spotlight-dor-resources 0.0.3 → 0.0.4
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.rubocop.yml +17 -2
- data/.rubocop_todo.yml +1 -186
- data/app/models/spotlight/resources/dor_resource.rb +1 -0
- data/app/models/spotlight/resources/purl.rb +4 -4
- data/app/models/spotlight/resources/searchworks.rb +4 -5
- data/lib/spotlight/dor/indexer.rb +187 -125
- data/lib/spotlight/dor/resources.rb +19 -7
- data/lib/spotlight/dor/resources/engine.rb +3 -2
- data/lib/spotlight/dor/resources/version.rb +2 -1
- data/spec/integration/gdor_integration_spec.rb +9 -9
- data/spec/{unit → lib}/spotlight/dor/indexer_spec.rb +209 -10
- data/spec/models/spotlight/resources/purl_spec.rb +45 -39
- data/spec/models/spotlight/resources/searchworks_spec.rb +47 -44
- data/spec/vcr_cassettes/gdor_indexing_integration_test/{should_have_a_doc_id.yml → has_a_doc_id.yml} +60 -40
- data/spec/vcr_cassettes/gdor_indexing_integration_test/{should_have_spotlight_data.yml → has_exhibit-specific_indexing.yml} +48 -32
- data/spec/vcr_cassettes/gdor_indexing_integration_test/{should_have_the_gdor_data.yml → has_spotlight_data.yml} +48 -32
- data/spec/vcr_cassettes/gdor_indexing_integration_test/{should_have_exhibit-specific_indexing.yml → has_the_gdor_data.yml} +48 -32
- data/spotlight-dor-resources.gemspec +22 -22
- metadata +11 -21
- data/spec/integration/indexer_integration_spec.rb +0 -28
- data/spec/vcr_cassettes/indexer_integration_tests/donor_tags/no_donor_tags_ssim_field_in_solr_doc_when_note_displayLabel_Donor_tags_not_in_MODS.yml +0 -1382
- data/spec/vcr_cassettes/indexer_integration_tests/donor_tags/solr_doc_has_donor_tags_ssim_field_when_note_displayLabel_Donor_tags_is_in_MODS.yml +0 -1602
- data/spec/vcr_cassettes/indexer_integration_tests/genre/no_genre_ssim_field_when_genre_not_in_MODS.yml +0 -6822
- data/spec/vcr_cassettes/indexer_integration_tests/genre/solr_doc_has_genre_ssim_field_when_genre_in_MODS.yml +0 -1390
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 8b1e1f0ddb963f81f8fd302a284ef53592146ef9
|
4
|
+
data.tar.gz: 03687ee4c19706fa434993ce33ad59beac3b522a
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 01970798dc91a7dd7b3f451244e71573d6c1129eb2c59a7f7758b3fdfb4d7d70c3d5bf8e3d99d1a7300b404dbc7142869898b1792729f8ab1099beb1f7113cd1
|
7
|
+
data.tar.gz: 121941bdd33d142d20f27bcacd25d112327324d5e1f84e0f190896bb9fa0c932562cdaeea2a29dae846d84b940085e3bcdd3b92a4f832854e74c24f46269621d
|
data/.rubocop.yml
CHANGED
@@ -4,9 +4,24 @@ require: rubocop-rspec
|
|
4
4
|
|
5
5
|
AllCops:
|
6
6
|
Exclude:
|
7
|
+
- 'Gemfile'
|
8
|
+
- 'bin/**/*'
|
9
|
+
- 'db/**/*'
|
10
|
+
- 'config/**/*'
|
7
11
|
- 'spec/internal/**/*'
|
8
12
|
- 'spec/test_app_templates/**/*'
|
13
|
+
- 'spec/spec_helper.rb'
|
14
|
+
- 'spec/teaspoon_env.rb'
|
15
|
+
- 'vendor/**/*'
|
16
|
+
RunRailsCops: true
|
9
17
|
|
10
|
-
Metrics/
|
18
|
+
Metrics/LineLength:
|
19
|
+
Max: 130
|
20
|
+
|
21
|
+
Style/StringLiterals:
|
22
|
+
Enabled: true
|
23
|
+
EnforcedStyle: single_quotes
|
24
|
+
|
25
|
+
RSpec/DescribeClass:
|
11
26
|
Exclude:
|
12
|
-
- '
|
27
|
+
- 'spec/integration/*'
