search_solr_tools 6.0.0 → 6.2.0
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/CHANGELOG.md +9 -1
- data/bin/search_solr_tools +1 -13
- data/lib/search_solr_tools/config/environments.yaml +0 -32
- data/lib/search_solr_tools/harvesters/base.rb +0 -1
- data/lib/search_solr_tools/helpers/solr_format.rb +0 -15
- data/lib/search_solr_tools/helpers/translate_spatial_coverage.rb +0 -1
- data/lib/search_solr_tools/version.rb +1 -1
- data/lib/search_solr_tools.rb +1 -2
- data/search_solr_tools.gemspec +6 -6
- metadata +14 -56
- data/lib/search_solr_tools/harvesters/adc.rb +0 -49
- data/lib/search_solr_tools/harvesters/ade_auto_suggest.rb +0 -46
- data/lib/search_solr_tools/harvesters/bcodmo.rb +0 -64
- data/lib/search_solr_tools/harvesters/data_one.rb +0 -49
- data/lib/search_solr_tools/harvesters/echo.rb +0 -52
- data/lib/search_solr_tools/harvesters/eol.rb +0 -51
- data/lib/search_solr_tools/harvesters/gtnp.rb +0 -67
- data/lib/search_solr_tools/harvesters/ices.rb +0 -58
- data/lib/search_solr_tools/harvesters/ncdc_paleo.rb +0 -62
- data/lib/search_solr_tools/harvesters/nmi.rb +0 -34
- data/lib/search_solr_tools/harvesters/nodc.rb +0 -75
- data/lib/search_solr_tools/harvesters/oai.rb +0 -62
- data/lib/search_solr_tools/harvesters/pdc.rb +0 -40
- data/lib/search_solr_tools/harvesters/r2r.rb +0 -61
- data/lib/search_solr_tools/harvesters/rda.rb +0 -35
- data/lib/search_solr_tools/harvesters/tdar.rb +0 -71
- data/lib/search_solr_tools/harvesters/usgs.rb +0 -76
- data/lib/search_solr_tools/helpers/csw_iso_query_builder.rb +0 -29
- data/lib/search_solr_tools/helpers/data_one_format.rb +0 -74
- data/lib/search_solr_tools/helpers/iso_to_solr.rb +0 -97
- data/lib/search_solr_tools/helpers/iso_to_solr_format.rb +0 -197
- data/lib/search_solr_tools/helpers/ncdc_paleo_format.rb +0 -61
- data/lib/search_solr_tools/helpers/query_builder.rb +0 -13
- data/lib/search_solr_tools/helpers/r2r_format.rb +0 -25
- data/lib/search_solr_tools/helpers/selectors.rb +0 -22
- data/lib/search_solr_tools/helpers/tdar_format.rb +0 -70
- data/lib/search_solr_tools/helpers/usgs_format.rb +0 -50
- data/lib/search_solr_tools/selectors/adc.rb +0 -96
- data/lib/search_solr_tools/selectors/data_one.rb +0 -96
- data/lib/search_solr_tools/selectors/echo_iso.rb +0 -112
- data/lib/search_solr_tools/selectors/ices_iso.rb +0 -108
- data/lib/search_solr_tools/selectors/ncdc_paleo.rb +0 -90
- data/lib/search_solr_tools/selectors/nmi.rb +0 -107
- data/lib/search_solr_tools/selectors/nodc_iso.rb +0 -108
- data/lib/search_solr_tools/selectors/pdc_iso.rb +0 -109
- data/lib/search_solr_tools/selectors/r2r.rb +0 -115
- data/lib/search_solr_tools/selectors/rda.rb +0 -107
- data/lib/search_solr_tools/selectors/tdar_opensearch.rb +0 -91
- data/lib/search_solr_tools/selectors/usgs_iso.rb +0 -107
- data/lib/search_solr_tools/translators/bcodmo_json.rb +0 -89
- data/lib/search_solr_tools/translators/eol_to_solr.rb +0 -84
- data/lib/search_solr_tools/translators/gtnp_json.rb +0 -59
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: afbece000e765162dcb622479c3b9da58b534e4609b83ccb194a719ae5bb9c03
|
4
|
+
data.tar.gz: 012cb04bdced985254701d6f50fc99340c909fcff8000263c184dd9519282bbf
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 9b0c53d6ff652e840a8014f1ee0e31db62ad5faba634cd7aebf66ac90cd9a6865953ed081169f4b52ca96bf3ac135954a26d8aa6e5e7aa2bc1cf3298f946f260
|
7
|
+
data.tar.gz: 3975437ccf9540013f5c74e71a8e6f676273aab0e4b8a9a78df853b0010120b660f06a7a6ae630d6b40c6c6b8e56fc01b7562a6e2127508218c716e5ff4b41fc
|
data/CHANGELOG.md
CHANGED
@@ -1,4 +1,12 @@
|
|
1
|
-
## v6.
|
1
|
+
## v6.2.0 (2022-07-18)
|
2
|
+
|
3
|
+
- Remove deprecated harvesters and associated tests, helpers, etc.
