feat(shapefile): placetype local using English fallback and non-latin…

… (latin) names (#48) * Placetype local using English fallback and () latin names --------- Co-authored-by: Peter Johnson <[email protected]>
pelias · Oct 2, 2023 · 03b8fb3 · 03b8fb3
1 parent 425fecb
commit 03b8fb3
Show file tree

Hide file tree

Showing 3 changed files with 41 additions and 9 deletions.
diff --git a/whosonfirst/feature.js b/whosonfirst/feature.js
@@ -50,9 +50,15 @@ feature.getLocalLanguages = (feat, type = 'official') => {
 // see: https://github.com/whosonfirst-data/whosonfirst-data/issues/2154
 feature.getPlacetypeLocal = (feat) => _.flatten(
   feature.getLocalLanguages(feat, 'official')
-    .map(lang => `label:${lang}_x_preferred_placetype`)
-    .concat(['wof:placetype_local'])
-    .map(prop => _.get(feat, `properties.${prop}`))
+    .concat(['eng']) // fallback to English when no other languages are available
+    .map(lang => {
+      const endonym = _.get(feat, `properties.label:${lang}_x_preferred_placetype`)
+
+      // When a localized placetype isn't in a latin char set then WOF often provides
+      // a transliteration we can include in a parenthetical, like: مقاطعة (muhafazah)
+      const latinized = _.get(feat, `properties.label:${lang}_latn_x_preferred_placetype`)
+      return (endonym && latinized) ? endonym + ` (${latinized})` : endonym
+    })
 ).filter(val => _.isString(val) && !_.isEmpty(val))
 
 module.exports = feature
diff --git a/whosonfirst/feature.test.js b/whosonfirst/feature.test.js
@@ -80,19 +80,45 @@ module.exports.getLocalLanguages = (test) => {
 module.exports.getPlacetypeLocal = (test) => {
   test('getPlacetypeLocal', (t) => {
     t.deepEqual([], feature.getPlacetypeLocal({}))
-    t.deepEqual(['LOCAL_PLACETYPE'], feature.getPlacetypeLocal({
+    t.deepEqual([], feature.getPlacetypeLocal({
       properties: { 'wof:placetype_local': 'LOCAL_PLACETYPE' }
     }))
-    t.deepEqual(['LOCAL_PLACETYPE'], feature.getPlacetypeLocal({
-      properties: { 'wof:placetype_local': ['LOCAL_PLACETYPE'] }
-    }))
-    t.deepEqual(['ENDONYM_PLACETYPE', 'LOCAL_PLACETYPE'], feature.getPlacetypeLocal({
+    t.deepEqual(['ENDONYM_PLACETYPE'], feature.getPlacetypeLocal({
       properties: {
         'wof:lang_x_official': 'und',
         'label:und_x_preferred_placetype': 'ENDONYM_PLACETYPE',
         'wof:placetype_local': 'LOCAL_PLACETYPE'
       }
     }))
+    // Makkah, Saudi Arabia
+    t.deepEqual(['مقاطعة (muhafazah)', 'region'], feature.getPlacetypeLocal({
+      properties: {
+        'wof:lang_x_official': 'ara',
+        'label:ara_x_preferred_placetype': 'مقاطعة',
+        'label:ara_latn_x_preferred_placetype': 'muhafazah',
+        'label:eng_x_preferred_placetype': 'region',
+        'wof:placetype': 'region'
+      }
+    }))
+    // Glasgow, Scotland, UK
+    t.deepEqual(['unitary district'], feature.getPlacetypeLocal({
+      properties: {
+        'wof:lang_x_official': 'und',
+        'label:eng_x_preferred_placetype': 'unitary district',
+        'wof:placetype_local': 'unitary district',
+        'wof:placetype': 'county'
+      }
+    }))
+    // Madrid, province in Spain
+    t.deepEqual(['provincia', 'province'], feature.getPlacetypeLocal({
+      properties: {
+        'wof:lang_x_official': 'spa',
+        'label:spa_x_preferred_placetype': 'provincia',
+        'label:eng_x_preferred_placetype': 'province',
+        'wof:placetype_local': 'autonomous community', // old junk data
+        'wof:placetype': 'region'
+      }
+    }))
     t.end()
   })
 }
diff --git a/whosonfirst/view/shapefile.js b/whosonfirst/view/shapefile.js
@@ -29,7 +29,7 @@ module.exports = (feat, params) => {
     parent_id: spr.parent_id,
     name: spr.name,
     placetype: spr.placetype,
-    placelocal: _.first(feature.getPlacetypeLocal(feat)) || '', // note: not a SPR field
+    placelocal: feature.getPlaceTypeLocal(feat),
     country: spr.country,
     repo: spr.repo,
     lat: spr.latitude,