cocina_display 0.4.0 → 0.6.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/README.md +21 -16
- data/lib/cocina_display/cocina_record.rb +33 -72
- data/lib/cocina_display/concerns/access.rb +71 -0
- data/lib/cocina_display/concerns/contributors.rb +60 -41
- data/lib/cocina_display/concerns/events.rb +51 -25
- data/lib/cocina_display/concerns/forms.rb +134 -0
- data/lib/cocina_display/concerns/identifiers.rb +11 -3
- data/lib/cocina_display/concerns/subjects.rb +91 -0
- data/lib/cocina_display/contributor.rb +66 -11
- data/lib/cocina_display/dates/date.rb +9 -8
- data/lib/cocina_display/dates/date_range.rb +8 -0
- data/lib/cocina_display/events/event.rb +78 -0
- data/lib/cocina_display/events/imprint.rb +101 -0
- data/lib/cocina_display/events/location.rb +56 -0
- data/lib/cocina_display/marc_relator_codes.rb +314 -0
- data/lib/cocina_display/subject.rb +127 -0
- data/lib/cocina_display/title_builder.rb +2 -1
- data/lib/cocina_display/utils.rb +45 -6
- data/lib/cocina_display/version.rb +1 -1
- data/script/find_records.rb +85 -0
- metadata +27 -11
- data/lib/cocina_display/imprint.rb +0 -123
|
@@ -0,0 +1,56 @@
|
|
|
1
|
+
require_relative "../marc_country_codes"
|
|
2
|
+
|
|
3
|
+
module CocinaDisplay
|
|
4
|
+
module Events
|
|
5
|
+
# A single location represented in a Cocina event, like a publication place.
|
|
6
|
+
class Location
|
|
7
|
+
attr_reader :cocina
|
|
8
|
+
|
|
9
|
+
# Initialize a Location object with Cocina structured data.
|
|
10
|
+
# @param cocina [Hash] The Cocina structured data for the location.
|
|
11
|
+
def initialize(cocina)
|
|
12
|
+
@cocina = cocina
|
|
13
|
+
end
|
|
14
|
+
|
|
15
|
+
# The name of the location.
|
|
16
|
+
# Decodes a MARC country code if present and no value was present.
|
|
17
|
+
# @return [String, nil]
|
|
18
|
+
def display_str
|
|
19
|
+
cocina["value"] || decoded_country
|
|
20
|
+
end
|
|
21
|
+
|
|
22
|
+
# Is there an unencoded value (name) for this location?
|
|
23
|
+
# @return [Boolean]
|
|
24
|
+
def unencoded_value?
|
|
25
|
+
cocina["value"].present?
|
|
26
|
+
end
|
|
27
|
+
|
|
28
|
+
private
|
|
29
|
+
|
|
30
|
+
# A code, like a MARC country code, representing the location.
|
|
31
|
+
# @return [String, nil]
|
|
32
|
+
def code
|
|
33
|
+
cocina["code"]
|
|
34
|
+
end
|
|
35
|
+
|
|
36
|
+
# Decoded country name if the location is encoded with a MARC country code.
|
|
37
|
+
# @return [String, nil]
|
|
38
|
+
def decoded_country
|
|
39
|
+
MARC_COUNTRY[code] if marc_country? && valid_country_code?
|
|
40
|
+
end
|
|
41
|
+
|
|
42
|
+
# Is this a decodable country code?
|
|
43
|
+
# Excludes blank values and "xx" (unknown) and "vp" (various places).
|
|
44
|
+
# @return [Boolean]
|
|
45
|
+
def valid_country_code?
|
|
46
|
+
code.present? && ["xx", "vp"].exclude?(code)
|
|
47
|
+
end
|
|
48
|
+
|
|
49
|
+
# Is this location encoded with a MARC country code?
|
|
50
|
+
# @return [Boolean]
|
|
51
|
+
def marc_country?
|
|
52
|
+
cocina.dig("source", "code") == "marccountry"