|
4
|
+
|
5
|
+
## v6.1.0 (2022-07-14)
|
6
|
+
|
7
|
+
- Updated a few other dependencies that weren't at the newest versions.
|
8
|
+
|
9
|
+
## v6.0.0 (2022-07-14)
|
2
10
|
|
3
11
|
- Updated Ruby to 3.2.2, updated gem dependencies to more recent versions.
|
4
12
|
|
data/bin/search_solr_tools
CHANGED
@@ -120,20 +120,8 @@ class SolrHarvestCLI < Thor
|
|
120
120
|
no_tasks do
|
121
121
|
def harvester_map
|
122
122
|
{
|
123
|
-
'bco_dmo' => SearchSolrTools::Harvesters::BcoDmo,
|
124
|
-
'adc' => SearchSolrTools::Harvesters::Adc,
|
125
|
-
'echo' => SearchSolrTools::Harvesters::Echo,
|
126
|
-
'ices' => SearchSolrTools::Harvesters::Ices,
|
127
|
-
'nmi' => SearchSolrTools::Harvesters::Nmi,
|
128
|
-
'nodc' => SearchSolrTools::Harvesters::Nodc,
|
129
|
-
'r2r' => SearchSolrTools::Harvesters::R2R,
|
130
|
-
'rda' => SearchSolrTools::Harvesters::Rda,
|
131
|
-
'usgs' => SearchSolrTools::Harvesters::Usgs,
|
132
|
-
'tdar' => SearchSolrTools::Harvesters::Tdar,
|
133
|
-
'pdc' => SearchSolrTools::Harvesters::Pdc,
|
134
123
|
'nsidc' => SearchSolrTools::Harvesters::NsidcJson,
|
135
|
-
'nsidc_auto_suggest' => SearchSolrTools::Harvesters::NsidcAutoSuggest
|
136
|
-
'ade_auto_suggest' => SearchSolrTools::Harvesters::AdeAutoSuggest
|
124
|
+
'nsidc_auto_suggest' => SearchSolrTools::Harvesters::NsidcAutoSuggest
|
137
125
|
}
|
138
126
|
end
|
139
127
|
|
@@ -3,38 +3,6 @@
|
|
3
3
|
:collection_name: nsidc_oai
|
4
4
|
:collection_path: solr
|
5
5
|
:port: 8983
|
6
|
-
:bcodmo_url: http://www.bco-dmo.org/nsidc/arctic-deployments.json
|
7
|
-
:adc_url: https://arcticdata.io/metacat/d1/mn/v2/query/solr/select?q=northBoundCoord:%5B45.0%20TO%2090.0%5D
|
8
|
-
:data_one_url: https://cn.dataone.org/cn/v1/query/solr/select?q=northBoundCoord:%5B45.0%20TO%2090.0%5D
|
9
|
-
:echo_url: https://api.echo.nasa.gov/catalog-rest/echo_catalog/datasets.echo10?bounding_box=-180,45,180,90
|
10
|
-
:gtnp:
|
11
|
-
- http://www.gtnpdatabase.org/rest/boreholes/json
|
12
|
-
- http://www.gtnpdatabase.org/rest/activelayers/json
|
13
|
-
:ices_url: http://geo.ices.dk/geonetwork/srv/en/csw
|
14
|
-
:ncdc_paleo_url: https://gis.ncdc.noaa.gov/gptpaleo/csw
|
15
|
-
:nmi_url: http://arcticdata.met.no/metamod/oai
|
16
|
-
:nodc_url: https://data.nodc.noaa.gov/geoportal/csw
|
17
|
-
:pdc_url: http://www.polardata.ca/oai/provider
|
18
|
-
:rda_url: https://rda.ucar.edu/cgi-bin/oai
|
19
|
-
:tdar_url: http://core.tdar.org/search/rss
|
20
|
-
:usgs_url: https://www.sciencebase.gov/catalog/item/527cf4ede4b0850ea05182ee/csw
|
21
|
-
:eol:
|
22
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.SHEBA.thredds.xml
|
23
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.SBI.thredds.xml
|
24
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.PacMARS.thredds.xml
|
25
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BASE.thredds.xml
|
26
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.ATLAS.thredds.xml
|
27
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.ARC_MIP.thredds.xml
|
28
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.AMTS.thredds.xml
|
29
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BOREAS.thredds.xml
|
30
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BeringSea.thredds.xml
|
31
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.ARCSS.thredds.xml
|
32
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BEST.thredds.xml
|
33
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BSIERP.thredds.xml
|
34
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BARROW.thredds.xml
|
35
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.DBO.thredds.xml
|
36
|
-
- http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.ITEX.thredds.xml
