270
+ end
271
+ end
272
+
273
+ def normalize_empty_attributes
274
+ ng_xml.xpath('//mods:*[@*=""]', mods: MODS_NS).each do |node|
275
+ node.each { |attr_name, attr_value| node.delete(attr_name) if attr_value.blank? }
276
+ end
277
+ end
278
+
279
+ def normalize_xml_space
280
+ ng_xml.xpath('//mods:*[@xml:space]', mods: MODS_NS).each do |node|
281
+ node.delete('space')
282
+ end
283
+ end
284
+
285
+ def normalize_language_term_type
286
+ ng_xml.xpath('//mods:languageTerm[not(@type)]', mods: MODS_NS).each do |node|
287
+ node['type'] = 'code'
288
+ end
289
+ end
290
+
291
+ def normalize_access_condition
292
+ ng_xml.xpath('//mods:accessCondition[@type="restrictionOnAccess"]', mods: MODS_NS).each do |node|
293
+ node['type'] = 'restriction on access'
294
+ end
295
+ ng_xml.xpath('//mods:accessCondition[@type="restrictionsOnAccess"]', mods: MODS_NS).each do |node|
296
+ node['type'] = 'restriction on access'
297
+ end
298
+ ng_xml.xpath('//mods:accessCondition[@type="useAndReproduction"]', mods: MODS_NS).each do |node|
299
+ node['type'] = 'use and reproduction'
300
+ end
301
+ end
302
+
303
+ def normalize_identifier_type_attr
304
+ ng_xml.xpath('//mods:identifier[@type]', mods: MODS_NS).each do |node|
305
+ node['type'] = normalized_identifier_type_for(node['type'])
306
+ end
307
+ ng_xml.xpath('//mods:nameIdentifier[@type]', mods: MODS_NS).each do |node|
308
+ node['type'] = normalized_identifier_type_for(node['type'])
309
+ end
310
+ ng_xml.xpath('//mods:recordIdentifier[@source]', mods: MODS_NS).each do |node|
311
+ node['source'] = normalized_identifier_type_for(node['source'])
312
+ end
313
+ end
314
+
315
+ def normalized_identifier_type_for(type)
316
+ cocina_type, _mods_type, identifier_source = Cocina::Models::Mapping::FromMods::IdentifierType.cocina_type_for_mods_type(type)
317
+
318
+ return Cocina::Models::Mapping::FromMods::IdentifierType.mods_type_for_cocina_type(cocina_type) if identifier_source
319
+
320
+ type
321
+ end
322
+
323
+ def normalize_location_physical_location
324
+ ng_xml.xpath('//mods:location', mods: MODS_NS).each do |location_node|
325
+ location_node.xpath('mods:physicalLocation|mods:url|mods:shelfLocator', mods: MODS_NS).each do |node|
326
+ next unless node.content.present? || node['xlink:href']
327
+
328
+ new_location = Nokogiri::XML::Node.new('location', ng_xml)
329
+ new_location << node
330
+ location_node.parent << new_location
331
+ end
332
+ location_node.remove
333
+ end
334
+ end
335
+
336
+ def normalize_empty_related_items
337
+ ng_xml.xpath('//mods:relatedItem/mods:part[count(mods:*)=1]/mods:detail[count(mods:*)=1]/mods:number[not(text())]',
338
+ mods: MODS_NS).each do |number_node|
339
+ number_node.parent.parent.remove
340
+ end
341
+ ng_xml.xpath('//mods:relatedItem[not(mods:*) and not(@xlink:href)]', mods: MODS_NS, xlink: XLINK_NS).each(&:remove)
342
+ end
343
+
344
+ def normalize_notes
345
+ ng_xml.xpath('//mods:note', mods: MODS_NS).each do |note_node|
346
+ if Cocina::Models::Mapping::ToMods::Note.note_type_to_abstract_type.include?(note_node['type']&.downcase) ||
347
+ Cocina::Models::Mapping::ToMods::Note.display_label_to_abstract_type.include?(note_node['displayLabel'])
348
+ note_node.delete('type') unless note_node['type']&.downcase == 'summary'
349
+ note_node.name = 'abstract'
350
+ end
351
+ if Cocina::Models::Mapping::ToMods::Note.display_label_to_abstract_type.include? note_node['displayLabel']
352
+ note_node['displayLabel'] =
