geomash 0.2.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/Rakefile +32 -0
- data/config/geomash.yml.sample +24 -0
- data/lib/geomash.rb +86 -0
- data/lib/geomash/constants.rb +478 -0
- data/lib/geomash/geonames.rb +145 -0
- data/lib/geomash/parser.rb +220 -0
- data/lib/geomash/standardizer.rb +250 -0
- data/lib/geomash/tgn.rb +673 -0
- data/lib/geomash/town_lookup.rb +19 -0
- data/lib/geomash/version.rb +3 -0
- data/test/geomash_test.rb +146 -0
- data/test/geonames_test.rb +24 -0
- data/test/parser_test.rb +33 -0
- data/test/standardizer_test.rb +26 -0
- data/test/test_helper.rb +16 -0
- data/test/tgn_test.rb +19 -0
- data/test/town_lookup_test.rb +11 -0
- metadata +210 -0
data/lib/geomash/tgn.rb
ADDED
@@ -0,0 +1,673 @@
|
|
1
|
+
# -*- coding: utf-8 -*-
|
2
|
+
module Geomash
|
3
|
+
class TGN
|
4
|
+
|
5
|
+
def self.tgn_enabled
|
6
|
+
Geomash.config[:tgn_enabled] || true
|
7
|
+
end
|
8
|
+
|
9
|
+
=begin
|
10
|
+
81010/nation
|
11
|
+
81175/state
|
12
|
+
81165/region
|
13
|
+
84251/neighborhood
|
14
|
+
83002/inhabited place
|
15
|
+
|
16
|
+
nations
|
17
|
+
<http://vocab.getty.edu/tgn/7012149> <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300128207>
|
18
|
+
|
19
|
+
States (political divisions):
|
20
|
+
<http://vocab.getty.edu/tgn/7007517> <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000776> .
|
21
|
+
|
22
|
+
Counties: (Suffolk - http://vocab.getty.edu/aat/300000771)
|
23
|
+
<http://vocab.getty.edu/tgn/1002923> <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000771> .
|
24
|
+
|
25
|
+
Neighborhood: (Boston)
|
26
|
+
<http://vocab.getty.edu/tgn/7013445> <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300008347> .
|
27
|
+
|
28
|
+
|
29
|
+
Provinces:
|
30
|
+
http://vocab.getty.edu/aat/300000774
|
31
|
+
|
32
|
+
Departments:
|
33
|
+
http://vocab.getty.edu/aat/300000772
|
34
|
+
|
35
|
+
Governates:
|
36
|
+
http://vocab.getty.edu/aat/300235093
|
37
|
+
|
38
|
+
Territories:
|
39
|
+
http://vocab.getty.edu/aat/300135982
|
40
|
+
|
41
|
+
+ http://vocab.getty.edu/resource/getty/search?q=territory&luceneIndex=Brief&indexDataset=AAT&_form=%2Fresource%2Fgetty%2Fsearch
|
42
|
+
|
43
|
+
dependent state:
|
44
|
+
http://vocab.getty.edu/aat/300387176
|
45
|
+
|
46
|
+
|
47
|
+
union territory:
|
48
|
+
http://vocab.getty.edu/aat/300387122
|
49
|
+
|
50
|
+
national district:
|
51
|
+
http://vocab.getty.edu/aat/300387081
|
52
|
+
|
53
|
+
|
54
|
+
Roxbury:
|
55
|
+
http://vocab.getty.edu/tgn/7015002.json
|
56
|
+
|
57
|
+
|
58
|
+
|
59
|
+
#South Carolina - http://vocab.getty.edu/tgn/7007712
|
60
|
+
|
61
|
+
SELECT ?object_identifier
|
62
|
+
WHERE
|
63
|
+
{
|
64
|
+
?x <http://purl.org/dc/elements/1.1/identifier> 7007712 .
|
65
|
+
?x <http://vocab.getty.edu/ontology#broaderPreferredExtended> ?parent_country .
|
66
|
+
{
|
67
|
+
SELECT ?parent_country ?identifier_country ?aat_place_id
|
68
|
+
WHERE {
|
69
|
+
?parent_country <http://purl.org/dc/elements/1.1/identifier> ?identifier_country .
|
70
|
+
?parent_country <http://vocab.getty.edu/ontology#placeTypePreferred> ?aat_place_id .
|
71
|
+
?parent_country <http://www.w3.org/2000/01/rdf-schema#label> ?country_label .
|
72
|
+
}
|
73
|
+
GROUP BY ?parent_country
|
74
|
+
}
|
75
|
+
}
|
76
|
+
GROUP BY ?object_identifier
|
77
|
+
|
78
|
+
primary_tgn_response = Typhoeus::Request.get("http://vocab.getty.edu/tgn/#{tgn_id}.json")
|
79
|
+
|
80
|
+
|
81
|
+
when 'http://vocab.getty.edu/ontology#placeTypePreferred'
|
82
|
+
place_type_base[:aat_id] = ntriple['Object']['value']
|
83
|
+
when 'http://www.w3.org/2004/02/skos/core#prefLabel'
|
84
|
+
if ntriple['Object']['xml:lang'].present? && ntriple['Object']['xml:lang'] == 'en'
|
85
|
+
place_type_base[:label_en] = ntriple['Object']['value']
|
86
|
+
else if ntriple['Object']['xml:lang'].blank?