|
37
|
-
:r2r_url: http://get.rvdata.us/services/cruise/
|
38
6
|
|
39
7
|
# Not using DCS API v2 here because not all retired datasets have their "retired"
|
40
8
|
# flag checked. For example, GLA01.033 is retired; GLA01.018 is not, but it
|
@@ -10,22 +10,7 @@ module SearchSolrTools
|
|
10
10
|
# rubocop:disable Metrics/ModuleLength
|
11
11
|
module SolrFormat
|
12
12
|
DATA_CENTER_NAMES = {
|
13
|
-
BCODMO: { short_name: 'BCO-DMO', long_name: 'Biological and Chemical Oceanography Data Management Office' },
|
14
|
-
ADC: { short_name: 'NSF ADC', long_name: 'NSF Arctic Data Center' },
|
15
|
-
DATA_ONE: { short_name: 'DataONE', long_name: 'DataONE' },
|
16
|
-
ECHO: { short_name: 'NASA ECHO', long_name: 'NASA Earth Observing System (EOS) Clearing House (ECHO)' },
|
17
|
-
EOL: { short_name: 'UCAR NCAR EOL', long_name: 'UCAR NCAR - Earth Observing Laboratory' },
|
18
|
-
GTNP: { short_name: 'GTN-P', long_name: 'Global Terrestrial Network for Permafrost' },
|
19
|
-
ICES: { short_name: 'ICES', long_name: 'International Council for the Exploration of the Sea' },
|
20
|
-
NCDC_PALEO: { short_name: 'NOAA WDS Paleo', long_name: 'NOAA’s National Centers for Environmental Information, World Data Service for Paleoclimatology' },
|
21
|
-
NMI: { short_name: 'Met.no', long_name: 'Norwegian Meteorological Institute' },
|
22
|
-
NODC: { short_name: 'NOAA NODC', long_name: 'NOAA National Oceanographic Data Center' },
|
23
13
|
NSIDC: { short_name: 'NSIDC', long_name: 'National Snow and Ice Data Center' },
|
24
|
-
PDC: { short_name: 'PDC', long_name: 'Polar Data Catalogue' },
|
25
|
-
R2R: { short_name: 'R2R', long_name: 'Rolling Deck to Repository' },
|
26
|
-
RDA: { short_name: 'UCAR NCAR RDA', long_name: 'UCAR NCAR Research Data Archive' },
|
27
|
-
TDAR: { short_name: 'tDAR', long_name: 'tDAR: The Digital Archaeological Record' },
|
28
|
-
USGS: { short_name: 'USGS ScienceBase', long_name: 'U.S. Geological Survey ScienceBase' }
|
29
14
|
}
|
30
15
|
|
31
16
|
NOT_SPECIFIED = 'Not specified'
|
data/lib/search_solr_tools.rb
CHANGED
@@ -1,10 +1,9 @@
|
|
1
1
|
require_relative 'search_solr_tools/config/environments'
|
2
2
|
require_relative 'search_solr_tools/version'
|
3
3
|
|
4
|
-
require_relative 'search_solr_tools/helpers/selectors'
|
5
4
|
require_relative 'search_solr_tools/helpers/harvest_status'
|
6
5
|
require_relative 'search_solr_tools/errors/harvest_error'
|
7
6
|
|
8
|
-
%w(
|
7
|
+
%w( harvesters translators ).each do |subdir|
|
9
8
|
Dir[File.join(__dir__, 'search_solr_tools', subdir, '*.rb')].each { |file| require file }
|
10
9
|
end
|
data/search_solr_tools.gemspec
CHANGED
@@ -28,9 +28,9 @@ Gem::Specification.new do |spec|
|
|
28
28
|
spec.add_runtime_dependency 'ffi-geos', '~> 2.4.0'
|
29
29
|
spec.add_runtime_dependency 'iso8601', '~> 0.13.0'
|
30
30
|
spec.add_runtime_dependency 'multi_json', '~> 1.15.0'
|
31
|
-
spec.add_runtime_dependency 'nokogiri', '~> 1.
|
31
|
+
spec.add_runtime_dependency 'nokogiri', '~> 1.15.3'
|
32
32
|
spec.add_runtime_dependency 'rest-client', '~> 2.1.0'
|
33
|
-
spec.add_runtime_dependency 'rgeo', '~>
|
33
|
+
spec.add_runtime_dependency 'rgeo', '~> 3.0.0'
|
34
34
|
spec.add_runtime_dependency 'rgeo-geojson', '~> 2.1.1'
|
35
35
|
spec.add_runtime_dependency 'rsolr', '~> 2.5.0'
|
36
36
|
spec.add_runtime_dependency 'thor', '~> 1.2.2'
|
@@ -39,11 +39,11 @@ Gem::Specification.new do |spec|
|
|
39
39
|
spec.add_development_dependency 'guard', '~> 2.18.0'
|
40
40
|
spec.add_development_dependency 'guard-rspec', '~> 4.7.3'
|
41
41
|
spec.add_development_dependency 'guard-rubocop', '~> 1.5.0'
|
42
|
-
spec.add_development_dependency 'listen', '3.
