f04bc944-f9b4-4a42-8b26-dcacd0e3e688

Committed 11 Mar 2025 10:27PM UTC coverage: 34.017% (-58.1%) from 92.162%

Build # f04bc944-f9b4-4a42-8b26-dcacd0e3e688

Build Type

Pull #2653

circleci

Committed by

christinach

Commit Message

Add new lc_subject_facet field.
Helps with the vocabulary work https://github.com/pulibrary/orangelight/pull/3386
In this new field we index only the lc subject heading and the subdivisions
So that when the user searches using the Details section, they can query solr for
all the subject headings and their divisions.

This is needed for the Subject browse Vocabulary work.
example: "lc_subject_facet": [
             "Booksellers and bookselling—Italy—Directories",
             "Booksellers and bookselling-Italy",
             "Booksellers and bookselling"
              ]

Pull Request Pull Request #2653: Add new lc_subject_facet field.

Run Details

1 of 3 new or added lines in 1 file covered. (33.33%)

2215 existing lines in 93 files now uncovered.

1294 of 3804 relevant lines covered (34.02%)

0.99 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

16.28

/marc_to_solr/lib/format/bib_types.rb

1	require 'traject/macros/marc21_semantics'	1✔
2	require_relative '../format/xv6xx'	1✔
3
4	# Determine the "types" of material represented by the bib record.
5	# The comments below come from the Ex Libris Aleph system and represent
6	# the logic used within it to determine types. This file is based
7	# on the logic use at the University of Michigan
8
9	class BibTypes	1✔
10	attr_reader :bib_format, :record	1✔
11
12	def initialize(bib_format, record)	1✔
UNCOV 13	@bib_format = bib_format	×
UNCOV 14	@record = record	×
15	# Memoize values, since many of them are used several times
UNCOV 16	@spec_vals = Hash.new { \|h, spec_string\| h[spec_string] = Traject::MarcExtractor.new(spec_string).extract(@record) }	×
17
18	# Need these a lot -- the sub x and v from any 6XX field
UNCOV 19	@xv6XX = XV6XX.new(@record)	×
20	end
21
22	def codes	1✔
UNCOV 23	codes = []	×
UNCOV 24	codes.concat self.video_types	×
UNCOV 25	codes.concat self.audio_types	×
UNCOV 26	codes.concat self.microform_types	×
UNCOV 27	codes.concat self.musical_score_types	×
UNCOV 28	codes.concat self.map_types	×
UNCOV 29	codes.concat self.serial_types	×
UNCOV 30	codes.concat self.mixed_types	×
UNCOV 31	codes.concat self.software_types	×
UNCOV 32	codes.concat self.statistics_types	×
UNCOV 33	codes.concat self.conference_types	×
UNCOV 34	codes.concat self.biography_types	×
UNCOV 35	codes.concat self.reference_types	×
UNCOV 36	codes.concat self.pp_types	×
UNCOV 37	codes.concat self.videogame_types	×
38
UNCOV 39	codes.uniq!	×
UNCOV 40	codes.compact!	×
41
UNCOV 42	codes	×
43	end
44
45	# Provide memoized values for on-the-fly created MarcExtractor
46	# objects
47	#
48	# @param [String] spec_string A Traject::MarcExtractor-compatible spec string
