From 2f2e353ecef52639a78cac3bc407ccfe64197ac9 Mon Sep 17 00:00:00 2001 From: Mario Date: Sat, 7 Oct 2023 16:00:34 +0000 Subject: use new lang detect library which supports much more languages --- .../src/LanguageDetection/Trainer.php | 50 ++++++++++++++++++++++ 1 file changed, 50 insertions(+) create mode 100644 vendor/patrickschur/language-detection/src/LanguageDetection/Trainer.php (limited to 'vendor/patrickschur/language-detection/src/LanguageDetection/Trainer.php') diff --git a/vendor/patrickschur/language-detection/src/LanguageDetection/Trainer.php b/vendor/patrickschur/language-detection/src/LanguageDetection/Trainer.php new file mode 100644 index 000000000..2bc5e6761 --- /dev/null +++ b/vendor/patrickschur/language-detection/src/LanguageDetection/Trainer.php @@ -0,0 +1,50 @@ + + * @package LanguageDetection + */ +class Trainer extends NgramParser +{ + /** + * Generates language profiles for all language files + * + * @param string $dirname Name of the directory where the translations files are located + * @return void + */ + public function learn(string $dirname = '') + { + if (empty($dirname)) + { + $dirname = __DIR__ . '/../../resources/*/*.txt'; + } + else if (!\is_dir($dirname) || !\is_readable($dirname)) + { + throw new \InvalidArgumentException('Provided directory could not be found or is not readable'); + } + else + { + $dirname = \rtrim($dirname, '/'); + $dirname .= '/*/*.txt'; + } + + /** @var \GlobIterator $txt */ + foreach (new \GlobIterator($dirname) as $txt) + { + $content = \mb_strtolower(\file_get_contents($txt->getPathname())); + + \file_put_contents( + \substr_replace($txt->getPathname(), 'php', -3), + \sprintf("getBasename('.txt') => $this->getNgrams($content) ], true)) + ); + } + } +} -- cgit v1.2.3