|
|
53
|
+
end
|
|
54
|
+
end
|
|
55
|
+
end
|
|
56
|
+
end
|
|
@@ -0,0 +1,314 @@
|
|
|
1
|
+
# Map of MARC relator codes to human-readable terms.
|
|
2
|
+
# https://www.loc.gov/marc/relators/relaterm.html
|
|
3
|
+
|
|
4
|
+
module CocinaDisplay
|
|
5
|
+
MARC_RELATOR = {
|
|
6
|
+
"abr" => "abridger",
|
|
7
|
+
"acp" => "art copyist",
|
|
8
|
+
"act" => "actor",
|
|
9
|
+
"adi" => "art director",
|
|
10
|
+
"adp" => "adapter",
|
|
11
|
+
"aft" => "author of afterword, colophon, etc.", # discontinued
|
|
12
|
+
"anc" => "announcer",
|
|
13
|
+
"anl" => "analyst",
|
|
14
|
+
"anm" => "animator",
|
|
15
|
+
"ann" => "annotator",
|
|
16
|
+
"ant" => "bibliographic antecedent",
|
|
17
|
+
"ape" => "appellee",
|
|
18
|
+
"apl" => "appellant",
|
|
19
|
+
"app" => "applicant",
|
|
20
|
+
"aqt" => "author in quotations or text abstracts",
|
|
21
|
+
"arc" => "architect",
|
|
22
|
+
"ard" => "artistic director",
|
|
23
|
+
"arr" => "arranger",
|
|
24
|
+
"art" => "artist",
|
|
25
|
+
"asg" => "assignee",
|
|
26
|
+
"asn" => "associated name",
|
|
27
|
+
"ato" => "autographer",
|
|
28
|
+
"att" => "attributed name",
|
|
29
|
+
"auc" => "auctioneer",
|
|
30
|
+
"aud" => "author of dialog",
|
|
31
|
+
"aue" => "audio engineer",
|
|
32
|
+
"aui" => "author of introduction, etc.", # discontinued
|
|
33
|
+
"aup" => "audio producer",
|
|
34
|
+
"aus" => "screenwriter",
|
|
35
|
+
"aut" => "author",
|
|
36
|
+
"bdd" => "binding designer",
|
|
37
|
+
"bjd" => "bookjacket designer",
|
|
38
|
+
"bka" => "book artist",
|
|
39
|
+
"bkd" => "book designer",
|
|
40
|
+
"bkp" => "book producer",
|
|
41
|
+
"blw" => "blurb writer",
|
|
42
|
+
"bnd" => "binder",
|
|
43
|
+
"bpd" => "bookplate designer",
|
|
44
|
+
"brd" => "broadcaster",
|
|
45
|
+
"brl" => "braille embosser",
|
|
46
|
+
"bsl" => "bookseller",
|
|
47
|
+
"cad" => "casting director",
|
|
48
|
+
"cas" => "caster",
|
|
49
|
+
"ccp" => "conceptor",
|
|
50
|
+
"chr" => "choreographer",
|
|
51
|
+
"clb" => "collaborator", # discontinued
|
|
52
|
+
"cli" => "client",
|
|
53
|
+
"cll" => "calligrapher",
|
|
54
|
+
"clr" => "colorist",
|
|
55
|
+
"clt" => "collotyper",
|
|
56
|
+
"cmm" => "commentator",
|
|
57
|
+
"cmp" => "composer",
|
|
58
|
+
"cmt" => "compositor",
|
|
59
|
+
"cnd" => "conductor",
|
|
60
|
+
"cng" => "cinematographer",
|
|
61
|
+
"cns" => "censor",
|
|
62
|
+
"coe" => "contestant-appellee",
|
|
63
|
+
"col" => "collector",
|
|
64
|
+
"com" => "compiler",
|
|
65
|
+
"con" => "conservator",
|
|
66
|
+
"cop" => "camera operator",
|
|
67
|
+
"cor" => "collection registrar",
|
|
68
|
+
"cos" => "contestant",
|
|
69
|
+
"cot" => "contestant-appellant",
|
|
70
|
+
"cou" => "court governed",
|
|
71
|
+
"cov" => "cover designer",
|
|
72
|
+
"cpc" => "copyright claimant",
|
|
73
|
+
"cpe" => "complainant-appellee",
|
|
74
|
+
"cph" => "copyright holder",
|
|
75
|
+
"cpl" => "complainant",
|
|
76
|
+
"cpt" => "complainant-appellant",
|
|
77
|
+
"cre" => "creator",
|
|
78
|
+
"crp" => "correspondent",
|
|
79
|
+
"crr" => "corrector",
|
|
80
|
+
"crt" => "court reporter",
|
|
81
|
+
"csl" => "consultant",
|
|
82
|
+
"csp" => "consultant to a project",
|
|
83
|
+
"cst" => "costume designer",
|
|
84
|
+
"ctb" => "contributor",
|
|
85
|
+
"cte" => "contestee-appellee",
|
|
86
|
+
"ctg" => "cartographer",
|
|
87
|
+
"ctr" => "contractor",
|
|
88
|
+
"cts" => "contestee",
|
|
89
|
+
"ctt" => "contestee-appellant",
|
|
90
|
+
"cur" => "curator",