49	# @return [Array<String>] The strings in the specified subfields/byterange/whatever
50	def [](spec_string)	1✔
UNCOV 51	@spec_vals[spec_string]	×
52	end
53
54	### Video stuff
55
56	# TYP VB Video (Blu-ray) 538## a MATCH Blu-ray
57	# TYP VB Video (Blu-ray) 007 F00-05 MATCH v???s
58	# TYP VB Video (Blu-ray) 250## a MATCH Blu-ray
59	# TYP VB Video (Blu-ray) 852## j MATCH video-b*
60	# TYP VB Video (Blu-ray) 852## j MATCH bd-rom*
61	#
62	# TYP VD Video (DVD) 538## a MATCH DVD*
63	# TYP VD Video (DVD) 007 F04-01 EQUAL v
64	# 007 F00-01 EQUAL v
65	# TYP VD Video (DVD) 007 F04-01 EQUAL v
66	# 008 F33-01 EQUAL v
67	# !
68	# ! Visual material: vHs
69	# TYP VH Video (VHS) 538## a MATCH VHS*
70	# TYP VH Video (VHS) 007 F04-01 EQUAL b
71	# 007 F00-01 EQUAL v
72	# TYP VH Video (VHS) 007 F04-01 EQUAL b
73	# 008 F33-01 EQUAL v
74	# !
75	# ! Visual materials: fiLm/video
76	# TYP VL Motion Picture 007 F00-01 EQUAL m
77	# TYP VL Motion Picture FMT F00-02 EQUAL VM
78	# 008 F33-01 EQUAL m
79
80	def video_types	1✔
UNCOV 81	types = []	×
82
UNCOV 83	types << 'VB' if self['538a:250a'].grep(/blu-ray/i).size > 0	×
UNCOV 84	types << 'VB' if self['007[0-5]'].grep(/v...s/i).size > 0	×
UNCOV 85	types << 'VB' if self['852j'].grep(/\A(?:bd-rom\|video-b)/i).size > 0	×
86
UNCOV 87	@record.fields('007').map { \|f\| f.value }.each do \|f\|	×
UNCOV 88	if (f[0] == 'v') \|\| self['008[33]'].include?('v')	×
UNCOV 89	types << 'VD' if f[4] == 'v'	×
UNCOV 90	types << 'VH' if f[4] == 'b'	×
91	end
92	end
93
UNCOV 94	types << 'VD' if self['538a'].grep(/\Advd(?!-rom)/i).size > 0	×
95
UNCOV 96	types << 'VH' if self['538a'].grep(/\AVHS/i).size > 0	×
97
UNCOV 98	types << 'VL' if self['007[0]'].include?('m')	×
UNCOV 99	types << 'VL' if (self.bib_format == 'VM') && self['008[33]'].include?('m')	×
100
UNCOV 101	types.uniq!	×
UNCOV 102	return types	×
103	end
104
105	# Audio/music
106	# ! Recording: Compact disc
107	# TYP RC Audio CD LDR F06-01 EQUAL [i,j]
108	# FMT F00-02 EQUAL MU
109	# 007 F01-01 EQUAL d
110	# 007 F12-01 EQUAL e
111	# TYP RC Audio CD 8524 j MATCH CD*
112	# 8524 b EQUAL MUSIC
113	# !
114	# ! Recording: LP record
115	# TYP RL Audio LP LDR F06-01 EQUAL [i,j]
116	# FMT F00-02 EQUAL MU
117	# 007 F01-01 EQUAL d
118	# 300 a MATCH SOUND DISC
119	# 300 b MATCH 33 1/3 RPM
120	#
121	# TYP RL Audio LP 8524 j MATCH LP*
122	# 8524 c EQUAL MUSI
123	# TYP RL Audio LP 8524 j MATCH LP*
124	# 8524 b EQUAL MUSIC
125	# !
126	# ! Recording: Music
127	# TYP RM Audio (music) LDR F06-01 EQUAL j
128	# FMT F00-02 EQUAL MU
129	# !
130	# ! Recording: Spoken word
131	# TYP RS Audio (spoken word) LDR F06-01 EQUAL i
132	# FMT F00-02 EQUAL MU
133	# !
134	# ! Recording: Undefined
135	# TYP RU Audio LDR F06-01 EQUAL [i,j]
136	# FMT F00-02 EQUAL MU
137	#
138
139	def audio_types	1✔
UNCOV 140	ldr6 = record.leader[6]	×
