HornsAndHooves-indirizzo 0.1.7.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +7 -0
- data/.travis.yml +5 -0
- data/Gemfile +13 -0
- data/HornsAndHooves-indirizzo.gemspec +28 -0
- data/LICENSE.txt +165 -0
- data/README.md +87 -0
- data/Rakefile +19 -0
- data/VERSION +1 -0
- data/lib/indirizzo.rb +1 -0
- data/lib/indirizzo/address.rb +86 -0
- data/lib/indirizzo/address_hash_extractor.rb +85 -0
- data/lib/indirizzo/city.rb +17 -0
- data/lib/indirizzo/constants.rb +653 -0
- data/lib/indirizzo/helper.rb +30 -0
- data/lib/indirizzo/match.rb +13 -0
- data/lib/indirizzo/number_helper.rb +29 -0
- data/lib/indirizzo/numbers.rb +55 -0
- data/lib/indirizzo/parser.rb +111 -0
- data/lib/indirizzo/street.rb +40 -0
- data/test/test_address.rb +299 -0
- data/test/test_constants.rb +55 -0
- data/test/test_helper.rb +3 -0
- data/test/test_number_helper.rb +10 -0
- data/test/test_numbers.rb +44 -0
- metadata +123 -0
|
@@ -0,0 +1,17 @@
|
|
|
1
|
+
module Indirizzo
|
|
2
|
+
class City
|
|
3
|
+
def self.city_parts(city)
|
|
4
|
+
strings = []
|
|
5
|
+
city.map do |string|
|
|
6
|
+
tokens = string.split(" ")
|
|
7
|
+
strings |= (0...tokens.length).to_a.reverse.map do |i|
|
|
8
|
+
(i...tokens.length).map {|j| tokens[i..j].join(" ")}
|
|
9
|
+
end.flatten
|
|
10
|
+
end
|
|
11
|
+
# Don't return strings that consist solely of abbreviations.
|
|
12
|
+
# NOTE: Is this a micro-optimization that has edge cases that will break?
|
|
13
|
+
# Answer: Yes, it breaks on "Prairie"
|
|
14
|
+
strings.reject { |s| Std_Abbr.key?(s) }.uniq
|
|
15
|
+
end
|
|
16
|
+
end
|
|
17
|
+
end
|
|
@@ -0,0 +1,653 @@
|
|
|
1
|
+
# coding: utf-8
|
|
2
|
+
require 'set'
|
|
3
|
+
require 'indirizzo/numbers'
|
|
4
|
+
|
|
5
|
+
module Indirizzo
|
|
6
|
+
class Map < Hash
|
|
7
|
+
# The Map class provides a two-way mapping between postal abbreviations
|
|
8
|
+
# and their fully written equivalents.
|
|
9
|
+
#attr_accessor :partial
|
|
10
|
+
attr_accessor :regexp
|
|
11
|
+
def self.[] (*items)
|
|
12
|
+
hash = super(*items)
|
|
13
|
+
hash.build_match
|
|
14
|
+
hash.keys.each {|k| hash[k.downcase] = hash.fetch(k)}
|
|
15
|
+
hash.values.each {|v| hash[v.downcase] = v}
|
|
16
|
+
hash.freeze
|
|
17
|
+
end
|
|
18
|
+
def build_match
|
|
19
|
+
@regexp = Regexp.new(
|
|
20
|
+
'\b(' + [keys,values].flatten.join("|") + ')\b',
|
|
21
|
+
Regexp::IGNORECASE)
|
|
22
|
+
end
|
|
23
|
+
def key? (key)
|
|
24
|
+
super(key.downcase)
|
|
25
|
+
end
|
|
26
|
+
def [] (key)
|
|
27
|
+
super(key.downcase)
|
|
28
|
+
end
|
|
29
|
+
end
|
|
30
|
+
|
|
31
|
+
# The Directional constant maps compass direction words in English and
|
|
32
|
+
# Spanish to their 1- or 2- letter abbreviations. See 2008 TIGER/Line
|
|
33
|
+
# technical documentation Appendix C for more details.
|
|
34
|
+
Directional = Map[
|
|
35
|
+
"North" => "N",
|
|
36
|
+
"South" => "S",
|
|
37
|
+
"East" => "E",
|
|
38
|
+
"West" => "W",
|
|
39
|
+
"Northeast" => "NE",
|
|
40
|
+
"Northwest" => "NW",
|
|
41
|
+
"Southeast" => "SE",
|
|
42
|
+
"Southwest" => "SW",
|
|
43
|
+
"Norte" => "N",
|
|
44
|
+
"Sur" => "S",
|
|
45
|
+
"Este" => "E",
|
|
46
|
+
"Oeste" => "O",
|
|
47
|
+
"Noreste" => "NE",
|
|
48
|
+
"Noroeste" => "NO",
|
|
49
|
+
"Sudeste" => "SE",
|
|
50
|
+
"Sudoeste" => "SO"
|
|
51
|
+
]