|
|
91
|
+
"cwt" => "commentator for written text",
|
|
92
|
+
"dbd" => "dubbing director",
|
|
93
|
+
"dbp" => "distribution place",
|
|
94
|
+
"dfd" => "defendant",
|
|
95
|
+
"dfe" => "defendant-appellee",
|
|
96
|
+
"dft" => "defendant-appellant",
|
|
97
|
+
"dgc" => "degree committee member",
|
|
98
|
+
"dgg" => "degree granting institution",
|
|
99
|
+
"dgs" => "degree supervisor",
|
|
100
|
+
"dis" => "dissertant",
|
|
101
|
+
"djo" => "dj",
|
|
102
|
+
"dln" => "delineator",
|
|
103
|
+
"dnc" => "dancer",
|
|
104
|
+
"dnr" => "donor",
|
|
105
|
+
"dpc" => "depicted",
|
|
106
|
+
"dpt" => "depositor",
|
|
107
|
+
"drm" => "draftsman",
|
|
108
|
+
"drt" => "director",
|
|
109
|
+
"dsr" => "designer",
|
|
110
|
+
"dst" => "distributor",
|
|
111
|
+
"dtc" => "data contributor",
|
|
112
|
+
"dte" => "dedicatee",
|
|
113
|
+
"dtm" => "data manager",
|
|
114
|
+
"dto" => "dedicator",
|
|
115
|
+
"dub" => "dubious author",
|
|
116
|
+
"edc" => "editor of compilation",
|
|
117
|
+
"edd" => "editorial director",
|
|
118
|
+
"edm" => "editor of moving image work",
|
|
119
|
+
"edt" => "editor",
|
|
120
|
+
"egr" => "engraver",
|
|
121
|
+
"elg" => "electrician",
|
|
122
|
+
"elt" => "electrotyper",
|
|
123
|
+
"eng" => "engineer",
|
|
124
|
+
"enj" => "enacting jurisdiction",
|
|
125
|
+
"etr" => "etcher",
|
|
126
|
+
"evp" => "event place",
|
|
127
|
+
"exp" => "expert",
|
|
128
|
+
"fac" => "facsimilist",
|
|
129
|
+
"fds" => "film distributor",
|
|
130
|
+
"fld" => "field director",
|
|
131
|
+
"flm" => "film editor",
|
|
132
|
+
"fmd" => "film director",
|
|
133
|
+
"fmk" => "filmmaker",
|
|
134
|
+
"fmo" => "former owner",
|
|
135
|
+
"fmp" => "film producer",
|
|
136
|
+
"fnd" => "funder",
|
|
137
|
+
"fon" => "founder",
|
|
138
|
+
"fpy" => "first party",
|
|
139
|
+
"frg" => "forger",
|
|
140
|
+
"gdv" => "game developer",
|
|
141
|
+
"gis" => "geographic information specialist",
|
|
142
|
+
"grt" => "graphic technician", # discontinued
|
|
143
|
+
"his" => "host institution",
|
|
144
|
+
"hnr" => "honoree",
|
|
145
|
+
"hst" => "host",
|
|
146
|
+
"ill" => "illustrator",
|
|
147
|
+
"ilu" => "illuminator",
|
|
148
|
+
"ink" => "inker",
|
|
149
|
+
"ins" => "inscriber",
|
|
150
|
+
"inv" => "inventor",
|
|
151
|
+
"isb" => "issuing body",
|
|
152
|
+
"itr" => "instrumentalist",
|
|
153
|
+
"ive" => "interviewee",
|
|
154
|
+
"ivr" => "interviewer",
|
|
155
|
+
"jud" => "judge",
|
|
156
|
+
"jug" => "jurisdiction governed",
|
|
157
|
+
"lbr" => "laboratory",
|
|
158
|
+
"lbt" => "librettist",
|
|
159
|
+
"ldr" => "laboratory director",
|
|
160
|
+
"led" => "lead",
|
|
161
|
+
"lee" => "libelee-appellee",
|
|
162
|
+
"lel" => "libelee",
|
|
163
|
+
"len" => "lender",
|
|
164
|
+
"let" => "libelee-appellant",
|
|
165
|
+
"lgd" => "lighting designer",
|
|
166
|
+
"lie" => "libelant-appellee",
|
|
167
|
+
"lil" => "libelant",
|
|
168
|
+
"lit" => "libelant-appellant",
|
|
169
|
+
"lsa" => "landscape architect",
|
|
170
|
+
"lse" => "licensee",
|
|
171
|
+
"lso" => "licensor",
|
|
172
|
+
"ltg" => "lithographer",
|
|
173
|
+
"ltr" => "letterer",
|
|
174
|
+
"lyr" => "lyricist",
|
|
175
|
+
"mcp" => "music copyist",
|
|
176
|
+
"mdc" => "metadata contact",
|
|
177
|
+
"med" => "medium",
|
|
178
|
+
"mfp" => "manufacture place",
|
|
179
|
+
"mfr" => "manufacturer",
|
|
180
|
+
"mka" => "makeup artist",
|
|
181
|
+
"mod" => "moderator",
|
|
182
|
+
"mon" => "monitor",
|
|
183
|