|
87
|
+
place_type_base[:label_default] = ntriple['Object']['value']
|
88
|
+
|
89
|
+
|
90
|
+
tgn_main_term_info = {}
|
91
|
+
broader_place_type_list = ["http://vocab.getty.edu/tgn/"#{tgn_id}]
|
92
|
+
|
93
|
+
primary_tgn_response = Typhoeus::Request.get("http://vocab.getty.edu/download/json", :params=>{:uri=>"http://vocab.getty.edu/tgn/#{tgn_id}.json"})
|
94
|
+
as_json_tgn_response = JSON.parse(primary_tgn_response.body)
|
95
|
+
|
96
|
+
as_json_tgn_response['results']['bindings'].each do |ntriple|
|
97
|
+
case ntriple['Predicate']['value']
|
98
|
+
when 'http://www.w3.org/2004/02/skos/core#prefLabel'
|
99
|
+
if ntriple['Object']['xml:lang'].present? && ntriple['Object']['xml:lang'] == 'en'
|
100
|
+
tgn_main_term_info[:label_en] = ntriple['Object']['value']
|
101
|
+
elsif ntriple['Object']['xml:lang'].blank?
|
102
|
+
tgn_main_term_info[:label_default] = ntriple['Object']['value']
|
103
|
+
end
|
104
|
+
when 'http://vocab.getty.edu/ontology#placeTypePreferred'
|
105
|
+
tgn_main_term_info[:aat_place] = ntriple['Object']['value']
|
106
|
+
when 'http://vocab.getty.edu/ontology#broaderPreferredExtended'
|
107
|
+
broader_place_type_list << ntriple['Object']['value']
|
108
|
+
end
|
109
|
+
|
110
|
+
end
|
111
|
+
|
112
|
+
query = "SELECT ?identifier_place ?place_label_default ?place_label_en ?aat_pref WHERE {"
|
113
|
+
|
114
|
+
broader_place_type_list.each do |place_uri|
|
115
|
+
query += %{{<#{place_uri}> <http://purl.org/dc/elements/1.1/identifier> ?identifier_place .
|
116
|
+
OPTIONAL {<#{place_uri}> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_en
|
117
|
+
FILTER langMatches( lang(?place_label_en), "en" )
|
118
|
+
}
|
119
|
+
OPTIONAL {<#{place_uri}> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_default
|
120
|
+
FILTER langMatches( lang(?place_label_default), "" )
|
121
|
+
}
|
122
|
+
<#{place_uri}> <http://vocab.getty.edu/ontology#placeTypePreferred> ?aat_pref
|
123
|
+
} UNION
|
124
|
+
}
|
125
|
+
end
|
126
|
+
|
127
|
+
query = query[0..-12]
|
128
|
+
query += ". } GROUP BY ?identifier_place ?place_label_default ?place_label_en ?aat_pref"
|
129
|
+
|
130
|
+
tgn_response_for_aat = Typhoeus::Request.get("http://vocab.getty.edu/sparql.json", :params=>{:query=>query})
|
131
|
+
as_json_tgn_response_for_aat = JSON.parse(tgn_response_for_aat.body)
|
132
|
+
|
133
|
+
as_json_tgn_response_for_aat["results"]["bindings"].each do |aat_response|
|
134
|
+
#aat_response['identifier_place']['value']
|
135
|
+
#aat_response['place_label_default']['value']
|
136
|
+
#....
|
137
|
+
end
|
138
|
+
|
139
|
+
|
140
|
+
|
141
|
+
|
142
|
+
|
143
|
+
EXAMPLE SPARQL:
|
144
|
+
|
145
|
+
SELECT ?identifier_place ?place_label_default ?place_label_en ?aat_pref
|
146
|
+
WHERE {
|
147
|
+
{<http://vocab.getty.edu/tgn/1000001> <http://purl.org/dc/elements/1.1/identifier> ?identifier_place .
|
148
|
+
OPTIONAL {<http://vocab.getty.edu/tgn/1000001> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_en
|
149
|
+
FILTER langMatches( lang(?place_label_en), "en" )
|
150
|
+
}
|
151
|
+
OPTIONAL {<http://vocab.getty.edu/tgn/1000001> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_default
|
152
|
+
FILTER langMatches( lang(?place_label_default), "" )
|
153
|
+
}
|
154
|
+
<http://vocab.getty.edu/tgn/1000001> <http://vocab.getty.edu/ontology#placeTypePreferred> ?aat_pref
|
155
|
+
} UNION
|
156
|
+
{<http://vocab.getty.edu/tgn/7012149> <http://purl.org/dc/elements/1.1/identifier> ?identifier_place .
