feat(solver): add MustNotPreceedFilter

parser/AddressParser.js

@@ -22,6 +22,7 @@ const ExclusiveCartesianSolver = require('../solver/ExclusiveCartesianSolver')
 const LeadingAreaDeclassifier = require('../solver/LeadingAreaDeclassifier')
 const MultiStreetSolver = require('../solver/MultiStreetSolver')
 const TokenDistanceFilter = require('../solver/TokenDistanceFilter')
+const MustNotPreceedFilter = require('../solver/MustNotPreceedFilter')
 const SubsetFilter = require('../solver/SubsetFilter')
 
 class AddressParser extends Parser {
@@ -61,6 +62,17 @@ class AddressParser extends Parser {
         new LeadingAreaDeclassifier(),
         new MultiStreetSolver(),
         new TokenDistanceFilter(),
+        new MustNotPreceedFilter('PostcodeClassification', 'HouseNumberClassification'),
+        new MustNotPreceedFilter('PostcodeClassification', 'StreetClassification'),
+        new MustNotPreceedFilter('LocalityClassification', 'HouseNumberClassification'),
+        new MustNotPreceedFilter('LocalityClassification', 'StreetClassification'),
+        new MustNotPreceedFilter('RegionClassification', 'HouseNumberClassification'),
+        new MustNotPreceedFilter('RegionClassification', 'StreetClassification'),
+        new MustNotPreceedFilter('CountryClassification', 'RegionClassification'),
+        new MustNotPreceedFilter('CountryClassification', 'LocalityClassification'),
+        new MustNotPreceedFilter('CountryClassification', 'PostcodeClassification'),
+        new MustNotPreceedFilter('CountryClassification', 'StreetClassification'),
+        new MustNotPreceedFilter('CountryClassification', 'HouseNumberClassification'),
         new SubsetFilter()
       ]
     )

solver/MustNotPreceedFilter.js

@@ -0,0 +1,28 @@
+// enforce that a object classification cannot preceed subject classification
+// @todo: handle case of multiple object classifications matching
+
+class MustNotPreceedFilter {
+  constructor (objectClassification, subjectClassification) {
+    this.classification = {
+      object: objectClassification,
+      subject: subjectClassification
+    }
+  }
+  solve (tokenizer) {
+    tokenizer.solution.forEach(s => {
+      let object = s.pair.filter(p => p.classification.constructor.name === this.classification.object)
+      let subject = s.pair.filter(p => p.classification.constructor.name === this.classification.subject)
+
+      // solution contains both object & subject classifications
+      if (object.length > 0 && subject.length > 0) {
+        // the object comes before the subject(s)
+        if (subject.some(p => p.span.start > object[0].span.end)) {
+          // remove the object classification from this solution
+          s.pair = s.pair.filter(p => p.classification.constructor.name !== this.classification.object)
+        }
+      }
+    })
+  }
+}
+
+module.exports = MustNotPreceedFilter

solver/MustNotPreceedFilter.test.js

@@ -0,0 +1,46 @@
+const Tokenizer = require('../tokenization/Tokenizer')
+const Span = require('../tokenization/Span')
+const PostcodeClassification = require('../classification/PostcodeClassification')
+const StreetClassification = require('../classification/StreetClassification')
+const Solution = require('./Solution')
+const SolutionPair = require('./SolutionPair')
+const MustNotPreceedFilter = require('./MustNotPreceedFilter')
+
+module.exports.tests = {}
+
+module.exports.tests.postcode_preceeds_street = (test) => {
+  test('postcode_preceeds_street: remove postcode', (t) => {
+    let tok = new Tokenizer()
+
+    let s1 = new Span('A')
+    s1.start = 0
+    s1.end = 1
+
+    let s2 = new Span('B')
+    s2.start = 3
+    s2.end = 4
+
+    let sp1 = new SolutionPair(s1, new PostcodeClassification(1.0))
+    let sp2 = new SolutionPair(s2, new StreetClassification(1.0))
+
+    tok.solution = [ new Solution([sp1, sp2]) ]
+
+    let c = new MustNotPreceedFilter('PostcodeClassification', 'StreetClassification')
+    c.solve(tok)
+
+    t.deepEquals(tok.solution.length, 1)
+    t.deepEquals(tok.solution[0].pair.length, 1)
+    t.deepEquals(tok.solution[0].pair[0], sp2)
+    t.end()
+  })
+}
+
+module.exports.all = (tape, common) => {
+  function test (name, testFunction) {
+    return tape(`MustNotPreceedFilter: ${name}`, testFunction)
+  }
+
+  for (var testCase in module.exports.tests) {
+    module.exports.tests[testCase](test, common)
+  }
+}

test/addressit.usa.test.js

@@ -64,11 +64,6 @@ const testcase = (test, common) => {
       { locality: 'Atlanta' }, { region: 'Georgia' },
       { postcode: '31035' }
     ],
-    [
-      { housenumber: '754' }, { street: 'Pharr Rd' },
-      { locality: 'Atlanta' }, { country: 'Georgia' },
-      { postcode: '31035' }
-    ],
     [
       { housenumber: '754' }, { street: 'Pharr Rd' },
       { locality: 'Georgia' }, { postcode: '31035' }