+
"mrb" => "marbler",
|
|
184
|
+
"mrk" => "markup editor",
|
|
185
|
+
"msd" => "musical director",
|
|
186
|
+
"mte" => "metal-engraver",
|
|
187
|
+
"mtk" => "minute taker",
|
|
188
|
+
"mup" => "music programmer",
|
|
189
|
+
"mus" => "musician",
|
|
190
|
+
"mxe" => "mixing engineer",
|
|
191
|
+
"nan" => "news anchor",
|
|
192
|
+
"nrt" => "narrator",
|
|
193
|
+
"onp" => "onscreen participant",
|
|
194
|
+
"opn" => "opponent",
|
|
195
|
+
"org" => "originator",
|
|
196
|
+
"orm" => "organizer",
|
|
197
|
+
"osp" => "onscreen presenter",
|
|
198
|
+
"oth" => "other",
|
|
199
|
+
"own" => "owner",
|
|
200
|
+
"pad" => "place of address",
|
|
201
|
+
"pan" => "panelist",
|
|
202
|
+
"pat" => "patron",
|
|
203
|
+
"pbd" => "publishing director",
|
|
204
|
+
"pbl" => "publisher",
|
|
205
|
+
"pdr" => "project director",
|
|
206
|
+
"pfr" => "proofreader",
|
|
207
|
+
"pht" => "photographer",
|
|
208
|
+
"plt" => "platemaker",
|
|
209
|
+
"pma" => "permitting agency",
|
|
210
|
+
"pmn" => "production manager",
|
|
211
|
+
"pnc" => "penciller",
|
|
212
|
+
"pop" => "printer of plates",
|
|
213
|
+
"ppm" => "papermaker",
|
|
214
|
+
"ppt" => "puppeteer",
|
|
215
|
+
"pra" => "praeses",
|
|
216
|
+
"prc" => "process contact",
|
|
217
|
+
"prd" => "production personnel",
|
|
218
|
+
"pre" => "presenter",
|
|
219
|
+
"prf" => "performer",
|
|
220
|
+
"prg" => "programmer",
|
|
221
|
+
"prm" => "printmaker",
|
|
222
|
+
"prn" => "production company",
|
|
223
|
+
"pro" => "producer",
|
|
224
|
+
"prp" => "production place",
|
|
225
|
+
"prs" => "production designer",
|
|
226
|
+
"prt" => "printer",
|
|
227
|
+
"prv" => "provider",
|
|
228
|
+
"pta" => "patent applicant",
|
|
229
|
+
"pte" => "plaintiff-appellee",
|
|
230
|
+
"ptf" => "plaintiff",
|
|
231
|
+
"pth" => "patent holder",
|
|
232
|
+
"ptt" => "plaintiff-appellant",
|
|
233
|
+
"pup" => "publication place",
|
|
234
|
+
"rap" => "rapporteur",
|
|
235
|
+
"rbr" => "rubricator",
|
|
236
|
+
"rcd" => "recordist",
|
|
237
|
+
"rce" => "recording engineer",
|
|
238
|
+
"rcp" => "addressee",
|
|
239
|
+
"rdd" => "radio director",
|
|
240
|
+
"red" => "redaktor",
|
|
241
|
+
"ren" => "renderer",
|
|
242
|
+
"res" => "researcher",
|
|
243
|
+
"rev" => "reviewer",
|
|
244
|
+
"rpc" => "radio producer",
|
|
245
|
+
"rps" => "repository",
|
|
246
|
+
"rpt" => "reporter",
|
|
247
|
+
"rpy" => "responsible party",
|
|
248
|
+
"rse" => "respondent-appellee",
|
|
249
|
+
"rsg" => "restager",
|
|
250
|
+
"rsp" => "respondent",
|
|
251
|
+
"rsr" => "restorationist",
|
|
252
|
+
"rst" => "respondent-appellant",
|
|
253
|
+
"rth" => "research team head",
|
|
254
|
+
"rtm" => "research team member",
|
|
255
|
+
"rxa" => "remix artist",
|
|
256
|
+
"sad" => "scientific advisor",
|
|
257
|
+
"sce" => "scenarist",
|
|
258
|
+
"scl" => "sculptor",
|
|
259
|
+
"scr" => "scribe",
|
|
260
|
+
"sde" => "sound engineer",
|
|
261
|
+
"sds" => "sound designer",
|
|
262
|
+
"sec" => "secretary",
|
|
263
|
+
"sfx" => "special effects provider",
|
|
264
|
+
"sgd" => "stage director",
|
|
265
|
+
"sgn" => "signer",
|
|
266
|
+
"sht" => "spporting host",
|
|
267
|
+
"sll" => "seller",
|
|
268
|
+
"sng" => "singer",
|
|
269
|
+
"spk" => "speaker",
|
|
270
|
+
"spn" => "sponsor",
|
|
271
|
+
"spy" => "second party",
|
|
272
|
+
"srv" => "surveyor",
|
|
273
|
+
"std" => "set designer",
|
|
274
|
+
"stg" => "setting",
|
|
275
|
+
"stl" => "storyteller",
|
|
276
|
