relaton-doi 1.14.2 → 1.14.4
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.github/workflows/rake.yml +1 -0
- data/.github/workflows/release.yml +3 -2
- data/.gitignore +1 -0
- data/.rubocop.yml +1 -1
- data/Gemfile +7 -2
- data/lib/relaton_doi/crossref.rb +11 -424
- data/lib/relaton_doi/parser.rb +773 -0
- data/lib/relaton_doi/version.rb +1 -1
- data/lib/relaton_doi.rb +1 -2
- data/relaton-doi.gemspec +6 -13
- metadata +12 -95
@@ -0,0 +1,773 @@
|
|
1
|
+
module RelatonDoi
|
2
|
+
class Parser
|
3
|
+
COUNTRIES = %w[USA].freeze
|
4
|
+
|
5
|
+
TYPES = {
|
6
|
+
"book-chapter" => "inbook",
|
7
|
+
"book-part" => "inbook",
|
8
|
+
"book-section" => "inbook",
|
9
|
+
"book-series" => "book",
|
10
|
+
"book-set" => "book",
|
11
|
+
"book-track" => "inbook",
|
12
|
+
"component" => "misc",
|
13
|
+
"database" => "dataset",
|
14
|
+
"dissertation" => "thesis",
|
15
|
+
"edited-book" => "book",
|
16
|
+
"grant" => "misc",
|
17
|
+
"journal-article" => "article",
|
18
|
+
"journal-issue" => "article",
|
19
|
+
"journal-volume" => "journal",
|
20
|
+
"monograph" => "book",
|
21
|
+
"other" => "misc",
|
22
|
+
"peer-review" => "article",
|
23
|
+
"posted-content" => "dataset",
|
24
|
+
"proceedings-article" => "inproceedings",
|
25
|
+
"proceedings-series" => "proceedings",
|
26
|
+
"reference-book" => "book",
|
27
|
+
"reference-entry" => "inbook",
|
28
|
+
"report-component" => "techreport",
|
29
|
+
"report-series" => "techreport",
|
30
|
+
"report" => "techreport",
|
31
|
+
}.freeze
|
32
|
+
|
33
|
+
REALATION_TYPES = {
|
34
|
+
"is-cited-by" => "isCitedIn",
|
35
|
+
"belongs-to" => "related",
|
36
|
+
"is-child-of" => "includedIn",
|
37
|
+
"is-expression-of" => "expressionOf",
|
38
|
+
"has-expression" => "hasExpression",
|
39
|
+
"is-manifestation-of" => "manifestationOf",
|
40
|
+
"is-manuscript-of" => "draftOf",
|
41
|
+
"has-manuscript" => "hasDraft",
|
42
|
+
"is-preprint-of" => "draftOf",
|
43
|
+
"has-preprint" => "hasDraft",
|
44
|
+
"is-replaced-by" => "obsoletedBy",
|
45
|
+
"replaces" => "obsoletes",
|
46
|
+
"is-translation-of" => "translatedFrom",
|
47
|
+
"has-translation" => "hasTranslation",
|
48
|
+
"is-version-of" => "editionOf",
|
49
|
+
"has-version" => "hasEdition",
|
50
|
+
"is-based-on" => "updates",
|
51
|
+
"is-basis-for" => "updatedBy",
|
52
|
+
"is-comment-on" => "commentaryOf",
|
53
|
+
"has-comment" => "hasCommentary",
|
54
|
+
"is-continued-by" => "hasSuccessor",
|
55
|
+
"continues" => "successorOf",
|
56
|
+
"is-derived-from" => "derives",
|
57
|
+
"has-derivation" => "derivedFrom",
|
58
|
+
"is-documented-by" => "describedBy",
|
59
|
+
"documents" => "describes",
|
60
|
+
"is-part-of" => "partOf",
|
61
|
+
"has-part" => "hasPart",
|
62
|
+
"is-review-of" => "reviewOf",
|
63
|
+
"has-review" => "hasReview",
|
64
|
+
"references" => "cites",
|
65
|
+
"is-referenced-by" => "isCitedIn",
|
66
|
+
"requires" => "hasComplement",
|
67
|
+
"is-required-by" => "complementOf",
|
68
|
+
"is-supplement-to" => "complementOf",
|
69
|
+
"is-supplemented-by" => "hasComplement",
|
70
|
+
}.freeze
|
71
|
+
|
72
|
+
ATTRS = %i[type fetched title docid date link abstract contributor place
|
73
|
+
doctype relation extent series medium].freeze
|
74
|
+
|
75
|
+
#
|
76
|
+
# Initialize instance.
|
77
|
+
#
|
78
|
+
# @param [Hash] src The source hash.
|
79
|
+
#
|
80
|
+
def initialize(src)