141
UNCOV 142	types = []	×
143
144	# Get the 8524* fields
UNCOV 145	f8524 = record.fields('852').select { \|f\| f.indicator1 == '4' }	×
146
147	# RC
UNCOV 148	if %w[i j].include?(ldr6) && (bib_format == 'MU')	×
149	@record.fields('007').map { \|f\| f.value }.each do \|f\|	×
150	if f[1] == 'd' && f[12] == 'e'	×
151	types << 'RC'	×
152	break	×
153	end
154	end
155	end
156
UNCOV 157	f8524.each do \|f\|	×
158	if (f['b']&.upcase == 'MUSIC') && (f['j'] =~ /\ACD/i)	×
159	types << 'RC'	×
160	break	×
161	end
162	end
163
164	# RL
165
UNCOV 166	if (bib_format == 'MU') && %w[i j].include?(ldr6) && self['007[1]'].include?('d')	×
167	record.fields('300').each do \|f\|	×
168	str = f.subfields.collect { \|s\| s.value }.join(' ')	×
169	if (str =~ /DISC/i) && str =~ %r{33 1/3 RPM}i	×
170	types << 'RL'	×
171	break	×
172	end
173	end
174	end
175
UNCOV 176	f8524.each do \|f\|	×
177	if (f['j'] =~ /\ALP/i) &&	×
178	((f['b'].upcase == 'MUSIC') \|\| (f['c'].upcase == 'MUSI'))	×
179	types << 'RL'	×
180	break	×
181	end
182	end
183
184	# RM
UNCOV 185	types << 'RM' if (ldr6 == 'j') && (bib_format == 'MU')	×
186
187	# RS
UNCOV 188	types << 'RS' if (ldr6 == 'i') && (bib_format == 'MU')	×
189
190	# RU
UNCOV 191	types << 'RU' if %w[i j].include?(ldr6) && (bib_format == 'MU')	×
192
UNCOV 193	types.uniq!	×
UNCOV 194	return types	×
195	end
196
197	# Microform
198	# ! MicroForms
199	# TYP WM Microform FMT F00-02 EQUAL BK
200	# 008 F23-01 EQUAL [a,b,c]
201	# TYP WM Microform FMT F00-02 EQUAL MU
202	# 008 F23-01 EQUAL [a,b,c]
203	# TYP WM Microform FMT F00-02 EQUAL SE
204	# 008 F23-01 EQUAL [a,b,c]
205	# TYP WM Microform FMT F00-02 EQUAL MX
206	# 008 F23-01 EQUAL [a,b,c]
207
208	# TYP WM Microform 245## h MATCH micro
209
210	# TYP WM Microform FMT F00-02 EQUAL MP
211	# 008 F29-01 EQUAL [a,b,c]
212	# TYP WM Microform FMT F00-02 EQUAL VM
213	# 008 F29-01 EQUAL [a,b,c]
214
215	def microform_types	1✔
UNCOV 216	return [] unless record['008']	×
217
UNCOV 218	types = ['WM']	×
UNCOV 219	f8_23 = record['008'].value[23]	×
UNCOV 220	return types if %w[BK MU SE MX].include?(bib_format) && %w[a b c].include?(f8_23)	×
221
UNCOV 222	f8_29 = record['008'].value[29]	×
UNCOV 223	return types if %w[MP VM].include?(bib_format) && %w[a b c].include?(f8_29)	×
224
UNCOV 225	return types if record['245'] && (record['245']['h'] =~ /micro/i)	×
226
227	# Nope. Not microform
UNCOV 228	return []	×
229	end
230
231	# ! Musical Score
232	# TYP MS Musical Score LDR F06-01 EQUAL [c,d]
233
234	def musical_score_types	1✔
UNCOV 235	types = []	×
UNCOV 236	types << 'MS' if %w[c d].include?(record.leader[6])	×
UNCOV 237	return types	×
238	end
239
240	# ! Maps: Numerous
241	# TYP MN Maps-Atlas FMT F00-02 EQUAL MP
242	# TYP MN Maps-Atlas LDR F06-01 EQUAL [e,f]
243	# TYP MN Maps-Atlas 007 F00-01 EQUAL a
244	# !
