implement russian detection

Author: faissaloux

src/Expectations/Profanity.php

@@ -3,6 +3,7 @@
 declare(strict_types=1);
 
 use JonPurvis\Profanify\Expectations\TargetedProfanity;
+use JonPurvis\Profanify\Support\RussianNormalizer;
 use Pest\Arch\Contracts\ArchExpectation;
 use Pest\Arch\Support\FileLineFinder;
 use PHPUnit\Architecture\Elements\ObjectDescription;
@@ -51,6 +52,13 @@ function (ObjectDescription $object) use (&$foundWords, $excluding, $including,
                 return true;
             }
 
+            // Check if russian
+            if (preg_match('/[А-Яа-яЁё]/u', $word)) {
+                $normalized = RussianNormalizer::normalizeRussianText($fileContents);
+
+                return str_contains($normalized, $word);
+            }
+
             preg_match_all('/[a-zA-Z]\w*/', $fileContents, $matches);
 
             foreach ($matches[0] as $token) {

src/Support/RussianNormalizer.php

@@ -19,50 +19,4 @@ public static function normalizeRussianText(string $text): string
 
         return preg_replace('/[^а-я]+/u', '', $text) ?: '';
     }
-
-    /**
-     * @return array<int, string>|null
-     */
-    public static function filterRussianProfanity(string $text): ?array
-    {
-        /** @var array<int, string> $profanities */
-        $profanities = include __DIR__.'/../Config/profanities/ru.php';
-
-        $normalized = self::normalizeRussianText($text);
-        $found = [];
-
-        foreach ($profanities as $bad) {
-            if ($bad !== '' && str_contains($normalized, (string) $bad)) {
-                $found[] = $bad;
-            }
-        }
-
-        return $found !== [] ? $found : null;
-    }
-
-    public static function assertNoRussianProfanity(string $filePath): void
-    {
-        $lines = file($filePath, FILE_IGNORE_NEW_LINES) ?: [];
-        /** @var array<int, string> $badWords */
-        $badWords = include __DIR__.'/../Config/profanities/ru.php';
-
-        $offended = [];
-
-        foreach ($lines as $num => $line) {
-            $norm = self::normalizeRussianText($line);
-
-            foreach ($badWords as $bad) {
-                if ($bad !== '' && str_contains($norm, (string) $bad)) {
-                    $offended[] = $num + 1;
-                    break;
-                }
-            }
-        }
-
-        if ($offended !== []) {
-            throw new \Exception(
-                sprintf('Profanity in %s, lines: %s', $filePath, implode(', ', $offended)),
-            );
-        }
-    }
 }

tests/toHaveProfanity.php

@@ -24,6 +24,10 @@
         ->toHaveNoProfanity();
 })->throws(ArchExpectationFailedException::class);
 
+it('fails if a file contains russian profanity', function () {
+    expect('Tests\Fixtures\HasExplicitRussianProfanity')->toHaveNoProfanity(language: 'ru');
+})->throws(ArchExpectationFailedException::class);
+
 it('fails if file contains profanity manually included', function () {
     expect('Tests\Fixtures\HasUncoveredProfanity')
         ->toHaveNoProfanity(including: ['dagnabbit']);