+
"stm" => "stage manager",
|
|
277
|
+
"stn" => "standards body",
|
|
278
|
+
"str" => "stereotyper",
|
|
279
|
+
"swd" => "software developer",
|
|
280
|
+
"tad" => "technical advisor",
|
|
281
|
+
"tau" => "television writer",
|
|
282
|
+
"tcd" => "technical director",
|
|
283
|
+
"tch" => "teacher",
|
|
284
|
+
"ths" => "thesis advisor",
|
|
285
|
+
"tld" => "television director",
|
|
286
|
+
"tlg" => "television guest",
|
|
287
|
+
"tlh" => "television host",
|
|
288
|
+
"tlp" => "television producer",
|
|
289
|
+
"trc" => "transcriber",
|
|
290
|
+
"trl" => "translator",
|
|
291
|
+
"tyd" => "type designer",
|
|
292
|
+
"tyg" => "typographer",
|
|
293
|
+
"uvp" => "university place",
|
|
294
|
+
"vac" => "voice actor",
|
|
295
|
+
"vdg" => "videographer",
|
|
296
|
+
"vfx" => "visual effects provider",
|
|
297
|
+
"voc" => "vocalist", # discontinued
|
|
298
|
+
"wac" => "writer of added commentary",
|
|
299
|
+
"wal" => "writer of added lyrics",
|
|
300
|
+
"wam" => "writer of accompanying material",
|
|
301
|
+
"wat" => "writer of added text",
|
|
302
|
+
"waw" => "writer of afterword",
|
|
303
|
+
"wdc" => "woodcutter",
|
|
304
|
+
"wde" => "wood engraver",
|
|
305
|
+
"wfs" => "writer of film story",
|
|
306
|
+
"wft" => "writer of intertitles",
|
|
307
|
+
"wfw" => "writer of foreword",
|
|
308
|
+
"win" => "writer of introduction",
|
|
309
|
+
"wit" => "witness",
|
|
310
|
+
"wpr" => "writer of preface",
|
|
311
|
+
"wst" => "writer of supplementary textual content",
|
|
312
|
+
"wts" => "writer of television story"
|
|
313
|
+
}
|
|
314
|
+
end
|
|
@@ -0,0 +1,127 @@
|
|
|
1
|
+
require_relative "utils"
|
|
2
|
+
require_relative "contributor"
|
|
3
|
+
require_relative "title_builder"
|
|
4
|
+
require_relative "dates/date"
|
|
5
|
+
|
|
6
|
+
module CocinaDisplay
|
|
7
|
+
# Base class for subjects in Cocina structured data.
|
|
8
|
+
class Subject
|
|
9
|
+
attr_reader :cocina
|
|
10
|
+
|
|
11
|
+
# Extract the type of the subject from the Cocina structured data.
|
|
12
|
+
# If no top-level type, uses the first structuredValue type.
|
|
13
|
+
# @param cocina [Hash] The Cocina structured data for the subject.
|
|
14
|
+
# @return [String, nil] The type of the subject, or nil if none
|
|
15
|
+
# @see https://github.com/sul-dlss/cocina-models/blob/main/docs/description_types.md#subject-types
|
|
16
|
+
def self.detect_type(cocina)
|
|
17
|
+
cocina["type"] || Utils.flatten_nested_values(cocina).pick("type")
|
|
18
|
+
end
|
|
19
|
+
|
|
20
|
+
# Choose and create the appropriate Subject subclass based on type.
|
|
21
|
+
# @param cocina [Hash] The Cocina structured data for the subject.
|
|
22
|
+
# @return [Subject]
|
|
23
|
+
# @see detect_type
|
|
24
|
+
def self.from_cocina(cocina)
|
|
25
|
+
case detect_type(cocina)
|
|
26
|
+
when "person", "family", "organization", "conference", "event", "name"
|
|
27
|
+
NameSubject.new(cocina)
|
|
28
|
+
when "title"
|
|
29
|
+
TitleSubject.new(cocina)
|
|
30
|
+
when "time"
|
|
31
|
+
TemporalSubject.new(cocina)
|
|
32
|
+
# TODO: special handling for geospatial subjects
|
|
33
|
+
# when "map coordinates", "bounding box coordinates", "point coordinates"
|
|
34
|
+
else
|
|
35
|
+
Subject.new(cocina)
|
|
36
|
+
end
|
|
37
|
+
end
|
|
38
|
+
|
|
39
|
+
# Initialize a Subject object with Cocina structured data.
|
|
40
|
+
# @param cocina [Hash] The Cocina structured data for the subject.
|
|
41
|
+
def initialize(cocina)
