Request: added Helpers::parseQualityList() and rewrote detectLanguage() on top of it

dg · dg · commit 80fe2a59177d · 2026-06-02T21:42:51.000+02:00
parseQualityList() is a generic parser for HTTP quality-value headers
(Accept, Accept-Language, Accept-Encoding, ...). detectLanguage() now
matches case-insensitively and returns the language code in the caller's
original casing.
diff --git a/src/Http/Helpers.php b/src/Http/Helpers.php
@@ -9,6 +9,7 @@
 
 use Nette;
 use Nette\Utils\DateTime;
+use function array_shift, arsort, explode, preg_match, strtolower, trim;
 
 
 /**
@@ -35,6 +36,39 @@ public static function formatDate(string|int|\DateTimeInterface $time): string
 	}
 
 
+	/**
+	 * Parses an HTTP quality-value list such as the Accept, Accept-Language or Accept-Encoding header
+	 * into tokens mapped to their q-factor, ordered by descending preference. Tokens are lowercased and
+	 * those explicitly rejected with q=0 are omitted.
+	 * @return array<string, float>  e.g. ['cs-cz' => 1.0, 'en' => 0.8]
+	 */
+	public static function parseQualityList(string $header): array
+	{
+		$list = [];
+		foreach (explode(',', $header) as $item) {
+			$params = explode(';', $item);
+			$token = strtolower(trim((string) array_shift($params)));
+			if ($token === '') {
+				continue;
+			}
+
+			$q = 1.0;
+			foreach ($params as $param) {
+				if (preg_match('#^\s*q\s*=\s*([0-9.]+)#i', $param, $m)) {
+					$q = min(1.0, (float) $m[1]); // q is capped at 1 per RFC 9110
+				}
+			}
+
+			if ($q > 0) {
+				$list[$token] = max($list[$token] ?? 0.0, $q); // a repeated token keeps its highest q
+			}
+		}
+
+		arsort($list); // stable since PHP 8.0, so equal q keeps header order
+		return $list;
+	}
+
+
 	/**
 	 * Checks whether an IP address falls within a CIDR block (e.g. '192.168.1.0/24').
 	 */
diff --git a/src/Http/Request.php b/src/Http/Request.php
@@ -8,7 +8,7 @@
 namespace Nette\Http;
 
 use Nette;
-use function array_change_key_case, base64_decode, count, explode, func_num_args, implode, preg_match, preg_match_all, rsort, strcasecmp, strtr;
+use function array_change_key_case, array_keys, base64_decode, count, explode, func_num_args, preg_match, str_starts_with, strcasecmp, strlen, strtolower, strtr, usort;
 
 
 /**
@@ -304,25 +304,18 @@ public function detectLanguage(array $langs): ?string
 			return null;
 		}
 
-		$s = strtolower($header);  // case insensitive
-		$s = strtr($s, '_', '-');  // cs_CZ means cs-CZ
-		rsort($langs);             // first more specific
-		preg_match_all('#(' . implode('|', $langs) . ')(?:-[^\s,;=]+)?\s*(?:;\s*q=([0-9.]+))?#', $s, $matches);
+		usort($langs, fn($a, $b) => strlen($b) <=> strlen($a)); // more specific first
+		$accepted = Helpers::parseQualityList(strtr($header, '_', '-')); // cs_CZ means cs-CZ
 
-		if (!$matches[0]) {
-			return null;
-		}
-
-		$max = 0;
-		$lang = null;
-		foreach ($matches[1] as $key => $value) {
-			$q = $matches[2][$key] === '' ? 1.0 : (float) $matches[2][$key];
-			if ($q > $max) {
-				$max = $q;
-				$lang = $value;
+		foreach (array_keys($accepted) as $token) {
+			foreach ($langs as $lang) {
+				$l = strtolower($lang);
+				if ($token === '*' || $token === $l || str_starts_with($token, $l . '-')) {
+					return $lang;
+				}
 			}
 		}
 
-		return $lang;
+		return null;
 	}
 }
diff --git a/tests/Http/Helpers.phpt b/tests/Http/Helpers.phpt
@@ -38,6 +38,33 @@ test('IPv6 address matching', function () {
 
 
 
+test('parseQualityList', function () {
+	Assert::same([], Helpers::parseQualityList(''));
+	Assert::same(['gzip' => 1.0], Helpers::parseQualityList('gzip'));
+
+	// ordered by descending q-factor, default q is 1.0
+	Assert::same(
+		['da' => 1.0, 'en-gb' => 0.8, 'en' => 0.7],
+		Helpers::parseQualityList('da, en-gb;q=0.8, en;q=0.7'),
+	);
+
+	// equal q keeps the header order (stable sort)
+	Assert::same(['en' => 1.0, 'cs' => 1.0], Helpers::parseQualityList('en, cs'));
+
+	// tokens are lowercased and trimmed, q=0 is dropped
+	Assert::same(
+		['text/html' => 0.9, '*/*' => 0.8],
+		Helpers::parseQualityList('TEXT/HTML ; q=0.9 , identity;q=0 , */* ;q=0.8'),
+	);
+
+	// a repeated token keeps its highest q
+	Assert::same(['de' => 0.9], Helpers::parseQualityList('de;q=0.9, de;q=0.1'));
+
+	// q is capped at 1
+	Assert::same(['a' => 1.0, 'b' => 0.9], Helpers::parseQualityList('a;q=5, b;q=0.9'));
+});
+
+
 test('date formatting', function () {
 	Assert::same('Tue, 15 Nov 1994 08:12:31 GMT', Helpers::formatDate('1994-11-15T08:12:31+0000'));
 	Assert::same('Tue, 15 Nov 1994 08:12:31 GMT', Helpers::formatDate('1994-11-15T10:12:31+0200'));
diff --git a/tests/Http/Request.detectLanguage.phpt b/tests/Http/Request.detectLanguage.phpt
@@ -29,6 +29,23 @@ test('language with quality weights', function () {
 });
 
 
+test('case-insensitive matching, original case is returned', function () {
+	$headers = ['Accept-Language' => 'cs_CZ, EN;q=0.8'];
+	$request = new Http\Request(new Http\UrlScript, headers: $headers);
+
+	Assert::same('CS', $request->detectLanguage(['CS', 'En']));
+	Assert::same('cs-cz', $request->detectLanguage(['en', 'cs-cz']));
+});
+
+
+test('wildcard matches any supported language', function () {
+	$request = new Http\Request(new Http\UrlScript, headers: ['Accept-Language' => 'fr, *;q=0.1']);
+
+	Assert::same('en', $request->detectLanguage(['en', 'cs'])); // fr unsupported, * falls back
+	Assert::null($request->detectLanguage([])); // nothing to fall back to
+});
+
+
 test('no Accept-Language header', function () {
 	$headers = [];
 	$request = new Http\Request(new Http\UrlScript, headers: $headers);