|
42
|
+
spec.add_development_dependency 'listen', '3.8.0'
|
43
43
|
spec.add_development_dependency 'rake', '~> 13.0.6'
|
44
|
-
spec.add_development_dependency 'rspec', '~> 3.
|
45
|
-
spec.add_development_dependency 'rubocop', '~> 1.
|
44
|
+
spec.add_development_dependency 'rspec', '~> 3.12.0'
|
45
|
+
spec.add_development_dependency 'rubocop', '~> 1.54.2'
|
46
46
|
spec.add_development_dependency 'rubocop-rake', '~> 0.6.0'
|
47
|
-
spec.add_development_dependency 'rubocop-rspec', '~> 2.
|
47
|
+
spec.add_development_dependency 'rubocop-rspec', '~> 2.22.0'
|
48
48
|
spec.add_development_dependency 'webmock', '~> 3.18.1'
|
49
49
|
end
|
metadata
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: search_solr_tools
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 6.
|
4
|
+
version: 6.2.0
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Chris Chalstrom
|
@@ -14,7 +14,7 @@ authors:
|
|
14
14
|
autorequire:
|
15
15
|
bindir: bin
|
16
16
|
cert_chain: []
|
17
|
-
date: 2023-07-
|
17
|
+
date: 2023-07-18 00:00:00.000000000 Z
|
18
18
|
dependencies:
|
19
19
|
- !ruby/object:Gem::Dependency
|
20
20
|
name: ffi-geos
|
@@ -64,14 +64,14 @@ dependencies:
|
|
64
64
|
requirements:
|
65
65
|
- - "~>"
|
66
66
|
- !ruby/object:Gem::Version
|
67
|
-
version: 1.
|
67
|
+
version: 1.15.3
|
68
68
|
type: :runtime
|
69
69
|
prerelease: false
|
70
70
|
version_requirements: !ruby/object:Gem::Requirement
|
71
71
|
requirements:
|
72
72
|
- - "~>"
|
73
73
|
- !ruby/object:Gem::Version
|
74
|
-
version: 1.
|
74
|
+
version: 1.15.3
|
75
75
|
- !ruby/object:Gem::Dependency
|
76
76
|
name: rest-client
|
77
77
|
requirement: !ruby/object:Gem::Requirement
|
@@ -92,14 +92,14 @@ dependencies:
|
|
92
92
|
requirements:
|
93
93
|
- - "~>"
|
94
94
|
- !ruby/object:Gem::Version
|
95
|
-
version:
|
95
|
+
version: 3.0.0
|
96
96
|
type: :runtime
|
97
97
|
prerelease: false
|
98
98
|
version_requirements: !ruby/object:Gem::Requirement
|
99
99
|
requirements:
|
100
100
|
- - "~>"
|
101
101
|
- !ruby/object:Gem::Version
|
102
|
-
version:
|
102
|
+
version: 3.0.0
|
103
103
|
- !ruby/object:Gem::Dependency
|
104
104
|
name: rgeo-geojson
|
105
105
|
requirement: !ruby/object:Gem::Requirement
|
@@ -204,14 +204,14 @@ dependencies:
|
|
204
204
|
requirements:
|
205
205
|
- - '='
|
206
206
|
- !ruby/object:Gem::Version
|
207
|
-
version: 3.
|
207
|
+
version: 3.8.0
|
208
208
|
type: :development
|
209
209
|
prerelease: false
|
210
210
|
version_requirements: !ruby/object:Gem::Requirement
|
211
211
|
requirements:
|
212
212
|
- - '='
|
213
213
|
- !ruby/object:Gem::Version
|
214
|
-
version: 3.
|
214
|
+
version: 3.8.0
|
215
215
|
- !ruby/object:Gem::Dependency
|
216
216
|
name: rake
|
217
217
|
requirement: !ruby/object:Gem::Requirement
|
@@ -232,28 +232,28 @@ dependencies:
|
|
232
232
|
requirements:
|
233
233
|
- - "~>"
|
234
234
|
- !ruby/object:Gem::Version
|
235
|
-
version: 3.
|
235
|
+
version: 3.12.0
|
236
236
|
type: :development
|
237
237
|
prerelease: false
|
238
238
|
version_requirements: !ruby/object:Gem::Requirement
|
239
239
|
requirements:
|
240
240
|
- - "~>"
|
241
241
|
- !ruby/object:Gem::Version
|
242
|
-
version: 3.
|
242
|
+
version: 3.12.0
|
243
243
|
- !ruby/object:Gem::Dependency
|
244
244
|
name: rubocop
|
245
245
|
requirement: !ruby/object:Gem::Requirement
|
246
246
|
requirements:
|
247
247
|
- - "~>"
|
248
248
|
- !ruby/object:Gem::Version
|
249
|
-
version: 1.