|
data/.rubocop_todo.yml
CHANGED
@@ -1,188 +1,3 @@
|
|
1
|
-
# This configuration was generated by
|
2
|
-
# `rubocop --auto-gen-config`
|
3
|
-
# on 2015-10-29 09:20:45 -0700 using RuboCop version 0.34.2.
|
4
|
-
# The point is for the user to remove these configuration records
|
5
|
-
# one by one as the offenses are removed from the code base.
|
6
|
-
# Note that changes in the inspected code, or installation of new
|
7
|
-
# versions of RuboCop, may require this file to be generated again.
|
8
1
|
|
9
|
-
|
10
|
-
# Offense count: 1
|
11
|
-
# Configuration parameters: CountComments.
|
12
|
-
Metrics/ClassLength:
|
13
|
-
Max: 102
|
14
|
-
|
15
|
-
# Offense count: 84
|
16
|
-
# Configuration parameters: AllowURI, URISchemes.
|
17
|
-
Metrics/LineLength:
|
18
|
-
Max: 171
|
19
|
-
|
20
|
-
# Offense count: 1
|
21
|
-
# Configuration parameters: CountComments.
|
22
|
-
Metrics/MethodLength:
|
23
|
-
Max: 13
|
24
|
-
|
25
|
-
# Offense count: 1
|
26
|
-
# Cop supports --auto-correct.
|
27
|
-
Performance/StringReplacement:
|
28
|
-
Exclude:
|
29
|
-
- 'lib/spotlight/dor/indexer.rb'
|
30
|
-
|
31
|
-
# Offense count: 2
|
32
|
-
RSpec/DescribeClass:
|
33
|
-
Exclude:
|
34
|
-
- 'spec/integration/gdor_integration_spec.rb'
|
35
|
-
- 'spec/integration/indexer_integration_spec.rb'
|
36
|
-
|
37
|
-
# Offense count: 4
|
38
|
-
RSpec/DescribedClass:
|
39
|
-
Exclude:
|
40
|
-
- 'spec/models/spotlight/resources/purl_spec.rb'
|
41
|
-
- 'spec/models/spotlight/resources/searchworks_spec.rb'
|
42
|
-
|
43
|
-
# Offense count: 26
|
44
|
-
# Configuration parameters: CustomTransform, IgnoredWords.
|
45
|
-
RSpec/ExampleWording:
|
46
|
-
Exclude:
|
47
|
-
- 'spec/integration/gdor_integration_spec.rb'
|
48
|
-
- 'spec/models/spotlight/resources/purl_spec.rb'
|
49
|
-
- 'spec/models/spotlight/resources/searchworks_spec.rb'
|
50
|
-
|
51
|
-
# Offense count: 4
|
52
|
-
# Cop supports --auto-correct.
|
53
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles.
|
54
|
-
Style/AndOr:
|
55
|
-
Exclude:
|
56
|
-
- 'lib/spotlight/dor/indexer.rb'
|
57
|
-
- 'spec/spec_helper.rb'
|
58
|
-
|
59
|
-
# Offense count: 11
|
60
|
-
# Cop supports --auto-correct.
|
61
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles.
|
62
|
-
Style/BracesAroundHashParameters:
|
63
|
-
Exclude:
|
64
|
-
- 'spec/models/spotlight/resources/purl_spec.rb'
|
65
|
-
- 'spec/models/spotlight/resources/searchworks_spec.rb'
|
66
|
-
|
67
|
-
# Offense count: 6
|
68
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles.
|
69
2
|
Style/ClassAndModuleChildren:
|
70
|
-
|
71
|
-
- 'app/models/spotlight/resources/dor_resource.rb'
|
72
|
-
- 'app/models/spotlight/resources/harvestdor.rb'
|
73
|
-
- 'app/models/spotlight/resources/purl.rb'
|
74
|
-
- 'app/models/spotlight/resources/searchworks.rb'
|
75
|
-
- 'lib/spotlight/dor/indexer.rb'
|
76
|
-
- 'lib/spotlight/dor/resources/engine.rb'