245	# ! Maps: One (commented out as per Judy Ahronheim as this TYP duplicates MN)
246	# !TYP MO Map FMT F00-02 EQUAL MP
247	# !TYP MO Map 007 F00-01 EQUAL a
248
249	def map_types	1✔
UNCOV 250	types = []	×
UNCOV 251	types << 'MN' if (bib_format == 'MP') \|\| %w[e f].include?(record.leader[6]) \|\| self['007[0]'].include?('a')	×
UNCOV 252	return types	×
253	end
254
255	# Serials
256	# ! serial: A Journal
257	# TYP AJ Journal FMT F00-02 EQUAL SE
258	# 008 F21-01 EQUAL p
259	# 008 F22-01 EQUAL [^,a,b,c,d,f,g,h,i,s,x,z,\|]
260	# 008 F29-01 EQUAL [0,\|]
261	# TYP AJ Journal FMT F00-02 EQUAL SE
262	# 008 F21-01 EQUAL [^,d,l,m,p,w,\|]
263	# 008 F22-01 EQUAL [^,a,b,c,d,f,g,h,i,s,x,z,\|]
264	# 008 F24-01 EQUAL [a,b,g,m,n,o,p,s,w,x,y,^]
265	# 008 F29-01 EQUAL [0,\|]
266	# !
267	# ! serial: A Newspaper
268	# TYP AN Newspaper FMT F00-02 EQUAL SE
269	# 008 F21-01 EQUAL n
270	# TYP AN Newspaper FMT F00-02 EQUAL SE
271	# 008 F22-01 EQUAL e
272	#
273	# ! serial: All, including serials with other FMT codes
274	# TYP SX All Serials LDR F07-01 EQUAL [b,s]
275
276	# Wrap it all up in serial_types
277	def serial_types	1✔
UNCOV 278	types = []	×
UNCOV 279	types << 'SX' if %w[b s].include?(record.leader[7])	×
UNCOV 280	types.concat journal_types	×
UNCOV 281	types.concat newspaper_types	×
UNCOV 282	types.uniq!	×
UNCOV 283	return types	×
284	end
285
286	def journal_types	1✔
UNCOV 287	types = []	×
288	# gotta be SE and have a 008
UNCOV 289	return types unless (bib_format == 'SE') && record['008']	×
290
291	# We need lots of chars from the 008
292	f8 = record['008'].value	×
293
294	if (f8[21] == 'p') &&	×
295	[' ', 'a', 'b', 'c', 'd', 'f', 'g', 'h', 'i', 's', 'x', 'z', '\|'].include?(f8[22]) &&
296	['0', '\|'].include?(f8[29])
297	types << 'AJ'	×
298	end
299
300	if [' ', 'd', 'l', 'm', 'p', 'w', '\|'].include?(f8[21]) &&	×
301	[' ', 'a', 'b', 'c', 'd', 'f', 'g', 'h', 'i', 's', 'x', 'z', '\|'].include?(f8[22]) &&
302	['a', 'b', 'g', 'm', 'n', 'o', 'p', 's', 'w', 'x', 'y', ' '].include?(f8[24]) &&
303	['0', '\|'].include?(f8[29])
304	types << 'AJ'	×
305	end
306
307	types.uniq!	×
308	return types	×
309	end
310
311	def newspaper_types	1✔
UNCOV 312	types = []	×
UNCOV 313	types << 'AN' if (bib_format == 'SE') && record['008'] &&	×
314	((record['008'].value[21] == 'n') \|\| (record['008'].value[22] == 'e'))	×
UNCOV 315	return types	×
316	end
317
318	# ! Mixed material: archi-V-e
319	# TYP MV Archive FMT F00-02 EQUAL MX
320	# TYP MV Archive LDR F08-01 EQUAL a
321	# !
322	# ! Mixed material: manuscript
323	# TYP MW Manuscript LDR F06-01 EQUAL [d,f,p,t]
324
325	def mixed_types	1✔
UNCOV 326	types = []	×
UNCOV 327	types << 'MV' if (bib_format == 'MX') \|\| (record.leader[8] == 'a')	×
UNCOV 328	types << 'MW' if %w[d f p t].include?(record.leader[6])	×
UNCOV 329	return types	×
330	end
331
332	# TYP CR CDROM 852## j MATCH cd-rom*
333	# TYP CR CDROM 852## j MATCH cdrom*