|
249
|
+
version: 1.54.2
|
250
250
|
type: :development
|
251
251
|
prerelease: false
|
252
252
|
version_requirements: !ruby/object:Gem::Requirement
|
253
253
|
requirements:
|
254
254
|
- - "~>"
|
255
255
|
- !ruby/object:Gem::Version
|
256
|
-
version: 1.
|
256
|
+
version: 1.54.2
|
257
257
|
- !ruby/object:Gem::Dependency
|
258
258
|
name: rubocop-rake
|
259
259
|
requirement: !ruby/object:Gem::Requirement
|
@@ -274,14 +274,14 @@ dependencies:
|
|
274
274
|
requirements:
|
275
275
|
- - "~>"
|
276
276
|
- !ruby/object:Gem::Version
|
277
|
-
version: 2.
|
277
|
+
version: 2.22.0
|
278
278
|
type: :development
|
279
279
|
prerelease: false
|
280
280
|
version_requirements: !ruby/object:Gem::Requirement
|
281
281
|
requirements:
|
282
282
|
- - "~>"
|
283
283
|
- !ruby/object:Gem::Version
|
284
|
-
version: 2.
|
284
|
+
version: 2.22.0
|
285
285
|
- !ruby/object:Gem::Dependency
|
286
286
|
name: webmock
|
287
287
|
requirement: !ruby/object:Gem::Requirement
|
@@ -321,59 +321,17 @@ files:
|
|
321
321
|
- lib/search_solr_tools/config/environments.rb
|
322
322
|
- lib/search_solr_tools/config/environments.yaml
|
323
323
|
- lib/search_solr_tools/errors/harvest_error.rb
|
324
|
-
- lib/search_solr_tools/harvesters/adc.rb
|
325
|
-
- lib/search_solr_tools/harvesters/ade_auto_suggest.rb
|
326
324
|
- lib/search_solr_tools/harvesters/auto_suggest.rb
|
327
325
|
- lib/search_solr_tools/harvesters/base.rb
|
328
|
-
- lib/search_solr_tools/harvesters/bcodmo.rb
|
329
|
-
- lib/search_solr_tools/harvesters/data_one.rb
|
330
|
-
- lib/search_solr_tools/harvesters/echo.rb
|
331
|
-
- lib/search_solr_tools/harvesters/eol.rb
|
332
|
-
- lib/search_solr_tools/harvesters/gtnp.rb
|
333
|
-
- lib/search_solr_tools/harvesters/ices.rb
|
334
|
-
- lib/search_solr_tools/harvesters/ncdc_paleo.rb
|
335
|
-
- lib/search_solr_tools/harvesters/nmi.rb
|
336
|
-
- lib/search_solr_tools/harvesters/nodc.rb
|
337
326
|
- lib/search_solr_tools/harvesters/nsidc_auto_suggest.rb
|
338
327
|
- lib/search_solr_tools/harvesters/nsidc_json.rb
|
339
|
-
- lib/search_solr_tools/harvesters/oai.rb
|
340
|
-
- lib/search_solr_tools/harvesters/pdc.rb
|
341
|
-
- lib/search_solr_tools/harvesters/r2r.rb
|
342
|
-
- lib/search_solr_tools/harvesters/rda.rb
|
343
|
-
- lib/search_solr_tools/harvesters/tdar.rb
|
344
|
-
- lib/search_solr_tools/harvesters/usgs.rb
|
345
328
|
- lib/search_solr_tools/helpers/bounding_box_util.rb
|
346
|
-
- lib/search_solr_tools/helpers/csw_iso_query_builder.rb
|
347
|
-
- lib/search_solr_tools/helpers/data_one_format.rb
|
348
329
|
- lib/search_solr_tools/helpers/facet_configuration.rb
|
349
330
|
- lib/search_solr_tools/helpers/harvest_status.rb
|
350
331
|
- lib/search_solr_tools/helpers/iso_namespaces.rb
|
351
|
-
- lib/search_solr_tools/helpers/iso_to_solr.rb
|
352
|
-
- lib/search_solr_tools/helpers/iso_to_solr_format.rb
|
353
|
-
- lib/search_solr_tools/helpers/ncdc_paleo_format.rb
|
354
|
-
- lib/search_solr_tools/helpers/query_builder.rb
|
355
|
-
- lib/search_solr_tools/helpers/r2r_format.rb
|
356
|
-
- lib/search_solr_tools/helpers/selectors.rb
|
357
332
|
- lib/search_solr_tools/helpers/solr_format.rb
|
358
|
-
- lib/search_solr_tools/helpers/tdar_format.rb
|
359
333
|
- lib/search_solr_tools/helpers/translate_spatial_coverage.rb
|
360
334
|
- lib/search_solr_tools/helpers/translate_temporal_coverage.rb
|
361
|
-
- lib/search_solr_tools/helpers/usgs_format.rb
|
362
|
-
- lib/search_solr_tools/selectors/adc.rb
|
363
|
-
- lib/search_solr_tools/selectors/data_one.rb
|
364
|
-
- lib/search_solr_tools/selectors/echo_iso.rb
|
365
|
-