|
|
42
|
+
@cocina = cocina
|
|
43
|
+
end
|
|
44
|
+
|
|
45
|
+
# The type of the subject.
|
|
46
|
+
# If no top-level type, uses the first structuredValue type.
|
|
47
|
+
# @return [String, nil] The type of the subject, or nil if none
|
|
48
|
+
# @see detect_type
|
|
49
|
+
def type
|
|
50
|
+
self.class.detect_type(cocina)
|
|
51
|
+
end
|
|
52
|
+
|
|
53
|
+
# A string representation of the subject, formatted for display.
|
|
54
|
+
# Concatenates any structured values with an appropriate delimiter.
|
|
55
|
+
# Subclasses may override this for more specific formatting.
|
|
56
|
+
# @return [String]
|
|
57
|
+
def display_str
|
|
58
|
+
Utils.compact_and_join(descriptive_values, delimiter: delimiter)
|
|
59
|
+
end
|
|
60
|
+
|
|
61
|
+
private
|
|
62
|
+
|
|
63
|
+
# Flatten any structured values into an array of Hashes with "value" keys.
|
|
64
|
+
# If no structured values, will return the top-level cocina data.
|
|
65
|
+
# @see Utils.flatten_nested_values
|
|
66
|
+
# @return [Array<Hash>] An array of Hashes representing all values.
|
|
67
|
+
def descriptive_values
|
|
68
|
+
Utils.flatten_nested_values(cocina).pluck("value")
|
|
69
|
+
end
|
|
70
|
+
|
|
71
|
+
# Delimiter to use for joining structured subject values.
|
|
72
|
+
# LCSH uses a comma (the default); catalog headings use " > ".
|
|
73
|
+
# @return [String]
|
|
74
|
+
def delimiter
|
|
75
|
+
if cocina["displayLabel"]&.downcase == "catalog heading"
|
|
76
|
+
" > "
|
|
77
|
+
else
|
|
78
|
+
", "
|
|
79
|
+
end
|
|
80
|
+
end
|
|
81
|
+
end
|
|
82
|
+
|
|
83
|
+
# A subject representing a named entity.
|
|
84
|
+
class NameSubject < Subject
|
|
85
|
+
attr_reader :name
|
|
86
|
+
|
|
87
|
+
# Initialize a NameSubject object with Cocina structured data.
|
|
88
|
+
# @param cocina [Hash] The Cocina structured data for the subject.
|
|
89
|
+
def initialize(cocina)
|
|
90
|
+
super
|
|
91
|
+
@name = Contributor::Name.new(cocina)
|
|
92
|
+
end
|
|
93
|
+
|
|
94
|
+
# Use the contributor name formatting rules for display.
|
|
95
|
+
# @return [String] The formatted name string, including life dates
|
|
96
|
+
# @see CocinaDisplay::Contributor::Name#display_str
|
|
97
|
+
def display_str
|
|
98
|
+
@name.display_str(with_date: true)
|
|
99
|
+
end
|
|
100
|
+
end
|
|
101
|
+
|
|
102
|
+
# A subject representing an entity with a title.
|
|
103
|
+
class TitleSubject < Subject
|
|
104
|
+
# Construct a title string to use for display.
|
|
105
|
+
# @see CocinaDisplay::TitleBuilder.build
|
|
106
|
+
# @note Unclear how often structured title subjects occur "in the wild".
|
|
107
|
+
# @return [String]
|
|
108
|
+
def display_str
|
|
109
|
+
TitleBuilder.build([cocina])
|
|
110
|
+
end
|
|
111
|
+
end
|
|
112
|
+
|
|
113
|
+
# A subject representing a date and/or time.
|
|
114
|
+
class TemporalSubject < Subject
|
|
115
|
+
attr_reader :date
|
|
116
|
+
|
|
117
|
+
def initialize(cocina)
|
|
118
|
+
super
|
|
119
|
+
@date = Dates::Date.from_cocina(cocina)
|
|
120
|
+
end
|
|
121
|
+
|
|
122
|
+
# @return [String] The formatted date/time string for display
|
|
123
|
+
def display_str
|
|
124
|
+
@date.qualified_value
|
|
125
|
+
end
|
|
126
|
+
end
|
|
127
|
+
end
|
|
@@ -53,7 +53,7 @@ module CocinaDisplay
|
|
|
53
53
|
def self.sort_title(titles, catalog_links: [])
|
|
54
54
|
part_label = catalog_links.find { |link| link["catalog"] == "folio" }&.fetch("partLabel", nil)
|
|
55
55
|
[new(strategy: :first, add_punctuation: false, only_one_parallel_value: false, part_label: part_label, sortable: true).build(titles)]
|
|
56
|
-
.flatten.compact.map { |title| title.gsub(/[[:punct:]]*/, "").strip }
|
|
56
|
+
.flatten.compact.map { |title| title.gsub(/[[:punct:]]*/, "").squeeze(" ").strip }
|
|
57
57
|
end
|
|
58
58
|
|
|
59
59
|
# @param strategy [Symbol] ":first" selects a single title value based on precedence of
|
|
@@ -116,6 +116,7 @@ module CocinaDisplay
|
|
|
116
116
|
end
|
|
117
117
|
|
|
118
118
|
def extract_title(cocina_title)
|
|
119
|
+
return if cocina_title.blank?