|
77
|
-
|
78
|
-
# Offense count: 7
|
79
|
-
# Configuration parameters: Exclude.
|
80
|
-
Style/Documentation:
|
81
|
-
Exclude:
|
82
|
-
- 'app/models/spotlight/resources/dor_resource.rb'
|
83
|
-
- 'app/models/spotlight/resources/purl.rb'
|
84
|
-
- 'app/models/spotlight/resources/searchworks.rb'
|
85
|
-
- 'lib/spotlight/dor/indexer.rb'
|
86
|
-
- 'lib/spotlight/dor/resources.rb'
|
87
|
-
- 'lib/spotlight/dor/resources/engine.rb'
|
88
|
-
- 'lib/spotlight/dor/resources/version.rb'
|
89
|
-
|
90
|
-
# Offense count: 2
|
91
|
-
Style/DoubleNegation:
|
92
|
-
Exclude:
|
93
|
-
- 'app/models/spotlight/resources/purl.rb'
|
94
|
-
- 'app/models/spotlight/resources/searchworks.rb'
|
95
|
-
|
96
|
-
# Offense count: 1
|
97
|
-
# Cop supports --auto-correct.
|
98
|
-
Style/EmptyLines:
|
99
|
-
Exclude:
|
100
|
-
- 'spec/spec_helper.rb'
|
101
|
-
|
102
|
-
# Offense count: 4
|
103
|
-
# Cop supports --auto-correct.
|
104
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles.
|
105
|
-
Style/EmptyLinesAroundClassBody:
|
106
|
-
Exclude:
|
107
|
-
- 'app/models/spotlight/resources/purl.rb'
|
108
|
-
- 'app/models/spotlight/resources/searchworks.rb'
|
109
|
-
- 'lib/spotlight/dor/resources/engine.rb'
|
110
|
-
|
111
|
-
# Offense count: 1
|
112
|
-
# Cop supports --auto-correct.
|
113
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles.
|
114
|
-
Style/EmptyLinesAroundModuleBody:
|
115
|
-
Exclude:
|
116
|
-
- 'lib/spotlight/dor/resources.rb'
|
117
|
-
|
118
|
-
# Offense count: 9
|
119
|
-
# Cop supports --auto-correct.
|
120
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles.
|
121
|
-
Style/MethodDefParentheses:
|
122
|
-
Enabled: false
|
123
|
-
|
124
|
-
# Offense count: 4
|
125
|
-
# Cop supports --auto-correct.
|
126
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles, AllowInnerSlashes.
|
127
|
-
Style/RegexpLiteral:
|
128
|
-
Exclude:
|
129
|
-
- 'app/models/spotlight/resources/purl.rb'
|
130
|
-
- 'app/models/spotlight/resources/searchworks.rb'
|
131
|
-
|
132
|
-
# Offense count: 2
|
133
|
-
# Cop supports --auto-correct.
|
134
|
-
Style/SpaceAfterComma:
|
135
|
-
Exclude:
|
136
|
-
- 'lib/spotlight/dor/indexer.rb'
|
137
|
-
|
138
|
-
# Offense count: 2
|
139
|
-
# Cop supports --auto-correct.
|
140
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles, EnforcedStyleForEmptyBraces, SpaceBeforeBlockParameters.
|
141
|
-
Style/SpaceInsideBlockBraces:
|
142
|
-
Enabled: false
|
143
|
-
|
144
|
-
# Offense count: 31
|
145
|
-
# Cop supports --auto-correct.
|
146
|
-
# Configuration parameters: EnforcedStyle, EnforcedStyleForEmptyBraces, SupportedStyles.
|
147
|
-
Style/SpaceInsideHashLiteralBraces:
|
148
|
-
Enabled: false
|
149
|
-
|
150
|
-
# Offense count: 122
|
151
|
-
# Cop supports --auto-correct.
|
152
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles.
|
153
|
-
Style/StringLiterals:
|
154
|
-
Enabled: false
|
155
|
-
|
156
|
-
# Offense count: 1
|
157
|
-
# Cop supports --auto-correct.
|
158
|
-
# Configuration parameters: IgnoredMethods.
|
159
|
-
Style/SymbolProc:
|
160
|
-
Exclude:
|
161
|
-
- 'lib/spotlight/dor/indexer.rb'
|
162
|
-
|
163
|
-
# Offense count: 2
|
164
|
-
# Cop supports --auto-correct.
|
165
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles.
