From 0e9068627661d7bb372c8840699ac2b2592426b5 Mon Sep 17 00:00:00 2001
From: Istvan Soos <istvan.soos@gmail.com>
Date: Wed, 2 Apr 2025 10:53:15 +0200
Subject: [PATCH] Slightly lower score for depluralized matches.

---
 app/lib/search/mem_index.dart                | 41 +++++++++++++++-----
 app/test/search/maps_test.dart               |  2 +-
 app/test/search/mem_index_test.dart          |  2 +-
 app/test/search/package_name_index_test.dart | 16 ++++++++
 4 files changed, 50 insertions(+), 11 deletions(-)

diff --git a/app/lib/search/mem_index.dart b/app/lib/search/mem_index.dart
index 364dd3eec5..1979272882 100644
--- a/app/lib/search/mem_index.dart
+++ b/app/lib/search/mem_index.dart
@@ -619,11 +619,20 @@ class PackageNameIndex {
     IndexedScore<String>? filterOnNonZeros,
   }) {
     assert(score.keys.length == _packageNames.length);
-    final singularWord = word.length <= 3 || !word.endsWith('s')
-        ? word
-        : word.substring(0, word.length - 1);
-    final lowercasedWord = singularWord.toLowerCase();
+    final lowercasedWord = word.toLowerCase();
     final collapsedWord = _removeUnderscores(lowercasedWord);
+
+    // Note: This is a very simple plurality check, which may not work wor many
+    //       cases, but provided a simple approximation.
+    //       The check should be done on the lowercased input value.
+    final lowercaseDepluralized =
+        lowercasedWord.length <= 3 || !lowercasedWord.endsWith('s')
+            ? null
+            : lowercasedWord.substring(0, lowercasedWord.length - 1);
+    final collapsedDepluralized = lowercaseDepluralized == null
+        ? null
+        : _removeUnderscores(lowercaseDepluralized);
+
     final parts =
         collapsedWord.length <= 3 ? [collapsedWord] : trigrams(collapsedWord);
     for (var i = 0; i < _data.length; i++) {
@@ -632,18 +641,24 @@ class PackageNameIndex {
       }
 
       final entry = _data[i];
-      if (entry.collapsed.length >= collapsedWord.length &&
-          entry.collapsed.contains(collapsedWord)) {
-        // also check for non-collapsed match
-        if (entry.lowercased.length >= lowercasedWord.length &&
-            entry.lowercased.contains(lowercasedWord)) {
+      // Check for direct substring match.
+      // TODO: Consider using trie or other substring index here.
+      if (entry._containsCollapsed(collapsedDepluralized ?? collapsedWord)) {
+        // most score for original non-collapsed match
+        if (entry._containsLowercased(lowercasedWord)) {
           score.setValue(i, 1.0);
           continue;
         }
 
+        // otherwise a slightly lower score for:
+        // - collapsed-only original match
+        // - non-collapsed depluralized match
+        // - collapsed depluralized match
         score.setValue(i, 0.99);
+
         continue;
       }
+
       var matched = 0;
       var unmatched = 0;
       final acceptThreshold = parts.length ~/ 2;
@@ -682,6 +697,14 @@ class _PkgNameData {
   final Set<String> trigrams;
 
   _PkgNameData(this.lowercased, this.collapsed, this.trigrams);
+
+  bool _containsLowercased(String value) {
+    return lowercased.length >= value.length && lowercased.contains(value);
+  }
+
+  bool _containsCollapsed(String value) {
+    return collapsed.length >= value.length && collapsed.contains(value);
+  }
 }
 
 extension on List<IndexedPackageHit> {
diff --git a/app/test/search/maps_test.dart b/app/test/search/maps_test.dart
index af69e3fcc1..5ebbb9e66c 100644
--- a/app/test/search/maps_test.dart
+++ b/app/test/search/maps_test.dart
@@ -30,7 +30,7 @@ void main() {
         'sdkLibraryHits': [],
         'packageHits': [
           {'package': 'maps', 'score': 1.0},
-          {'package': 'map', 'score': 1.0},
+          {'package': 'map', 'score': 0.99},
         ],
       });
     });
diff --git a/app/test/search/mem_index_test.dart b/app/test/search/mem_index_test.dart
index 1e751e7c0b..71119f1e93 100644
--- a/app/test/search/mem_index_test.dart
+++ b/app/test/search/mem_index_test.dart
@@ -562,8 +562,8 @@ server.dart adds a small, prescriptive server (PicoServer) that can be configure
       final match2 = index.search(
           ServiceSearchQuery.parse(query: 'apps', order: SearchOrder.text));
       expect(match2.packageHits.map((e) => e.toJson()), [
-        {'package': 'app', 'score': 1.0},
         {'package': 'apps', 'score': 1.0},
+        {'package': 'app', 'score': 0.99},
       ]);
     });
 
diff --git a/app/test/search/package_name_index_test.dart b/app/test/search/package_name_index_test.dart
index 2ec47ed645..198fb88def 100644
--- a/app/test/search/package_name_index_test.dart
+++ b/app/test/search/package_name_index_test.dart
@@ -127,4 +127,20 @@ void main() {
       });
     });
   });
+
+  group('redis', () {
+    final index = PackageNameIndex([
+      'redis',
+      'x_redis_client',
+      'credit_union',
+    ]);
+
+    test('redis', () {
+      expect(index.search('redis'), {
+        'redis': 1.0,
+        'x_redis_client': 1.0,
+        'credit_union': 0.99,
+      });
+    });
+  });
 }