From 89ef34a41c56c4f4c4f867177319416395a4dea4 Mon Sep 17 00:00:00 2001
From: Peter Johnson <insomnia@rcpt.at>
Date: Mon, 5 Dec 2016 19:28:25 +0100
Subject: [PATCH] analyze: remove common mandarin labels

---
 lib/analyze.js      |  6 +++++-
 test/lib/analyze.js | 11 ++++++++---
 2 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/lib/analyze.js b/lib/analyze.js
index e25c9655..5cf4dd43 100644
--- a/lib/analyze.js
+++ b/lib/analyze.js
@@ -51,9 +51,13 @@ function housenumber( num ){
                   .replace(' 1/2', '½')
                   .replace(' 3/4', '¾');
 
-  // remove common labels
+  // remove common english labels
   number = number.replace(/\s+(apartment|apt|lot|space|ste|suite|unit)\s+/gi, '');
 
+  // remove common mandarin labels
+  // see: https://eastasiastudent.net/china/mandarin/postal-address/
+  number = number.replace(/(号|號|室|宅|楼)/g, '');
+
   // remove spaces from housenumber. eg: '2 A' -> '2A'
   number = number.replace(/\s+/g, '').toLowerCase();
 
diff --git a/test/lib/analyze.js b/test/lib/analyze.js
index cb2a23d0..67aa3e8f 100644
--- a/test/lib/analyze.js
+++ b/test/lib/analyze.js
@@ -59,8 +59,9 @@ module.exports.analyze.housenumber = function(test) {
     t.false(isNaN(analyze.housenumber('326 1/2')), 'half suffix');
     t.false(isNaN(analyze.housenumber('8¼')), 'quarter suffix');
     t.false(isNaN(analyze.housenumber('4701 #B')), 'hash delimited suffix');
-    t.false(isNaN(analyze.housenumber('1434 UNIT #B')), 'remove common label "unit"');
-    t.false(isNaN(analyze.housenumber('1434 SUITE #B')), 'remove common label "suite"');
+    t.false(isNaN(analyze.housenumber('1434 UNIT #B')), 'remove common english label "unit"');
+    t.false(isNaN(analyze.housenumber('1434 SUITE #B')), 'remove common english label "suite"');
+    t.false(isNaN(analyze.housenumber('158號')), 'remove common mandarin label');
     t.end();
   });
 
@@ -161,13 +162,17 @@ module.exports.analyze.housenumber = function(test) {
   //   t.equal(float, 5285.03);
   //   t.end();
   // });
-  test('housenumber: removes common labels', function(t) {
+  test('housenumber: removes common english labels', function(t) {
     var float = analyze.housenumber('4701 UNIT #B');
     t.equal(float, 4701.06);
     var float2 = analyze.housenumber('1 APT A');
     t.equal(float2, 1.03);
     t.end();
   });
+  test('housenumber: removes common mandarin labels', function(t) {
+    t.equal(analyze.housenumber('158號'), 158.00);
+    t.end();
+  });
 };
 module.exports.analyze.housenumberFloatToString = function(test) {
   test('housenumberFloatToString: invalid', function(t) {