feat(names): Detect near duplicate names

This handles the case where one alt name is a substring fully contained in another.
pelias · Sep 24, 2021 · 481144e · 481144e
1 parent 71b4b89
commit 481144e
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 20 deletions.
diff --git a/stream/tag_mapper.js b/stream/tag_mapper.js
@@ -43,7 +43,7 @@ module.exports = function(){
           if (trimmed_value)  {
             if (!trimmed_name ) {
               doc.setName('default', trim( tags[key]));
-            } else {
+            } else if(!trimmed_name.includes(trimmed_value)) {
               doc.setNameAlias('default', trim( tags[key]));
             }
           }

diff --git a/test/fixtures/combined_vancouver_queens.json b/test/fixtures/combined_vancouver_queens.json
@@ -856355,16 +856355,10 @@
     "_type": "_doc",
     "data": {
       "name": {
-        "default": [
-          "Mountain Equipment Co-op (MEC)",
-          "MEC"
-        ]
+        "default": "Mountain Equipment Co-op (MEC)"
       },
       "phrase": {
-        "default": [
-          "Mountain Equipment Co-op (MEC)",
-          "MEC"
-        ]
+        "default": "Mountain Equipment Co-op (MEC)"
       },
       "address_parts": {
         "number": "212",
@@ -892394,8 +892388,7 @@
     "_type": "_doc",
     "data": {
       "name": {
-        "default": "IPOH Asian House"
-      },
+        "default": "IPOH Asian House" },
       "phrase": {
         "default": "IPOH Asian House"
       },
@@ -892429,16 +892422,10 @@
     "_type": "_doc",
     "data": {
       "name": {
-        "default": [
-          "On Lok Restaurant & Wun Tun House",
-          "On Lok"
-        ]
+        "default": "On Lok Restaurant & Wun Tun House"
       },
       "phrase": {
-        "default": [
-          "On Lok Restaurant & Wun Tun House",
-          "On Lok"
-        ]
+        "default":  "On Lok Restaurant & Wun Tun House"
       },
       "address_parts": {
         "number": "2010",
@@ -956784,4 +956771,4 @@
       "bounding_box": "{\"min_lat\":49.2174915,\"max_lat\":49.2194865,\"min_lon\":-123.2018987,\"max_lon\":-123.1991481}"
     }
   }
-]
+]
diff --git a/test/stream/tag_mapper.js b/test/stream/tag_mapper.js
@@ -112,6 +112,38 @@ module.exports.tests.osm_names = function(test, common) {
   });
 };
 
+// Reject alt names that are a substring of the main name
+module.exports.tests.substring_alt_name = function(test, common) {
+  var doc = new Document('a','b',1);
+  doc.setMeta('tags', { 'name': 'test place', 'alt_name': 'test pl' });
+
+  test('rejects - substring alt name', function(t) {
+    var stream = mapper();
+    stream.pipe( through.obj( function( doc, enc, next ){
+      t.deepEqual(doc.name, { default: 'test place' }, 'substring name removed');
+      t.end(); // test will fail if not called (or called twice).
+      next();
+    }));
+    stream.write(doc);
+  });
+};
+
+// Reject alt names that are a substring of the main name, even if they appear before the name in list of tags
+module.exports.tests.substring_alt_name2 = function(test, common) {
+  var doc = new Document('a','b',1);
+  doc.setMeta('tags', { 'alt_name': 'test pl', name: 'test place'});
+
+  test('rejects - substring alt name', function(t) {
+    var stream = mapper();
+    stream.pipe( through.obj( function( doc, enc, next ){
+      t.deepEqual(doc.name, { default: 'test place' }, 'substring name removed');
+      t.end(); // test will fail if not called (or called twice).
+      next();
+    }));
+    stream.write(doc);
+  });
+};
+
 // Cover the case of a tag key being 'name:' eg. { 'name:': 'foo' }
 // Not to be confused with { 'name': 'foo' } (note the extraneous colon)
 module.exports.tests.extraneous_colon = function(test, common) {