|
166
|
-
Style/TrailingBlankLines:
|
167
|
-
Exclude:
|
168
|
-
- 'spec/models/spotlight/resources/searchworks_spec.rb'
|
169
|
-
- 'spec/spec_helper.rb'
|
170
|
-
|
171
|
-
# Offense count: 12
|
172
|
-
# Cop supports --auto-correct.
|
173
|
-
Style/TrailingWhitespace:
|
174
|
-
Exclude:
|
175
|
-
- 'lib/spotlight/dor/resources.rb'
|
176
|
-
- 'spec/models/spotlight/resources/searchworks_spec.rb'
|
177
|
-
|
178
|
-
# Offense count: 1
|
179
|
-
# Cop supports --auto-correct.
|
180
|
-
# Configuration parameters: ExactNameMatch, AllowPredicates, AllowDSLWriters, IgnoreClassMethods, Whitelist.
|
181
|
-
Style/TrivialAccessors:
|
182
|
-
Exclude:
|
183
|
-
- 'lib/spotlight/dor/indexer.rb'
|
184
|
-
|
185
|
-
# Offense count: 4
|
186
|
-
# Configuration parameters: EnforcedStyle, SupportedStyles.
|
187
|
-
Style/VariableName:
|
188
|
-
Enabled: false
|
3
|
+
Enabled: false
|
@@ -1,14 +1,14 @@
|
|
1
1
|
module Spotlight::Resources
|
2
|
+
# Resource provider for PURL pages
|
2
3
|
class Purl < Spotlight::Resources::DorResource
|
3
4
|
self.weight = -1000
|
4
5
|
|
5
|
-
def self.can_provide?
|
6
|
-
|
6
|
+
def self.can_provide?(res)
|
7
|
+
res.url.match(%r{^https?://purl.stanford.edu/}).present?
|
7
8
|
end
|
8
9
|
|
9
10
|
def doc_id
|
10
|
-
url.match(
|
11
|
+
url.match(%r{^https?://purl.stanford.edu/([^#/\.]+)})[1]
|
11
12
|
end
|
12
|
-
|
13
13
|
end
|
14
14
|
end
|
@@ -1,15 +1,14 @@
|
|
1
1
|
module Spotlight::Resources
|
2
|
+
# Resource provider for searchworks pages
|
2
3
|
class Searchworks < Spotlight::Resources::DorResource
|
3
|
-
|
4
4
|
self.weight = -1000
|
5
5
|
|
6
|
-
def self.can_provide?
|
7
|
-
|
6
|
+
def self.can_provide?(res)
|
7
|
+
res.url.match(%r{^https?://searchworks[^\.]*.stanford.edu/}).present?
|
8
8
|
end
|
9
9
|
|
10
10
|
def doc_id
|
11
|
-
url.match(
|
11
|
+
url.match(%r{^https?://searchworks[^\.]*.stanford.edu/.*view/([^/\.#]+)})[1]
|
12
12
|
end
|
13
|
-
|
14
13
|
end
|
15
14
|
end
|
@@ -1,171 +1,233 @@
|
|
1
|
+
# rubocop:disable Metrics/ClassLength
|
1
2
|
# external gems
|
2
3
|
require 'gdor/indexer'
|
3
4
|
require 'solrizer'
|
4
|
-
|
5
|
+
|
5
6
|
module Spotlight::Dor
|
7
|
+
# Base class to harvest from DOR via harvestdor gem
|
6
8
|
class Indexer < GDor::Indexer
|
7
|
-
# add contentMetadata fields
|
8
|
-
before_index do |sdb, solr_doc|
|
9
|
-
Solrizer.insert_field(solr_doc, 'content_metadata_type', sdb.public_xml.xpath("/publicObject/contentMetadata/@type").text, :symbol, :displayable)
|
10
|
-
|
11
|
-
sdb.public_xml.xpath("/publicObject/contentMetadata").xpath('resource/file[@mimetype="image/jp2"]').each do |node|
|
12
|
-
file_id = node.attr('id').gsub(".jp2", '')
|
13
|
-
|
14
|
-
if node.attr('id') =~ /jp2$/ and !solr_doc[Solrizer.solr_name('content_metadata_first_image_file_name', :displayable)]
|
15
|
-
Solrizer.insert_field(solr_doc, 'content_metadata_first_image_file_name', file_id, :displayable)
|
16
