feat(unit): more permissive unit patterns (#96)

pelias · Jun 3, 2020 · 4667240 · 4667240
1 parent c0dc786
commit 4667240
Show file tree

Hide file tree

Showing 3 changed files with 113 additions and 11 deletions.
diff --git a/classifier/UnitClassifier.js b/classifier/UnitClassifier.js
@@ -1,19 +1,38 @@
 const WordClassifier = require('./super/WordClassifier')
 const UnitClassification = require('../classification/UnitClassification')
 
-class UnitClassifier extends WordClassifier {
-  each (span) {
-    // skip spans which do not contain numbers
-    if (!span.contains.numerals) { return }
+const AllNumbersRegExp = /^#?\d+$/
+const SingleLetterRegExp = /^#?[A-Za-z]$/
+const NumbersThenLetterRegExp = /^#?\d+[A-Za-z]$/
+const LetterThenNumbersRegExp = /^#?[A-Za-z]\d+$/
+
+// based on https://stackoverflow.com/questions/9213237/combining-regular-expressions-in-javascript
+function combineRegExps (...args) {
+  var components = []
+
+  args.forEach((arg) => {
+    components = components.concat(arg._components || arg.source)
+  })
 
-    if (/^\d+$/.test(span.body)) {
-      let prev = span.graph.findOne('prev')
+  var combined = new RegExp('(?:' + components.join(')|(?:') + ')')
+  return combined
+}
+
+const combinedUnitRegexp = combineRegExps(
+  AllNumbersRegExp,
+  SingleLetterRegExp,
+  NumbersThenLetterRegExp,
+  LetterThenNumbersRegExp
+)
 
-      // Unit must be preceded by unit type
-      if (!prev || !prev.classifications.hasOwnProperty('UnitTypeClassification')) {
-        return
-      }
+class UnitClassifier extends WordClassifier {
+  each (span) {
+    const prev = span.graph.findOne('prev')
+    const hasPrevUnitToken = prev && prev.classifications.hasOwnProperty('UnitTypeClassification')
 
+    // If the previous token in a unit word, like apt or suite
+    // and this token is something like A2, 3b, 120, A, label it as a unit (number)
+    if (hasPrevUnitToken && combinedUnitRegexp.test(span.body)) {
       span.classify(new UnitClassification(1))
     }
   }

diff --git a/classifier/UnitClassifier.test.js b/classifier/UnitClassifier.test.js
@@ -33,6 +33,21 @@ module.exports.tests.without_unit_type = (test) => {
     t.deepEqual(s.classifications, { })
     t.end()
   })
+  test('letter and number without unit type', (t) => {
+    let s = classify('a2')
+    t.deepEqual(s.classifications, { })
+    t.end()
+  })
+  test('single letter without unit type', (t) => {
+    let s = classify('a')
+    t.deepEqual(s.classifications, { })
+    t.end()
+  })
+  test('number with # without unit type', (t) => {
+    let s = classify('#22')
+    t.deepEqual(s.classifications, { })
+    t.end()
+  })
 }
 
 module.exports.tests.with_unit_type = (test) => {
@@ -48,7 +63,22 @@ module.exports.tests.with_unit_type = (test) => {
   })
   test('number and letter with unit type', (t) => {
     let s = classify('2020a', 'unit')
-    t.deepEqual(s.classifications, { })
+    t.deepEqual(s.classifications, { UnitClassification: new UnitClassification(1.0) })
+    t.end()
+  })
+  test('letter and number with unit type', (t) => {
+    let s = classify('a2', 'unit')
+    t.deepEqual(s.classifications, { UnitClassification: new UnitClassification(1.0) })
+    t.end()
+  })
+  test('single letter with unit type', (t) => {
+    let s = classify('a', 'unit')
+    t.deepEqual(s.classifications, { UnitClassification: new UnitClassification(1.0) })
+    t.end()
+  })
+  test('number with # with unit type', (t) => {
+    let s = classify('#22', 'unit')
+    t.deepEqual(s.classifications, { UnitClassification: new UnitClassification(1.0) })
     t.end()
   })
 }

diff --git a/test/address.usa.test.js b/test/address.usa.test.js
@@ -73,6 +73,59 @@ const testcase = (test, common) => {
   assert('1389a IA 42 IA', [{ housenumber: '1389a' }, { street: 'IA 42' }, { region: 'IA' }], true)
 
   assert('1111 MD 760, Lusby, MD, USA', [{ housenumber: '1111' }, { street: 'MD 760' }, { locality: 'Lusby' }, { region: 'MD' }, { country: 'USA' }], true)
+
+  // unit + unit number tests
+  assert('52 Ten Eyck St Apt 3 Brooklyn NY', [
+    { housenumber: '52' }, { street: 'Ten Eyck St' },
+    { unit_type: 'Apt' },
+    { unit: '3' },
+    { locality: 'Brooklyn' },
+    { region: 'NY' }
+  ])
+
+  assert('52 Ten Eyck St Apt 3b Brooklyn NY', [
+    { housenumber: '52' }, { street: 'Ten Eyck St' },
+    { unit_type: 'Apt' },
+    { unit: '3b' },
+    { locality: 'Brooklyn' },
+    { region: 'NY' }
+  ])
+
+  assert('52 Ten Eyck St Apt 3B Brooklyn NY', [
+    { housenumber: '52' }, { street: 'Ten Eyck St' },
+    { unit_type: 'Apt' },
+    { unit: '3B' },
+    { locality: 'Brooklyn' },
+    { region: 'NY' }
+  ])
+
+  assert('52 Ten Eyck St Apt #3b Brooklyn NY', [
+    { housenumber: '52' }, { street: 'Ten Eyck St' },
+    { unit_type: 'Apt' },
+    { unit: '#3b' },
+    { locality: 'Brooklyn' },
+    { region: 'NY' }
+  ])
+
+  assert('52 Ten Eyck St 3 Brooklyn NY', [
+    { housenumber: '52' }, { street: 'Ten Eyck St' },
+    { locality: 'Brooklyn' },
+    { region: 'NY' }
+  ])
+
+  assert('52 Ten Eyck St 3 Brooklyn NY', [
+    { housenumber: '52' }, { street: 'Ten Eyck St' },
+    { locality: 'Brooklyn' },
+    { region: 'NY' }
+  ])
+
+  assert('6 Montague Terrace Apt A2 Brooklyn NY', [
+    { housenumber: '6' }, { street: 'Montague Terrace' },
+    { unit_type: 'Apt' },
+    { unit: 'A2' },
+    { locality: 'Brooklyn' },
+    { region: 'NY' }
+  ])
 }
 
 module.exports.all = (tape, common) => {