From 0ed71c534a289ea25e88f9ac4e5095714bd0806f Mon Sep 17 00:00:00 2001
From: Peter Johnson <insomnia@rcpt.at>
Date: Thu, 6 Jun 2019 18:50:26 +0200
Subject: [PATCH] feat: autocomplete jitter fixes (#44)

* fix: remove unused scheme

* fix: improve street_name scheme

* feat(CompositeClassifier): improvements to composite classifier, see PR notes

* feat(test): add autocomplete jitter test cases
---
 classifier/CompositeClassifier.js | 16 +++++++++++++++-
 classifier/scheme/street.js       | 15 ---------------
 classifier/scheme/street_name.js  |  6 ++++--
 test/address.usa.test.js          | 18 ++++++++++++++++++
 4 files changed, 37 insertions(+), 18 deletions(-)

diff --git a/classifier/CompositeClassifier.js b/classifier/CompositeClassifier.js
index 4807856d..be5e8dd0 100644
--- a/classifier/CompositeClassifier.js
+++ b/classifier/CompositeClassifier.js
@@ -49,6 +49,11 @@ class CompositeClassifier extends SectionClassifier {
   }
 
   each (section) {
+    let phrases = section.graph.findAll('phrase')
+
+    // sort phrases so shorter phrases are matched first
+    phrases.sort((a, b) => a.norm.length - b.norm.length)
+
     this.schemes.forEach(s => {
       // invalid scheme
       if (!Array.isArray(s.scheme)) { return }
@@ -57,7 +62,6 @@ class CompositeClassifier extends SectionClassifier {
       let candidates = []
 
       // compute candidate lists
-      let phrases = section.graph.findAll('phrase')
       candidates = s.scheme.map(s => phrases.filter(this.match.bind(null, s)))
 
       // no candidates were found for one or more schemes
@@ -79,6 +83,16 @@ class CompositeClassifier extends SectionClassifier {
           } else if (prev && curr.graph.findOne('child:first').graph.findOne('prev') !== prev.graph.findOne('child:last')) {
             return false
           }
+
+          // avoid adding tokens to the front of a street classification
+          // that begins with a street prefix.
+          // eg. 'A + Ave B' (ave is both a valid prefix & suffix)
+          if (next && next.classifications.hasOwnProperty('StreetClassification')) {
+            let firstChild = next.graph.findOne('child')
+            if (firstChild && firstChild.classifications.hasOwnProperty('StreetPrefixClassification')) {
+              return false
+            }
+          }
         }
         return true
       })
diff --git a/classifier/scheme/street.js b/classifier/scheme/street.js
index 374fbb32..eeb14418 100644
--- a/classifier/scheme/street.js
+++ b/classifier/scheme/street.js
@@ -303,21 +303,6 @@ module.exports = [
       }
     ]
   },
-  {
-    // West Main Street
-    confidence: 0.84,
-    Class: StreetClassification,
-    scheme: [
-      {
-        is: ['DirectionalClassification'],
-        not: ['StreetClassification', 'IntersectionClassification']
-      },
-      {
-        is: ['StreetClassification'],
-        not: ['DirectionalClassification']
-      }
-    ]
-  },
   {
     // Main Street West
     confidence: 0.88,
diff --git a/classifier/scheme/street_name.js b/classifier/scheme/street_name.js
index 0f88b653..0a8b5b7b 100644
--- a/classifier/scheme/street_name.js
+++ b/classifier/scheme/street_name.js
@@ -7,7 +7,8 @@ module.exports = [
     Class: StreetNameClassification,
     scheme: [
       {
-        is: ['StopWordClassification']
+        is: ['StopWordClassification'],
+        not: ['DirectionalClassification']
       },
       {
         is: ['AlphaClassification', 'PersonClassification'],
@@ -25,7 +26,8 @@ module.exports = [
         not: ['StreetClassification', 'IntersectionClassification', 'StopWordClassification', 'StreetPrefixClassification']
       },
       {
-        is: ['StopWordClassification']
+        is: ['StopWordClassification'],
+        not: ['DirectionalClassification']
       },
       {
         is: ['AlphaClassification', 'PersonClassification'],
diff --git a/test/address.usa.test.js b/test/address.usa.test.js
index aa41ed1d..bac0c07b 100644
--- a/test/address.usa.test.js
+++ b/test/address.usa.test.js
@@ -30,6 +30,24 @@ const testcase = (test, common) => {
 
   // postcode not allowed in first position otherwise
   assert('90210 Foo', [])
+
+  // autocomplete street name jitter
+  // note: we are only testing the street name stays the same throughout
+  assert('N FISKE AVE', [{ street: 'N FISKE AVE' }], true)
+  assert('N FISKE AVE P', [{ street: 'N FISKE AVE' }], true)
+  assert('N FISKE AVE Po', [{ street: 'N FISKE AVE' }, { region: 'Po' }], true)
+  assert('N FISKE AVE Por', [{ street: 'N FISKE AVE' }, { region: 'Por' }], true)
+  assert('N FISKE AVE Port', [{ street: 'N FISKE AVE' }, { locality: 'Port' }], true)
+  assert('N FISKE AVE Portl', [{ street: 'N FISKE AVE' }], true)
+  assert('N FISKE AVE Portla', [{ street: 'N FISKE AVE' }], true)
+  assert('N FISKE AVE Portlan', [{ street: 'N FISKE AVE' }], true)
+  assert('N FISKE AVE Portland', [{ street: 'N FISKE AVE' }, { locality: 'Portland' }], true)
+  assert('N DWIGHT AVE Portland O', [{ street: 'N DWIGHT AVE' }, { locality: 'Portland' }], true)
+  assert('N DWIGHT AVE Portland Or', [{ street: 'N DWIGHT AVE' }, { locality: 'Portland' }, { region: 'Or' }], true)
+  assert('N DWIGHT AVE Portland Ore', [{ street: 'N DWIGHT AVE' }, { locality: 'Portland' }], true)
+  assert('N DWIGHT AVE Portland Oreg', [{ street: 'N DWIGHT AVE' }, { locality: 'Portland' }], true)
+  assert('N DWIGHT AVE Portland Orego', [{ street: 'N DWIGHT AVE' }, { locality: 'Portland' }], true)
+  assert('N DWIGHT AVE Portland Oregon', [{ street: 'N DWIGHT AVE' }, { locality: 'Portland' }, { region: 'Oregon' }], true)
 }
 
 module.exports.all = (tape, common) => {