Unit parsing

[missinglink]

The following addresses with unit numbers are not able to accurately detect the 'B' as a unit.

221 B Baker St
(0.99) ➜ [ { housenumber: '221' }, { street: 'B Baker St' } ]

221/B Baker St
(0.48) ➜ [ { street: 'Baker St' } ]

[missinglink]

I started looking at adding a UnitClassification but it's not as simple as I had hoped :(

new file mode 100644
index 0000000..0c6fc02
--- /dev/null
+++ b/classification/UnitClassification.js
@@ -0,0 +1,11 @@
+const Classification = require('./Classification')
+
+class UnitClassification extends Classification {
+  constructor (confidence, meta) {
+    super(confidence, meta)
+    this.public = true
+    this.label = 'unit'
+  }
+}
+
+module.exports = UnitClassification
diff --git a/classification/UnitClassification.test.js b/classification/UnitClassification.test.js
new file mode 100644
index 0000000..bdf44e7
--- /dev/null
+++ b/classification/UnitClassification.test.js
@@ -0,0 +1,24 @@
+const Classification = require('./UnitClassification')
+
+module.exports.tests = {}
+
+module.exports.tests.constructor = (test) => {
+  test('constructor', (t) => {
+    let c = new Classification()
+    t.false(c.public)
+    t.equals(c.label, 'unit')
+    t.equals(c.confidence, 1.0)
+    t.deepEqual(c.meta, {})
+    t.end()
+  })
+}
+
+module.exports.all = (tape, common) => {
+  function test (name, testFunction) {
+    return tape(`UnitClassification: ${name}`, testFunction)
+  }
+
+  for (var testCase in module.exports.tests) {
+    module.exports.tests[testCase](test, common)
+  }
+}
diff --git a/classifier/UnitClassifier.js b/classifier/UnitClassifier.js
new file mode 100644
index 0000000..c3e12c3
--- /dev/null
+++ b/classifier/UnitClassifier.js
@@ -0,0 +1,12 @@
+const WordClassifier = require('./super/WordClassifier')
+const UnitClassification = require('../classification/UnitClassification')
+
+class UnitClassifier extends WordClassifier {
+  each (span) {
+    if (/^[a-zA-Z\u0400-\u04FF]{1}$/.test(span.body)) {
+      span.classify(new UnitClassification(1))
+    }
+  }
+}
+
+module.exports = UnitClassifier
diff --git a/classifier/UnitClassifier.test.js b/classifier/UnitClassifier.test.js
new file mode 100644
index 0000000..68e5089
--- /dev/null
+++ b/classifier/UnitClassifier.test.js
@@ -0,0 +1,45 @@
+const UnitClassifier = require('./UnitClassifier')
+const UnitClassification = require('../classification/UnitClassification')
+const Span = require('../tokenization/Span')
+const classifier = new UnitClassifier()
+
+module.exports.tests = {}
+
+function classify (body) {
+  let s = new Span(body)
+  classifier.each(s)
+  return s
+}
+
+module.exports.tests.alpha = (test) => {
+  test('UnitClassification: English letter uppercase', (t) => {
+    let s = classify('A')
+    t.deepEqual(s.classifications, { UnitClassification: new UnitClassification(1.0) })
+    t.end()
+  })
+  test('UnitClassification: English letter lowercase', (t) => {
+    let s = classify('a')
+    t.deepEqual(s.classifications, { UnitClassification: new UnitClassification(1.0) })
+    t.end()
+  })
+  test('UnitClassification: Cyrillic', (t) => {
+    let s = classify('в')
+    t.deepEqual(s.classifications, { UnitClassification: new UnitClassification(1.0) })
+    t.end()
+  })
+  test('UnitClassification: Cyrillic', (t) => {
+    let s = classify('б')
+    t.deepEqual(s.classifications, { UnitClassification: new UnitClassification(1.0) })
+    t.end()
+  })
+}
+
+module.exports.all = (tape, common) => {
+  function test (name, testFunction) {
+    return tape(`UnitClassifier: ${name}`, testFunction)
+  }
+
+  for (var testCase in module.exports.tests) {
+    module.exports.tests[testCase](test, common)
+  }
+}
diff --git a/parser/AddressParser.js b/parser/AddressParser.js
index d070631..8e8aa7d 100644
--- a/parser/AddressParser.js
+++ b/parser/AddressParser.js
@@ -2,6 +2,7 @@ const Parser = require('./Parser')
 const AlphaNumericClassifier = require('../classifier/AlphaNumericClassifier')
 const TokenPositionClassifier = require('../classifier/TokenPositionClassifier')
 const HouseNumberClassifier = require('../classifier/HouseNumberClassifier')
+const UnitClassifier = require('../classifier/UnitClassifier')
 const PostcodeClassifier = require('../classifier/PostcodeClassifier')
 const StreetPrefixClassifier = require('../classifier/StreetPrefixClassifier')
 const StreetSuffixClassifier = require('../classifier/StreetSuffixClassifier')
@@ -42,6 +43,7 @@ class AddressParser extends Parser {

         // word classifiers
         new HouseNumberClassifier(),
+        new UnitClassifier(),
         new PostcodeClassifier(),
         new StreetPrefixClassifier(),
         new StreetSuffixClassifier(),

[Joxit]

Is this the wanted result ?

[ { housenumber: '221' }, { unit: 'B' }, { street: 'B Baker St' } ]

[missinglink]

Sorry for the delay, the desired result would be:

housenumber: 221
unit: B
street: Baker St