|
157
|
+
OPTIONAL {<http://vocab.getty.edu/tgn/7012149> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_en
|
158
|
+
FILTER langMatches( lang(?place_label_en), "en" )
|
159
|
+
}
|
160
|
+
OPTIONAL {<http://vocab.getty.edu/tgn/7012149> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_default
|
161
|
+
FILTER langMatches( lang(?place_label_default), "" )
|
162
|
+
}
|
163
|
+
<http://vocab.getty.edu/tgn/7012149> <http://vocab.getty.edu/ontology#placeTypePreferred> ?aat_pref
|
164
|
+
} UNION
|
165
|
+
{<http://vocab.getty.edu/tgn/7029392> <http://purl.org/dc/elements/1.1/identifier> ?identifier_place .
|
166
|
+
OPTIONAL {<http://vocab.getty.edu/tgn/7029392> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_en
|
167
|
+
FILTER langMatches( lang(?place_label_en), "en" )
|
168
|
+
}
|
169
|
+
OPTIONAL {<http://vocab.getty.edu/tgn/7029392> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_default
|
170
|
+
FILTER langMatches( lang(?place_label_default), "" )
|
171
|
+
}
|
172
|
+
<http://vocab.getty.edu/tgn/7012149> <http://vocab.getty.edu/ontology#placeTypePreferred> ?aat_pref
|
173
|
+
} .
|
174
|
+
|
175
|
+
|
176
|
+
}
|
177
|
+
GROUP BY ?identifier_place ?place_label_default ?place_label_en ?aat_pref
|
178
|
+
|
179
|
+
|
180
|
+
|
181
|
+
|
182
|
+
=end
|
183
|
+
|
184
|
+
def self.get_tgn_data(tgn_id)
|
185
|
+
return nil if Geomash::TGN.tgn_enabled != true
|
186
|
+
|
187
|
+
tgn_id = tgn_id.strip
|
188
|
+
|
189
|
+
tgn_main_term_info = {}
|
190
|
+
broader_place_type_list = []
|
191
|
+
|
192
|
+
primary_tgn_response = Typhoeus::Request.get("http://vocab.getty.edu/download/json", :params=>{:uri=>"http://vocab.getty.edu/tgn/#{tgn_id}.json"})
|
193
|
+
|
194
|
+
return nil if(primary_tgn_response.response_code == 404) #Couldn't find TGN... FIXME: additional check needed if TGN is down?
|
195
|
+
|
196
|
+
as_json_tgn_response = JSON.parse(primary_tgn_response.body)
|
197
|
+
|
198
|
+
#There is a bug with some TGN JSON files currently. Example: http://vocab.getty.edu/tgn/7014203.json . Per an email
|
199
|
+
# with Getty, this is a hackish workaround for now.
|
200
|
+
if(as_json_tgn_response['results'].blank?)
|
201
|
+
query = %{
|
202
|
+
PREFIX tgn: <http://vocab.getty.edu/tgn/>
|
203
|
+
PREFIX gvp: <http://vocab.getty.edu/ontology#>
|
204
|
+
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
|
205
|
+
PREFIX dct: <http://purl.org/dc/terms/>
|
206
|
+
PREFIX bibo: <http://purl.org/ontology/bibo/>
|
207
|
+
PREFIX skosxl: <http://www.w3.org/2008/05/skos-xl#>
|
208
|
+
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
|
209
|
+
PREFIX iso: <http://purl.org/iso25964/skos-thes#>
|
210
|
+
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
|
211
|
+
PREFIX schema: <http://schema.org/>
|
212
|
+
CONSTRUCT {
|
213
|
+
?s ?p1 ?o1.
|
214
|
+
?ac ?p2 ?o2.
|
215
|
+
?t ?p3 ?o3.
|
216
|
+
?ss ?p4 ?o4.
|
217
|
+
?ts ?p6 ?o6.
|
218
|
+
?st ?p7 ?o7.
|
219
|
+
?ar ?p8 ?o8.
|
220
|
+
?l1 ?p9 ?o9.
|
221
|
+
?l2 ?pA ?oA.
|
222
|
+
?pl ?pB ?oB.
|
223
|
+
?ge ?pC ?oC.
|
224
|
+
} WHERE {
|
225
|
+
BIND (tgn:#{tgn_id} as ?s)
|
226
|
+
{?s ?p1 ?o1 FILTER(!isBlank(?o1) &&
|
227
|
+
!(?p1 in (gvp:narrowerExtended, skos:narrowerTransitive, skos:semanticRelation)))}
|
228
|
+
UNION {?s skos:changeNote ?ac. ?ac ?p2 ?o2}
|
229
|
+
UNION {?s dct:source ?ss. ?ss a bibo:DocumentPart. ?ss ?p4 ?o4}
|
230
|
+
UNION {?s skos:scopeNote|skosxl:prefLabel|skosxl:altLabel ?t.
|
231
|
+
{?t ?p3 ?o3 FILTER(!isBlank(?o3))}
|
232
|
+
UNION {?t dct:source ?ts. ?ts a bibo:DocumentPart. ?ts ?p6 ?o6}}
|
233
|
+
UNION {?st rdf:subject ?s. ?st ?p7 ?o7}
|
234
|
+
UNION {?s skos:member/^rdf:first ?l1. ?l1 ?p9 ?o9}
|
235
|
+
UNION {?s iso:subordinateArray ?ar FILTER NOT EXISTS {?ar skosxl:prefLabel ?t1}.