334	# TYP CR CDROM 852## j MATCH cd-rom*
335	# TYP CS Software 852## j MATCH software*
336
337	def software_types	1✔
UNCOV 338	types = []	×
UNCOV 339	self['852j'].each do \|j\|	×
UNCOV 340	types << 'CR' if /\Acd-?rom/i.match?(j)	×
UNCOV 341	types << 'CS' if /\Asoftware/i.match?(j)	×
342	end
UNCOV 343	types.uniq!	×
UNCOV 344	return types	×
345	end
346
347	# ! X (no icon) - Conference
348	# TYP XC Conference 008 F29-01 EQUAL 1
349	# TYP XC Conference 111## EXIST
350	# TYP XC Conference 711## EXIST
351	# TYP XC Conference 811## EXIST
352	# TYP XC Conference FMT F00-02 EQUAL CF
353	# 006 F00-01 EQUAL [a,s]
354	# 006 F12-01 EQUAL 1
355	# TYP XC Conference FMT F00-02 EQUAL MU
356	# 008 F30-01 EQUAL c
357	# TYP XC Conference FMT F00-02 EQUAL MU
358	# 008 F31-01 EQUAL c
359	# ! additional types defined for vufind extract
360	# TYP XC Conference 6#### xv MATCH congresses
361
362	def conference_types	1✔
363	# Get the easy stuff done first
364
UNCOV 365	return ['XC'] if (record['008'] && (record['008'].value[29] == '1')) \|\| record.fields(%w[111 711 811]).size > 0	×
366
UNCOV 367	if (bib_format == 'CF')	×
368	@record.fields('006').map { \|f\| f.value }.each do \|f\|	×
369	return ['XC'] if %w[a s].include?(f[0]) && (f[12] == '1')	×
370	end
371	end
372
UNCOV 373	if (bib_format == 'MU') && record['008'] &&	×
374	(record['008'].value[30..31] =~ /c/)	×
375	return ['XC']	×
376	end
377
UNCOV 378	return ['XC'] if @xv6XX.match?(/congresses/i)	×
379
380	# Nope.
UNCOV 381	return []	×
382	end
383
384	# ! X (no icon) - Statistics
385	# TYP XS Statistics 650## x MATCH Statistic*
386	# TYP XS Statistics 6#### x MATCH Statistic*
387	# TYP XS Statistics 6#### v MATCH Statistic*
388	# TYP XS Statistics FMT F00-02 EQUAL BK
389	# 008 F24-01 EQUAL s
390	# TYP XS Statistics FMT F00-02 EQUAL BK
391	# 008 F25-01 EQUAL s
392	# TYP XS Statistics FMT F00-02 EQUAL BK
393	# 008 F26-01 EQUAL s
394	# TYP XS Statistics FMT F00-02 EQUAL BK
395	# 008 F27-01 EQUAL s
396
397	def statistics_types	1✔
UNCOV 398	return ['XS'] if bib_format == 'BK' && (record['008'] && record['008'].value[24..27] =~ /s/)	×
399
UNCOV 400	return ['XS'] if @xv6XX.match?(/\AStatistic/i)	×
401
402	# Nope
UNCOV 403	return []	×
404	end
405
406	# TYP EN Encyclopedias 6#### xv MATCH encyclopedias
407	# TYP EN Encyclopedias 008 F24-01 EQUAL e
408	# TYP EN Encyclopedias 006 F07-01 EQUAL e
409	#
410	# TYP DI Dictionaries 6#### xv MATCH dictionaries
411	# TYP DI Dictionaries 008 F24-01 EQUAL d
412	# TYP DI Dictionaries 006 F07-01 EQUAL d
413	#
414	# TYP DR Directories 6#### xv MATCH directories
415	# TYP DR Directories 008 F24-01 EQUAL r
416	# TYP DR Directories 006 F07-01 EQUAL d
417
418	def reference_types	1✔
UNCOV 419	types = []	×
420
421	# Will need the 008[24] and 006[7]
UNCOV 422	f8_24 = self['008[24]']	×
UNCOV 423	f6_7 = self['006[7]']	×