|
81
|
+
@src = src
|
82
|
+
@item = {}
|
83
|
+
end
|
84
|
+
|
85
|
+
#
|
86
|
+
# Initialize instance and parse the source hash.
|
87
|
+
#
|
88
|
+
# @param [Hash] src The source hash.
|
89
|
+
#
|
90
|
+
# @return [RelatonBib::BibliographicItem, RelatonIetf::IetfBibliographicItem,
|
91
|
+
# RelatonBipm::BipmBibliographicItem, RelatonIeee::IeeeBibliographicItem,
|
92
|
+
# RelatonNist::NistBibliographicItem] The bibitem.
|
93
|
+
#
|
94
|
+
def self.parse(src)
|
95
|
+
new(src).parse
|
96
|
+
end
|
97
|
+
|
98
|
+
#
|
99
|
+
# Parse the source hash.
|
100
|
+
#
|
101
|
+
# @return [RelatonBib::BibliographicItem, RelatonIetf::IetfBibliographicItem,
|
102
|
+
# RelatonBipm::BipmBibliographicItem, RelatonIeee::IeeeBibliographicItem,
|
103
|
+
# RelatonNist::NistBibliographicItem] The bibitem.
|
104
|
+
#
|
105
|
+
def parse
|
106
|
+
ATTRS.each { |m| @item[m] = send "parse_#{m}" }
|
107
|
+
create_bibitem @src["DOI"], @item
|
108
|
+
end
|
109
|
+
|
110
|
+
#
|
111
|
+
# Create a bibitem from the bibitem hash.
|
112
|
+
#
|
113
|
+
# @param [String] doi The DOI.
|
114
|
+
# @param [Hash] bibitem The bibitem hash.
|
115
|
+
#
|
116
|
+
# @return [RelatonBib::BibliographicItem, RelatonIetf::IetfBibliographicItem,
|
117
|
+
# RelatonBipm::BipmBibliographicItem, RelatonIeee::IeeeBibliographicItem,
|
118
|
+
# RelatonNist::NistBibliographicItem] The bibitem.
|
119
|
+
#
|
120
|
+
def create_bibitem(doi, bibitem) # rubocop:disable Metrics/CyclomaticComplexity
|
121
|
+
case doi
|
122
|
+
when /\/nist/ then RelatonNist::NistBibliographicItem.new(**bibitem)
|
123
|
+
when /\/rfc\d+/ then RelatonIetf::IetfBibliographicItem.new(**bibitem)
|
124
|
+
when /\/0026-1394\// then RelatonBipm::BipmBibliographicItem.new(**bibitem)
|
125
|
+
when /\/ieee/ then RelatonIeee::IeeeBibliographicItem.new(**bibitem)
|
126
|
+
else RelatonBib::BibliographicItem.new(**bibitem)
|
127
|
+
end
|
128
|
+
end
|
129
|
+
|
130
|
+
#
|
131
|
+
# Parse the type.
|
132
|
+
#
|
133
|
+
# @return [String] The type.
|
134
|
+
#
|
135
|
+
def parse_type
|
136
|
+
TYPES[@src["type"]] || @src["type"]
|
137
|
+
end
|
138
|
+
|
139
|
+
#
|
140
|
+
# Parse the document type
|
141
|
+
#
|
142
|
+
# @return [String] The document type.
|
143
|
+
#
|
144
|
+
def parse_doctype
|
145
|
+
@src["type"]
|
146
|
+
end
|
147
|
+
|
148
|
+
#
|
149
|
+
# Parse the fetched date.
|
150
|
+
#
|
151
|
+
# @return [String] The fetched date.
|
152
|
+
#
|
153
|
+
def parse_fetched
|
154
|
+
Date.today.to_s
|
155
|
+
end
|
156
|
+
|
157
|
+
#
|
158
|
+
# Parse titles from the source hash.
|
159
|
+
#
|
160
|
+
# @return [Array<Hash>] The titles.
|
161
|
+
#
|
162
|
+
def parse_title # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
|
163
|
+
if @src["title"].is_a?(Array) && @src["title"].any?
|
164
|
+
main_sub_titles
|
165
|
+
elsif @src["project"].is_a?(Array) && @src["project"].any?