353
+ note_node['displayLabel'].capitalize
354
+ end
355
+ end
356
+ end
357
+
358
+ def normalize_abstracts
359
+ ng_xml.xpath('/mods:mods/mods:abstract', mods: MODS_NS).each do |abstract_node|
360
+ if Cocina::Models::Mapping::ToMods::Note.note_type_to_abstract_type.include? abstract_node['type']&.downcase
361
+ abstract_node['type'] =
362
+ abstract_node['type'].downcase
363
+ end
364
+ if Cocina::Models::Mapping::ToMods::Note.display_label_to_abstract_type.include? abstract_node['displayLabel']
365
+ abstract_node['displayLabel'] =
366
+ abstract_node['displayLabel'].capitalize
367
+ end
368
+ abstract_node.delete('type') if abstract_node['type'] == 'abstract'
369
+ end
370
+ end
371
+
372
+ def normalize_usage_primary
373
+ normalize_usage_primary_for(ng_xml.root)
374
+ ng_xml.xpath('/mods:mods/mods:relatedItem', mods: ModsNormalizer::MODS_NS).each do |related_item_node|
375
+ normalize_usage_primary_for(related_item_node)
376
+ end
377
+ ng_xml.xpath('//mods:subject', mods: ModsNormalizer::MODS_NS).each { |subject_node| normalize_usage_primary_for(subject_node) }
378
+ end
379
+
380
+ def normalize_usage_primary_for(base_node)
381
+ %w[genre language classification subject titleInfo typeOfResource name].each do |node_name|
382
+ primary_nodes = base_node.xpath("mods:#{node_name}[@usage=\"primary\"]", mods: MODS_NS)
383
+ next if primary_nodes.size < 2
384
+
385
+ primary_nodes[1..].each { |primary_node| primary_node.delete('usage') }
386
+ end
387
+ end
388
+
389
+ def normalize_related_item_attributes
390
+ ng_xml.xpath('/mods:mods/mods:relatedItem[@lang or @script]', mods: ModsNormalizer::MODS_NS).each do |related_item_node|
391
+ related_item_node.delete('lang')
392
+ related_item_node.delete('script')
393
+ end
394
+ end
395
+
396
+ def blank_ng_xml
397
+ Nokogiri::XML(<<~XML
398
+ <mods xmlns="http://www.loc.gov/mods/v3"#{' '}
399
+ xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"#{' '}
400
+ version="3.6"#{' '}
401
+ xsi:schemaLocation="http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-3-6.xsd" />
402
+ XML
403
+ )
404
+ end
405
+ end
406
+ end
407
+ end
408
+ end
409
+ end
@@ -0,0 +1,27 @@
1
+ # frozen_string_literal: true
2
+
3
+ module Cocina
4
+ module Models
5
+ module Mapping
6
+ # Utility methods for generating purl links
7
+ class Purl
8
+ class_attribute :base_url, default: 'https://purl.stanford.edu'
9
+
10
+ def self.for(druid:)
11
+ return nil if druid.nil?
12
+
13
+ "#{base_url}/#{druid.delete_prefix('druid:')}"
14
+ end
15
+
16
+ def self.purl?(node)
17
+ node.start_with?("https://#{host}") || node.start_with?("http://#{host}")
18
+ end
19
+
20
+ # the purl without the protocol part
21
+ def self.host
22
+ @host ||= base_url.sub(%r{^https?://}, '')
23
+ end
24
+ end
25
+ end
26
+ end
27
+ end
@@ -0,0 +1,155 @@
1
+ # frozen_string_literal: true
2
+
3
+ module Cocina
4
+ module Models
5
+ module Mapping
6
+ module ToMods
7
+ # Maps the Access subschema from cocina to MODS XML
8
+ class Access # rubocop:disable Metrics/ClassLength
9
+ # @params [Nokogiri::XML::Builder] xml
10
+ # @params [Cocina::Models::Access] access
11
+ # @params [string] purl
12
+ def self.write(xml:, access:, purl:)
13
+ new(xml: xml, access: access, purl: purl).write
14
+ end
15
+
16
+ def initialize(xml:, access:, purl:)
17
+ @xml = xml
18
+ @access = access
19
+ @purl = purl
20
+ end
21
+
22
+ def write
23
+ write_purl unless purl.nil?
24
+ return if access.nil?