- lib/search_solr_tools/selectors/ices_iso.rb
|
366
|
-
- lib/search_solr_tools/selectors/ncdc_paleo.rb
|
367
|
-
- lib/search_solr_tools/selectors/nmi.rb
|
368
|
-
- lib/search_solr_tools/selectors/nodc_iso.rb
|
369
|
-
- lib/search_solr_tools/selectors/pdc_iso.rb
|
370
|
-
- lib/search_solr_tools/selectors/r2r.rb
|
371
|
-
- lib/search_solr_tools/selectors/rda.rb
|
372
|
-
- lib/search_solr_tools/selectors/tdar_opensearch.rb
|
373
|
-
- lib/search_solr_tools/selectors/usgs_iso.rb
|
374
|
-
- lib/search_solr_tools/translators/bcodmo_json.rb
|
375
|
-
- lib/search_solr_tools/translators/eol_to_solr.rb
|
376
|
-
- lib/search_solr_tools/translators/gtnp_json.rb
|
377
335
|
- lib/search_solr_tools/translators/nsidc_json.rb
|
378
336
|
- lib/search_solr_tools/version.rb
|
379
337
|
- search_solr_tools.gemspec
|
@@ -1,49 +0,0 @@
|
|
1
|
-
require_relative 'base'
|
2
|
-
|
3
|
-
module SearchSolrTools
|
4
|
-
module Harvesters
|
5
|
-
class Adc < Base
|
6
|
-
def initialize(env = 'development', die_on_failure = false)
|
7
|
-
super
|
8
|
-
@page_size = 250
|
9
|
-
@translator = Helpers::IsoToSolr.new :adc
|
10
|
-
end
|
11
|
-
|
12
|
-
def harvest_and_delete
|
13
|
-
puts "Running harvest of adc catalog from #{metadata_url}"
|
14
|
-
super(method(:harvest_adc_into_solr), "data_centers:\"#{Helpers::SolrFormat::DATA_CENTER_NAMES[:ADC][:long_name]}\"")
|
15
|
-
end
|
16
|
-
|
17
|
-
def harvest_adc_into_solr
|
18
|
-
start = 0
|
19
|
-
while (entries = get_results_from_adc(start)) && (entries.length > 0)
|
20
|
-
begin
|
21
|
-
insert_solr_docs(get_docs_with_translated_entries_from_adc(entries))
|
22
|
-
rescue => e
|
23
|
-
puts "ERROR: #{e}\n\n"
|
24
|
-
raise e if @die_on_failure
|
25
|
-
end
|
26
|
-
start += @page_size
|
27
|
-
end
|
28
|
-
end
|
29
|
-
|
30
|
-
def get_results_from_adc(start)
|
31
|
-
get_results(build_request(start, @page_size), './response/result/doc')
|
32
|
-
end
|
33
|
-
|
34
|
-
def metadata_url
|
35
|
-
SolrEnvironments[@environment][:adc_url]
|
36
|
-
end
|
37
|
-
|
38
|
-
def get_docs_with_translated_entries_from_adc(entries)
|
39
|
-
entries.map do |e|
|
40
|
-
create_new_solr_add_doc_with_child(@translator.translate(e).root)
|
41
|
-
end
|
42
|
-
end
|
43
|
-
|
44
|
-
def build_request(start = 0, max_records = 100)
|
45
|
-
"#{metadata_url}&start=#{start}&rows=#{max_records}"
|
46
|
-
end
|
47
|
-
end
|
48
|
-
end
|
49
|
-
end
|
@@ -1,46 +0,0 @@
|
|
1
|
-
require_relative 'auto_suggest'
|
2
|
-
|
3
|
-
module SearchSolrTools
|
4
|
-
module Harvesters
|
5
|
-
class AdeAutoSuggest < AutoSuggest
|
6
|
-
def harvest_and_delete
|
7
|
-
puts 'Building auto-suggest indexes for ADE'
|
8
|
-
super(method(:harvest), 'source:"ADE"', @env_settings[:auto_suggest_collection_name])
|
9
|
-
end
|
10
|
-
|
11
|
-
def harvest
|
12
|
-
url = "#{solr_url}/#{@env_settings[:collection_name]}/select?q=*%3A*&fq=source%3AADE&fq=spatial:[45.0,-180.0+TO+90.0,180.0]&rows=0&wt=json&indent=true&facet=true&facet.mincount=1&facet.sort=count&facet.limit=-1"
|
13
|
-
super url, fields
|
14
|
-
end
|
15
|
-
|
16
|
-
def fields
|
17
|
-
{
|
18
|
-
'full_keywords_and_parameters' => { weight: 2, source: 'ADE', creator: method(:keyword_creator) },
|
19
|
-
'full_authors' => { weight: 1, source: 'ADE', creator: method(:author_creator) }
|
20
|
-
}
|
21
|
-
end
|
22
|
-
|
23
|
-
def split_creator(value, count, field_weight, source, split_regex)
|
24
|
-
add_docs = []
|
25
|
-
value.downcase.split(split_regex).each do |v|
|
26
|
-
v = v.strip.chomp('/')
|
27
|
-
add_docs.concat(ade_length_limit_creator(v, count, field_weight, source)) unless v.nil? || v.empty?
|
28
|
-
end
|
29
|
-
add_docs
|
30
|
-
end
|
31
|
-
|
32
|
-
def keyword_creator(value, count, field_weight, source)
|
33
|
-
split_creator value, count, field_weight, source, %r{/ [\/ \>]+ /}
|
34
|
-
end
|
35
|
-
|
36
|
-
def author_creator(value, count, field_weight, source)
|
37
|
-
split_creator value, count, field_weight, source, %r{/;/}
|
38
|
-
end
|
39
|
-
|
40
|
-
def ade_length_limit_creator(value, count, field_weight, source)
|
41
|
-
return [] if value.length > 80
|
42
|
-
standard_add_creator value, count, field_weight, source
|
43
|
-
end
|
44
|
-
end
|
45
|
-
end
|
46
|
-
end
|
@@ -1,64 +0,0 @@
|
|
1
|
-
require_relative 'base'
|
2
|
-
require 'json'
|
3
|
-
require 'rest-client'
|
4
|
-
|
5
|
-
module SearchSolrTools
|
6
|
-
module Harvesters
|
7
|
-
# Harvests data from BcoDmo endpoint, translates and adds it to solr
|
8
|
-
class BcoDmo < Base
|
9
|
-
def initialize(env = 'development', die_on_failure = false)
|
10
|
-
super env, die_on_failure
|
11
|
-
@translator = Translators::BcodmoJsonToSolr.new
|
12
|
-
@wkt_parser = RGeo::WKRep::WKTParser.new(nil, {}) # (factory_generator_=nil,
|
13
|
-
end
|
14
|
-
|
15
|
-
def harvest_and_delete
|
16
|
-
puts "Running harvest of BCO-DMO catalog from #{bcodmo_url}"
|
17
|
-
super(method(:harvest_bcodmo_into_solr), "data_centers:\"#{Helpers::SolrFormat::DATA_CENTER_NAMES[:BCODMO][:long_name]}\"")
|
18
|
-
end
|
19
|
-
|
20
|
-
def bcodmo_url
|
21
|
-
SolrEnvironments[@environment][:bcodmo_url]
|
22
|
-
end
|
23
|
-
|
24
|
-
def harvest_bcodmo_into_solr
|
25
|
-
result = translate_bcodmo
|
26
|
-
insert_solr_docs(result[:add_docs], Base::JSON_CONTENT_TYPE)
|
27
|
-
|
28
|
-
errors_exist = result[:failure_ids].length > 0
|
29
|
-
fail 'Failed to harvest some records from BCO-DMO' if errors_exist && @die_on_failure
|
30
|
-
end
|
31
|
-
|
32
|
-
def translate_bcodmo
|
33
|
-
documents = []
|
34
|
-
failure_ids = []
|
35
|
-
request_json(SolrEnvironments[@environment][:bcodmo_url]).each do |record|
|
36
|
-
geometry = request_json(record['geometryUrl'])
|
37
|
-
results = parse_record(record, geometry)
|
38
|
-
results[:documents].each { |d| documents << d }
|
39
|
-
results[:failure_ids].each { |id| failure_ids << id }
|
40
|
-
end
|
41
|
-
{ add_docs: documents, failure_ids: failure_ids }
|
42
|
-
end
|
43
|
-
|
44
|
-
def request_json(url)
|
45
|
-
puts "Request: #{url}"
|
46
|
-
JSON.parse(RestClient.get(url))
|
47
|
-
end
|
48
|
-
|
49
|
-
def parse_record(record, geometry)
|
50
|
-
documents = []
|
51
|
-
failure_ids = []
|
52
|
-
begin
|
53
|
-
JSON.parse(RestClient.get(record['datasets'])).each do |dataset|
|
54
|
-
documents << { 'add' => { 'doc' => @translator.translate(dataset, record, geometry) } }
|
55
|
-
end
|
56
|
-
rescue => e
|
57
|
-
puts "Failed to add record #{record['id']} with error #{e} (#{e.message}) : #{e.backtrace.join("\n")}"
|
58
|
-
failure_ids << record['id']
|
59
|
-
end
|
60
|
-
{ documents: documents, failure_ids: failure_ids }
|
61
|
-
end
|
62
|
-
end
|
63
|
-
end
|
64
|
-
end
|
@@ -1,49 +0,0 @@
|
|
1
|
-
require_relative 'base'
|
2
|
-
|
3
|
-
module SearchSolrTools
|
4
|
-
module Harvesters
|
5
|
-
class DataOne < Base
|
6
|
-
def initialize(env = 'development', die_on_failure = false)
|
7
|
-
super
|
8
|
-
@page_size = 250
|
9
|
-
@translator = Helpers::IsoToSolr.new :data_one
|
10
|
-
end
|
11
|
-
|
12
|
-
def harvest_and_delete
|
13
|
-
puts "Running harvest of dataONE catalog from #{metadata_url}"
|
14
|
-
super(method(:harvest_data_one_into_solr), "data_centers:\"#{Helpers::SolrFormat::DATA_CENTER_NAMES[:DATA_ONE][:long_name]}\"")
|
15
|
-
end
|
16
|
-
|
17
|
-
def harvest_data_one_into_solr
|
18
|
-
start = 0
|
19
|
-
while (entries = get_results_from_data_one(start)) && (entries.length > 0)
|
20
|
-
begin
|
21
|
-
insert_solr_docs(get_docs_with_translated_entries_from_data_one(entries))
|
22
|
-
rescue => e
|
23
|
-
puts "ERROR: #{e}\n\n"
|
24
|
-
raise e if @die_on_failure
|
25
|
-
end
|
26
|
-
start += @page_size
|
27
|
-
end
|
28
|
-
end
|
29
|
-
|
30
|
-
def get_results_from_data_one(start)
|
31
|
-
get_results(build_request(start, @page_size), './response/result/doc')
|
32
|
-
end
|
33
|
-
|
34
|
-
def metadata_url
|
35
|
-
SolrEnvironments[@environment][:data_one_url]
|
36
|
-
end
|
37
|
-
|
38
|
-
def get_docs_with_translated_entries_from_data_one(entries)
|
39
|
-
entries.map do |e|
|
40
|
-
create_new_solr_add_doc_with_child(@translator.translate(e).root)
|
41
|
-
end
|
42
|
-
end
|
43
|
-
|
44
|
-
def build_request(start = 0, max_records = 100)
|
45
|
-
"#{metadata_url}&start=#{start}&rows=#{max_records}"
|
46
|
-
end
|
47
|
-
end
|
48
|
-
end
|
49
|
-
end
|
@@ -1,52 +0,0 @@
|
|
1
|
-
require_relative 'base'
|
2
|
-
|
3
|
-
module SearchSolrTools
|
4
|
-
module Harvesters
|
5
|
-
# Harvests data from ECHO and inserts it into Solr after it has been translated
|
6
|
-
class Echo < Base
|
7
|
-
def initialize(env = 'development', die_on_failure = false)
|
8
|
-
super env, die_on_failure
|
9
|
-
@page_size = 100
|
10
|
-
@translator = Helpers::IsoToSolr.new :echo
|
11
|
-
end
|
12
|
-
|
13
|
-
def harvest_and_delete
|
14
|
-
puts "Running harvest of ECHO catalog from #{echo_url}"
|
15
|
-
super(method(:harvest_echo_into_solr), "data_centers:\"#{Helpers::SolrFormat::DATA_CENTER_NAMES[:ECHO][:long_name]}\"")
|
16
|
-
end
|
17
|
-
|
18
|
-
# get translated entries from ECHO and add them to Solr
|
19
|
-
# this is the main entry point for the class
|
20
|
-
def harvest_echo_into_solr
|
21
|
-
page_num = 1
|
22
|
-
while (entries = get_results_from_echo(page_num)) && (entries.length > 0)
|
23
|
-
begin
|
24
|
-
insert_solr_docs get_docs_with_translated_entries_from_echo(entries)
|
25
|
-
rescue => e
|
26
|
-
puts "ERROR: #{e}\n\n"
|
27
|
-
raise e if @die_on_failure
|
28
|
-
end
|
29
|
-
page_num += 1
|
30
|
-
end
|
31
|
-
end
|
32
|
-
|
33
|
-
def echo_url
|
34
|
-
SolrEnvironments[@environment][:echo_url]
|
35
|
-
end
|
36
|
-
|
37
|
-
def get_results_from_echo(page_num)
|
38
|
-
get_results build_request(@page_size, page_num), './/results/result', 'application/echo10+xml'
|
39
|
-
end
|
40
|
-
|
41
|
-
def get_docs_with_translated_entries_from_echo(entries)
|
42
|
-
entries.map do |entry|
|
43
|
-
create_new_solr_add_doc_with_child(@translator.translate(entry).root)
|
44
|
-
end
|
45
|
-
end
|
46
|
-
|
47
|
-
def build_request(max_records = '25', page_num = '1')
|
48
|
-
echo_url + '&page_size=' + max_records.to_s + '&page_num=' + page_num.to_s
|
49
|
-
end
|
50
|
-
end
|
51
|
-
end
|
52
|
-
end
|