|
-
Solrizer.insert_field(solr_doc, 'content_metadata_first_image_width', node.xpath('./imageData/@width').text, :displayable)
|
17
|
-
Solrizer.insert_field(solr_doc, 'content_metadata_first_image_height', node.xpath('./imageData/@height').text, :displayable)
|
18
|
-
end
|
19
|
-
|
20
|
-
Solrizer.insert_field(solr_doc, 'content_metadata_image_iiif_info', "https://stacks.stanford.edu/image/iiif/#{solr_doc[:id]}%2F#{file_id}/info.json", :displayable)
|
21
|
-
Solrizer.insert_field(solr_doc, 'thumbnail_square_url', "https://stacks.stanford.edu/image/#{solr_doc[:id]}/#{file_id}_square", :displayable)
|
22
|
-
Solrizer.insert_field(solr_doc, 'thumbnail_url', "https://stacks.stanford.edu/image/#{solr_doc[:id]}/#{file_id}_thumb", :displayable)
|
23
|
-
Solrizer.insert_field(solr_doc, 'large_image_url', "https://stacks.stanford.edu/image/#{solr_doc[:id]}/#{file_id}_large", :displayable)
|
24
|
-
Solrizer.insert_field(solr_doc, 'full_image_url', "https://stacks.stanford.edu/image/#{solr_doc[:id]}/#{file_id}_full", :displayable)
|
25
|
-
end
|
26
|
-
end
|
27
|
-
|
28
9
|
# tweak author_sort field from stanford-mods
|
29
10
|
before_index do |_sdb, solr_doc|
|
30
|
-
solr_doc[:author_sort] &&= solr_doc[:author_sort].
|
11
|
+
solr_doc[:author_sort] &&= solr_doc[:author_sort].tr("\uFFFF", "\uFFFD")
|
31
12
|
end
|
32
13
|
|
33
14
|
# add fields from raw mods
|
34
|
-
before_index :add_box
|
35
|
-
# see comment with add_donor_tags about Feigenbaum specific donor tags data
|
36
|
-
before_index :add_donor_tags
|
37
15
|
before_index :add_genre
|
38
|
-
before_index :add_folder
|
39
|
-
before_index :add_folder_name
|
40
|
-
before_index :add_series
|
41
|
-
before_index :mods_cartographics_indexing
|
42
16
|
|
43
|
-
def
|
44
|
-
@solr_client
|
45
|
-
end
|
46
|
-
|
47
|
-
def solr_document resource
|
17
|
+
def solr_document(resource)
|
48
18
|
doc_hash = super
|
49
19
|
run_hook :before_index, resource, doc_hash
|
50
20
|
doc_hash
|
51
21
|
end
|
52
22
|
|
53
|
-
def resource
|
23
|
+
def resource(druid)
|
54
24
|
Harvestdor::Indexer::Resource.new harvestdor, druid
|
55
25
|
end
|
56
26
|
|
57
27
|
private
|
58
28
|
|
59
|
-
# add
|
60
|
-
|
61
|
-
|
62
|
-
def add_box(sdb, solr_doc)
|
63
|
-
# see spec for data from actual collections
|
64
|
-
# _location.physicalLocation should find top level and relatedItem
|
65
|
-
box_num = sdb.smods_rec._location.physicalLocation.map do |node|
|
66
|
-
val = node.text
|
67
|
-
# note that this will also find Flatbox or Flat-box
|
68
|
-
match_data = val.match(/Box ?:? ?([^,|(Folder)]+)/i)
|
69
|
-
match_data[1].strip if match_data.present?
|
70
|
-
end.compact
|
71
|
-
|
72
|
-
solr_doc['box_ssi'] = box_num.first
|
29
|
+
# add plain MODS <genre> element data, not the SearchWorks genre values
|
30
|
+
def add_genre(sdb, solr_doc)
|
31
|
+
insert_field solr_doc, 'genre', sdb.smods_rec.genre.content, :symbol # this is a _ssim field