|
236
|
+
{?ar ?p8 ?o8}
|
237
|
+
UNION {?ar skos:member/^rdf:first ?l2. ?l2 ?pA ?oA}}
|
238
|
+
UNION {?s foaf:focus ?pl.
|
239
|
+
{?pl ?pB ?oB}
|
240
|
+
UNION {?pl schema:geo ?ge. ?ge ?pC ?oC}}
|
241
|
+
}
|
242
|
+
}
|
243
|
+
|
244
|
+
query = query.squish
|
245
|
+
|
246
|
+
primary_tgn_response = Typhoeus::Request.post("http://vocab.getty.edu/sparql.json", :body=>{:query=>query})
|
247
|
+
as_json_tgn_response = JSON.parse(primary_tgn_response.body)
|
248
|
+
end
|
249
|
+
|
250
|
+
#FIXME: Temporary hack to determine more cases of non-blank/english place name conflicts that require resolution.
|
251
|
+
label_remaining_check = false
|
252
|
+
|
253
|
+
as_json_tgn_response['results']['bindings'].each do |ntriple|
|
254
|
+
case ntriple['Predicate']['value']
|
255
|
+
when 'http://www.w3.org/2004/02/skos/core#prefLabel'
|
256
|
+
if ntriple['Object']['xml:lang'].present? && ntriple['Object']['xml:lang'] == 'en'
|
257
|
+
tgn_main_term_info[:label_en] = ntriple['Object']['value']
|
258
|
+
elsif ntriple['Object']['xml:lang'].present? && ntriple['Object']['xml:lang'] == 'zh-latn-pinyin'
|
259
|
+
tgn_main_term_info[:label_other] = ntriple['Object']['value']
|
260
|
+
elsif ntriple['Object']['xml:lang'].blank?
|
261
|
+
tgn_main_term_info[:label_default] = ntriple['Object']['value']
|
262
|
+
else
|
263
|
+
label_remaining_check = true if tgn_main_term_info[:label_remaining].present?
|
264
|
+
tgn_main_term_info[:label_remaining] = ntriple['Object']['value']
|
265
|
+
end
|
266
|
+
when 'http://www.w3.org/2004/02/skos/core#altLabel'
|
267
|
+
if ntriple['Object']['xml:lang'].present? && ntriple['Object']['xml:lang'] == 'en'
|
268
|
+
tgn_main_term_info[:label_alt] = ntriple['Object']['value']
|
269
|
+
end
|
270
|
+
when 'http://vocab.getty.edu/ontology#placeTypePreferred'
|
271
|
+
tgn_main_term_info[:aat_place] = ntriple['Object']['value']
|
272
|
+
when 'http://schema.org/latitude'
|
273
|
+
tgn_main_term_info[:latitude] = ntriple['Object']['value']
|
274
|
+
when 'http://schema.org/longitude'
|
275
|
+
tgn_main_term_info[:longitude] = ntriple['Object']['value']
|
276
|
+
when 'http://vocab.getty.edu/ontology#broaderPreferredExtended'
|
277
|
+
broader_place_type_list << ntriple['Object']['value']
|
278
|
+
end
|
279
|
+
|
280
|
+
end
|
281
|
+
|
282
|
+
# coordinates
|
283
|
+
coords = nil
|
284
|
+
if tgn_main_term_info[:latitude].present?
|
285
|
+
coords = {}
|
286
|
+
coords[:latitude] = tgn_main_term_info[:latitude]
|
287
|
+
coords[:longitude] = tgn_main_term_info[:longitude]
|
288
|
+
coords[:combined] = tgn_main_term_info[:latitude] + ',' + tgn_main_term_info[:longitude]
|
289
|
+
end
|
290
|
+
|
291
|
+
hier_geo = {}
|
292
|
+
#Default term to best label language...
|
293
|
+
tgn_term = tgn_main_term_info[:label_en]
|
294
|
+
tgn_term ||= tgn_main_term_info[:label_default]
|
295
|
+
tgn_term ||= tgn_main_term_info[:label_other]
|
296
|
+
tgn_term ||= tgn_main_term_info[:label_alt]
|
297
|
+
if tgn_term.blank?
|
298
|
+
if label_remaining_check
|
299
|
+
raise "Could not determine a single label for TGN: " + tgn_id
|
300
|
+
else
|
301
|
+
tgn_term = tgn_main_term_info[:label_remaining]
|
302
|
+
end
|
303
|
+
end
|
304
|
+
|
305
|
+
tgn_term_type = tgn_main_term_info[:aat_place].split('/').last
|
306
|
+
|
307
|
+
#Initial Term
|
308
|
+
if tgn_term.present? && tgn_term_type.present?