|
|
52
|
+
|
|
53
|
+
# The Prefix_Qualifier constant maps feature prefix qualifiers to their
|
|
54
|
+
# abbreviations. See 2008 TIGER/Line technical documentation Appendix D.
|
|
55
|
+
Prefix_Qualifier = Map[
|
|
56
|
+
"Alternate" => "Alt",
|
|
57
|
+
"Business" => "Bus",
|
|
58
|
+
"Bypass" => "Byp",
|
|
59
|
+
"Extended" => "Exd",
|
|
60
|
+
"Historic" => "Hst",
|
|
61
|
+
"Loop" => "Lp",
|
|
62
|
+
"Old" => "Old",
|
|
63
|
+
"Private" => "Pvt",
|
|
64
|
+
"Public" => "Pub",
|
|
65
|
+
"Spur" => "Spr",
|
|
66
|
+
]
|
|
67
|
+
|
|
68
|
+
# The Suffix_Qualifier constant maps feature suffix qualifiers to their
|
|
69
|
+
# abbreviations. See 2008 TIGER/Line technical documentation Appendix D.
|
|
70
|
+
Suffix_Qualifier = Map[
|
|
71
|
+
"Access" => "Acc",
|
|
72
|
+
"Alternate" => "Alt",
|
|
73
|
+
"Business" => "Bus",
|
|
74
|
+
"Bypass" => "Byp",
|
|
75
|
+
"Connector" => "Con",
|
|
76
|
+
"Extended" => "Exd",
|
|
77
|
+
"Extension" => "Exn",
|
|
78
|
+
"Loop" => "Lp",
|
|
79
|
+
"Private" => "Pvt",
|
|
80
|
+
"Public" => "Pub",
|
|
81
|
+
"Scenic" => "Scn",
|
|
82
|
+
"Spur" => "Spr",
|
|
83
|
+
"Ramp" => "Rmp",
|
|
84
|
+
"Underpass" => "Unp",
|
|
85
|
+
"Overpass" => "Ovp",
|
|
86
|
+
]
