refactor(parser): improved removal of unit numbers from $subject and …

…$admin
pelias · Sep 3, 2020 · ac26263 · ac26263
1 parent 80e10c1
commit ac26263
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 2 deletions.
diff --git a/sanitizer/_text_pelias_parser.js b/sanitizer/_text_pelias_parser.js
@@ -137,9 +137,10 @@ function parse (clean) {
   // '    VVVV NN SSSSSSS AAAAAA PPPPP      '
   let mask = solution.mask(t);
 
-  // the entire input text as seen by the parser with any postcode classification(s) removed
+  // the entire input text as seen by the parser with any postcode and unit
+  // classification(s) removed
   let body = t.span.body.split('')
-    .map((c, i) => (mask[i] !== 'P') ? c : ' ')
+    .map((c, i) => !/[PU]/.test(mask[i]) ? c : ' ')
     .join('');
 
   // same as $body above but with consecutive whitespace squashed and trimmed.

diff --git a/test/unit/sanitizer/_text_pelias_parser.js b/test/unit/sanitizer/_text_pelias_parser.js
@@ -419,6 +419,16 @@ module.exports.tests.text_parser = function (test, common) {
   cases.push(['e8 1dn', { subject: 'e8 1dn' }, true]);
   // cases.push(['london e8 1dn', { subject: 'e8 1dn' }, true]); // issue
 
+  // unit number between address and locality
+  cases.push(['7750 Kennedy Rd #2A Markham', {
+    subject: '7750 Kennedy Rd',
+    housenumber: '7750',
+    street: 'Kennedy Rd',
+    unit: '#2A',
+    locality: 'Markham',
+    admin: 'Markham',
+  }]);
+
   cases.forEach(testcase => {
     let input = testcase[0];
     let expected = testcase[1];