|
166
|
+
project_titles
|
167
|
+
elsif @src["container-title"].is_a?(Array) && @src["container-title"].size > 1
|
168
|
+
@src["container-title"][0..-2].map { |t| create_title t }
|
169
|
+
else []
|
170
|
+
end
|
171
|
+
end
|
172
|
+
|
173
|
+
#
|
174
|
+
# Parse main and subtitle from the source hash.
|
175
|
+
#
|
176
|
+
# @return [Array<Hash>] The titles.
|
177
|
+
#
|
178
|
+
def main_sub_titles
|
179
|
+
title = @src["title"].map { |t| create_title t }
|
180
|
+
RelatonBib.array(@src["subtitle"]).each { |t| title << create_title(t, "subtitle") }
|
181
|
+
RelatonBib.array(@src["short-title"]).each { |t| title << create_title(t, "short") }
|
182
|
+
title
|
183
|
+
end
|
184
|
+
|
185
|
+
#
|
186
|
+
# Fetch titles from the projects.
|
187
|
+
#
|
188
|
+
# @return [Array<Hash>] The titles.
|
189
|
+
#
|
190
|
+
def project_titles
|
191
|
+
RelatonBib.array(@src["project"]).reduce([]) do |memo, proj|
|
192
|
+
memo + RelatonBib.array(proj["project-title"]).map { |t| create_title t["title"] }
|
193
|
+
end
|
194
|
+
end
|
195
|
+
|
196
|
+
#
|
197
|
+
# Create a title from the title and type.
|
198
|
+
#
|
199
|
+
# @param [String] title The title content.
|
200
|
+
# @param [String] type The title type. Defaults to "main".
|
201
|
+
#
|
202
|
+
# @return [RelatonBib::TypedTitleString] The title.
|
203
|
+
#
|
204
|
+
def create_title(title, type = "main")
|
205
|
+
cnt = str_cleanup title
|
206
|
+
RelatonBib::TypedTitleString.new type: type, content: cnt, script: "Latn"
|
207
|
+
end
|
208
|
+
|
209
|
+
#
|
210
|
+
# Parse a docid from the source hash.
|
211
|
+
#
|
212
|
+
# @return [Array<RelatonBib::DocumentIdentifier>] The docid.
|
213
|
+
#
|
214
|
+
def parse_docid
|
215
|
+
%w[DOI ISBN ISSN].each_with_object([]) do |type, obj|
|
216
|
+
prm = type == "DOI"
|
217
|
+
RelatonBib.array(@src[type]).each do |id|
|
218
|
+
t = issn_type(type, id)
|
219
|
+
obj << RelatonBib::DocumentIdentifier.new(type: t, id: id, primary: prm)
|
220
|
+
end
|
221
|
+
end
|
222
|
+
end
|
223
|
+
|
224
|
+
#
|
225
|
+
# Create an ISSN type if it's an ISSN ID.
|
226
|
+
#
|
227
|
+
# @param [String] type identifier type
|
228
|
+
# @param [String] id identifier
|
229
|
+
#
|
230
|
+
# @return [String] identifier type
|
231
|
+
#
|
232
|
+
def issn_type(type, id)
|
233
|
+
return type unless type == "ISSN"
|
234
|
+
|
235
|
+
t = @src["issn-type"]&.find { |it| it["value"] == id }&.dig("type")
|
236
|
+
t ? "issn.#{t}" : type.downcase
|
237
|
+
end
|
238
|
+
|
239
|
+
#
|
240
|
+
# Parce dates from the source hash.
|
241
|
+
#
|
242
|
+
# @return [Array<RelatonBib::BibliographicDate>] The dates.
|
243
|
+
#
|
244
|
+
def parse_date # rubocop:disable Metrics/CyclomaticComplexity
|
245
|
+
dates = %w[issued published approved].each_with_object([]) do |type, obj|
|
246
|
+
next unless @src.dig(type, "date-parts")&.first&.compact&.any?
|
247
|
+
|
248
|
+
obj << RelatonBib::BibliographicDate.new(type: type, on: date_type(type))
|
249
|
+
end
|
250
|
+
if dates.none?
|
251
|
+
dates << RelatonBib::BibliographicDate.new(type: "created", on: date_type("created"))
|
252
|
+
end
|
253
|
+
dates
|
254
|
+
end
|
255
|
+
|
256
|
+
#
|
257
|
+
# Join date parts into a string.
|
258
|
+
#
|
259
|
+
# @param [String] type The date type.
|
260
|
+
#
|
261
|
+
# @return [String] The date string.
|
262
|
+
#
|
263
|
+
def date_type(type)
|
264
|
+
@src[type]["date-parts"][0].map { |d| d.to_s.rjust(2, "0") }.join "-"