|
|
87
|
+
|
|
88
|
+
# The Prefix_Canonical constant maps canonical TIGER/Line street type
|
|
89
|
+
# prefixes to their abbreviations. This list is the subset of the list from
|
|
90
|
+
# 2008 TIGER/Line technical documentation Appendix E that was extracted from
|
|
91
|
+
# a TIGER/Line database import.
|
|
92
|
+
Prefix_Canonical = {
|
|
93
|
+
"Arcade" => "Arc",
|
|
94
|
+
"Autopista" => "Autopista",
|
|
95
|
+
"Avenida" => "Ave",
|
|
96
|
+
"Avenue" => "Ave",
|
|
97
|
+
"Boulevard" => "Blvd",
|
|
98
|
+
"Bulevar" => "Bulevar",
|
|
99
|
+
"Bureau of Indian Affairs Highway" => "BIA Hwy",
|
|
100
|
+
"Bureau of Indian Affairs Road" => "BIA Rd",
|
|
101
|
+
"Bureau of Indian Affairs Route" => "BIA Rte",
|
|
102
|
+
"Bureau of Land Management Road" => "BLM Rd",
|
|
103
|
+
"Bypass" => "Byp",
|
|
104
|
+
"Calle" => "Cll",
|
|
105
|
+
"Calleja" => "Calleja",
|
|
106
|
+
"Callejón" => "Callejón",
|
|
107
|
+
"Caminito" => "Cmt",
|
|
108
|
+
"Camino" => "Cam",
|
|
109
|
+
"Carretera" => "Carr",
|
|
110
|
+
"Cerrada" => "Cer",
|
|
111
|
+
"Círculo" => "Cír",
|
|
112
|
+
"Commons" => "Cmns",
|
|
113
|
+
"Corte" => "Corte",
|
|
114
|
+
"County Highway" => "Co Hwy",
|
|
115
|
+
"County Lane" => "Co Ln",
|
|
116
|
+
"County Road" => "Co Rd",
|
|
117
|
+
"County Route" => "Co Rte",
|
|
118
|
+
"County State Aid Highway" => "Co St Aid Hwy",
|
|
119
|
+
"County Trunk Highway" => "Co Trunk Hwy",
|
|
120
|
+
"County Trunk Road" => "Co Trunk Rd",
|
|
121
|
+
"Court" => "Ct",
|
|
122
|
+
"Delta Road" => "Delta Rd",
|
|
123
|
+
"District of Columbia Highway" => "DC Hwy",
|
|
124
|
+
"Driveway" => "Driveway",
|
|
125
|
+
"Entrada" => "Ent",
|
|
126
|
+
"Expreso" => "Expreso",
|
|
127
|
+
"Expressway" => "Expy",
|
|
128
|
+
"Farm Road" => "Farm Rd",
|
|
129
|
+
"Farm-to-Market Road" => "FM",
|
|
130
|
+
"Fire Control Road" => "Fire Cntrl Rd",
|
|
131
|
+
"Fire District Road" => "Fire Dist Rd",
|
|
132
|
+
"Fire Lane" => "Fire Ln",
|
|
133
|
+
"Fire Road" => "Fire Rd",
|
|
134
|
+
"Fire Route" => "Fire Rte",
|
|
135
|
+
"Fire Trail" => "Fire Trl",
|
|
136
|
+
"Forest Highway" => "Forest Hwy",
|
|
137
|
+
"Forest Road" => "Forest Rd",
|
|
138
|
+
"Forest Route" => "Forest Rte",
|
|
139
|
+
"Forest Service Road" => "FS Rd",
|
|
140
|
+
"Highway" => "Hwy",
|
|
141
|
+
"Indian Route" => "Indian Rte",
|
|
142
|
+
"Indian Service Route" => "Indian Svc Rte",
|
|
143
|
+
"Interstate Highway" => "I-",
|
|
144
|
+
"Lane" => "Ln",
|
|
145
|
+
"Logging Road" => "Logging Rd",
|
|
146
|
+
"Loop" => "Loop",
|
|
147
|
+
"National Forest Development Road" => "Nat For Dev Rd",
|
|
148
|
+
"Navajo Service Route" => "Navajo Svc Rte",
|
|
149
|
+
"Parish Road" => "Parish Rd",
|
|
150
|
+
"Pasaje" => "Pasaje",
|
|
151
|
+
"Paseo" => "Pso",
|
|
152
|
+
"Passage" => "Psge",
|
|
153
|
+
"Placita" => "Pla",
|
|
154
|
+
"Plaza" => "Plz",
|
|
155
|
+
"Point" => "Pt",
|
|
156
|
+
"Puente" => "Puente",
|
|
157
|
+
"Ranch Road" => "Ranch Rd",
|
|
158
|
+
"Ranch to Market Road" => "RM",
|
|
159
|
+
"Reservation Highway" => "Resvn Hwy",
|
|
160
|
+
"Road" => "Rd",
|
|
161
|
+
"Route" => "Rte",
|
|
162
|
+
"Row" => "Row",
|
|
163
|
+
"Rue" => "Rue",
|
|
164
|
+
"Ruta" => "Ruta",
|
|
165
|
+
"Sector" => "Sec",
|
|
166
|
+
"Sendero" => "Sendero",
|
|
167
|
+
"Service Road" => "Svc Rd",
|
|
168
|
+
"Skyway" => "Skwy",
|
|
169
