feat(FillSolver): Add FillSolver, it will add unclassified tokens to the street

This will be used only when StreetPrefixClassification is used.

Remove Paris from regions and add cité in street_types.
Paris is always used as a locality

Author: Joxit

classifier/scheme/street_name.js

@@ -33,6 +33,24 @@ module.exports = [
       }
     ]
   },
+  {
+    // du 4 septembre
+    confidence: 0.5,
+    Class: StreetNameClassification,
+    scheme: [
+      {
+        is: ['StopWordClassification']
+      },
+      {
+        is: ['NumericClassification'],
+        not: ['PostcodeClassification']
+      },
+      {
+        is: ['AlphaClassification'],
+        not: ['StreetClassification', 'IntersectionClassification', 'LocalityClassification']
+      }
+    ]
+  },
   {
     // dos Fiéis de Deus
     confidence: 0.5,

parser/AddressParser.js

@@ -29,6 +29,7 @@ const TokenDistanceFilter = require('../solver/TokenDistanceFilter')
 const MustNotPreceedFilter = require('../solver/MustNotPreceedFilter')
 const MustNotFollowFilter = require('../solver/MustNotFollowFilter')
 const SubsetFilter = require('../solver/SubsetFilter')
+const FillSolver = require('../solver/FillSolver')
 
 class AddressParser extends Parser {
   constructor (options) {
@@ -106,7 +107,8 @@ class AddressParser extends Parser {
         new MustNotFollowFilter('LocalityClassification', 'RegionClassification'),
         new MustNotFollowFilter('LocalityClassification', 'CountryClassification'),
         new TokenDistanceFilter(),
-        new SubsetFilter()
+        new SubsetFilter(),
+        new FillSolver()
       ],
       options
     )

resources/pelias/dictionaries/libpostal/fr/street_types.txt

@@ -0,0 +1,2 @@
+cité|cite
+cités|cites

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

@@ -0,0 +1,2 @@
+# This is not used as region
+!paris

solver/FillSolver.js

@@ -0,0 +1,23 @@
+class FillSolver {
+  solve (tokenizer) {
+    tokenizer.solution.forEach(solution => {
+      // Get streets from this solution
+      const streets = solution.pair.filter(p => p.classification.constructor.name === 'StreetClassification')
+      // Get all nodes that are not in the solution
+      const missings = tokenizer.section[0].graph.edges.child.filter(c => !solution.pair.some(p => p.span.intersects(c)))
+
+      missings.forEach(missing => {
+        const street = streets.find(s => s.span.end === missing.start - 1)
+        const prefix = street && street.span.graph.findOne('child:first')
+        if (prefix && prefix.classifications.StreetPrefixClassification) {
+          street.span.setBody(`${street.span.body} ${missing.body}`)
+          street.span.graph.add('child', missing)
+          street.span.graph.remove('child:last', street.span.graph.findOne('child:last'))
+          street.span.graph.add('child:last', missing)
+        }
+      })
+    })
+  }
+}
+
+module.exports = FillSolver

test/address.fra.test.js

@@ -79,6 +79,22 @@ const testcase = (test, common) => {
   assert('Mery-Sur-Oise', [
     { locality: 'Mery-Sur-Oise' }
   ], true)
+
+  assert('16 Rue Des Petits Carreaux', [
+    { housenumber: '16' }, { street: 'Rue Des Petits Carreaux' }
+  ], true)
+
+  assert('16 Rue Des Petits Carreaux Paris', [
+    { housenumber: '16' }, { street: 'Rue Des Petits Carreaux' }, { locality: 'Paris' }
+  ], true)
+
+  assert('4 Cité Du Cardinal Lemoine 75005', [
+    { housenumber: '4' }, { street: 'Cité Du Cardinal Lemoine' }, { postcode: '75005' }
+  ], true)
+
+  assert('32 Rue Du 4 Septembre', [
+    { housenumber: '32' }, { street: 'Rue Du 4 Septembre' }
+  ], true)
 }
 
 module.exports.all = (tape, common) => {