|
265
|
+
end
|
266
|
+
|
267
|
+
#
|
268
|
+
# Parse links from the source hash.
|
269
|
+
#
|
270
|
+
# @return [Array<RelatonBib::TypedUri>] The links.
|
271
|
+
#
|
272
|
+
def parse_link # rubocop:disable Metrics/MethodLength, Metrics/AbcSize, Metrics/CyclomaticComplexity
|
273
|
+
disprefered_links = %w[similarity-checking text-mining]
|
274
|
+
links = []
|
275
|
+
if @src["URL"]
|
276
|
+
links << RelatonBib::TypedUri.new(type: "DOI", content: @src["URL"])
|
277
|
+
end
|
278
|
+
[@src["link"], @src.dig("resource", "primary")].flatten.compact.each do |l|
|
279
|
+
next if disprefered_links.include? l["intended-application"]
|
280
|
+
|
281
|
+
type = case l["URL"]
|
282
|
+
when /\.pdf$/ then "pdf"
|
283
|
+
# when /\/rfc\d+$|iopscience\.iop\.org|ieeexplore\.ieee\.org/
|
284
|
+
else "src"
|
285
|
+
end
|
286
|
+
links << RelatonBib::TypedUri.new(type: type, content: l["URL"]) # if type
|
287
|
+
end
|
288
|
+
links
|
289
|
+
end
|
290
|
+
|
291
|
+
#
|
292
|
+
# Parse abstract from the source hash.
|
293
|
+
#
|
294
|
+
# @return [Array<RelatonBib::FormattedString>] The abstract.
|
295
|
+
#
|
296
|
+
def parse_abstract
|
297
|
+
return [] unless @src["abstract"]
|
298
|
+
|
299
|
+
content = @src["abstract"]
|
300
|
+
abstract = RelatonBib::FormattedString.new(
|
301
|
+
content: content, language: "en", script: "Latn", format: "text/html",
|
302
|
+
)
|
303
|
+
[abstract]
|
304
|
+
end
|
305
|
+
|
306
|
+
#
|
307
|
+
# Parse contributors from the source hash.
|
308
|
+
#
|
309
|
+
# @return [Array<RelatonBib::ContributionInfo>] The contributors.
|
310
|
+
#
|
311
|
+
def parse_contributor
|
312
|
+
contribs = author_investigators
|
313
|
+
contribs += authors_editors_translators
|
314
|
+
contribs += contribs_from_parent(contribs)
|
315
|
+
contribs << contributor(org_publisher, "publisher")
|
316
|
+
contribs += org_aurhorizer
|
317
|
+
contribs + org_enabler
|
318
|
+
end
|
319
|
+
|
320
|
+
#
|
321
|
+
# Create authors investigators from the source hash.
|
322
|
+
#
|
323
|
+
# @return [Array<RelatonBib::ContributionInfo>] The authors investigators.
|
324
|
+
#
|
325
|
+
def author_investigators
|
326
|
+
RelatonBib.array(@src["project"]).reduce([]) do |memo, proj|
|
327
|
+
memo + create_investigators(proj, "lead-investigator") +
|
328
|
+
create_investigators(proj, "investigator")
|
329
|
+
end
|
330
|
+
end
|
331
|
+
|
332
|
+
#
|
333
|
+
# Create investigators from the project.
|
334
|
+
#
|
335
|
+
# @param [Hash] project The project hash.
|
336
|
+
# @param [String] type The investigator type. "lead-investigator" or "investigator".
|
337
|
+
#
|
338
|
+
# @return [Array<RelatonBib::ContributionInfo>] The investigators.
|
339
|
+
#
|
340
|
+
def create_investigators(project, type)
|
341
|
+
description = type.gsub("-", " ")
|
342
|
+
RelatonBib.array(project[type]).map do |inv|
|
343
|
+
contributor(create_person(inv), "author", description)
|
344
|
+
end
|
345
|
+
end
|
346
|
+
|
347
|
+
#
|
348
|
+
# Create authors editors translators from the source hash.
|
349
|
+
#
|
350
|
+
# @return [Array<RelatonBib::ContributionInfo>] The authors editors translators.
|
351
|
+
#
|
352
|
+
def authors_editors_translators
|
353
|
+
%w[author editor translator].each_with_object([]) do |type, a|
|
354
|
+
@src[type]&.each do |c|
|
355
|
+
contrib = if c["family"]
|
356
|
+
create_person(c)
|
357
|
+
else
|
358
|
+
RelatonBib::Organization.new(name: str_cleanup(c["name"]))
|
359
|
+
end
|
360
|
+
a << contributor(contrib, type)
|
361
|
+
end
|
362
|
+
end
|
363
|
+
end
|
364
|
+
|
365
|
+
#
|
366
|
+
# Fetch authors and editors from parent if they are not present in the book part.
|
367
|
+
#
|
368
|
+
# @param [Array<RelatonBib::ContributionInfo>] contribs present contributors
|
369
|
+
#
|
370
|
+
# @return [Array<RelatonBib::ContributionInfo>] contributors with authors and editors from parent
|
371
|
+
#
|
372
|
+
def contribs_from_parent(contribs) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
|
373
|
+
return [] unless %w[inbook inproceedings dataset].include?(parse_type) && @src["container-title"]
|
374
|
+
|
375
|
+
has_authors = contribs.any? { |c| c.role&.any? { |r| r.type == "author" } }
|
376
|
+
has_editors = contribs.any? { |c| c.role&.any? { |r| r.type == "editor" } }
|
377
|
+
return [] if has_authors && has_editors
|
378
|
+
|
379
|
+
create_authors_editors(has_authors, "author")