|
+
"Square" => "Sq",
|
|
170
|
+
"State Forest Service Road" => "St FS Rd",
|
|
171
|
+
"State Highway" => "State Hwy",
|
|
172
|
+
"State Loop" => "State Loop",
|
|
173
|
+
"State Road" => "State Rd",
|
|
174
|
+
"State Route" => "State Rte",
|
|
175
|
+
"State Spur" => "State Spur",
|
|
176
|
+
"State Trunk Highway" => "St Trunk Hwy",
|
|
177
|
+
"Terrace" => "Ter",
|
|
178
|
+
"Town Highway" => "Town Hwy",
|
|
179
|
+
"Town Road" => "Town Rd",
|
|
180
|
+
"Township Highway" => "Twp Hwy",
|
|
181
|
+
"Township Road" => "Twp Rd",
|
|
182
|
+
"Trail" => "Trl",
|
|
183
|
+
"Tribal Road" => "Tribal Rd",
|
|
184
|
+
"Tunnel" => "Tunl",
|
|
185
|
+
"US Forest Service Highway" => "USFS Hwy",
|
|
186
|
+
"US Forest Service Road" => "USFS Rd",
|
|
187
|
+
"US Highway" => "US Hwy",
|
|
188
|
+
"US Route" => "US Rte",
|
|
189
|
+
"Vereda" => "Ver",
|
|
190
|
+
"Via" => "Via",
|
|
191
|
+
"Vista" => "Vis",
|
|
192
|
+
}
|
|
193
|
+
|
|
194
|
+
# The Prefix_Alternate constant maps alternate prefix street types to
|
|
195
|
+
# their canonical abbreviations. This list was merged in from the USPS
|
|
196
|
+
# list at http://www.usps.com/ncsc/lookups/abbr_suffix.txt.
|
|
197
|
+
Prefix_Alternate = {
|
|
198
|
+
"Av" => "Ave",
|
|
199
|
+
"Aven" => "Ave",
|
|
200
|
+
"Avenu" => "Ave",
|
|
201
|
+
"Avenue" => "Ave",
|
|
202
|
+
"Avn" => "Ave",
|
|
203
|
+
"Avnue" => "Ave",
|
|
204
|
+
"Boul" => "Blvd",
|
|
205
|
+
"Boulv" => "Blvd",
|
|
206
|
+
"Bypa" => "Byp",
|
|
207
|
+
"Bypas" => "Byp",
|
|
208
|
+
"Byps" => "Byp",
|
|
209
|
+
"Crt" => "Ct",
|
|
210
|
+
"Exp" => "Expy",
|
|
211
|
+
"Expr" => "Expy",
|
|
212
|
+
"Express" => "Expy",
|
|
213
|
+
"Expw" => "Expy",
|
|
214
|
+
"Highwy" => "Hwy",
|
|
215
|
+
"Hiway" => "Hwy",
|
|
216
|
+
"Hiwy" => "Hwy",
|
|
217
|
+
"Hway" => "Hwy",
|
|
218
|
+
#"La" => "Ln", # causes problems with Spanglish place names
|
|
219
|
+
"Lanes" => "Ln",
|
|
220
|
+
"Loops" => "Loop",
|
|
221
|
+
"Plza" => "Plz",
|
|
222
|
+
"Sqr" => "Sq",
|
|
223
|
+
"Sqre" => "Sq",
|
|
224
|
+
"Squ" => "Sq",
|
|
225
|
+
"Terr" => "Ter",
|
|
226
|
+
"Tr" => "Trl",
|
|
227
|
+
"Trails" => "Trl",
|
|
228
|
+
"Trls" => "Trl",
|
|
229
|
+
"Tunel" => "Tunl",
|
|
230
|
+
"Tunls" => "Tunl",
|
|
231
|
+
"Tunnels" => "Tunl",
|
|
232
|
+
"Tunnl" => "Tunl",
|
|
233
|
+
"Vdct" => "Via",
|
|
234
|
+
"Viadct" => "Via",
|
|
235
|
+
"Viaduct" => "Via",
|
|
236
|
+
"Vist" => "Vis",
|
|
237
|
+
"Vst" => "Vis",
|
|
238
|
+
"Vsta" => "Vis"
|
|
239
|
+
}
|
|
240
|
+
|
|
241
|
+
# The Prefix_Type constant merges the canonical prefix type abbreviations
|
|
242
|
+
# with their USPS accepted alternates.
|
|
243
|
+
Prefix_Type = Map[ Prefix_Canonical.merge(Prefix_Alternate) ]
|
|
244
|
+
|
|
245
|
+
# The Suffix_Canonical constant maps canonical TIGER/Line street type
|
|
246
|
+
# suffixes to their abbreviations. This list is the subset of the list from
|
|
247
|
+
# 2008 TIGER/Line technical documentation Appendix E that was extracted from
|
|
248
|
+
# a TIGER/Line database import.
|
|
249
|
+
Suffix_Canonical = {
|
|
250
|
+
"Alley" => "Aly",
|
|
251
|
+
"Arcade" => "Arc",
|
|
252
|
+
"Avenida" => "Ave",
|
|
253
|
+
"Avenue" => "Ave",
|
|
254
|
+
"Beltway" => "Beltway",
|
|
255
|
+
"Boulevard" => "Blvd",
|
|
256
|
+
"Bridge" => "Brg",
|
|
257
|
+
"Bypass" => "Byp",
|
|
258
|
+
"Causeway" => "Cswy",
