Using the php-dfa-sensitive Library for Sensitive Word Detection in PHP Projects

The article introduces a PHP extension called php-dfa-sensitive , which provides DFA‑based sensitive‑word filtering capabilities for applications that need to validate user signatures or messages.

Installation is performed via Composer:

composer require lustre/php-dfa-sensitive

After installing, a new service class SensitiveWords should be placed under the app/Services directory. The class encapsulates a singleton instance of DfaFilter\SensitiveHelper and loads default dictionaries located in storage/dict (e.g., bk.txt , fd.txt , ms.txt , etc.).

<?php
namespace App\Services;
use DfaFilter\SensitiveHelper;

class SensitiveWords {
    protected static $handle = null;
    private function __construct() {}
    private function __clone() {}
    public static function getInstance($word_path = []) {
        if (!self::$handle) {
            $default_path = [
                storage_path('dict/bk.txt'),
                storage_path('dict/fd.txt'),
                storage_path('dict/ms.txt'),
                storage_path('dict/qt.txt'),
                storage_path('dict/sq.txt'),
                storage_path('dict/tf.txt'),
            ];
            $paths = array_merge($default_path, $word_path);
            self::$handle = SensitiveHelper::init();
            if (!empty($paths)) {
                foreach ($paths as $path) {
                    self::$handle->setTreeByFile($path);
                }
            }
        }
        return self::$handle;
    }
    public static function isLegal($content) {
        return self::getInstance()->islegal($content);
    }
    public static function replace($content, $replace_char = '', $repeat = false, $match_type = 1) {
        return self::getInstance()->replace($content, $replace_char, $repeat, $match_type);
    }
    public static function mark($content, $start_tag, $end_tag, $match_type = 1) {
        return self::getInstance()->mark($content, $start_tag, $end_tag, $match_type);
    }
    public static function getBadWord($content, $match_type = 1, $word_num = 0) {
        return self::getInstance()->getBadWord($content, $match_type, $word_num);
    }
}

In application code you can now call the static methods, for example to retrieve any illegal words:

$bad_word = SensitiveWords::getBadWord($content);
if (!empty($bad_word)) {
    throw new \Exception('包含敏感词:' . current($bad_word));
}

Place your custom word lists in storage/dict and add them to the array passed to getInstance if needed. The article concludes with a reminder that the content is sourced from the internet.