feat(country_codes): add dictionaries for country codes (#26)

Author: missinglink

.gitignore

@@ -6,7 +6,7 @@ resources/whosonfirst/dictionaries/*/*.txt
 
 # whitelist certain files
 !resources/whosonfirst/dictionaries/country/wof:country.txt
-!resources/whosonfirst/dictionaries/country/wof:shortcode.txt
+!resources/whosonfirst/dictionaries/country/wof:country_alpha3.txt
 !resources/whosonfirst/dictionaries/country/name:eng_x_preferred.txt
 !resources/whosonfirst/dictionaries/dependency/wof:shortcode.txt
 !resources/whosonfirst/dictionaries/dependency/name:eng_x_preferred.txt

classification/CountryClassification.js

@@ -2,7 +2,7 @@ const Classification = require('./Classification')
 
 class CountryClassification extends Classification {
   constructor (confidence, meta) {
-    super(confidence, meta)
+    super(0.9, meta)
     this.public = true
     this.label = 'country'
   }

classification/CountryClassification.test.js

@@ -7,7 +7,7 @@ module.exports.tests.constructor = (test) => {
     let c = new Classification()
     t.true(c.public)
     t.equals(c.label, 'country')
-    t.equals(c.confidence, 1.0)
+    t.equals(c.confidence, 0.9)
     t.deepEqual(c.meta, {})
     t.end()
   })

classifier/WhosOnFirstClassifier.js

@@ -24,7 +24,7 @@ const placetypes = {
   //   classifications: [AreaClassification, DependencyClassification]
   // },
   'country': {
-    files: ['name:eng_x_preferred.txt'],
+    files: ['name:eng_x_preferred.txt', 'wof:country.txt', 'wof:country_alpha3.txt'],
     classifications: [AreaClassification, CountryClassification]
   }
 }
@@ -104,7 +104,16 @@ class WhosOnFirstClassifier extends PhraseClassifier {
 
     Object.keys(placetypes).forEach(placetype => {
       if (this.tokens[placetype].has(span.norm)) {
-        placetypes[placetype].classifications.forEach(Class => span.classify(new Class(1.0)))
+        // do not classify tokens if they already have a 'StopWordClassification'
+        if (
+          span.classifications.hasOwnProperty('StopWordClassification') || (
+            span.graph.length('child') > 0 &&
+            span.graph.findOne('child').classifications.hasOwnProperty('StopWordClassification')
+          )
+        ) { return }
+
+        // classify tokens
+        placetypes[placetype].classifications.forEach(Class => span.classify(new Class()))
       }
     })
   }

resources/pelias/dictionaries/whosonfirst/locality/name:eng_x_preferred.txt

@@ -10,6 +10,7 @@ sf
 !temple
 !airport
 !deli
+!us
 # remove any localities which share a name with a US state
 !alabama
 !alaska

resources/whosonfirst/dictionaries/country/wof:country_alpha3.txt

@@ -0,0 +1,216 @@
+mac
+bhr
+btn
+lso
+pan
+mmr
+sxm
+vat
+gha
+irq
+idn
+omn
+bfa
+jam
+esh
+mdg
+tza
+afg
+kor
+guf
+bwa
+lao
+zwe
+cmr
+ben
+lbr
+sau
+eth
+com
+ecu
+vut
+ner
+tto
+ago
+bdi
+gnq
+tha
+abw
+qat
+rwa
+mdv
+kaz
+lka
+isr
+ven
+dji
+hnd
+tcd
+ken
+plw
+grd
+bhs
+and
+png
+reu
+pry
+mus
+khm
+irn
+sen
+lca
+spm
+mkd
+som
+mwi
+gtm
+caf
+tur
+blr
+guy
+kwt
+twn
+alb
+gab
+syr
+glp
+jor
+jpn
+fsm
+hti
+cyp
+mng
+cuw
+sur
+civ
+dza
+ton
+imn
+npl
+sle
+ind
+bgd
+hkg
+cri
+brb
+cck
+yem
+dma
+arg
+phl
+ury
+tjk
+col
+per
+myt
+atg
+mtq
+lbn
+nam
+cog
+slv
+ggy
+kna
+mli
+zmb
+vct
+blz
+are
+egy
+tls
+slb
+jey
+nic
+gmb
+sgp
+tuv
+bih
+bol
+uga
+lby
+swz
+prk
+cod
+uzb
+tgo
+ssd
+pak
+syc
+mhl
+mne
+tkm
+cub
+mrt
+wsm
+rus
+gin
+mar
+chn
+tun
+kir
+dom
+ata
+aze
+cpv
+ant
+moz
+nga
+mys
+nru
+brn
+sdn
+fji
+gnb
+kgz
+vnm
+stp
+arm
+eri
+aut
+aus
+bel
+bgr
+bra
+can
+che
+chl
+cze
+deu
+dnk
+esp
+est
+fin
+fra
+gbr
+geo
+grc
+grl
+hrv
+hun
+irl
+isl
+ita
+xkx
+lie
+ltu
+lux
+lva
+mco
+mda
+mex
+mlt
+nld
+nor
+nzl
+pol
+prt
+pse
+rou
+srb
+smr
+svk
+svn
+swe
+usa
+ukr
+zaf

test/addressit.usa.test.js

@@ -94,7 +94,7 @@ const testcase = (test, common) => {
     { place: 'Mt Tabor Park' },
     { housenumber: '6220' }, { street: 'SE Salmon St' },
     { locality: 'Portland' }, { region: 'OR' },
-    { postcode: '97215' }
+    { postcode: '97215' }, { country: 'USA' }
   ], true)
 
   // assert('Mt Tabor Park', [], true)

test/functional.test.js

@@ -83,9 +83,8 @@ const testcase = (test, common) => {
 
   // do not parse 'aus' as a locality if it follows a region
   assert('new south wales aus', [
-    [{ region: 'new south wales' }],
-    [{ locality: 'south wales' }]]
-  )
+    { region: 'new south wales' }, { country: 'aus' }
+  ], true)
 }
 
 module.exports.all = (tape, common) => {