|
|
259
|
+
"Circle" => "Cir",
|
|
260
|
+
"Common" => "Cmn",
|
|
261
|
+
"Commons" => "Cmns",
|
|
262
|
+
"Corners" => "Cors",
|
|
263
|
+
"Court" => "Ct",
|
|
264
|
+
"Courts" => "Cts",
|
|
265
|
+
"Crescent" => "Cres",
|
|
266
|
+
"Crest" => "Crst",
|
|
267
|
+
"Crossing" => "Xing",
|
|
268
|
+
"Cutoff" => "Cutoff",
|
|
269
|
+
"Drive" => "Dr",
|
|
270
|
+
"Driveway" => "Driveway",
|
|
271
|
+
"Esplanade" => "Esplanade",
|
|
272
|
+
"Estates" => "Ests",
|
|
273
|
+
"Expressway" => "Expy",
|
|
274
|
+
"Forest Highway" => "Forest Hwy",
|
|
275
|
+
"Fork" => "Frk",
|
|
276
|
+
"Four-Wheel Drive Trail" => "4WD Trl",
|
|
277
|
+
"Freeway" => "Fwy",
|
|
278
|
+
"Grade" => "Grade",
|
|
279
|
+
"Heights" => "Hts",
|
|
280
|
+
"Highway" => "Hwy",
|
|
281
|
+
"Jeep Trail" => "Jeep Trl",
|
|
282
|
+
"Landing" => "Lndg",
|
|
283
|
+
"Lane" => "Ln",
|
|
284
|
+
"Logging Road" => "Logging Rd",
|
|
285
|
+
"Loop" => "Loop",
|
|
286
|
+
"Motorway" => "Mtwy",
|
|
287
|
+
"Oval" => "Oval",
|
|
288
|
+
"Overpass" => "Opas",
|
|
289
|
+
"Parkway" => "Pkwy",
|
|
290
|
+
"Pass" => "Pass",
|
|
291
|
+
"Passage" => "Psge",
|
|
292
|
+
"Path" => "Path",
|
|
293
|
+
"Pike" => "Pike",
|
|
294
|
+
"Place" => "Pl",
|
|
295
|
+
"Plaza" => "Plz",
|
|
296
|
+
"Point" => "Pt",
|
|
297
|
+
"Pointe" => "Pointe",
|
|
298
|
+
"Promenade" => "Promenade",
|
|
299
|
+
"Railroad" => "RR",
|
|
300
|
+
"Railway" => "Rlwy",
|
|
301
|
+
"Ramp" => "Ramp",
|
|
302
|
+
"River" => "Riv",
|
|
303
|
+
"Road" => "Rd",
|
|
304
|
+
"Roadway" => "Roadway",
|
|
305
|
+
"Route" => "Rte",
|
|
306
|
+
"Row" => "Row",
|
|
307
|
+
"Rue" => "Rue",
|
|
308
|
+
"Service Road" => "Svc Rd",
|
|
309
|
+
"Skyway" => "Skwy",
|
|
310
|
+
"Spur" => "Spur",
|
|
311
|
+
"Square" => "Sq",
|
|
312
|
+
"Stravenue" => "Stra",
|
|
313
|
+
"Street" => "St",
|
|
314
|
+
"Strip" => "Strip",
|
|
315
|
+
"Terrace" => "Ter",
|
|
316
|
+
"Thoroughfare" => "Thoroughfare",
|
|
317
|
+
"Tollway" => "Tollway",
|
|
318
|
+
"Trace" => "Trce",
|
|
319
|
+
"Trafficway" => "Trfy",
|
|
320
|
+
"Trail" => "Trl",
|
|
321
|
+
"Trolley" => "Trolley",
|
|
322
|
+
"Truck Trail" => "Truck Trl",
|
|
323
|
+
"Tunnel" => "Tunl",
|
|
324
|
+
"Turnpike" => "Tpke",
|
|
325
|
+
"Viaduct" => "Viaduct",
|
|
326
|
+
"View" => "Vw",
|
|
327
|
+
"Vista" => "Vis",
|
|
328
|
+
"Walk" => "Walk",
|
|
329
|
+
"Walkway" => "Walkway",
|
|
330
|
+
"Way" => "Way",
|
|
331
|
+
}
|
|
332
|
+
|
|
333
|
+
# The Suffix_Alternate constant maps alternate suffix street types to
|
|
334
|
+
# their canonical abbreviations. This list was merged in from the USPS
|
|
335
|
+
# list at http://www.usps.com/ncsc/lookups/abbr_suffix.txt.
|
|
336
|
+
Suffix_Alternate = {
|
|
337
|
+
"Allee" => "Aly",
|
|
338
|
+
"Ally" => "Aly",
|
|
339
|
+
"Av" => "Ave",
|
|
340
|
+
"Aven" => "Ave",
|
|
341
|
+
"Avenu" => "Ave",
|
|
342
|
+
"Avenue" => "Ave",
|
|
343
|
+
"Avn" => "Ave",
|
|
344
|
+
"Avnue" => "Ave",
|
|
345
|
+
"Boul" => "Blvd",
|
|
346
|
+
"Boulv" => "Blvd",
|
|
347
|
+
"Brdge" => "Brg",
|
|
348
|
+
"Bypa" => "Byp",
|
|
349
|
+
"Bypas" => "Byp",
|
|
350
|
+
"Byps" => "Byp",
|
|
351
|
+
"Causway" => "Cswy",
|
|
352
|
+
"Circ" => "Cir",
|
|
353
|
+
"Circl" => "Cir",
|
|
354
|
+
"Crcl" => "Cir",
|
|
355
|
+
"Crcle" => "Cir",
|
|
356
|
+
"Crecent" => "Cres",
|
|
357
|
+
"Cresent" => "Cres",
|
|
358
|
+
