very minor tweaks to the conflation algorithm

src/stage2-preprocess/preprocessNZGB.ts

@@ -34,7 +34,7 @@ type TempObject = {
 
 // these transformations must be kept to a bare minimum
 const transformName = (_nzgbName: string, type: NameType) => {
-  let nzgbName = _nzgbName.replace(/ pa$/i, ' Pā');
+  let nzgbName = _nzgbName.trim().replace(/ pa$/i, ' Pā');
 
   // for train stations, OSM doesn't include the suffix in the name
   if (type === 'Railway Station') {
@@ -59,6 +59,9 @@ async function csvToTemp(): Promise<{ out: TempObject; ety: EtymologyReport }> {
         if (!(index % 1000)) process.stdout.write('.');
         index += 1;
 
+        // "Discontinued" don't exist or completely irrelevant
+        if (data.status.endsWith('Discontinued')) return;
+
         /** cause of the BOM character at the start of the csv file we do this */
         const ref = +(data.name_id || data['\uFEFFname_id' as 'name_id']);
 

src/stage3-conflate/compareFeatures/compareFeatures.ts

@@ -117,7 +117,9 @@ export function compareFeatures(
   const threshold =
     osm.osmId[0] === 'n'
       ? DISTANCE_APART_THRESHOLD_NODE
-      : DISTANCE_APART_THRESHOLD_AREA;
+      : osm.osmId[0] === 'w'
+        ? DISTANCE_APART_THRESHOLD_AREA
+        : Infinity; // relations are not checked for distance
   if (metresAway > threshold && !DONT_TRY_TO_MOVE.has(nzgb.type)) {
     tagChanges.__action = 'move';
   }

src/stage3-conflate/getPresetTags.ts

@@ -23,12 +23,12 @@ export function getPresetTags(place: NZGBFeature): {
     return {
       all: preset.subseaTags,
       match: preset.subseaTags,
-      acceptTags: preset.acceptTags,
+      acceptTags: [...(preset.acceptTags || []), preset.onLandTags],
     };
   }
   return {
     all: preset.onLandTags,
     match: preset.onLandTags,
-    acceptTags: preset.acceptTags,
+    acceptTags: [...(preset.acceptTags || []), preset.subseaTags],
   };
 }