From 5b03f386fe6d1a94e46b30a6dd8ec685022dd0d6 Mon Sep 17 00:00:00 2001
From: ghost <localhost>
Date: Tue, 29 Aug 2023 15:38:32 +0300
Subject: [PATCH] implement semantic search by stopwords settings

---
 src/config/app.php.example |  9 +++++++++
 src/library/sphinx.php     | 27 ++++++++++++++++-----------
 src/public/magnet.php      |  5 +++--
 3 files changed, 28 insertions(+), 13 deletions(-)

diff --git a/src/config/app.php.example b/src/config/app.php.example
index c459053..d336e1b 100644
--- a/src/config/app.php.example
+++ b/src/config/app.php.example
@@ -83,6 +83,15 @@ define('MAGNET_EDITOR_LOCK_TIMEOUT', 60*60);
 define('MAGNET_META_TITLE_MIN_LENGTH', 10);
 define('MAGNET_META_DESCRIPTION_MIN_LENGTH', 0);
 
+define('MAGNET_STOP_WORDS_SIMILAR',
+  [
+    'series',
+    'season',
+    'discography',
+    // ...
+  ]
+);
+
 // Comment
 define('COMMENT_DEFAULT_APPROVED', false);
 define('COMMENT_DEFAULT_PUBLIC', false);
diff --git a/src/library/sphinx.php b/src/library/sphinx.php
index e36a8ef..b56eee0 100644
--- a/src/library/sphinx.php
+++ b/src/library/sphinx.php
@@ -11,20 +11,20 @@ class Sphinx {
     $this->_sphinx->setAttribute(PDO::ATTR_DEFAULT_FETCH_MODE, PDO::FETCH_OBJ);
   }
 
-  public function searchMagnetsTotal(string $keyword, string $mode = 'default') : int
+  public function searchMagnetsTotal(string $keyword, string $mode = 'default', array $stopWords = []) : int
   {
     $query = $this->_sphinx->prepare('SELECT COUNT(*) AS `total` FROM `magnet` WHERE MATCH(?)');
 
     $query->execute(
       [
-        self::_match($keyword, $mode)
+        self::_match($keyword, $mode, $stopWords)
       ]
     );
 
     return $query->fetch()->total;
   }
 
-  public function searchMagnets(string $keyword, int $start, int $limit, int $maxMatches, string $mode = 'default')
+  public function searchMagnets(string $keyword, int $start, int $limit, int $maxMatches, string $mode = 'default', array $stopWords = [])
   {
     $query = $this->_sphinx->prepare("SELECT *
 
@@ -40,14 +40,14 @@ class Sphinx {
 
     $query->execute(
       [
-        self::_match($keyword, $mode)
+        self::_match($keyword, $mode, $stopWords)
       ]
     );
 
     return $query->fetchAll();
   }
 
-  private static function _match(string $keyword, string $mode = 'default') : string
+  private static function _match(string $keyword, string $mode = 'default', array $stopWords = []) : string
   {
     $keyword = trim($keyword);
 
@@ -67,15 +67,17 @@ class Sphinx {
 
         $result = [];
 
-        foreach ((array) explode(' ', $keyword) as $i => $value)
+        $keyword = preg_replace('/[\d]/ui', ' ', $keyword);
+        $keyword = preg_replace('/[\s]+/ui', ' ', $keyword);
+        $keyword = trim($keyword);
+
+        foreach ((array) explode(' ', $keyword) as $value)
         {
           if (mb_strlen($value) > 5)
           {
-            $result[] = sprintf('@metaTitle "%s" | @dn "%s"', $value, $value);
-
-            if ($i > 3)
+            if (!in_array(mb_strtolower($value), $stopWords))
             {
-              break;
+              $result[] = sprintf('@metaTitle "%s" | @dn "%s"', $value, $value);
             }
           }
         }
@@ -97,7 +99,10 @@ class Sphinx {
 
         foreach ((array) explode(' ', $keyword) as $value)
         {
-          $result[] = sprintf('@"*%s*"', $value);
+          if (!in_array(mb_strtolower($value), $stopWords))
+          {
+            $result[] = sprintf('@"*%s*"', $value);
+          }
         }
 
         return implode(' | ', $result);
diff --git a/src/public/magnet.php b/src/public/magnet.php
index 9b0eb42..8655968 100644
--- a/src/public/magnet.php
+++ b/src/public/magnet.php
@@ -371,7 +371,7 @@ echo '<?xml version="1.0" encoding="UTF-8"?>' . PHP_EOL ?>
                     </span>
                   </div>
                 </div>
-                <?php if ($similarMagnetsTotal = $sphinx->searchMagnetsTotal($magnet->metaTitle ? $magnet->metaTitle : $magnet->dn, 'similar')) { ?>
+                <?php if ($similarMagnetsTotal = $sphinx->searchMagnetsTotal($magnet->metaTitle ? $magnet->metaTitle : $magnet->dn, 'similar', MAGNET_STOP_WORDS_SIMILAR)) { ?>
                   <?php if ($similarMagnetsTotal > 1) { // skip current magnet ?>
                     <div class="padding-y-8 padding-x-16">
                       <a name="similar"></a>
@@ -384,7 +384,8 @@ echo '<?xml version="1.0" encoding="UTF-8"?>' . PHP_EOL ?>
                                         0,
                                         10,
                                         $similarMagnetsTotal,
-                                        'similar'
+                                        'similar',
+                                        MAGNET_STOP_WORDS_SIMILAR
                                       ) as $result) { ?>
                           <?php if ($magnet = $db->getMagnet($result->magnetid)) { ?>
                             <?php if ($result->magnetid != $response->magnet->magnetId && // skip current magnet