|
309
|
+
case tgn_term_type
|
310
|
+
when '300128176' #continent
|
311
|
+
hier_geo[:continent] = tgn_term
|
312
|
+
when '300128207' #nations
|
313
|
+
hier_geo[:country] = tgn_term
|
314
|
+
when '300000774' #province
|
315
|
+
hier_geo[:province] = tgn_term
|
316
|
+
when '300236112', '300182722', '300387194', '300387052' #region, union, semi-independent political entity
|
317
|
+
hier_geo[:region] = tgn_term
|
318
|
+
when '300000776', '300000772', '300235093' #state, department, governorate
|
319
|
+
hier_geo[:state] = tgn_term
|
320
|
+
when '300387081' #national district
|
321
|
+
if tgn_term == 'District of Columbia'
|
322
|
+
hier_geo[:state] = tgn_term
|
323
|
+
else
|
324
|
+
hier_geo[:territory] = tgn_term
|
325
|
+
end
|
326
|
+
when '300135982', '300387176', '300387122' #territory, dependent state, union territory
|
327
|
+
hier_geo[:territory] = tgn_term
|
328
|
+
when '300000771' #county
|
329
|
+
hier_geo[:county] = tgn_term
|
330
|
+
when '300008347' #inhabited place
|
331
|
+
hier_geo[:city] = tgn_term
|
332
|
+
when '300000745' #neighborhood
|
333
|
+
hier_geo[:city_section] = tgn_term
|
334
|
+
when '300008791', '300387062' #island
|
335
|
+
hier_geo[:island] = tgn_term
|
336
|
+
when '300387575', '300387346', '300167671', '300387178', '300387082' #'81101/area', '22101/general region', '83210/deserted settlement', '81501/historical region', '81126/national division'
|
337
|
+
hier_geo[:area] = tgn_term
|
338
|
+
else
|
339
|
+
non_hier_geo = tgn_term
|
340
|
+
end
|
341
|
+
|
342
|
+
#Broader places
|
343
|
+
#FIXME: could parse xml:lang instead of the three optional clauses now... didn't expect places to lack a default preferred label.
|
344
|
+
|
345
|
+
query = "SELECT ?identifier_place ?place_label_default ?place_label_en ?place_label_remaining ?aat_pref WHERE {"
|
346
|
+
|
347
|
+
broader_place_type_list.each do |place_uri|
|
348
|
+
query += %{{<#{place_uri}> <http://purl.org/dc/elements/1.1/identifier> ?identifier_place .
|
349
|
+
OPTIONAL {<#{place_uri}> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_en
|
350
|
+
FILTER langMatches( lang(?place_label_en), "en" )
|
351
|
+
}
|
352
|
+
OPTIONAL {<#{place_uri}> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_default
|
353
|
+
FILTER langMatches( lang(?place_label_default), "" )
|
354
|
+
}
|
355
|
+
OPTIONAL {<#{place_uri}> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_latn_pinyin
|
356
|
+
FILTER langMatches( lang(?place_label_latn_pinyin), "zh-latn-pinyin" )
|
357
|
+
}
|
358
|
+
OPTIONAL {<#{place_uri}> <http://www.w3.org/2004/02/skos/core#altLabel> ?place_label_alt
|
359
|
+
FILTER langMatches( lang(?place_label_alt), "en" )
|
360
|
+
}
|
361
|
+
OPTIONAL {<#{place_uri}> <http://www.w3.org/2004/02/skos/core#prefLabel> ?place_label_remaining
|
362
|
+
FILTER(!langMatches( lang(?place_label_remaining), "" ) && !langMatches( lang(?place_label_remaining), "en" ) && !langMatches( lang(?place_label_remaining), "zh-latn-pinyin" ))
|
363
|
+
}
|
364
|
+
<#{place_uri}> <http://vocab.getty.edu/ontology#placeTypePreferred> ?aat_pref
|
365
|
+
} UNION
|
366
|
+
}
|
367
|
+
end
|
368
|
+
|
369
|
+
query = query[0..-12]
|
370
|
+
query += ". } GROUP BY ?identifier_place ?place_label_default ?place_label_en ?place_label_latn_pinyin ?place_label_alt ?place_label_remaining ?aat_pref"
|
371
|
+
query = query.squish
|
372
|
+
|
373
|
+
tgn_response_for_aat = Typhoeus::Request.post("http://vocab.getty.edu/sparql.json", :body=>{:query=>query})
|
374
|
+
as_json_tgn_response_for_aat = JSON.parse(tgn_response_for_aat.body)
|
375
|
+
|
376
|
+
as_json_tgn_response_for_aat["results"]["bindings"].each do |aat_response|
|
377
|
+
tgn_term_type = aat_response['aat_pref']['value'].split('/').last
|
378
|
+
|
379
|
+
if aat_response['place_label_en'].present? && aat_response['place_label_en']['value'] != '-'
|
380
|
+
tgn_term = aat_response['place_label_en']['value']
|
381
|
+
elsif aat_response['place_label_default'].present? && aat_response['place_label_default']['value'] != '-'
|
382
|
+
tgn_term = aat_response['place_label_default']['value']