|
380
|
+
end
|
381
|
+
|
382
|
+
#
|
383
|
+
# Fetch parent item from Crossref.
|
384
|
+
#
|
385
|
+
# @return [Hash, nil] parent item
|
386
|
+
#
|
387
|
+
def parent_item # rubocop:disable Metrics/AbcSize
|
388
|
+
@parent_item ||= begin
|
389
|
+
query = [@src["container-title"][0], fetch_year].compact.join "+"
|
390
|
+
filter = "type:#{%w[book book-set edited-book monograph reference-book].join ',type:'}"
|
391
|
+
resp = Faraday.get "https://api.crossref.org/works?query=#{query}&filter=#{filter}"
|
392
|
+
json = JSON.parse resp.body
|
393
|
+
json["message"]["items"].detect { |i| i["title"].include? @src["container-title"][0] }
|
394
|
+
end
|
395
|
+
end
|
396
|
+
|
397
|
+
#
|
398
|
+
# Create authors and editors from parent item.
|
399
|
+
#
|
400
|
+
# @param [Boolean] has true if authors or editors are present in the book part
|
401
|
+
# @param [String] type "author" or "editor"
|
402
|
+
#
|
403
|
+
# @return [Array<RelatonBib::ContributionInfo>] authors or editors
|
404
|
+
#
|
405
|
+
def create_authors_editors(has, type)
|
406
|
+
return [] if has || !parent_item
|
407
|
+
|
408
|
+
RelatonBib.array(parent_item[type]).map { |a| contributor(create_person(a), type) }
|
409
|
+
end
|
410
|
+
|
411
|
+
#
|
412
|
+
# Cerate an organization publisher from the source hash.
|
413
|
+
#
|
414
|
+
# @return [RelatonBib::Organization] The organization.
|
415
|
+
#
|
416
|
+
def org_publisher
|
417
|
+
pbr = @src["institution"]&.detect do |i|
|
418
|
+
@src["publisher"].include?(i["name"]) ||
|
419
|
+
i["name"].include?(@src["publisher"])
|
420
|
+
end
|
421
|
+
a = pbr["acronym"]&.first if pbr
|
422
|
+
RelatonBib::Organization.new name: str_cleanup(@src["publisher"]), abbreviation: a
|
423
|
+
end
|
424
|
+
|
425
|
+
#
|
426
|
+
# Clean up trailing punctuation and whitespace from a string.
|
427
|
+
#
|
428
|
+
# @param [String] str The string to clean up.
|
429
|
+
#
|
430
|
+
# @return [String] The cleaned up string.
|
431
|
+
#
|
432
|
+
def str_cleanup(str)
|
433
|
+
str.strip.sub(/[,\/\s]+$/, "").sub(/\s:$/, "")
|
434
|
+
end
|
435
|
+
|
436
|
+
#
|
437
|
+
# Parse authorizer contributor from the source hash.
|
438
|
+
#
|
439
|
+
# @return [Array<RelatonBib::ContributionInfo>] The authorizer contributor.
|
440
|
+
#
|
441
|
+
def org_aurhorizer
|
442
|
+
return [] unless @src["standards-body"]
|
443
|
+
|
444
|
+
name, acronym = @src["standards-body"].values_at("name", "acronym")
|
445
|
+
org = RelatonBib::Organization.new name: name, abbreviation: acronym
|
446
|
+
[contributor(org, "authorizer")]
|
447
|
+
end
|
448
|
+
|
449
|
+
#
|
450
|
+
# Parse enabler contributor from the source hash.
|
451
|
+
#
|
452
|
+
# @return [Array<RelatonBib::ContributionInfo>] The enabler contributor.
|
453
|
+
#
|
454
|
+
def org_enabler
|
455
|
+
RelatonBib.array(@src["project"]).each_with_object([]) do |proj, memo|
|
456
|
+
proj["funding"].each do |f|
|
457
|
+
memo << create_enabler(f.dig("funder", "name"))
|
458
|
+
end
|
459
|
+
end + RelatonBib.array(@src["funder"]).map { |f| create_enabler f["name"] }
|
460
|
+
end
|
461
|
+
|
462
|
+
#
|
463
|
+
# Create enabler contributor with type "enabler".
|
464
|
+
#
|
465
|
+
# @param [String] name <description>
|
466
|
+
#
|
467
|
+
# @return [RelatonBib::ContributionInfo] The enabler contributor.
|
468
|
+
#
|
469
|
+
def create_enabler(name)
|
470
|
+
org = RelatonBib::Organization.new name: name
|
471
|
+
contributor(org, "enabler")