"Crscnt" => "Cres",
|
|
359
|
+
"Crsent" => "Cres",
|
|
360
|
+
"Crsnt" => "Cres",
|
|
361
|
+
"Crssing" => "Xing",
|
|
362
|
+
"Crssng" => "Xing",
|
|
363
|
+
"Crt" => "Ct",
|
|
364
|
+
"Driv" => "Dr",
|
|
365
|
+
"Drv" => "Dr",
|
|
366
|
+
"Exp" => "Expy",
|
|
367
|
+
"Expr" => "Expy",
|
|
368
|
+
"Express" => "Expy",
|
|
369
|
+
"Expw" => "Expy",
|
|
370
|
+
"Freewy" => "Fwy",
|
|
371
|
+
"Frway" => "Fwy",
|
|
372
|
+
"Frwy" => "Fwy",
|
|
373
|
+
"Height" => "Hts",
|
|
374
|
+
"Hgts" => "Hts",
|
|
375
|
+
"Highwy" => "Hwy",
|
|
376
|
+
"Hiway" => "Hwy",
|
|
377
|
+
"Hiwy" => "Hwy",
|
|
378
|
+
"Ht" => "Hts",
|
|
379
|
+
"Hway" => "Hwy",
|
|
380
|
+
"La" => "Ln",
|
|
381
|
+
"Lanes" => "Ln",
|
|
382
|
+
"Lndng" => "Lndg",
|
|
383
|
+
"Loops" => "Loop",
|
|
384
|
+
"Ovl" => "Oval",
|
|
385
|
+
"Parkways" => "Pkwy",
|
|
386
|
+
"Parkwy" => "Pkwy",
|
|
387
|
+
"Paths" => "Path",
|
|
388
|
+
"Pikes" => "Pike",
|
|
389
|
+
"Pkway" => "Pkwy",
|
|
390
|
+
"Pkwys" => "Pkwy",
|
|
391
|
+
"Pky" => "Pkwy",
|
|
392
|
+
"Plza" => "Plz",
|
|
393
|
+
"Rivr" => "Riv",
|
|
394
|
+
"Rvr" => "Riv",
|
|
395
|
+
"Spurs" => "Spur",
|
|
396
|
+
"Sqr" => "Sq",
|
|
397
|
+
"Sqre" => "Sq",
|
|
398
|
+
"Squ" => "Sq",
|
|
399
|
+
"Str" => "St",
|
|
400
|
+
"Strav" => "Stra",
|
|
401
|
+
"Strave" => "Stra",
|
|
402
|
+
"Straven" => "Stra",
|
|
403
|
+
"Stravn" => "Stra",
|
|
404
|
+
"Strt" => "St",
|
|
405
|
+
"Strvn" => "Stra",
|
|
406
|
+
"Strvnue" => "Stra",
|
|
407
|
+
"Terr" => "Ter",
|
|
408
|
+
"Tpk" => "Tpke",
|
|
409
|
+
"Tr" => "Trl",
|
|
410
|
+
"Traces" => "Trce",
|
|
411
|
+
"Trails" => "Trl",
|
|
412
|
+
"Trls" => "Trl",
|
|
413
|
+
"Trnpk" => "Tpke",
|
|
414
|
+
"Trpk" => "Tpke",
|
|
415
|
+
"Tunel" => "Tunl",
|
|
416
|
+
"Tunls" => "Tunl",
|
|
417
|
+
"Tunnels" => "Tunl",
|
|
418
|
+
"Tunnl" => "Tunl",
|
|
419
|
+
"Turnpk" => "Tpke",
|
|
420
|
+
"Vist" => "Vis",
|
|
421
|
+
"Vst" => "Vis",
|
|
422
|
+
"Vsta" => "Vis",
|
|
423
|
+
"Walks" => "Walk",
|
|
424
|
+
"Wy" => "Way",
|
|
425
|
+
}
|
|
426
|
+
|
|
427
|
+
# The Suffix_Type constant merges the canonical suffix type abbreviations
|
|
428
|
+
# with their USPS accepted alternates.
|
|
429
|
+
Suffix_Type = Map[ Suffix_Canonical.merge(Suffix_Alternate) ]
|
|
430
|
+
|
|
431
|
+
# The Unit_Type constant lists acceptable USPS unit type abbreviations
|
|
432
|
+
# from http://www.usps.com/ncsc/lookups/abbr_sud.txt.
|
|
433
|
+
Unit_Type = Map[
|
|
434
|
+
"Apartment" => "Apt",
|
|
435
|
+
"Basement" => "Bsmt",
|
|
436
|
+
"Building" => "Bldg",
|
|
437
|
+
"Department"=> "Dept",
|
|
438
|
+
"Floor" => "Fl",
|
|
439
|
+
"Front" => "Frnt",
|
|
440
|
+
"Hangar" => "Hngr",
|
|
441
|
+
"Lobby" => "Lbby",
|
|
442
|
+
"Lot" => "Lot",
|
|
443
|
+
"Lower" => "Lowr",
|
|
444
|
+
"Office" => "Ofc",
|
|
445
|
+
"Penthouse" => "Ph",
|
|
446
|
+
"Pier" => "Pier",
|
|
447
|
+
"Rear" => "Rear",
|
|
448
|
+
"Room" => "Rm",
|
|
449
|
+
"Side" => "Side",
|
|
450
|
+
"Slip" => "Slip",
|
|
451
|
+
"Space" => "Spc",
|
|
452
|
+
"Stop" => "Stop",
|
|
453
|
+
"Suite" => "Ste",
|
|
454
|
+
"Trailer" => "Trlr",
|
|
455
|
+
"Unit" => "Unit",
|
|
456
|
+
"Upper" => "Uppr",
|
|
457
|
+
]
|
|
458
|
+
|
|
459
|
+
Std_Abbr = Map[
|
|
460
|
+
[Directional, Prefix_Qualifier, Suffix_Qualifier,
|
|
461
|
+
Prefix_Type, Suffix_Type].inject({}) {|x,y|x.merge y}
|
|
462
|
+
]