|
383
|
+
elsif aat_response['place_label_latn_pinyin'].present? && aat_response['place_label_latn_pinyin']['value'] != '-'
|
384
|
+
tgn_term = aat_response['place_label_latn_pinyin']['value']
|
385
|
+
elsif aat_response['place_label_alt'].present? && aat_response['place_label_alt']['value'] != '-'
|
386
|
+
tgn_term = aat_response['place_label_alt']['value']
|
387
|
+
else
|
388
|
+
tgn_term = aat_response['place_label_remaining']['value']
|
389
|
+
end
|
390
|
+
|
391
|
+
case tgn_term_type
|
392
|
+
when '300128176' #continent
|
393
|
+
hier_geo[:continent] = tgn_term
|
394
|
+
when '300128207' #nation
|
395
|
+
hier_geo[:country] = tgn_term
|
396
|
+
when '300000774' #province
|
397
|
+
hier_geo[:province] = tgn_term
|
398
|
+
when '300236112', '300182722', '300387194', '300387052' #region, union, semi-independent political entity
|
399
|
+
hier_geo[:region] = tgn_term
|
400
|
+
when '300000776', '300000772', '300235093' #state, department, governorate
|
401
|
+
hier_geo[:state] = tgn_term
|
402
|
+
when '300387081' #national district
|
403
|
+
if tgn_term == 'District of Columbia'
|
404
|
+
hier_geo[:state] = tgn_term
|
405
|
+
else
|
406
|
+
hier_geo[:territory] = tgn_term
|
407
|
+
end
|
408
|
+
when '300135982', '300387176', '300387122' #territory, dependent state, union territory
|
409
|
+
hier_geo[:territory] = tgn_term
|
410
|
+
when '300000771' #county
|
411
|
+
hier_geo[:county] = tgn_term
|
412
|
+
when '300008347' #inhabited place
|
413
|
+
hier_geo[:city] = tgn_term
|
414
|
+
when '300000745' #neighborhood
|
415
|
+
hier_geo[:city_section] = tgn_term
|
416
|
+
when '300008791', '300387062' #island
|
417
|
+
hier_geo[:island] = tgn_term
|
418
|
+
when '300387575', '300387346', '300167671', '300387178', '300387082' #'81101/area', '22101/general region', '83210/deserted settlement', '81501/historical region', '81126/national division'
|
419
|
+
hier_geo[:area] = tgn_term
|
420
|
+
end
|
421
|
+
end
|
422
|
+
|
423
|
+
tgn_data = {}
|
424
|
+
tgn_data[:coords] = coords
|
425
|
+
tgn_data[:hier_geo] = hier_geo.length > 0 ? hier_geo : nil
|
426
|
+
tgn_data[:non_hier_geo] = non_hier_geo ? non_hier_geo : nil
|
427
|
+
|
428
|
+
else
|
429
|
+
|
430
|
+
tgn_data = nil
|
431
|
+
|
432
|
+
end
|
433
|
+
|
434
|
+
return tgn_data
|
435
|
+
|
436
|
+
end
|
437
|
+
|
438
|
+
|
439
|
+
def self.tgn_id_from_geo_hash(geo_hash)
|
440
|
+
return nil if Geomash::TGN.tgn_enabled != true
|
441
|
+
|
442
|
+
geo_hash = geo_hash.clone
|
443
|
+
|
444
|
+
max_retry = 3
|
445
|
+
sleep_time = 60 # In seconds
|
446
|
+
retry_count = 0
|
447
|
+
|
448
|
+
return_hash = {}
|
449
|
+
|
450
|
+
state_part = geo_hash[:state_part]
|
451
|
+
|
452
|
+
country_code = Geomash::Constants::COUNTRY_TGN_LOOKUP[geo_hash[:country_part]][:tgn_id] unless Geomash::Constants::COUNTRY_TGN_LOOKUP[geo_hash[:country_part]].blank?
|
453
|
+
country_code ||= ''
|
454
|
+
|
455
|
+
|
456
|
+
country_part = Geomash::Constants::COUNTRY_TGN_LOOKUP[geo_hash[:country_part]][:tgn_country_name] unless Geomash::Constants::COUNTRY_TGN_LOOKUP[geo_hash[:country_part]].blank?
|
457
|
+
country_part ||= geo_hash[:country_part]
|
458
|
+
country_part ||= ''
|
459
|
+
|
460
|
+
city_part = geo_hash[:city_part]
|
461
|
+
|
462
|
+
neighborhood_part = geo_hash[:neighborhood_part]
|
463
|
+
|
464
|
+
|
465
|
+
|
466
|
+
if city_part.blank? && state_part.blank?
|
467
|
+
# Limit to nations
|
468
|
+
query = %{SELECT ?object_identifier
|
469
|
+
WHERE
|
470
|
+
{
|
471
|
+
?x <http://purl.org/dc/elements/1.1/identifier> ?object_identifier .
|
472
|
+
?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300128207> .
|
473
|
+
?x <http://www.w3.org/2004/02/skos/core#prefLabel> ?object_label .
|
474
|
+
FILTER regex(?object_label, "^#{country_part}$", "i" )
|
475
|
+
}}
|
476
|
+
elsif state_part.present? && city_part.blank? && country_code == 7012149
|
477
|
+
#Limit to states
|
478
|
+
query = %{SELECT ?object_identifier
|
479
|
+
WHERE
|
480
|
+
{
|
481
|
+
?x <http://purl.org/dc/elements/1.1/identifier> ?object_identifier .