|
472
|
+
end
|
473
|
+
|
474
|
+
#
|
475
|
+
# Create contributor from an entity and a role type.
|
476
|
+
#
|
477
|
+
# @param [RelatonBib::Person, RelatonBib::Organization] entity The entity.
|
478
|
+
# @param [String] type The role type.
|
479
|
+
#
|
480
|
+
# @return [RelatonBib::ContributionInfo] The contributor.
|
481
|
+
#
|
482
|
+
def contributor(entity, type, descriprion = nil)
|
483
|
+
role = { type: type }
|
484
|
+
role[:description] = [descriprion] if descriprion
|
485
|
+
RelatonBib::ContributionInfo.new(entity: entity, role: [role])
|
486
|
+
end
|
487
|
+
|
488
|
+
#
|
489
|
+
# Create a person from a person hash.
|
490
|
+
#
|
491
|
+
# @param [Hash] person The person hash.
|
492
|
+
#
|
493
|
+
# @return [RelatonBib::Person] The person.
|
494
|
+
#
|
495
|
+
def create_person(person)
|
496
|
+
RelatonBib::Person.new(
|
497
|
+
name: create_person_name(person),
|
498
|
+
affiliation: create_affiliation(person),
|
499
|
+
identifier: person_id(person),
|
500
|
+
)
|
501
|
+
end
|
502
|
+
|
503
|
+
#
|
504
|
+
# Create person affiliations from a person hash.
|
505
|
+
#
|
506
|
+
# @param [Hash] person The person hash.
|
507
|
+
#
|
508
|
+
# @return [Array<RelatonBib::Affiliation>] The affiliations.
|
509
|
+
#
|
510
|
+
def create_affiliation(person)
|
511
|
+
(person["affiliation"] || []).map do |a|
|
512
|
+
org = RelatonBib::Organization.new(name: a["name"])
|
513
|
+
RelatonBib::Affiliation.new organization: org
|
514
|
+
end
|
515
|
+
end
|
516
|
+
|
517
|
+
#
|
518
|
+
# Create a person full name from a person hash.
|
519
|
+
#
|
520
|
+
# @param [Hash] person The person hash.
|
521
|
+
#
|
522
|
+
# @return [RelatonBib::FullName] The full name.
|
523
|
+
#
|
524
|
+
def create_person_name(person)
|
525
|
+
surname = titlecase(person["family"])
|
526
|
+
sn = RelatonBib::LocalizedString.new(surname, "en", "Latn")
|
527
|
+
RelatonBib::FullName.new(
|
528
|
+
surname: sn, forename: forename(person), addition: nameaddition(person),
|
529
|
+
completename: completename(person), prefix: nameprefix(person)
|
530
|
+
)
|
531
|
+
end
|
532
|
+
|
533
|
+
#
|
534
|
+
# Capitalize the first letter of each word in a string except for words that
|
535
|
+
# are 2 letters or less.
|
536
|
+
#
|
537
|
+
# @param [<Type>] str <description>
|
538
|
+
#
|
539
|
+
# @return [<Type>] <description>
|
540
|
+
#
|
541
|
+
def titlecase(str)
|
542
|
+
str.split.map do |s|
|
543
|
+
if s.size > 2 && s.upcase == s && !/\.&/.match?(s)
|
544
|
+
s.capitalize
|
545
|
+
else
|
546
|
+
s
|
547
|
+
end
|
548
|
+
end.join " "
|
549
|
+
end
|
550
|
+
|
551
|
+
#
|
552
|
+
# Create a person name prefix from a person hash.
|
553
|
+
#
|
554
|
+
# @param [Hash] person The person hash.
|
555
|
+
#
|
556
|
+
# @return [Array<RelatonBib::LocalizedString>] The name prefix.
|
557
|
+
#
|
558
|
+
def nameprefix(person)
|
559
|
+
return [] unless person["prefix"]
|
560
|
+
|
561
|
+
[RelatonBib::LocalizedString.new(person["prefix"], "en", "Latn")]
|
562
|
+
end
|
563
|
+
|
564
|
+
#
|
565
|
+
# Create a complete name from a person hash.
|
566
|
+
#
|
567
|
+
# @param [Hash] person The person hash.
|
568
|
+
#
|
569
|
+
# @return [RelatonBib::LocalizedString] The complete name.
|
570
|
+
#
|
571
|
+
def completename(person)
|
572
|
+
return unless person["name"]
|
573
|
+
|
574
|
+
RelatonBib::LocalizedString.new(person["name"], "en", "Latn")
