Merge pull request #24 from ginkelsoft-development/feature/minimum-prefix-length

add configurable minimum prefix length for search queries

Author: ginkelsoft-development

config/encrypted-search.php

@@ -36,7 +36,7 @@
     |--------------------------------------------------------------------------
     |
     | The maximum number of prefix levels to generate for prefix-based search.
-    | For example, the term “wietse” would generate:
+    | For example, the term "wietse" would generate:
     |   ["w", "wi", "wie", "wiet", "wiets", "wietse"]
     |
     | Increasing this value improves search precision for short terms, but
@@ -45,6 +45,29 @@
     */
     'max_prefix_depth' => 6,
 
+    /*
+    |--------------------------------------------------------------------------
+    | Minimum Prefix Length
+    |--------------------------------------------------------------------------
+    |
+    | The minimum number of characters required for prefix-based searches.
+    | This prevents overly broad matches from very short search terms.
+    |
+    | For example, with min_prefix_length = 3:
+    | - Searching for "Wi" (2 chars) will return no results
+    | - Searching for "Wil" (3 chars) will work normally
+    |
+    | This helps prevent performance issues and reduces false positives
+    | when users search for very short terms like "a" or "de".
+    |
+    | Recommended values:
+    | - 2: Allow two-character searches (more flexible, more false positives)
+    | - 3: Require three characters (good balance)
+    | - 4: Require four characters (very precise, less flexible)
+    |
+    */
+    'min_prefix_length' => env('ENCRYPTED_SEARCH_MIN_PREFIX', 3),
+
     /*
     |--------------------------------------------------------------------------
     | Automatic Indexing of Encrypted Casts

src/Support/Tokens.php

@@ -69,21 +69,27 @@ public static function exact(string $normalized, string $pepper): string
      * These prefix hashes can be used to implement fast "starts-with"
      * queries while maintaining cryptographic privacy.
      *
-     * Example: "alex" with maxDepth=3 yields tokens for "a", "al", "ale".
+     * Only prefixes at or above the minimum length (from config) are generated.
+     * This prevents overly broad matches from very short search terms.
+     *
+     * Example: "alex" with maxDepth=4, minLength=2 yields tokens for "al", "ale", "alex".
+     * (skips "a" because it's below minimum length)
      *
      * @param string $normalized
      *     The normalized (lowercase, diacritic-free) string.
      * @param int $maxDepth
      *     The maximum number of prefix characters to hash.
      * @param string $pepper
      *     A secret application-level random string from configuration.
+     * @param int $minLength
+     *     The minimum prefix length to generate (default: 1 for backwards compatibility).
      *
      * @return string[]
      *     An array of hex-encoded SHA-256 prefix tokens.
      *
      * @throws \RuntimeException if pepper is empty
      */
-    public static function prefixes(string $normalized, int $maxDepth, string $pepper): array
+    public static function prefixes(string $normalized, int $maxDepth, string $pepper, int $minLength = 1): array
     {
         if (empty($pepper)) {
             throw new \RuntimeException(
@@ -96,7 +102,10 @@ public static function prefixes(string $normalized, int $maxDepth, string $peppe
         $len = mb_strlen($normalized, 'UTF-8');
         $depth = min($maxDepth, $len);
 
-        for ($i = 1; $i <= $depth; $i++) {
+        // Start from minimum length instead of 1
+        $start = max(1, $minLength);
+
+        for ($i = $start; $i <= $depth; $i++) {
             $prefix = mb_substr($normalized, 0, $i, 'UTF-8');
             $out[] = hash('sha256', $prefix . $pepper);
         }

src/Traits/HasEncryptedSearchIndex.php

@@ -73,6 +73,7 @@ public function updateSearchIndex(): void
 
         $pepper = (string) config('encrypted-search.search_pepper', '');
         $max = (int) config('encrypted-search.max_prefix_depth', 6);
+        $min = (int) config('encrypted-search.min_prefix_length', 1);
         $useElastic = config('encrypted-search.elasticsearch.enabled', false);
 
         $rows = [];
@@ -108,7 +109,7 @@ public function updateSearchIndex(): void
 
             // Generate prefix-based tokens
             if (!empty($modes['prefix'])) {
-                foreach (Tokens::prefixes($normalized, $max, $pepper) as $token) {
+                foreach (Tokens::prefixes($normalized, $max, $pepper, $min) as $token) {
                     $rows[] = [
                         'model_type' => static::class,
                         'model_id'   => $this->getKey(),
@@ -275,18 +276,30 @@ public function scopeEncryptedExact(Builder $query, string $field, string $term)
     public function scopeEncryptedPrefix(Builder $query, string $field, string $term): Builder
     {
         $pepper = (string) config('encrypted-search.search_pepper', '');
+        $minLength = (int) config('encrypted-search.min_prefix_length', 1);
         $normalized = Normalizer::normalize($term);
 
         if (!$normalized) {
             return $query->whereRaw('1=0');
         }
 
+        // Check if search term meets minimum length requirement
+        if (mb_strlen($normalized, 'UTF-8') < $minLength) {
+            return $query->whereRaw('1=0');
+        }
+
         $tokens = Tokens::prefixes(
             $normalized,
             (int) config('encrypted-search.max_prefix_depth', 6),
-            $pepper
+            $pepper,
+            $minLength
         );
 
+        // If no tokens generated (term too short), return no results
+        if (empty($tokens)) {
+            return $query->whereRaw('1=0');
+        }
+
         // Check if Elasticsearch is enabled
         if (config('encrypted-search.elasticsearch.enabled', false)) {
             $modelIds = $this->searchElasticsearch($field, $tokens, 'prefix');

tests/Feature/EncryptedSearchIntegrationTest.php

@@ -67,6 +67,9 @@ protected function setUp(): void
         // Disable Elasticsearch during tests (we test DB index)
         config()->set('encrypted-search.elasticsearch.enabled', false);
 
+        // Set minimum prefix length to 1 for backwards compatibility in basic tests
+        config()->set('encrypted-search.min_prefix_length', 1);
+
         // Ensure Eloquent events are active (boot model & dispatcher)
         \Illuminate\Database\Eloquent\Model::unsetEventDispatcher();
         \Illuminate\Database\Eloquent\Model::setEventDispatcher(app('events'));

tests/Feature/HasEncryptedSearchIndexEdgeCasesTest.php

@@ -48,6 +48,7 @@ protected function setUp(): void
 
         config()->set('encrypted-search.elasticsearch.enabled', false);
         config()->set('encrypted-search.search_pepper', 'test-pepper-secret');
+        config()->set('encrypted-search.min_prefix_length', 1);
 
         \Illuminate\Database\Eloquent\Model::unsetEventDispatcher();
         \Illuminate\Database\Eloquent\Model::setEventDispatcher(app('events'));