From 1a4bcf875570ddd824bdfe12c9684eb51c575340 Mon Sep 17 00:00:00 2001
From: Simon Vieille <simon.vieille@trinaps.com>
Date: Sun, 24 Sep 2023 19:35:27 +0200
Subject: [PATCH] upgrade search engine

---
 src/Repository/Blog/PostRepositoryQuery.php | 38 +++++++++++++--------
 1 file changed, 24 insertions(+), 14 deletions(-)

diff --git a/src/Repository/Blog/PostRepositoryQuery.php b/src/Repository/Blog/PostRepositoryQuery.php
index 5f3d3b3..6b3e04e 100644
--- a/src/Repository/Blog/PostRepositoryQuery.php
+++ b/src/Repository/Blog/PostRepositoryQuery.php
@@ -54,19 +54,12 @@ class PostRepositoryQuery extends RepositoryQuery
         ;
     }
 
-    protected function filterHandler(string $name, $value)
-    {
-        if ('category' === $name) {
-            $this->inCategory($value);
-        }
-    }
-
     public function search(?string $keywords, ?string $tag)
     {
         $keywords = explode(' ', $keywords);
 
-        $filterWords = fn($keyword) => trim($keyword) !== '' && preg_match('/[a-zA-Z]+/', $keyword) && mb_strlen($keyword) > 2;
-        $filter = fn($keyword) => trim($keyword) !== '' && preg_match('/[a-zA-Z]+/', $keyword) && mb_strlen($keyword) > 2;
+        $filterWords = fn ($keyword) => '' !== trim($keyword) && preg_match('/[a-zA-Z]+/', $keyword);
+        $filter = fn ($keyword) => '' !== trim($keyword) && preg_match('/[a-zA-Z]+/', $keyword);
 
         $keywords = array_filter($keywords, $filter);
 
@@ -83,7 +76,8 @@ class PostRepositoryQuery extends RepositoryQuery
                 WHERE
                     post.status = 1 AND
                     post.published_at < :date
-            ');
+            '
+            );
 
             $statement = $query->execute([
                 ':date' => (new \DateTime())->format('Y-m-d H:i:s'),
@@ -104,17 +98,19 @@ class PostRepositoryQuery extends RepositoryQuery
                 $words = array_filter($words, $filterWords);
 
                 foreach ($keywords as $keyword) {
-                    if(str_contains(mb_strtolower($v['content']), mb_strtolower($keyword))) {
+                    if (str_contains(mb_strtolower($v['content']), mb_strtolower($keyword))) {
                         $similarity = 99;
 
                         if (isset($matches[$v['id']])) {
                             $matches[$v['id']]['similarity'] += $similarity;
+                            ++$matches[$v['id']]['count'];
                         } else {
                             $matches[$v['id']] = [
                                 'id' => $v['id'],
                                 'title' => $v['title'],
                                 'published_at' => $v['published_at'],
                                 'similarity' => $similarity,
+                                'count' => 1,
                             ];
                         }
                     }
@@ -125,16 +121,18 @@ class PostRepositoryQuery extends RepositoryQuery
 
                             if (isset($matches[$v['id']])) {
                                 $matches[$v['id']]['similarity'] += $similarity;
+                                ++$matches[$v['id']]['count'];
                             } else {
                                 $matches[$v['id']] = [
                                     'id' => $v['id'],
                                     'title' => $v['title'],
                                     'published_at' => $v['published_at'],
                                     'similarity' => $similarity,
+                                    'count' => 1,
                                 ];
                             }
                         } else {
-                            $lev  = levenshtein($word, $keyword);
+                            $lev = levenshtein($word, $keyword);
                             $similarity = 100 - ($lev * 100 / mb_strlen($word));
 
                             if ($similarity > 70) {
@@ -146,6 +144,7 @@ class PostRepositoryQuery extends RepositoryQuery
                                         'title' => $v['title'],
                                         'published_at' => $v['published_at'],
                                         'similarity' => $similarity,
+                                        'count' => 1,
                                     ];
                                 }
                             }
@@ -154,7 +153,11 @@ class PostRepositoryQuery extends RepositoryQuery
                 }
             }
 
-            usort($matches, function($a, $b) {
+            $matches = array_filter($matches, function($match) use ($keywords) {
+                return (100 * $match['count'] / count($keywords)) > 80;
+            });
+
+            usort($matches, function ($a, $b) {
                 if ($a['similarity'] > $b['similarity']) {
                     return -1;
                 }
@@ -163,7 +166,7 @@ class PostRepositoryQuery extends RepositoryQuery
                     return 1;
                 }
 
-                return ($a['published_at'] <> $b['published_at']) * -1;
+                return ($a['published_at'] != $b['published_at']) * -1;
             });
 
             $ids = array_column($matches, 'id');
@@ -188,4 +191,11 @@ class PostRepositoryQuery extends RepositoryQuery
 
         return $this;
     }
+
+    protected function filterHandler(string $name, $value)
+    {
+        if ('category' === $name) {
+            $this->inCategory($value);
+        }
+    }
 }