|
|
463
|
+
|
|
464
|
+
# The Name_Abbr constant maps common toponym abbreviations to their
|
|
465
|
+
# full word equivalents. This list was constructed partly by hand, and
|
|
466
|
+
# partly by matching USPS alternate abbreviations with feature names
|
|
467
|
+
# found in the TIGER/Line dataset.
|
|
468
|
+
Name_Abbr = Map[
|
|
469
|
+
"Av" => "Avenue",
|
|
470
|
+
"Ave" => "Avenue",
|
|
471
|
+
"Blvd" => "Boulevard",
|
|
472
|
+
"Bot" => "Bottom",
|
|
473
|
+
"Boul" => "Boulevard",
|
|
474
|
+
"Boulv" => "Boulevard",
|
|
475
|
+
"Br" => "Branch",
|
|
476
|
+
"Brg" => "Bridge",
|
|
477
|
+
"Canyn" => "Canyon",
|
|
478
|
+
"Cen" => "Center",
|
|
479
|
+
"Cent" => "Center",
|
|
480
|
+
"Cir" => "Circle",
|
|
481
|
+
"Circ" => "Circle",
|
|
482
|
+
"Ck" => "Creek",
|
|
483
|
+
"Cnter" => "Center",
|
|
484
|
+
"Cntr" => "Center",
|
|
485
|
+
"Cnyn" => "Canyon",
|
|
486
|
+
"Cor" => "Corner",
|
|
487
|
+
"Cors" => "Corners",
|
|
488
|
+
"Cp" => "Camp",
|
|
489
|
+
"Cr" => "Creek",
|
|
490
|
+
"Crcl" => "Circle",
|
|
491
|
+
"Crcle" => "Circle",
|
|
492
|
+
"Cres" => "Crescent",
|
|
493
|
+
"Crscnt" => "Crescent",
|
|
494
|
+
"Ct" => "Court",
|
|
495
|
+
"Ctr" => "Center",
|
|
496
|
+
"Cts" => "Courts",
|
|
497
|
+
"Cyn" => "Canyon",
|
|
498
|
+
"Div" => "Divide",
|
|
499
|
+
"Dr" => "Drive",
|
|
500
|
+
"Dv" => "Divide",
|
|
501
|
+
"Est" => "Estate",
|
|
502
|
+
"Ests" => "Estates",
|
|
503
|
+
"Ext" => "Extension",
|
|
504
|
+
"Extn" => "Extension",
|
|
505
|
+
"Extnsn" => "Extension",
|
|
506
|
+
"Forests" => "Forest",
|
|
507
|
+
"Forg" => "Forge",
|
|
508
|
+
"Frg" => "Forge",
|
|
509
|
+
"Ft" => "Fort",
|
|
510
|
+
"Gatewy" => "Gateway",
|
|
511
|
+
"Gdn" => "Garden",
|
|
512
|
+
"Gdns" => "Gardens",
|
|
513
|
+
"Gtwy" => "Gateway",
|
|
514
|
+
"Harb" => "Harbor",
|
|
515
|
+
"Hbr" => "Harbor",
|
|
516
|
+
"Height" => "Heights",
|
|
517
|
+
"Hgts" => "Heights",
|
|
518
|
+
"Highwy" => "Highway",
|
|
519
|
+
"Hiway" => "Highway",
|
|
520
|
+
"Hiwy" => "Highway",
|
|
521
|
+
"Holws" => "Hollow",
|
|
522
|
+
"Ht" => "Heights",
|
|
523
|
+
"Hway" => "Highway",
|
|
524
|
+
"Hwy" => "Highway",
|
|
525
|
+
"Is" => "Island",
|
|
526
|
+
"Iss" => "Islands",
|
|
527
|
+
"Jct" => "Junction",
|
|
528
|
+
"Jction" => "Junction",
|
|
529
|
+
"Jctn" => "Junction",
|
|
530
|
+
"Junctn" => "Junction",
|
|
531
|
+
"Juncton" => "Junction",
|
|
532
|
+
"Ldg" => "Lodge",
|
|
533
|
+
"Lgt" => "Light",
|
|
534
|
+
"Lndg" => "Landing",
|
|
535
|
+
"Lodg" => "Lodge",
|
|
536
|
+
"Loops" => "Loop",
|
|
537
|
+
"Mt" => "Mount",
|
|
538
|
+
"Mtin" => "Mountain",
|
|
539
|
+
"Mtn" => "Mountain",
|
|
540
|
+
"Orch" => "Orchard",
|
|
541
|
+
"Parkwy" => "Parkway",
|
|
542
|
+
"Pk" => "Park",
|
|
543
|
+
"Pkway" => "Parkway",
|
|
544
|
+
"Pkwy" => "Parkway",
|
|
545
|
+
"Pky" => "Parkway",
|
|
546
|
+
"Pl" => "Place",
|
|
547
|
+
"Pnes" => "Pines",
|
|
548
|
+
"Pr" => "Prairie",
|
|
549
|
+
"Prr" => "Prairie",
|
|
550
|
+
"Pt" => "Point",
|
|
551
|
+
"Pts" => "Points",
|
|
552
|
+
"Rdg" => "Ridge",
|
|
553
|
+
"Riv" => "River",
|
|
554
|
+
"Rnchs" => "Ranch",
|
|
555
|
+
"Spg" => "Spring",
|
|
556
|
+
"Spgs" => "Springs",
|
|
557
|
+
"Spng" => "Spring",
|
|
558
|
+
"Spngs" => "Springs",
|
|
559
|
+
"Sq" => "Square",
|
|
560
|
+
"Squ" => "Square",
|
|
561
|
+
# "St" => "Saint",
|
|
562
|
+
"Sta" => "Station",
|
|
563
|
+
"Statn" => "Station",
|
|
564
|
+
"Ste" => "Sainte",
|
|
565
|
+
"Stn" => "Station",
|
|
566
|
+
"Str" => "Street",
|
|
567
|
+
"Ter" => "Terrace",
|
|
568
|
+
"Terr" => "Terrace",
|
|
569
|
+
"Tpk" => "Turnpike",
|
|
570
|
+
"Tpke" => "Turnpike",
|
|
571
|
+
"Tr" => "Trail",
|
|
572
|
+
"Trls" => "Trail",
|
|
573
|
+
"Trpk" => "Turnpike",
|
|
574
|
+
"Tunls" => "Tunnel",
|
|
575
|
+
"Un" => "Union",
|
|
576
|
+
"Vill" => "Village",
|
|
577
|
+
"Villag" => "Village",
|
|
578
|
+
"Villg" => "Village",
|
|
579
|
+
"Vis" => "Vista",
|
|
580
|
+
"Vlg" => "Village",
|
|
581
|
+
"Vlgs" => "Villages",
|
|
582
|
+
"Wls" => "Wells",
|
|
583
|
+
"Wy" => "Way",
|
|
584
|
+
"Xing" => "Crossing",
|
|
585
|
+
]
