fix: Merge pull request #1091 from pelias/joxit-saint-fixes

Improved Saint/Sainte synonym handling

sanitizer/_city_name_standardizer.js

@@ -1,20 +1,18 @@
 const _ = require('lodash');
 
-// matches 'ft', 'mt', 'saint', and 'sainte' on word boundary
-const mountSaintFort = /\b([fm]t|sainte?)\b/g;
+// matches 'ft', 'mt' on word boundary
+const mountFort = /\b([fm]t)\b/g;
 
 const transliterations = {
   'mt':     'mount',
-  'ft':     'fort',
-  'saint':  'st',
-  'sainte': 'ste'
+  'ft':     'fort'
 };
 
 function transliterate(match) {
   return _.get(transliterations, match);
 }
 
-// transliterate ft/mt/saint/sainte to fort/mount/st/ste, respectively
+// transliterate ft/mt to fort/mount, respectively
 function _sanitize(raw, clean) {
   // error & warning messages
   // this function doesn't add any error or warning messages
@@ -22,16 +20,16 @@ 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
+    // eg input: Ft. st Louis
+    // after 1.  ft  st louis
     // after 2.  fort  st louis
     // after 3.  fort st 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
-    const transliterated = periods_removed.replace(mountSaintFort, transliterate);
+    // 2.  transliterate 'ft'->'fort', etc
+    const transliterated = periods_removed.replace(mountFort, transliterate);
 
     // 3.  reduce whitespace sequences that can occur when removing periods down to a single space
     const whitespace_normalized = _.trimEnd(transliterated.replace(/\s+/, ' '));

test/unit/sanitizer/_city_name_standardizer.js

@@ -48,82 +48,6 @@ module.exports.tests.text_parser = function(test, common) {
 
   });
 
-  test('\'saint\' should be abbreviated to \'st\' 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',
-        county: 'saint county value',
-        state: 'saint state value',
-        postalcode: 'saint postalcode value',
-        country: 'saint country value'
-      }
-    };
-
-    const expected_clean = {
-      parsed_text: {
-        query: 'saint query value',
-        neighbourhood: 'saint neighbourhood value',
-        borough: 'saint borough value',
-        city: 'st city st value st',
-        county: 'saint county value',
-        state: 'saint state value',
-        postalcode: 'saint postalcode value',
-        country: 'saint country value'
-      }
-    };
-
-    const messages = sanitizer.sanitize(raw, clean);
-
-    t.deepEquals(clean, expected_clean);
-    t.deepEquals(messages.errors, [], 'no errors');
-    t.deepEquals(messages.warnings, [], 'no warnings');
-    t.end();
-
-  });
-
-  test('\'sainte\' should be abbreviated to \'ste\' 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',
-        county: 'sainte county value',
-        state: 'sainte state value',
-        postalcode: 'sainte postalcode value',
-        country: 'sainte country value'
-      }
-    };
-
-    const expected_clean = {
-      parsed_text: {
-        query: 'sainte query value',
-        neighbourhood: 'sainte neighbourhood value',
-        borough: 'sainte borough value',
-        city: 'ste city ste value ste',
-        county: 'sainte county value',
-        state: 'sainte state value',
-        postalcode: 'sainte postalcode value',
-        country: 'sainte country value'
-      }
-    };
-
-    const messages = sanitizer.sanitize(raw, clean);
-
-    t.deepEquals(clean, expected_clean);
-    t.deepEquals(messages.errors, [], 'no errors');
-    t.deepEquals(messages.warnings, [], 'no warnings');
-    t.end();
-
-  });
-
   test('\'ft\' should be expanded to \'fort\' wherever it appears in the city', function(t) {
     const raw = {};
 
@@ -200,18 +124,18 @@ 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\' should be expanded', function(t) {
     const raw = {};
 
     const clean = {
       parsed_text: {
-        city: 'mt. ft saint sainte mt ft.'
+        city: 'mt. ft mt ft.'
       }
     };
 
     const expected_clean = {
       parsed_text: {
-        city: 'mount fort st ste mount fort'
+        city: 'mount fort mount fort'
       }
     };