|
482
|
+
?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000776> .
|
483
|
+
?x <http://www.w3.org/2000/01/rdf-schema#label> ?object_label .
|
484
|
+
FILTER regex(?object_label, "^#{state_part}$", "i" )
|
485
|
+
|
486
|
+
?x <http://vocab.getty.edu/ontology#broaderPreferredExtended> <http://vocab.getty.edu/tgn/7012149> .
|
487
|
+
}}
|
488
|
+
elsif state_part.present? && city_part.blank?
|
489
|
+
#Limit to regions
|
490
|
+
|
491
|
+
query = %{SELECT ?object_identifier
|
492
|
+
WHERE
|
493
|
+
{
|
494
|
+
?x <http://purl.org/dc/elements/1.1/identifier> ?object_identifier .
|
495
|
+
{?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000774>} UNION
|
496
|
+
{?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000772>} UNION
|
497
|
+
{?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300235093>} UNION
|
498
|
+
{?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300135982>} UNION
|
499
|
+
{?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300387176>} UNION
|
500
|
+
{?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300387122>} UNION
|
501
|
+
{?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000776>} UNION
|
502
|
+
{?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300236112>} UNION
|
503
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300008347>} UNION
|
504
|
+
{?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300387081>} .
|
505
|
+
?x <http://www.w3.org/2000/01/rdf-schema#label> ?object_label .
|
506
|
+
FILTER regex(?object_label, "^#{state_part}$", "i" )
|
507
|
+
?x <http://vocab.getty.edu/ontology#broaderPreferredExtended> ?parent_country .
|
508
|
+
{
|
509
|
+
SELECT ?parent_country ?identifier_country
|
510
|
+
WHERE {
|
511
|
+
?parent_country <http://purl.org/dc/elements/1.1/identifier> ?identifier_country .
|
512
|
+
?parent_country <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300128207> .
|
513
|
+
?parent_country <http://www.w3.org/2000/01/rdf-schema#label> ?country_label .
|
514
|
+
FILTER regex(?country_label, "^#{country_part}$", "i" )
|
515
|
+
}
|
516
|
+
|
517
|
+
}
|
518
|
+
}
|
519
|
+
GROUP BY ?object_identifier
|
520
|
+
}
|
521
|
+
|
522
|
+
#FIXME Temporary: For Geomash.parse('Aknīste (Latvia)', true), seems to be a neighborhood placed in state
|
523
|
+
# {?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300008347>} UNION
|
524
|
+
elsif state_part.present? && city_part.present? && neighborhood_part.blank?
|
525
|
+
#Limited to only inhabited places at the moment...
|
526
|
+
query = %{SELECT ?object_identifier
|
527
|
+
WHERE
|
528
|
+
{
|
529
|
+
?x <http://purl.org/dc/elements/1.1/identifier> ?object_identifier .
|
530
|
+
?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300008347> .
|
531
|
+
?x <http://www.w3.org/2000/01/rdf-schema#label> ?object_label .
|
532
|
+
FILTER regex(?object_label, "^#{city_part}$", "i" )
|
533
|
+
?x <http://vocab.getty.edu/ontology#broaderPreferredExtended> ?parent_country .
|
534
|
+
{
|
535
|
+
SELECT ?parent_country ?identifier_country
|
536
|
+
WHERE {
|
537
|
+
?parent_country <http://purl.org/dc/elements/1.1/identifier> ?identifier_country .
|
538
|
+
?parent_country <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300128207> .
|
539
|
+
?parent_country <http://www.w3.org/2000/01/rdf-schema#label> ?country_label .
|
540
|
+
FILTER regex(?country_label, "^#{country_part}$", "i" )
|
541
|
+
}
|
542
|
+
|
543
|
+
}
|
544
|
+
?x <http://vocab.getty.edu/ontology#broaderPreferredExtended> ?parent_state .
|
545
|
+
{
|
546
|
+
SELECT ?parent_state ?identifier_state
|
547
|
+
WHERE {
|
548
|
+
?parent_state <http://purl.org/dc/elements/1.1/identifier> ?identifier_state .
|
549
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000774>} UNION
|
550
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000772>} UNION
|
551
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300235093>} UNION
|
552
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300135982>} UNION
|
553
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300387176>} UNION
|
554
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300387122>} UNION
|
555
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000776>} UNION
|
556
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300236112>} UNION
|
557
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300008347>} UNION
|
558
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300387081>} .
|
559
|
+
?parent_state <http://www.w3.org/2000/01/rdf-schema#label> ?state_label .
|
560
|
+
FILTER regex(?state_label, "^#{state_part}$", "i" )
|
561
|
+
}
|
562
|
+
|
563
|
+
}
|
564
|
+
|
565
|
+
}
|
566
|
+
GROUP BY ?object_identifier
|
567
|
+
}
|
568
|
+
|
569
|
+
|
570
|
+
elsif state_part.present? && city_part.present? && neighborhood_part.present?
|
571
|
+
#Limited to only to neighborhoods currently...