|
73
32
|
end
|
74
33
|
|
75
|
-
|
76
|
-
|
77
|
-
|
78
|
-
|
79
|
-
|
80
|
-
|
81
|
-
|
34
|
+
concerning :PhysicalLocation do
|
35
|
+
included do
|
36
|
+
before_index :add_box
|
37
|
+
before_index :add_folder
|
38
|
+
before_index :add_location
|
39
|
+
before_index :add_series
|
40
|
+
end
|
82
41
|
|
83
|
-
|
84
|
-
|
85
|
-
|
86
|
-
|
87
|
-
|
88
|
-
|
89
|
-
|
90
|
-
|
91
|
-
|
92
|
-
|
93
|
-
|
94
|
-
|
95
|
-
|
96
|
-
|
97
|
-
|
98
|
-
|
99
|
-
|
100
|
-
|
101
|
-
|
102
|
-
|
103
|
-
|
104
|
-
|
42
|
+
# add the box number to solr_doc as box_ssi field (note: single valued!)
|
43
|
+
# data in location/physicalLocation or in relatedItem/location/physicalLocation
|
44
|
+
# TODO: push this up to stanford-mods gem? or should it be hierarchical series/box/folder?
|
45
|
+
def add_box(sdb, solr_doc)
|
46
|
+
# see spec for data from actual collections
|
47
|
+
# _location.physicalLocation should find top level and relatedItem
|
48
|
+
box_num = sdb.smods_rec._location.physicalLocation.map do |node|
|
49
|
+
val = node.text
|
50
|
+
# note that this will also find Flatbox or Flat-box
|
51
|
+
match_data = val.match(/Box ?:? ?([^,|(Folder)]+)/i)
|
52
|
+
match_data[1].strip if match_data.present?
|
53
|
+
end.compact
|
54
|
+
|
55
|
+
solr_doc['box_ssi'] = box_num.first
|
56
|
+
end
|
57
|
+
|
58
|
+
# add the folder number to solr_doc as folder_ssi field (note: single valued!)
|
59
|
+
# data in location/physicalLocation or in relatedItem/location/physicalLocation
|
60
|
+
# TODO: push this up to stanford-mods gem? or should it be hierarchical series/box/folder?
|
61
|
+
def add_folder(sdb, solr_doc)
|
62
|
+
# see spec for data from actual collections
|
63
|
+
# _location.physicalLocation should find top level and relatedItem
|
64
|
+
folder_num = sdb.smods_rec._location.physicalLocation.map do |node|
|
65
|
+
val = node.text
|
66
|
+
|
67
|
+
match_data = if val =~ /\|/
|
68
|
+
# we assume the data is pipe-delimited, and may contain commas within values
|
69
|
+
val.match(/Folder ?:? ?([^|]+)/)
|
70
|
+
else
|
71
|
+
# the data should be comma-delimited, and may not contain commas within values
|
72
|
+
val.match(/Folder ?:? ?([^,]+)/)
|
73
|
+
end
|
74
|
+
|
75
|
+
match_data[1].strip if match_data.present?
|
76
|
+
end.compact
|
77
|
+
|
78
|
+
solr_doc['folder_ssi'] = folder_num.first
|
79
|
+
end
|
105
80
|
|
106
|
-
|
107
|
-
|
108
|
-
|
109
|
-
#
|
110
|
-
|
111
|
-
|
112
|
-
|
81
|
+
# add the physicalLocation as location_ssi field (note: single valued!)
|
82
|
+
# but only if it has series, box or folder data
|
83
|
+
# data in location/physicalLocation or in relatedItem/location/physicalLocation
|
84
|
+
# TODO: push this up to stanford-mods gem? or should it be hierarchical series/box/folder?
|
85
|
+
def add_location(sdb, solr_doc)
|
86
|
+
# see spec for data from actual collections
|
87
|
+
# _location.physicalLocation should find top level and relatedItem
|
88
|
+
loc = sdb.smods_rec._location.physicalLocation.map do |node|
|
89
|
+
node.text if node.text.match(/.*(Series)|(Accession)|(Folder)|(Box).*/i)
|
90
|
+
end.compact
|
91
|
+
|
92
|
+
solr_doc['location_ssi'] = loc.first
|
93
|
+
end
|
94
|
+
|
95
|
+
# add the series/accession 'number' to solr_doc as series_ssi field (note: single valued!)
|
96
|
+
# data in location/physicalLocation or in relatedItem/location/physicalLocation
|
97
|
+
# TODO: push this up to stanford-mods gem? or should it be hierarchical series/box/folder?
|
98
|
+
def add_series(sdb, solr_doc)
|
99
|
+
# see spec for data from actual collections
|
100
|
+
# _location.physicalLocation should find top level and relatedItem
|
101
|
+
series_num = sdb.smods_rec._location.physicalLocation.map do |node|
|
102
|
+
val = node.text
|
103
|
+
# feigenbaum uses 'Accession'
|
104
|
+
match_data = val.match(/(?:(?:Series)|(?:Accession)):? ([^,|]+)/i)
|
105
|
+
match_data[1].strip if match_data.present?
