From 4b1f352e542581e5c73752bd90ab674dff6b4161 Mon Sep 17 00:00:00 2001
From: Peter Johnson <738069+missinglink@users.noreply.github.com>
Date: Mon, 9 Mar 2026 15:11:47 +0100
Subject: [PATCH] feat(dedupe): squash empire and country

---
 helper/diffPlaces.js           |  19 ++++
 test/unit/middleware/dedupe.js | 168 +++++++++++++++++++++++++++++++++
 2 files changed, 187 insertions(+)

diff --git a/helper/diffPlaces.js b/helper/diffPlaces.js
index 0a3c195b7..136c90db7 100644
--- a/helper/diffPlaces.js
+++ b/helper/diffPlaces.js
@@ -95,6 +95,14 @@ function isParentHierarchyDifferent(item1, item2){
     }
   }
 
+  // special case to consider empires and country as the same for deduplication purposes
+  if (
+    (item1.layer === 'empire' && item2.layer === 'country') ||
+    (item1.layer === 'country' && item2.layer === 'empire')
+  ) {
+    return false;
+  }
+
   // special handling of postal codes, which we consider to be strictly
   // unique within a single country/dependency regardless of the rest of
   // the hierarchy (ie. we ignore other parent properties)
@@ -351,6 +359,17 @@ function layerDependentNormalization(names, layer) {
     });
   }
 
+  // empire / country USA synonyms
+  if (layer === 'empire' || layer === 'country') {
+    _.forEach(names, (value, lang) => {
+      copy[lang] = field.getArrayValue(value).map(name => {
+        return name
+          .replace(/^(united states) of america$/i, '$1')
+          .trim();
+      });
+    });
+  }
+
   // county
   if( layer === 'county' ){
     _.forEach(names, (value, lang) => {
diff --git a/test/unit/middleware/dedupe.js b/test/unit/middleware/dedupe.js
index f8872ee4b..d3ea2178e 100644
--- a/test/unit/middleware/dedupe.js
+++ b/test/unit/middleware/dedupe.js
@@ -821,6 +821,174 @@ module.exports.tests.priority = function(test, common) {
     });
   });
 
+  test('real-world test New Zealand: empire vs country', function (t) {
+    var req = {
+      clean: {
+        text: 'New Zealand',
+        size: 100
+      }
+    };
+    var res = {
+      data:  [
+        {
+          'name': {
+            'default': 'New Zealand'
+          },
+          'source': 'whosonfirst',
+          'source_id': '136253053',
+          'layer': 'empire',
+          'parent': {
+            'empire_id': 136253053
+          },
+        },
+        {
+          'name': {
+            'default': [ 'New Zealand', 'Aotearoa' ],
+          },
+          'source': 'whosonfirst',
+          'source_id': '85633345',
+          'layer': 'country',
+          'parent': {
+            'continent_id': 102191583,
+            'country_id': 85633345
+          },
+        }
+      ]
+    };
+
+    dedupe(req, res, function () {
+      t.equal(res.data.length, 1, 'results have fewer items than before');
+      t.equal(res.data[0].layer, 'country', 'empire result removed');
+      t.end();
+    });
+  });
+
+  test('real-world test New Zealand: empire vs country - inverted order', function (t) {
+    var req = {
+      clean: {
+        text: 'New Zealand',
+        size: 100
+      }
+    };
+    var res = {
+      data:  [
+        {
+          'name': {
+            'default': [ 'New Zealand', 'Aotearoa' ],
+          },
+          'source': 'whosonfirst',
+          'source_id': '85633345',
+          'layer': 'country',
+          'parent': {
+            'continent_id': 102191583,
+            'country_id': 85633345
+          },
+        },
+        {
+          'name': {
+            'default': 'New Zealand'
+          },
+          'source': 'whosonfirst',
+          'source_id': '136253053',
+          'layer': 'empire',
+          'parent': {
+            'empire_id': 136253053
+          },
+        }
+      ]
+    };
+
+    dedupe(req, res, function () {
+      t.equal(res.data.length, 1, 'results have fewer items than before');
+      t.equal(res.data[0].layer, 'country', 'empire result removed');
+      t.end();
+    });
+  });
+
+  test('real-world test United States: empire vs country', function (t) {
+    var req = {
+      clean: {
+        text: 'United States',
+        size: 100
+      }
+    };
+    var res = {
+      data:  [
+        {
+          'name': {
+            'default': 'United States'
+          },
+          'source': 'whosonfirst',
+          'source_id': '85633793',
+          'layer': 'country',
+          'parent': {
+            'empire_id': 136253057,
+            'country_id': 85633793
+          },
+        },
+        {
+          'name': {
+            'default': 'United States of America',
+          },
+          'source': 'whosonfirst',
+          'source_id': '136253057',
+          'layer': 'empire',
+          'parent': {
+            'empire_id': 136253057,
+          },
+        }
+      ]
+    };
+
+    dedupe(req, res, function () {
+      t.equal(res.data.length, 1, 'results have fewer items than before');
+      t.equal(res.data[0].layer, 'country', 'empire result removed');
+      t.end();
+    });
+  });
+
+  test('real-world test United States: empire vs country - inverted order', function (t) {
+    var req = {
+      clean: {
+        text: 'United States',
+        size: 100
+      }
+    };
+    var res = {
+      data:  [
+        {
+          'name': {
+            'default': 'United States of America',
+          },
+          'source': 'whosonfirst',
+          'source_id': '136253057',
+          'layer': 'empire',
+          'parent': {
+            'empire_id': 136253057,
+          },
+        },
+        {
+          'name': {
+            'default': 'United States'
+          },
+          'source': 'whosonfirst',
+          'source_id': '85633793',
+          'layer': 'country',
+          'parent': {
+            'empire_id': 136253057,
+            'country_id': 85633793
+          },
+        }
+      ]
+    };
+
+    dedupe(req, res, function () {
+      t.equal(res.data.length, 1, 'results have fewer items than before');
+      t.equal(res.data[0].layer, 'country', 'empire result removed');
+      t.end();
+    });
+  });
+
   test('A->B B->C dependency graph', function (t) {
     var req = {
       clean: {