|
|
119
120
|
title_values = if cocina_title["value"]
|
|
120
121
|
cocina_title["value"]
|
|
121
122
|
elsif cocina_title["structuredValue"].present?
|
data/lib/cocina_display/utils.rb
CHANGED
|
@@ -19,15 +19,54 @@ module CocinaDisplay
|
|
|
19
19
|
end.delete_suffix(delimiter)
|
|
20
20
|
end
|
|
21
21
|
|
|
22
|
-
# Recursively flatten structured values in Cocina metadata.
|
|
23
|
-
# Returns a list of hashes representing the "leaf" nodes with
|
|
22
|
+
# Recursively flatten structured, parallel, and grouped values in Cocina metadata.
|
|
23
|
+
# Returns a list of hashes representing the "leaf" nodes with +value+ key.
|
|
24
24
|
# @return [Array<Hash>] List of node hashes with "value" present
|
|
25
|
-
|
|
25
|
+
# @param cocina [Hash] The Cocina structured data to flatten
|
|
26
|
+
# @param output [Array] Used for recursion, should be empty on first call
|
|
27
|
+
# @example simple value
|
|
28
|
+
# cocina = { "value" => "John Doe", "type" => "name" }
|
|
29
|
+
# Utils.flatten_nested_values(cocina)
|
|
30
|
+
# #=> [{"value" => "John Doe", "type" => "name"}]
|
|
31
|
+
# @example structured values
|
|
32
|
+
# cocina = { "structuredValue" => [{"value" => "foo"}, {"value" => "bar"}] }
|
|
33
|
+
# Utils.flatten_nested_values(cocina)
|
|
34
|
+
# #=> [{"value" => "foo"}, {"value" => "bar"}]
|
|
35
|
+
# @example parallel structured and simple values
|
|
36
|
+
# cocina = { "parallelValue" => [{"value" => "foo" }, { "structuredValue" => [{"value" => "bar"}, {"value" => "baz"}] }] }
|
|
37
|
+
# Utils.flatten_nested_values(cocina)
|
|
38
|
+
# #=> [{"value" => "foo"}, {"value" => "foo"}, {"value" => "baz"}]
|
|
39
|
+
def self.flatten_nested_values(cocina, output = [])
|
|
26
40
|
return [cocina] if cocina["value"].present?
|
|
27
|
-
return cocina.flat_map { |node|
|
|
28
|
-
return output unless (structured_values = Array(cocina["structuredValue"])).present?
|
|
41
|
+
return cocina.flat_map { |node| flatten_nested_values(node, output) } if cocina.is_a?(Array)
|
|
29
42
|
|
|
30
|
-
|
|
43
|
+
nested_values = Array(cocina["structuredValue"]) + Array(cocina["parallelValue"]) + Array(cocina["groupedValue"])
|
|
44
|
+
return output unless nested_values.any?
|
|
45
|
+
|
|
46
|
+
nested_values.flat_map { |node| flatten_nested_values(node, output) }
|
|
47
|
+
end
|
|
48
|
+
|
|
49
|
+
# Recursively remove empty values from a hash, including nested hashes and arrays.
|
|
50
|
+
# @param hash [Hash] The hash to process
|
|
51
|
+
# @param output [Hash] Used for recursion, should be empty on first call
|
|
52
|
+
# @return [Hash] The hash with empty values removed
|
|
53
|
+
# @example
|
|
54
|
+
# hash = { "name" => "", "age" => nil, "address => { "city" => "Anytown", "state" => [] } }
|
|
55
|
+
# # Utils.remove_empty_values(hash)
|
|
56
|
+
# #=> { "address" => { "city" => "Anytown" } }
|
|
57
|
+
def self.deep_compact_blank(hash, output = {})
|
|
58
|
+
hash.each do |key, value|
|
|
59
|
+
if value.is_a?(Hash)
|
|
60
|
+
nested = deep_compact_blank(value)
|
|
61
|
+
output[key] = nested unless nested.empty?
|
|
62
|
+
elsif value.is_a?(Array)
|
|
63
|
+
compacted_array = value.map { |v| deep_compact_blank(v) }.reject(&:blank?)
|
|
64
|
+
output[key] = compacted_array unless compacted_array.empty?
|
|
65
|
+
elsif value.present?
