From 2f2e353ecef52639a78cac3bc407ccfe64197ac9 Mon Sep 17 00:00:00 2001
From: Mario <mario@mariovavti.com>
Date: Sat, 7 Oct 2023 16:00:34 +0000
Subject: use new lang detect library which supports much more languages

---
 .../src/LanguageDetection/Language.php             | 102 +++++++++++++++++++++
 1 file changed, 102 insertions(+)
 create mode 100644 vendor/patrickschur/language-detection/src/LanguageDetection/Language.php

(limited to 'vendor/patrickschur/language-detection/src/LanguageDetection/Language.php')

diff --git a/vendor/patrickschur/language-detection/src/LanguageDetection/Language.php b/vendor/patrickschur/language-detection/src/LanguageDetection/Language.php
new file mode 100644
index 000000000..c369a3367
--- /dev/null
+++ b/vendor/patrickschur/language-detection/src/LanguageDetection/Language.php
@@ -0,0 +1,102 @@
+<?php
+
+declare(strict_types = 1);
+
+namespace LanguageDetection;
+
+/**
+ * Class Language
+ *
+ * @copyright Patrick Schur
+ * @license https://opensource.org/licenses/mit-license.html MIT
+ * @author Patrick Schur <patrick_schur@outlook.de>
+ * @package LanguageDetection
+ */
+class Language extends NgramParser
+{
+    /**
+     * @var array<string, array<string, int>>
+     */
+    protected $tokens = [];
+
+    /**
+     * Loads all language files
+     *
+     * @param array $lang List of ISO 639-1 codes, that should be used in the detection phase
+     * @param string $dirname Name of the directory where the translations files are located
+     */
+    public function __construct(array $lang = [], string $dirname = '')
+    {
+        if (empty($dirname))
+        {
+            $dirname = __DIR__ . '/../../resources/*/*.php';
+        }
+        else if (!\is_dir($dirname) || !\is_readable($dirname))
+        {
+            throw new \InvalidArgumentException('Provided directory could not be found or is not readable');
+        }
+        else
+        {
+            $dirname = \rtrim($dirname, '/');
+            $dirname .= '/*/*.php';
+        }
+
+        $isEmpty = empty($lang);
+        $tokens = [];
+
+        foreach (\glob($dirname) as $file)
+        {
+            if ($isEmpty || \in_array(\basename($file, '.php'), $lang))
+            {
+                $tokens += require $file;
+            }
+        }
+
+        foreach ($tokens as $lang => $value) {
+            $this->tokens[$lang] = \array_flip($value);
+        }
+    }
+
+    /**
+     * Detects the language from a given text string
+     *
+     * @param string $str
+     * @return LanguageResult
+     */
+    public function detect(string $str): LanguageResult
+    {
+        $str = \mb_strtolower($str);
+
+        $samples = $this->getNgrams($str);
+
+        $result = [];
+
+        if (\count($samples) > 0)
+        {
+            foreach ($this->tokens as $lang => $value)
+            {
+                $index = $sum = 0;
+
+                foreach ($samples as $v)
+                {
+                    if (isset($value[$v]))
+                    {
+                        $x = $index++ - $value[$v];
+                        $y = $x >> (PHP_INT_SIZE * 8);
+                        $sum += ($x + $y) ^ $y;
+                        continue;
+                    }
+
+                    $sum += $this->maxNgrams;
+                    ++$index;
+                }
+
+                $result[$lang] = 1 - ($sum / ($this->maxNgrams * $index));
+            }
+
+            \arsort($result, SORT_NUMERIC);
+        }
+
+        return new LanguageResult($result);
+    }
+}
-- 
cgit v1.2.3