From ca92db88264fc19e7fae58b447ae2a5306437688 Mon Sep 17 00:00:00 2001
From: ghost <localhost>
Date: Sat, 28 Oct 2023 01:00:00 +0300
Subject: [PATCH] add transliteration support in search requests #13

---
 README.md              |  1 +
 composer.json          |  3 +-
 src/library/filter.php | 62 +++++++++++++++++++++++++++++++++++++++++-
 3 files changed, 64 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 474b4ad..f172535 100644
--- a/README.md
+++ b/README.md
@@ -282,6 +282,7 @@ See also: [SQLite tree](https://github.com/YGGverse/YGGo/tree/sqliteway)
 * Engine sources [MIT License](https://github.com/YGGverse/YGGo/blob/main/LICENSE)
 * Home page animation by [alvarotrigo](https://codepen.io/alvarotrigo/pen/GRvYNax)
 * CLI logo by [patorjk.com](https://patorjk.com/software/taag/#p=display&f=Slant&t=YGGo!)
+* Transliteration by [php-translit](https://github.com/ashtokalo/php-translit)
 * Identicons by [jdenticon](https://github.com/dmester/jdenticon-php)
 
 #### Feedback
diff --git a/composer.json b/composer.json
index a1b4111..af3b2e8 100644
--- a/composer.json
+++ b/composer.json
@@ -8,7 +8,8 @@
         "yggverse/cache": ">=0.3.0",
         "symfony/dom-crawler": "^6.3",
         "symfony/css-selector": "^6.3",
-        "jdenticon/jdenticon": "^1.0"
+        "jdenticon/jdenticon": "^1.0",
+        "ashtokalo/php-translit": "^0.2.0"
     },
     "license": "MIT",
     "autoload": {
diff --git a/src/library/filter.php b/src/library/filter.php
index d066b80..fcee673 100644
--- a/src/library/filter.php
+++ b/src/library/filter.php
@@ -113,7 +113,12 @@ class Filter {
 
           if (mb_strlen($word) > 1) {
 
-            $words[] = sprintf('(%s*)', $word);
+            $words[] = sprintf('%s*', $word);
+
+            foreach (self::_generateWordForms($word) as $wordForm)
+            {
+              $words[] = sprintf('(*%s*)', $wordForm);
+            }
           }
         }
 
@@ -132,4 +137,59 @@ class Filter {
 
     return $texts[(($number % 100) > 4 && ($number % 100) < 20) ? 2 : $cases[min($number % 10, 5)]];
   }
+
+  // Word forms generator to improve search results
+  // e.g. transliteration rules for latin filenames
+  private static function _generateWordForms(
+    string $keyword,
+    // #13 supported locales:
+    // https://github.com/ashtokalo/php-translit
+    array  $transliteration = [
+      'be',
+      'bg',
+      'el',
+      'hy',
+      'kk',
+      'mk',
+      'ru',
+      'ka',
+      'uk'
+    ],
+    // Additional char forms
+    array $charForms =
+    [
+      'c' => 'k',
+      'k' => 'c',
+    ]
+  ): array
+  {
+    $wordForms = [];
+
+    // Apply transliteration
+    foreach ($transliteration as $locale)
+    {
+      $wordForms[] = \ashtokalo\translit\Translit::object()->convert(
+        $keyword,
+        $locale
+      );
+    }
+
+    // Apply char forms
+    foreach ($wordForms as $wordForm)
+    {
+      foreach ($charForms as $from => $to)
+      {
+        $wordForms[] = str_replace(
+          $from,
+          $to,
+          $wordForm
+        );
+      }
+    }
+
+    // Remove duplicates
+    return array_unique(
+      $wordForms
+    );
+  }
 }
\ No newline at end of file