|
|
66
|
+
output[key] = value
|
|
67
|
+
end
|
|
68
|
+
end
|
|
69
|
+
output
|
|
31
70
|
end
|
|
32
71
|
end
|
|
33
72
|
end
|
|
@@ -0,0 +1,85 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
# This script is a simple, brute-force method for finding records that
|
|
4
|
+
# exhibit certain characteristics in the public Cocina JSON for testing.
|
|
5
|
+
#
|
|
6
|
+
# It queries purl-fetcher for all DRUIDs released to a specific target and
|
|
7
|
+
# then fetches each corresponding public Cocina record from PURL and examines it.
|
|
8
|
+
#
|
|
9
|
+
# You need to be on VPN to do this, as the purl-fetcher API is only accessible
|
|
10
|
+
# from within the Stanford network.
|
|
11
|
+
#
|
|
12
|
+
# To use, modify any of the noted items below, then run:
|
|
13
|
+
# $ bundle exec ruby script/find_records.rb
|
|
14
|
+
#
|
|
15
|
+
# You can exit early with Ctrl-C, and it will report how many records were
|
|
16
|
+
# checked before exiting. Running through an entire target will take awhile,
|
|
17
|
+
# on the order of 30 minutes or more.
|
|
18
|
+
|
|
19
|
+
require "benchmark"
|
|
20
|
+
require "pp"
|
|
21
|
+
require "purl_fetcher/client"
|
|
22
|
+
require "cocina_display"
|
|
23
|
+
require "cocina_display/utils"
|
|
24
|
+
|
|
25
|
+
# This should correspond to one of the release targets available in purl-fetcher,
|
|
26
|
+
# i.e. "Searchworks", "Earthworks", etc.
|
|
27
|
+
RELEASE_TARGET = "Searchworks"
|
|
28
|
+
|
|
29
|
+
# Modify this expression to match the JSON path you want to search, or just
|
|
30
|
+
# modify the `examine_record` method directly.
|
|
31
|
+
PATH_EXPR = "$..[?length(@.groupedValue) > 0]"
|
|
32
|
+
|
|
33
|
+
# Modify this method as needed to change what you're looking for in each record.
|
|
34
|
+
# It takes a CocinaRecord object and should return an array of [path, result] pairs.
|
|
35
|
+
def examine_record(record)
|
|
36
|
+
record.path(PATH_EXPR).map { |value, _node, _key, path| [path, CocinaDisplay::Utils.deep_compact_blank(value)] }
|
|
37
|
+
end
|
|
38
|
+
|
|
39
|
+
# Track total records in target and how many we've seen
|
|
40
|
+
released_to_target = []
|
|
41
|
+
processed_records = 0
|
|
42
|
+
|
|
43
|
+
# Handle Ctrl-C gracefully
|
|
44
|
+
Signal.trap("INT") do
|
|
45
|
+
puts "\nExiting after processing #{processed_records} records."
|
|
46
|
+
exit
|
|
47
|
+
end
|
|
48
|
+
|
|
49
|
+
# Fetch everything from purl-fetcher; note that this is one single HTTP request
|
|
50
|
+
# that returns a massive JSON response – it can be quite slow
|
|
51
|
+
puts "Finding records released to #{RELEASE_TARGET}..."
|
|
52
|
+
client = PurlFetcher::Client::Reader.new
|
|
53
|
+
query_time = Benchmark.realtime do
|
|
54
|
+
client.released_to(RELEASE_TARGET).each do |record|
|
|
55
|
+
released_to_target << record["druid"].delete_prefix("druid:")
|
|
56
|
+
end
|
|
57
|
+
rescue Faraday::ConnectionFailed => e
|
|
58
|
+
puts "Connection failed: #{e.message}; are you on VPN?"
|
|
59
|
+
exit 1
|
|
60
|
+
end
|
|
61
|
+
puts "Found #{released_to_target.size} records released to #{RELEASE_TARGET} in #{query_time.round(2)} seconds"
|
|
62
|
+
|
|
63
|
+
# Iterate through the list of DRUIDs and fetch each one from PURL, creating a
|
|
64
|
+
# CocinaRecord object. Then call our examine_record method on it and if
|
|
65
|
+
# anything was returned, print the DRUID and the results.
|
|
66
|
+
released_to_target.each do |druid|
|
|
67
|
+
begin
|
|
68
|
+
cocina_record = CocinaDisplay::CocinaRecord.fetch(druid)
|
|
69
|
+
processed_records += 1
|
|
70
|
+
rescue => e
|
|
71
|
+
puts "Error fetching record #{druid}: #{e.message}"
|
|
72
|
+
next
|
|
73
|
+
end
|
|
74
|
+
|
|
75
|
+
results = examine_record(cocina_record)
|
|
76
|
+
next if results.empty?
|
|
77
|
+
|
|
78
|
+
puts "Druid: #{druid}"
|
|
79
|
+
results.each do |path, result|
|
|
80
|
+
puts " Path: #{path}"
|
|
81
|
+
puts " Result: #{result.pretty_inspect}\n"
|
|
82
|
+
end
|
|
83
|
+
|
|
84
|
+
puts "-" * 80
|
|
85
|
+
end
|