|
|
586
|
+
|
|
587
|
+
# The State constant maps US state and territory names to their 2-letter
|
|
588
|
+
# USPS abbreviations.
|
|
589
|
+
State = Map[
|
|
590
|
+
"Alabama" => "AL",
|
|
591
|
+
"Alaska" => "AK",
|
|
592
|
+
"American Samoa" => "AS",
|
|
593
|
+
"Arizona" => "AZ",
|
|
594
|
+
"Arkansas" => "AR",
|
|
595
|
+
"California" => "CA",
|
|
596
|
+
"Colorado" => "CO",
|
|
597
|
+
"Connecticut" => "CT",
|
|
598
|
+
"Delaware" => "DE",
|
|
599
|
+
"District of Columbia" => "DC",
|
|
600
|
+
"Washington DC" => "DC",
|
|
601
|
+
"Washington D.C." => "DC",
|
|
602
|
+
"Federated States of Micronesia" => "FM",
|
|
603
|
+
"Florida" => "FL",
|
|
604
|
+
"Georgia" => "GA",
|
|
605
|
+
"Guam" => "GU",
|
|
606
|
+
"Hawaii" => "HI",
|
|
607
|
+
"Idaho" => "ID",
|
|
608
|
+
"Illinois" => "IL",
|
|
609
|
+
"Indiana" => "IN",
|
|
610
|
+
"Iowa" => "IA",
|
|
611
|
+
"Kansas" => "KS",
|
|
612
|
+
"Kentucky" => "KY",
|
|
613
|
+
"Louisiana" => "LA",
|
|
614
|
+
"Maine" => "ME",
|
|
615
|
+
"Marshall Islands" => "MH",
|
|
616
|
+
"Maryland" => "MD",
|
|
617
|
+
"Massachusetts" => "MA",
|
|
618
|
+
"Michigan" => "MI",
|
|
619
|
+
"Minnesota" => "MN",
|
|
620
|
+
"Mississippi" => "MS",
|
|
621
|
+
"Missouri" => "MO",
|
|
622
|
+
"Montana" => "MT",
|
|
623
|
+
"Nebraska" => "NE",
|
|
624
|
+
"Nevada" => "NV",
|
|
625
|
+
"New Hampshire" => "NH",
|
|
626
|
+
"New Jersey" => "NJ",
|
|
627
|
+
"New Mexico" => "NM",
|
|
628
|
+
"New York" => "NY",
|
|
629
|
+
"North Carolina" => "NC",
|
|
630
|
+
"North Dakota" => "ND",
|
|
631
|
+
"Northern Mariana Islands" => "MP",
|
|
632
|
+
"Ohio" => "OH",
|
|
633
|
+
"Oklahoma" => "OK",
|
|
634
|
+
"Oregon" => "OR",
|
|
635
|
+
"Palau" => "PW",
|
|
636
|
+
"Pennsylvania" => "PA",
|
|
637
|
+
"Puerto Rico" => "PR",
|
|
638
|
+
"Rhode Island" => "RI",
|
|
639
|
+
"South Carolina" => "SC",
|
|
640
|
+
"South Dakota" => "SD",
|
|
641
|
+
"Tennessee" => "TN",
|
|
642
|
+
"Texas" => "TX",
|
|
643
|
+
"Utah" => "UT",
|
|
644
|
+
"Vermont" => "VT",
|
|
645
|
+
"Virgin Islands" => "VI",
|
|
646
|
+
"Virginia" => "VA",
|
|
647
|
+
"Washington" => "WA",
|
|
648
|
+
"West Virginia" => "WV",
|
|
649
|
+
"Wisconsin" => "WI",
|
|
650
|
+
"Wyoming" => "WY"
|
|
651
|
+
]
|
|
652
|
+
|
|
653
|
+
end
|