|
106
|
+
end.compact
|
107
|
+
|
108
|
+
solr_doc['series_ssi'] = series_num.first
|
109
|
+
end
|
113
110
|
end
|
114
111
|
|
115
|
-
|
116
|
-
|
117
|
-
|
112
|
+
concerning :ContentMetadata do
|
113
|
+
included do
|
114
|
+
before_index :add_content_metadata_fields
|
115
|
+
end
|
116
|
+
|
117
|
+
def add_content_metadata_fields(sdb, solr_doc)
|
118
|
+
content_metadata = sdb.public_xml.at_xpath('/publicObject/contentMetadata')
|
119
|
+
return unless content_metadata.present?
|
120
|
+
|
121
|
+
Solrizer.insert_field(solr_doc, 'content_metadata_type', content_metadata['type'], :symbol, :displayable)
|
122
|
+
|
123
|
+
images = content_metadata.xpath('resource/file[@mimetype="image/jp2"]').select { |node| node.attr('id') =~ /jp2$/ }
|
124
|
+
|
125
|
+
add_thumbnail_fields(images.first, solr_doc) if images.first
|
126
|
+
|
127
|
+
images.each do |image|
|
128
|
+
add_image_fields(image, solr_doc)
|
129
|
+
end
|
130
|
+
end
|
131
|
+
|
132
|
+
private
|
133
|
+
|
134
|
+
def add_thumbnail_fields(node, solr_doc)
|
135
|
+
file_id = node.attr('id').gsub('.jp2', '')
|
136
|
+
image_data = node.at_xpath('./imageData')
|
137
|
+
|
138
|
+
Solrizer.insert_field(solr_doc, 'content_metadata_first_image_file_name', file_id, :displayable)
|
139
|
+
Solrizer.insert_field(solr_doc, 'content_metadata_first_image_width', image_data['width'], :displayable)
|
140
|
+
Solrizer.insert_field(solr_doc, 'content_metadata_first_image_height', image_data['height'], :displayable)
|
141
|
+
end
|
142
|
+
|
143
|
+
def add_image_fields(node, solr_doc)
|
144
|
+
file_id = node.attr('id').gsub('.jp2', '')
|
145
|
+
base_url = stacks_iiif_url(solr_doc[:id], file_id)
|
146
|
+
|
147
|
+
Solrizer.insert_field(solr_doc, 'content_metadata_image_iiif_info', "#{base_url}/info.json", :displayable)
|
148
|
+
Solrizer.insert_field(solr_doc, 'thumbnail_square_url', "#{base_url}/square/100,100/0/default.jpg", :displayable)
|
149
|
+
Solrizer.insert_field(solr_doc, 'thumbnail_url', "#{base_url}/full/!400,400/0/default.jpg", :displayable)
|
150
|
+
Solrizer.insert_field(solr_doc, 'large_image_url', "#{base_url}/full/pct:25/0/default.jpg", :displayable)
|
151
|
+
Solrizer.insert_field(solr_doc, 'full_image_url', "#{base_url}/full/full/0/default.jpg", :displayable)
|
152
|
+
end
|
153
|
+
|
154
|
+
def stacks_iiif_url(druid, file_name)
|
155
|
+
"#{Spotlight::Dor::Resources::Engine.config.stacks_iiif_url}/#{druid}%2F#{file_name}"