424
UNCOV 425	types << 'EN' if (f8_24.include? 'e') \|\| (f6_7.include? 'e')	×
426
UNCOV 427	if f6_7.include? 'd'	×
UNCOV 428	types << 'DI'	×
UNCOV 429	types << 'DR'	×
430	end
431
UNCOV 432	types << 'DI' if f8_24.include? 'd'	×
433
UNCOV 434	types << 'DR' if f8_24.include? 'r'	×
435
UNCOV 436	types << 'EN' if @xv6XX.match?(/encyclopedias/i)	×
UNCOV 437	types << 'DI' if @xv6XX.match?(/dictionaries/i)	×
UNCOV 438	types << 'DR' if @xv6XX.match?(/directories/i)	×
439
UNCOV 440	types.uniq!	×
UNCOV 441	return types	×
442	end
443
444	# TYP BI Biography 6#### xv MATCH biography
445	# TYP BI Biography 6#### xv MATCH diaries
446	# TYP BI Biography 008 F34-01 EQUAL [a,b,c]
447	# TYP BI Biography 006 F17-01 EQUAL [a,b,c]
448
449	def biography_types	1✔
UNCOV 450	return ['BI'] if record['008'] && %w[a b c].include?(record['008'].value[34])	×
UNCOV 451	return ['BI'] if (%w[a b c] & self['006[17]']).size > 0	×
452
UNCOV 453	return ['BI'] if @xv6XX.match?(/(?:biography\|diaries)/i)	×
454
455	# Nope
UNCOV 456	return []	×
457	end
458
459	# TYP PP Photographs & Pictorial Works 6#### xv MATCH pictorial works
460	# TYP PP Photographs & Pictorial Works 6#### xv MATCH views
461	# TYP PP Photographs & Pictorial Works 6#### xv MATCH photographs
462	# TYP PP Photographs & Pictorial Works 6#### xv MATCH in art
463	# TYP PP Photographs & Pictorial Works 6#### xv MATCH aerial views
464	# TYP PP Photographs & Pictorial Works 6#### xv MATCH aerial photographs
465	# TYP PP Photographs & Pictorial Works 6#### xv MATCH art
466	# TYP PP Photographs & Pictorial Works 6#### xv MATCH cariacatures and cartoons
467	# TYP PP Photographs & Pictorial Works 6#### xv MATCH comic books
468	# TYP PP Photographs & Pictorial Works 6#### xv MATCH illustrations
469	# TYP PP Photographs & Pictorial Works 6#### xv MATCH drawings
470	# TYP PP Photographs & Pictorial Works 6#### xv MATCH slides
471
472	class << self	1✔
473	attr_accessor :pp_regexp	1✔
474	end
475
476	self.pp_regexp = Regexp.union ['pictorial works',	1✔
477	'views',
478	'photographs',
479	'in art',
480	'aerial views',
481	'aerial photographs',
482	'cariacatures and cartoons',
483	'comic books',
484	'illustrations',
485	'drawings',
486	'slides'].map { \|s\| Regexp.new('\b' + s + '\b', true) }	11✔
487	self.pp_regexp = Regexp.union(self.pp_regexp, /\bart\b/i)	1✔
488
489	def pp_types	1✔
UNCOV 490	if @xv6XX.match? self.class.pp_regexp	×
UNCOV 491	return ['PP']	×
492	else
UNCOV 493	return []	×
494	end
495	end
496
497	# TYP VG Video Games FMT F00-02 EQUAL CF
498	# 008 F26-01 EQUAL g
499
500	def videogame_types	1✔
UNCOV 501	if (bib_format == 'CF') && (self['008[26]'].include? 'g')	×
502	return ['VG']	×
503	else
UNCOV 504	return []	×
505	end
506	end
507	end

pulibrary / bibdata / f04bc944-f9b4-4a42-8b26-dcacd0e3e688

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous