search_solr_tools 6.1.0 → 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 +4 -0
- 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
- metadata +2 -44
- 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
@@ -1,84 +0,0 @@
|
|
1
|
-
require 'search_solr_tools'
|
2
|
-
require_relative '../helpers/solr_format'
|
3
|
-
require_relative '../helpers/translate_temporal_coverage'
|
4
|
-
require_relative '../helpers/translate_spatial_coverage'
|
5
|
-
require_relative '../helpers/bounding_box_util'
|
6
|
-
|
7
|
-
module SearchSolrTools
|
8
|
-
module Translators
|
9
|
-
# Translates an EOL THREDDS dataset link set into a SOLR json ingest record
|
10
|
-
class EolToSolr
|
11
|
-
# rubocop:disable Metrics/AbcSize
|
12
|
-
# rubocop:disable Metrics/MethodLength
|
13
|
-
|
14
|
-
def translate(title_metadata, dataset_metadata)
|
15
|
-
temporal_coverage_values = Helpers::TranslateTemporalCoverage.translate_coverages get_time_coverages(dataset_metadata)
|
16
|
-
rev_date = dataset_metadata.xpath('//xmlns:date[@type="metadataCreated"]').text
|
17
|
-
geospatial_coverage = parse_geospatial_coverages(dataset_metadata)
|
18
|
-
{
|
19
|
-
'title' => title_metadata.xpath('//xmlns:dataset').first['name'],
|
20
|
-
'authoritative_id' => title_metadata.xpath('//xmlns:dataset').first['ID'],
|
21
|
-
'data_centers' => Helpers::SolrFormat::DATA_CENTER_NAMES[:EOL][:long_name],
|
22
|
-
'facet_data_center' => "#{Helpers::SolrFormat::DATA_CENTER_NAMES[:EOL][:long_name]} | #{Helpers::SolrFormat::DATA_CENTER_NAMES[:EOL][:short_name]}",
|
23
|
-
'summary' => dataset_metadata.xpath('//xmlns:documentation[@type="summary"]').text,
|
24
|
-
'temporal_coverages' => temporal_coverage_values['temporal_coverages'],
|
25
|
-
'temporal_duration' => temporal_coverage_values['temporal_duration'],
|
26
|
-
'temporal' => temporal_coverage_values['temporal'],
|
27
|
-
'facet_temporal_duration' => temporal_coverage_values['facet_temporal_duration'],
|
28
|
-
'last_revision_date' => rev_date.empty? ? Helpers::SolrFormat.date_str(DateTime.now) : Helpers::SolrFormat.date_str(rev_date),
|
29
|
-
'source' => 'ADE',
|
30
|
-
'keywords' => dataset_metadata.xpath('//xmlns:keyword').map(&:text),
|
31
|
-
'authors' => dataset_metadata.xpath('//xmlns:contributor[@role="author"]').map { |node| parse_eol_authors(node.text) }.join(', '),
|
32
|
-
'dataset_url' => eol_dataset_url(dataset_metadata),
|
33
|
-
'facet_spatial_coverage' => Helpers::BoundingBoxUtil.box_global?(geospatial_coverage),
|
34
|
-
'facet_spatial_scope' => Helpers::SolrFormat.get_spatial_scope_facet_with_bounding_box(geospatial_coverage),
|
35
|
-
'spatial_coverages' => %i(south west north east).map { |d| geospatial_coverage[d] }.join(' '),
|
36
|
-
'spatial_area' => spatial_coverage_to_spatial_area(geospatial_coverage),
|
37
|
-
'spatial' => %i(west south east north).map { |d| geospatial_coverage[d] }.join(' ')
|
38
|
-
}
|
39
|
-
end
|
40
|
-
|
41
|
-
def eol_dataset_url(node)
|
42
|
-
node.xpath('//xmlns:documentation[@xlink:href]').each do |doc|
|
43
|
-
return doc['xlink:href'] if doc['xlink:href'].match('http://data.eol.ucar.edu/codiac/dss/id=(\S*)')
|
44
|
-
end
|
45
|
-
rescue Nokogiri::XML::XPath::SyntaxError
|
46
|
-
puts "Warning - no documentation URL found in the following node: #{node.to_html}"
|
47
|
-
end
|
48
|
-
|
49
|
-
def parse_eol_authors(author)
|
50
|
-
if author.include?(' AT ') && author.include?(' dot ')
|
51
|
-
author = author[0..author.rindex(',') - 1]
|
52
|
-
end
|
53
|
-
author
|
54
|
-
end
|
55
|
-
|
56
|
-
def get_time_coverages(doc)
|
57
|
-
doc.xpath('//xmlns:timeCoverage').map do |node|
|
58
|
-
{ 'start' => node.xpath('./xmlns:start').text,
|
59
|
-
'end' => node.xpath('./xmlns:end').text }
|
60
|
-
end
|
61
|
-
end
|
62
|
-
|
63
|
-
def spatial_coverage_to_spatial_area(coverage)
|
64
|
-
return if [:north, :south].any? { |x| coverage[x].nil? }
|
65
|
-
coverage[:north].abs - coverage[:south].abs
|
66
|
-
end
|
67
|
-
|
68
|
-
def parse_geospatial_coverages(doc)
|
69
|
-
node = doc.xpath('//xmlns:geospatialCoverage')
|
70
|
-
south = node.xpath('./xmlns:northsouth/xmlns:start').text.to_f
|
71
|
-
north = south + node.xpath('./xmlns:northsouth/xmlns:size').text.to_f
|
72
|
-
west = node.xpath('./xmlns:eastwest/xmlns:start').text.to_f
|
73
|
-
east = west + node.xpath('./xmlns:eastwest/xmlns:size').text.to_f
|
74
|
-
# EOL uses out-of-range east-west values to represent bounding boxes
|
75
|
-
# that cross the date line. For any box with a value out of range,
|
76
|
-
# adjust the east/west value to lie within the -180 to 180 range.
|
77
|
-
east -= 360 if east > 180
|
78
|
-
west += 360 if west < -180
|
79
|
-
|
80
|
-
{ east: east, west: west, north: north, south: south }
|
81
|
-
end
|
82
|
-
end
|
83
|
-
end
|
84
|
-
end
|
@@ -1,59 +0,0 @@
|
|
1
|
-
require 'json'
|
2
|
-
require 'rest-client'
|
3
|
-
require 'rgeo/geo_json'
|
4
|
-
|
5
|
-
require 'search_solr_tools'
|
6
|
-
require_relative '../helpers/solr_format'
|
7
|
-
require_relative '../helpers/translate_temporal_coverage'
|
8
|
-
require_relative '../helpers/translate_spatial_coverage'
|
9
|
-
|
10
|
-
module SearchSolrTools
|
11
|
-
module Translators
|
12
|
-
# Translates GTN-P json to solr json format
|
13
|
-
class GtnpJsonToSolr
|
14
|
-
# rubocop:disable AbcSize
|
15
|
-
def translate(json_doc, json_record)
|
16
|
-
json_geo = json_doc['geo'].nil? ? json_doc['coordinates'] : json_doc['geo']['coordinates']
|
17
|
-
concatenated_name = "#{json_record['title']} - #{json_doc['name']}"
|
18
|
-
spatial_values = translate_geometry json_geo
|
19
|
-
{
|
20
|
-
'title' => concatenated_name,
|
21
|
-
'authoritative_id' => concatenated_name,
|
22
|
-
'data_centers' => Helpers::SolrFormat::DATA_CENTER_NAMES[:GTNP][:long_name],
|
23
|
-
'facet_data_center' => "#{Helpers::SolrFormat::DATA_CENTER_NAMES[:GTNP][:long_name]} | #{Helpers::SolrFormat::DATA_CENTER_NAMES[:GTNP][:short_name]}",
|
24
|
-
'summary' => json_record['abstract'].to_s,
|
25
|
-
'dataset_url' => json_doc['link'],
|
26
|
-
'source' => 'ADE',
|
27
|
-
'facet_spatial_scope' => spatial_values[:spatial_scope_facet],
|
28
|
-
'spatial_coverages' => spatial_values[:spatial_display],
|
29
|
-
'spatial_area' => spatial_values[:spatial_area],
|
30
|
-
'spatial' => spatial_values[:spatial_index],
|
31
|
-
'temporal_coverages' => Helpers::SolrFormat::NOT_SPECIFIED,
|
32
|
-
'facet_temporal_duration' => Helpers::SolrFormat::NOT_SPECIFIED,
|
33
|
-
'authors' => parse_people(json_doc)
|
34
|
-
}
|
35
|
-
end
|
36
|
-
|
37
|
-
def parse_people(json_doc)
|
38
|
-
people_found = []
|
39
|
-
return people_found unless json_doc.key?('citation') && json_doc['citation'].key?('contacts')
|
40
|
-
citation = json_doc['citation']
|
41
|
-
citation['contacts'].each do |person|
|
42
|
-
people_found << "#{person['givenName']} #{person['familyName']}"
|
43
|
-
end
|
44
|
-
people_found
|
45
|
-
end
|
46
|
-
|
47
|
-
def translate_geometry(json_geom)
|
48
|
-
geo_string = "{\"type\":\"Point\",\"coordinates\":[#{json_geom['longitude']},#{json_geom['latitude']}]}"
|
49
|
-
geometry = RGeo::GeoJSON.decode(geo_string, json_parser: :json)
|
50
|
-
{
|
51
|
-
spatial_display: Helpers::TranslateSpatialCoverage.geojson_to_spatial_display_str([geometry]),
|
52
|
-
spatial_index: Helpers::TranslateSpatialCoverage.geojson_to_spatial_index_str([geometry]),
|
53
|
-
spatial_area: Helpers::TranslateSpatialCoverage.geojson_to_spatial_area([geometry]),
|
54
|
-
spatial_scope_facet: Helpers::TranslateSpatialCoverage.geojson_to_spatial_scope_facet([geometry])
|
55
|
-
}
|
56
|
-
end
|
57
|
-
end
|
58
|
-
end
|
59
|
-
end
|