Fix pattern normalization form.

oldnomad · Oct 14, 2023 · 89e8103 · 89e8103
1 parent 672b7b4
commit 89e8103
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 4 deletions.
diff --git a/lib/Calibre/CalibreSearch.php b/lib/Calibre/CalibreSearch.php
@@ -16,6 +16,11 @@
  * Search implementation.
  */
 class CalibreSearch {
+	/**
+	 * Default Unicode form used for search.
+	 */
+	private const DEFAULT_FORM = Normalizer::NFKC;
+
 	/**
 	 * Construct an instance.
 	 *
@@ -33,7 +38,7 @@ private function __construct(private string $pattern) {
 	private static function removeDiacritics(string $text): string {
 		$text = normalizer_normalize($text, Normalizer::NFD);
 		$text = preg_replace('/[\p{M}]/u', '', $text);
-		return normalizer_normalize($text, Normalizer::NFKC);
+		return normalizer_normalize($text, self::DEFAULT_FORM);
 	}
 
 	/**
@@ -48,7 +53,7 @@ private static function appendHaystack(array &$haystack, mixed $text): void {
 		if (!is_string($text) || $text === '') {
 			return;
 		}
-		$text = normalizer_normalize($text, Normalizer::NFKC);
+		$text = normalizer_normalize($text, self::DEFAULT_FORM);
 		array_push($haystack, $text);
 		$textNoMarks = self::removeDiacritics($text);
 		if ($textNoMarks !== $text) {
@@ -94,9 +99,8 @@ private static function createPattern(string $terms): ?string {
 		if ($terms === '') {
 			return null;
 		}
-		/* @var string */
 		$dataEsc = str_replace('/', '\/', $terms);
-		return '/'.$dataEsc.'/inS';
+		return '/'.normalizer_normalize($dataEsc, self::DEFAULT_FORM).'/inS';
 	}
 
 	/**

diff --git a/tests/unit/SearchTest.php b/tests/unit/SearchTest.php
@@ -53,6 +53,16 @@ public function testSimple(): void {
 			...self::BOOK_DEFAULTS,
 			'title' => 'This is a Símple Tést',
 		])), 'Simple search in title -- diacritics-insensitive');
+		$filter = CalibreSearch::searchBooks(normalizer_normalize('símple', Normalizer::NFC));
+		$this->assertTrue(call_user_func($filter, $this->createCalibreItem(CalibreBook::class, $this->db, [
+			...self::BOOK_DEFAULTS,
+			'title' => 'This is a Símple Tést',
+		])), 'Simple search in title -- diacritics-insensitive (NFC)');
+		$filter = CalibreSearch::searchBooks(normalizer_normalize('símple', Normalizer::NFD));
+		$this->assertTrue(call_user_func($filter, $this->createCalibreItem(CalibreBook::class, $this->db, [
+			...self::BOOK_DEFAULTS,
+			'title' => 'This is a Símple Tést',
+		])), 'Simple search in title -- diacritics-insensitive (NFD)');
 	}
 
 	public function testRegex(): void {