|
575
|
+
end
|
576
|
+
|
577
|
+
#
|
578
|
+
# Create a forename from a person hash.
|
579
|
+
#
|
580
|
+
# @param [Hash] person The person hash.
|
581
|
+
#
|
582
|
+
# @return [Array<RelatonBib::LocalizedString>] The forename.
|
583
|
+
#
|
584
|
+
def forename(person)
|
585
|
+
return [] unless person["given"]
|
586
|
+
|
587
|
+
fname = titlecase(person["given"])
|
588
|
+
[RelatonBib::Forename.new(content: fname, language: "en", script: "Latn")]
|
589
|
+
end
|
590
|
+
|
591
|
+
#
|
592
|
+
# Create an addition from a person hash.
|
593
|
+
#
|
594
|
+
# @param [Hash] person The person hash.
|
595
|
+
#
|
596
|
+
# @return [Array<RelatonBib::LocalizedString>] The addition.
|
597
|
+
#
|
598
|
+
def nameaddition(person)
|
599
|
+
return [] unless person["suffix"]
|
600
|
+
|
601
|
+
[RelatonBib::LocalizedString.new(person["suffix"], "en", "Latn")]
|
602
|
+
end
|
603
|
+
|
604
|
+
#
|
605
|
+
# Create a person identifier from a person hash.
|
606
|
+
#
|
607
|
+
# @param [Hash] person The person hash.
|
608
|
+
#
|
609
|
+
# @return [Array<RelatonBib::PersonIdentifier>] The person identifier.
|
610
|
+
#
|
611
|
+
def person_id(person)
|
612
|
+
return [] unless person["ORCID"]
|
613
|
+
|
614
|
+
[RelatonBib::PersonIdentifier.new("orcid", person["ORCID"])]
|
615
|
+
end
|
616
|
+
|
617
|
+
#
|
618
|
+
# Parse a place from the source hash.
|
619
|
+
#
|
620
|
+
# @return [Array<RelatonBib::Place>] The place.
|
621
|
+
#
|
622
|
+
def parse_place # rubocop:disable Metrics/MethodLength, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity, Metrics/AbcSize
|
623
|
+
pub_location = @src["publisher-location"] || fetch_location
|
624
|
+
return [] unless pub_location
|
625
|
+
|
626
|
+
pls1, pls2 = pub_location.split(", ")
|
627
|
+
pls1 = str_cleanup pls1
|
628
|
+
pls2 &&= str_cleanup pls2
|
629
|
+
if COUNTRIES.include? pls2
|
630
|
+
country = RelatonBib::Place::RegionType.new(name: pls2)
|
631
|
+
[RelatonBib::Place.new(city: pls1, country: [country])]
|
632
|
+
elsif pls2 && pls2 == pls2&.upcase
|
633
|
+
region = RelatonBib::Place::RegionType.new(name: pls2)
|
634
|
+
[RelatonBib::Place.new(city: pls1, region: [region])]
|
635
|
+
elsif pls1 == pls2 || pls2.nil? || pls2.empty?
|
636
|
+
[RelatonBib::Place.new(city: pls1)]
|
637
|
+
else
|
638
|
+
[RelatonBib::Place.new(city: pls1), RelatonBib::Place.new(city: pls2)]
|
639
|
+
end
|
640
|
+
end
|
641
|
+
|
642
|
+
#
|
643
|
+
# Fetch location from container.
|
644
|
+
#
|
645
|
+
# @return [String, nil] The location.
|
646
|
+
#
|
647
|
+
def fetch_location # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity
|
648
|
+
title = @item[:title].first&.title&.content
|
649
|
+
qparts = [title, fetch_year, @src["publisher"]]
|
650
|
+
query = CGI.escape qparts.compact.join("+").gsub(" ", "+")
|
651
|
+
filter = "type:#{%w[book-chapter book-part book-section book-track].join(',type:')}"
|
652
|
+
resp = Faraday.get "https://api.crossref.org/works?query=#{query}&filter=#{filter}"
|
653
|
+
json = JSON.parse resp.body
|
654
|
+
json["message"]["items"].detect do |i|
|
655
|
+
i["publisher-location"] && i["container-title"].include?(title)
|
656
|
+
end&.dig("publisher-location")
|
657
|
+
end
|
658
|
+
|
659
|
+
#
|
660
|
+
# Parse relations from the source hash.
|
661
|
+
#
|
662
|
+
# @return [Array<RelatonBib::DocumentRelation>] The relations.
|
663
|
+
#
|
664
|
+
def parse_relation # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
|
665
|
+
rels = included_in_relation
|
666
|
+
@src["relation"].each_with_object(rels) do |(k, v), a|
|
667
|
+
type, desc = relation_type k
|
668
|
+
RelatonBib.array(v).each do |r|
|
669
|
+
rel_item = Crossref.get_by_id r["id"]
|
670
|
+
title = rel_item["title"].map { |t| create_title t }
|
671
|
+
docid = RelatonBib::DocumentIdentifier.new(id: r["id"], type: "DOI")
|
672
|
+
bib = create_bibitem r["id"], title: title, docid: [docid]
|
673
|
+
a << RelatonBib::DocumentRelation.new(type: type, description: desc, bibitem: bib)