|
572
|
+
query = %{SELECT ?object_identifier
|
573
|
+
WHERE
|
574
|
+
{
|
575
|
+
?x <http://purl.org/dc/elements/1.1/identifier> ?object_identifier .
|
576
|
+
?x <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000745> .
|
577
|
+
?x <http://www.w3.org/2000/01/rdf-schema#label> ?object_label .
|
578
|
+
FILTER regex(?object_label, "^#{neighborhood_part}$", "i" )
|
579
|
+
?x <http://vocab.getty.edu/ontology#broaderPreferredExtended> ?parent_country .
|
580
|
+
{
|
581
|
+
SELECT ?parent_country ?identifier_country
|
582
|
+
WHERE {
|
583
|
+
?parent_country <http://purl.org/dc/elements/1.1/identifier> ?identifier_country .
|
584
|
+
?parent_country <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300128207> .
|
585
|
+
?parent_country <http://www.w3.org/2000/01/rdf-schema#label> ?country_label .
|
586
|
+
FILTER regex(?country_label, "^#{country_part}$", "i" )
|
587
|
+
}
|
588
|
+
|
589
|
+
}
|
590
|
+
?x <http://vocab.getty.edu/ontology#broaderPreferredExtended> ?parent_state .
|
591
|
+
{
|
592
|
+
SELECT ?parent_state ?identifier_state
|
593
|
+
WHERE {
|
594
|
+
?parent_state <http://purl.org/dc/elements/1.1/identifier> ?identifier_state .
|
595
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000774>} UNION
|
596
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000772>} UNION
|
597
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300235093>} UNION
|
598
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300135982>} UNION
|
599
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300387176>} UNION
|
600
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300387122>} UNION
|
601
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300000776>} UNION
|
602
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300236112>} UNION
|
603
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300008347>} UNION
|
604
|
+
{?parent_state <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300387081>} .
|
605
|
+
?parent_state <http://www.w3.org/2000/01/rdf-schema#label> ?state_label .
|
606
|
+
FILTER regex(?state_label, "^#{state_part}$", "i" )
|
607
|
+
}
|
608
|
+
|
609
|
+
}
|
610
|
+
|
611
|
+
?x <http://vocab.getty.edu/ontology#broaderPreferredExtended> ?parent_city .
|
612
|
+
{
|
613
|
+
SELECT ?parent_city ?identifier_city
|
614
|
+
WHERE {
|
615
|
+
?parent_city <http://purl.org/dc/elements/1.1/identifier> ?identifier_city .
|
616
|
+
?parent_city <http://vocab.getty.edu/ontology#placeTypePreferred> <http://vocab.getty.edu/aat/300008347> .
|
617
|
+
?parent_city <http://www.w3.org/2000/01/rdf-schema#label> ?city_label .
|
618
|
+
FILTER regex(?city_label, "^#{city_part}$", "i" )
|
619
|
+
}
|
620
|
+
|
621
|
+
}
|
622
|
+
|
623
|
+
}
|
624
|
+
GROUP BY ?object_identifier
|
625
|
+
}
|
626
|
+
|
627
|
+
|
628
|
+
else
|
629
|
+
return nil
|
630
|
+
end
|
631
|
+
|
632
|
+
begin
|
633
|
+
|
634
|
+
if retry_count > 0
|
635
|
+
sleep(sleep_time)
|
636
|
+
end
|
637
|
+
retry_count = retry_count + 1
|
638
|
+
|
639
|
+
query = query.squish
|
640
|
+
tgn_response = Typhoeus::Request.get("http://vocab.getty.edu/sparql.json", :params=>{:query=>query})
|
641
|
+
|
642
|
+
end until (tgn_response.code != 500 || retry_count == max_retry)
|
643
|
+
|
644
|
+
|
645
|
+
|
646
|
+
|
647
|
+
unless tgn_response.code == 500
|
648
|
+
as_json = JSON.parse(tgn_response.body)
|
649
|
+
|
650
|
+
#This is ugly and needs to be redone to achieve better recursive...
|
651
|
+
if as_json["results"]["bindings"].present? && as_json["results"]["bindings"].first["object_identifier"].present?
|
652
|
+
return_hash[:id] = as_json["results"]["bindings"].first["object_identifier"]["value"]
|
653
|
+
return_hash[:rdf] = "http://vocab.getty.edu/tgn/#{return_hash[:id]}.rdf"
|
654
|
+
else
|
655
|
+
return nil
|
656
|
+
end
|
657
|
+
end
|
658
|
+
|
659
|
+
if tgn_response.code == 500
|
660
|
+
raise 'TGN Server appears to not be responding for Geographic query: ' + query
|
661
|
+
end
|
662
|
+
|
663
|
+
if return_hash.present?
|
664
|
+
return_hash[:original_string_differs] = Geomash::Standardizer.parsed_and_original_check(geo_hash)
|
665
|
+
return return_hash
|
666
|
+
else
|
667
|
+
return nil
|
668
|
+
end
|
669
|
+
end
|
670
|
+
|
671
|
+
|
672
|
+
end
|
673
|
+
end
|