25
+
26
+ write_access_conditions if access
27
+
28
+ Array(access.url).each do |url|
29
+ xml.location do
30
+ write_url(url)
31
+ end
32
+ end
33
+
34
+ write_physical_locations
35
+ write_digital_locations
36
+ write_shelf_locators
37
+ write_access_contact_locations
38
+ end
39
+
40
+ private
41
+
42
+ attr_reader :xml, :access, :purl
43
+
44
+ def write_physical_locations
45
+ Array(access.physicalLocation).reject do |physical_location|
46
+ shelf_locator?(physical_location)
47
+ end.each do |physical_location|
48
+ xml.location do
49
+ xml.physicalLocation physical_location.value || physical_location.code,
50
+ descriptive_attrs(physical_location)
51
+ end
52
+ end
53
+ end
54
+
55
+ def write_digital_locations
56
+ Array(access.digitalLocation).select do |digital_location|
57
+ digital_location.type == 'discovery'
58
+ end.each do |digital_location|
59
+ xml.location do
60
+ xml.physicalLocation digital_location.value || digital_location.code,
61
+ descriptive_attrs(digital_location)
62
+ end
63
+ end
64
+ end
65
+
66
+ def write_access_contact_locations
67
+ Array(access.accessContact).each do |access_contact|
68
+ if access_contact.type == 'email'
69
+ xml.note access_contact.value, descriptive_attrs(access_contact).merge({ type: 'contact' })
70
+ else
71
+ xml.location do
72
+ xml.physicalLocation access_contact.value || access_contact.code,
73
+ { type: 'repository' }.merge(descriptive_attrs(access_contact))
74
+ end
75
+ end
76
+ end
77
+ end
78
+
79
+ def write_shelf_locators
80
+ Array(access.physicalLocation).select do |physical_location|
81
+ shelf_locator?(physical_location)
82
+ end.each do |physical_location|
83
+ xml.location do
84
+ xml.shelfLocator physical_location.value
85
+ end
86
+ end
87
+ end
88
+
89
+ def write_url(url)
90
+ url_attrs = {}.tap do |attrs|
91
+ attrs[:usage] = 'primary display' if url.status == 'primary'
92
+ attrs[:displayLabel] = url.displayLabel
93
+ attrs[:note] = url.note.first.value if url.note.present?
94
+ end.compact
95
+ xml.url url.value, url_attrs
96
+ end
97
+
98
+ def primary_url_is_not_purl?
99
+ Array(access&.url).any? { |url| url.status == 'primary' }
100
+ end
101
+
102
+ def write_purl
103
+ purl_attrs = {}.tap do |attrs|
104
+ attrs[:note] = find_note_value(nil)
105
+ attrs[:usage] = 'primary display' unless primary_url_is_not_purl?
106
+ attrs[:displayLabel] = find_note_value('display label')
107
+ end.compact
108
+
109
+ xml.location do
110
+ xml.url purl, purl_attrs
111
+ end
112
+ end
113
+
114
+ def find_note_value(note_type)
115
+ Array(access&.note).find do |note|
116
+ note.type == note_type && purl_note?(note)
117
+ end&.value
118
+ end
119
+
120
+ def descriptive_attrs(cocina)
121
+ {
122
+ valueURI: cocina.uri,
123
+ authorityURI: cocina.source&.uri,
124
+ authority: cocina.source&.code,
125
+ script: cocina.valueLanguage&.valueScript&.code,
126
+ lang: cocina.valueLanguage&.code,
127
+ type: cocina.type,
128
+ displayLabel: cocina.displayLabel,
129
+ 'xlink:href' => cocina.valueAt
130
+ }.compact
131
+ end
132
+
133
+ def shelf_locator?(physical_location)
134
+ physical_location.type == 'shelf locator'
135
+ end
136
+
137
+ def write_access_conditions
138
+ Array(access.note).reject { |note| purl_note?(note) }.each do |note|
139
+ attributes = {
140
+ type: note.type == 'access restriction' ? 'restriction on access' : note.type,
141
+ displayLabel: note.displayLabel,
142
+ 'xlink:href' => note.valueAt
143
+ }.compact
144
+ xml.accessCondition note.value, attributes
145
+ end
146
+ end
147
+
148
+ def purl_note?(note)
149
+ Array(note.appliesTo).any? { |applies_to| applies_to.value == 'purl' }
150
+ end
151
+ end
152
+ end
153
+ end
154
+ end
155
+ end