|
156
|
+
end
|
118
157
|
end
|
119
158
|
|
120
|
-
|
121
|
-
|
122
|
-
|
123
|
-
|
124
|
-
#
|
125
|
-
|
126
|
-
|
127
|
-
|
128
|
-
|
129
|
-
|
130
|
-
|
131
|
-
|
132
|
-
|
133
|
-
|
159
|
+
concerning :FeigenbaumSpecificFields do
|
160
|
+
# These fields were specifically for the Feigenbaum exhibit. It is very
|
161
|
+
# likely it will go ununsed by other projects, but should be benign (since this field will not be created if
|
162
|
+
# this specific MODs note is not found.). Future work could refactor this to
|
163
|
+
# only create these fields on an as-needed basis.
|
164
|
+
|
165
|
+
included do
|
166
|
+
before_index :add_donor_tags
|
167
|
+
before_index :add_folder_name
|
168
|
+
end
|
169
|
+
|
170
|
+
def add_donor_tags(sdb, solr_doc)
|
171
|
+
donor_tags = sdb.smods_rec.note.select { |n| n.displayLabel == 'Donor tags' }.map(&:content)
|
172
|
+
insert_field solr_doc, 'donor_tags', donor_tags, :symbol # this is a _ssim field
|
173
|
+
end
|
174
|
+
|
175
|
+
# add the folder name to solr_doc as folder_name_ssi field (note: single valued!)
|
176
|
+
# data is specific to Feigenbaum collection and is in <note type='preferred citation'>
|
177
|
+
def add_folder_name(sdb, solr_doc)
|
178
|
+
# see spec for data examples
|
179
|
+
preferred_citation = sdb.smods_rec.note.select { |n| n.type_at == 'preferred citation' }.map(&:content)
|
180
|
+
match_data = preferred_citation.first.match(/Title: +(.+)/i) if preferred_citation.present?
|
181
|
+
solr_doc['folder_name_ssi'] = match_data[1].strip if match_data.present?
|
182
|
+
end
|
134
183
|
end
|
135
184
|
|
136
|
-
|
137
|
-
|
138
|
-
|
185
|
+
concerning :CartographicIndexing do
|
186
|
+
included do
|
187
|
+
before_index :mods_cartographics_indexing
|
188
|
+
end
|
139
189
|
|
140
|
-
|
141
|
-
|
190
|
+
def mods_cartographics_indexing(sdb, solr_doc)
|
191
|
+
coordinates = Array(sdb.smods_rec.subject.cartographics.coordinates)
|
142
192
|
|
143
|
-
|
193
|
+
insert_field(solr_doc, 'coordinates', coordinates.map(&:text), :stored_searchable)
|
144
194
|
|
145
|
-
|
195
|
+
solr_doc['point_bbox'] ||= []
|
196
|
+
solr_doc['point_bbox'] += coords_to_bboxes(coordinates)
|
197
|
+
end
|
198
|
+
|
199
|
+
private
|
146
200
|
|
147
|
-
|
148
|
-
|
201
|
+
def coords_to_bboxes(coordinates)
|
202
|
+
coordinates.select { |n| n.text =~ /^\(.*\)$/ }.map do |n|
|
203
|
+
coord_to_bbox(n.text)
|
204
|
+
end
|
205
|
+
end
|
149
206
|
|
150
|
-
|
151
|
-
|
207
|
+
def coord_to_bbox(coord)
|
208
|
+
bbox = coord.delete('(').delete(')')
|
209
|
+
|
210
|
+
lng, lat = bbox.split('/')
|
211
|
+
|
212
|
+
min_x, max_x = lng.split('--').map { |x| coord_to_decimal(x) }
|
213
|
+
max_y, min_y = lat.split('--').map { |y| coord_to_decimal(y) }
|
214
|
+
"#{min_x} #{min_y} #{max_x} #{max_y}"
|
152
215
|
end
|
153
|
-
end
|
154
|
-
# rubocop:enable Metrics/AbcSize
|
155
216
|
|
156
|
-
|
157
|
-
|
158
|
-
|
159
|
-
|
217
|
+
def coord_to_decimal(point)
|
218
|
+
regex = /(?<dir>[NESW])\s*(?<deg>\d+)°(?:(?<sec>\d+)ʹ)?/
|
219
|
+
match = regex.match(point)
|
220
|
+
dec = 0
|
160
221
|
|
161
|
-
|
162
|
-
|
163
|
-
|
222
|
+
dec += match['deg'].to_i
|
223
|
+
dec += match['sec'].to_f / 60
|
224
|
+
dec = -1 * dec if match['dir'] == 'W' || match['dir'] == 'S'
|
164
225
|
|
165
|
-
|
226
|
+
dec
|
227
|
+
end
|
166
228
|
end
|
167
229
|
|
168
|
-
def insert_field
|
230
|
+
def insert_field(solr_doc, field, values, *args)
|
169
231
|
Array(values).each do |v|
|
170
232
|
Solrizer.insert_field solr_doc, field, v, *args
|
171
233
|
end
|