Fix for #1077, fail to search Saint.. cities in structured queries

pelias · Feb 28, 2018 · d187e7d · d187e7d
1 parent c56336d
commit d187e7d
Show file tree

Hide file tree

Showing 2 changed files with 17 additions and 17 deletions.
diff --git a/sanitizer/_city_name_standardizer.js b/sanitizer/_city_name_standardizer.js
@@ -1,13 +1,13 @@
 const _ = require('lodash');
 
 // matches 'ft', 'mt', 'saint', and 'sainte' on word boundary
-const mountSaintFort = /\b([fm]t|sainte?)\b/g;
+const mountSaintFort = /\b([fm]t|ste?)\b/g;
 
 const transliterations = {
   'mt':     'mount',
   'ft':     'fort',
-  'saint':  'st',
-  'sainte': 'ste'
+  'st':  'saint',
+  'ste': 'sainte'
 };
 
 function transliterate(match) {
@@ -22,15 +22,15 @@ function _sanitize(raw, clean) {
 
   // only try to transliterate if there is a city in parsed_text
   if (!_.isEmpty(_.get(clean, 'parsed_text.city'))) {
-    // eg input: Ft. Saint Louis
-    // after 1.  ft  saint louis
-    // after 2.  fort  st louis
-    // after 3.  fort st louis
+    // eg input: Ft. st Louis
+    // after 1.  ft  st louis
+    // after 2.  fort  saint louis
+    // after 3.  fort saint louis
 
     // 1.  remove '.' that could abbreviate ft and mt (makes transliteration regex easier)
     const periods_removed = _.toLower(clean.parsed_text.city).replace(/\b(mt|ft)\./g, '$1 ');
 
-    // 2.  transliterate 'saint'->'st', etc
+    // 2.  transliterate 'st'->'saint', etc
     const transliterated = periods_removed.replace(mountSaintFort, transliterate);
 
     // 3.  reduce whitespace sequences that can occur when removing periods down to a single space

diff --git a/test/unit/sanitizer/_city_name_standardizer.js b/test/unit/sanitizer/_city_name_standardizer.js
@@ -48,15 +48,15 @@ module.exports.tests.text_parser = function(test, common) {
 
   });
 
-  test('\'saint\' should be abbreviated to \'st\' wherever it appears in the city', function(t) {
+  test('\'st\' should be expanded to \'saint\' wherever it appears in the city', function(t) {
     const raw = {};
 
     const clean = {
       parsed_text: {
         query: 'saint query value',
         neighbourhood: 'saint neighbourhood value',
         borough: 'saint borough value',
-        city: 'SainT city sAiNt value saInt',
+        city: 'st city ST value St',
         county: 'saint county value',
         state: 'saint state value',
         postalcode: 'saint postalcode value',
@@ -69,7 +69,7 @@ module.exports.tests.text_parser = function(test, common) {
         query: 'saint query value',
         neighbourhood: 'saint neighbourhood value',
         borough: 'saint borough value',
-        city: 'st city st value st',
+        city: 'saint city saint value saint',
         county: 'saint county value',
         state: 'saint state value',
         postalcode: 'saint postalcode value',
@@ -86,15 +86,15 @@ module.exports.tests.text_parser = function(test, common) {
 
   });
 
-  test('\'sainte\' should be abbreviated to \'ste\' wherever it appears in the city', function(t) {
+  test('\'ste\' should be expanded to \'sainte\' wherever it appears in the city', function(t) {
     const raw = {};
 
     const clean = {
       parsed_text: {
         query: 'sainte query value',
         neighbourhood: 'sainte neighbourhood value',
         borough: 'sainte borough value',
-        city: 'SaintE city sAinTe value saINte',
+        city: 'ste city STE value StE',
         county: 'sainte county value',
         state: 'sainte state value',
         postalcode: 'sainte postalcode value',
@@ -107,7 +107,7 @@ module.exports.tests.text_parser = function(test, common) {
         query: 'sainte query value',
         neighbourhood: 'sainte neighbourhood value',
         borough: 'sainte borough value',
-        city: 'ste city ste value ste',
+        city: 'sainte city sainte value sainte',
         county: 'sainte county value',
         state: 'sainte state value',
         postalcode: 'sainte postalcode value',
@@ -200,18 +200,18 @@ module.exports.tests.text_parser = function(test, common) {
 
   });
 
-  test('mixture of \'mt\', \'ft\', \'saint\', and \'sainte\' should be expanded/abbreviated', function(t) {
+  test('mixture of \'mt\', \'ft\', \'st\', and \'st\' should be expanded', function(t) {
     const raw = {};
 
     const clean = {
       parsed_text: {
-        city: 'mt. ft saint sainte mt ft.'
+        city: 'mt. ft st ste mt ft.'
       }
     };
 
     const expected_clean = {
       parsed_text: {
-        city: 'mount fort st ste mount fort'
+        city: 'mount fort saint sainte mount fort'
       }
     };