|
674
|
+
end
|
675
|
+
end
|
676
|
+
end
|
677
|
+
|
678
|
+
#
|
679
|
+
# Transform crossref relation type to relaton relation type.
|
680
|
+
#
|
681
|
+
# @param [String] crtype The crossref relation type.
|
682
|
+
#
|
683
|
+
# @return [Array<String>] The relaton relation type and description.
|
684
|
+
#
|
685
|
+
def relation_type(crtype)
|
686
|
+
type = REALATION_TYPES[crtype] || begin
|
687
|
+
desc = RelatonBib::FormattedString.new(content: crtype)
|
688
|
+
"related"
|
689
|
+
end
|
690
|
+
[type, desc]
|
691
|
+
end
|
692
|
+
|
693
|
+
#
|
694
|
+
# Create included in relation.
|
695
|
+
#
|
696
|
+
# @return [Array<RelatonBib::DocumentRelation>] The relations.
|
697
|
+
#
|
698
|
+
def included_in_relation
|
699
|
+
types = %w[
|
700
|
+
book book-chapter book-part book-section book-track dataset journal-issue
|
701
|
+
journal-value proceedings-article reference-entry report-component
|
702
|
+
]
|
703
|
+
return [] unless @src["container-title"] && types.include?(@src["type"])
|
704
|
+
|
705
|
+
@src["container-title"].map do |ct|
|
706
|
+
contrib = create_authors_editors false, "editor"
|
707
|
+
bib = RelatonBib::BibliographicItem.new(title: [content: ct], contributor: contrib)
|
708
|
+
RelatonBib::DocumentRelation.new(type: "includedIn", bibitem: bib)
|
709
|
+
end
|
710
|
+
end
|
711
|
+
|
712
|
+
#
|
713
|
+
# Fetch year from the source hash.
|
714
|
+
#
|
715
|
+
# @return [String] The year.
|
716
|
+
#
|
717
|
+
def fetch_year
|
718
|
+
d = @src["published"] || @src["approved"] || @src["created"]
|
719
|
+
d["date-parts"][0][0]
|
720
|
+
end
|
721
|
+
|
722
|
+
#
|
723
|
+
# Parse an extent from the source hash.
|
724
|
+
#
|
725
|
+
# @return [Array<RelatonBib::Locality>] The extent.
|
726
|
+
#
|
727
|
+
def parse_extent # rubocop:disable Metrics/AbcSize
|
728
|
+
extent = []
|
729
|
+
extent << RelatonBib::Locality.new("volume", @src["volume"]) if @src["volume"]
|
730
|
+
extent << RelatonBib::Locality.new("issue", @src["issue"]) if @src["issue"]
|
731
|
+
if @src["page"]
|
732
|
+
from, to = @src["page"].split("-")
|
733
|
+
extent << RelatonBib::Locality.new("page", from, to)
|
734
|
+
end
|
735
|
+
extent.any? ? [RelatonBib::LocalityStack.new(extent)] : []
|
736
|
+
end
|
737
|
+
|
738
|
+
#
|
739
|
+
# Parse a series from the source hash.
|
740
|
+
#
|
741
|
+
# @return [Arrey<RelatonBib::Series>] The series.
|
742
|
+
#
|
743
|
+
def parse_series # rubocop:disable Metrics/AbcSize, Metrics/MethodLength, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
|
744
|
+
types = %w[inbook incollection inproceedings]
|
745
|
+
return [] if !@src["container-title"] || types.include?(@item[:type]) || @src["type"] == "report-component"
|
746
|
+
|
747
|
+
con_ttl = if main_sub_titles.any? || project_titles.any?
|
748
|
+
@src["container-title"]
|
749
|
+
elsif @src["container-title"].size > 1
|
750
|
+
sct = @src["short-container-title"]&.last
|
751
|
+
abbrev = RelatonBib::LocalizedString.new sct if sct
|
752
|
+
@src["container-title"][-1..-1]
|
753
|
+
else []
|
754
|
+
end
|
755
|
+
con_ttl.map do |ct|
|
756
|
+
title = RelatonBib::TypedTitleString.new content: ct
|
757
|
+
RelatonBib::Series.new title: title, abbreviation: abbrev
|
758
|
+
end
|
759
|
+
end
|
760
|
+
|
761
|
+
#
|
762
|
+
# Parse a medium from the source hash.
|
763
|
+
#
|
764
|
+
# @return [RelatonBib::Mediub, nil] The medium.
|
765
|
+
#
|
766
|
+
def parse_medium
|
767
|
+
genre = @src["degree"]&.first
|
768
|
+
return unless genre
|
769
|
+
|
770
|
+
RelatonBib::Medium.new genre: genre
|
771
|
+
end
|
772
|
+
end
|
773
|
+
end
|