feat: Strict locale negotiation (codeigniter4#9360)

* feat: Add parameter for locale comparison

* feat: Update language negotiator

* tests: Update `NegotiateTest`

* docs: Update userguide

* fix: Rename feature option

* feat: Add `getBestLocaleMatch()`

* tests: Update `Negotiate` tests

* docs: Update docs

* fix: Rename feature option to back

* docs: Update the description for Negotiation

* fix: Update with the renamed option

* fix: Delete checking *.*

* fix: Move code to  file

Author: neznaika0

app/Config/Feature.php

@@ -26,4 +26,12 @@ class Feature extends BaseConfig
      * If false, `limit(0)` returns no records. (the behavior of 3.1.9 or later in version 3.x.)
      */
     public bool $limitZeroAsAll = true;
+
+    /**
+     * Use strict location negotiation.
+     *
+     * By default, the locale is selected based on a loose comparison of the language code (ISO 639-1)
+     * Enabling strict comparison will also consider the region code (ISO 3166-1 alpha-2).
+     */
+    public bool $strictLocaleNegotiation = false;
 }

system/HTTP/Negotiate.php

@@ -14,6 +14,7 @@
 namespace CodeIgniter\HTTP;
 
 use CodeIgniter\HTTP\Exceptions\HTTPException;
+use Config\Feature;
 
 /**
  * Class Negotiate
@@ -122,11 +123,15 @@ public function encoding(array $supported = []): string
      * types the application says it supports, and the types requested
      * by the client.
      *
-     * If no match is found, the first, highest-ranking client requested
+     * If strict locale negotiation is disabled and no match is found, the first, highest-ranking client requested
      * type is returned.
      */
     public function language(array $supported): string
     {
+        if (config(Feature::class)->strictLocaleNegotiation) {
+            return $this->getBestLocaleMatch($supported, $this->request->getHeaderLine('accept-language'));
+        }
+
         return $this->getBestMatch($supported, $this->request->getHeaderLine('accept-language'), false, false, true);
     }
 
@@ -189,6 +194,69 @@ protected function getBestMatch(
         return $strictMatch ? '' : $supported[0];
     }
 
+    /**
+     * Try to find the best matching locale. It supports strict locale comparison.
+     *
+     * If Config\App::$supportedLocales have "en-US" and "en-GB" locales, they can be recognized
+     * as two different locales. This method checks first for the strict match, then fallback
+     * to the most general locale (in this case "en") ISO 639-1 and finally to the locale variant
+     * "en-*" (ISO 639-1 plus "wildcard" for ISO 3166-1 alpha-2).
+     *
+     * If nothing from above is matched, then it returns the first option from the $supportedLocales array.
+     *
+     * @param list<string> $supportedLocales App-supported values
+     * @param ?string      $header           Compatible 'Accept-Language' header string
+     */
+    protected function getBestLocaleMatch(array $supportedLocales, ?string $header): string
+    {
+        if ($supportedLocales === []) {
+            throw HTTPException::forEmptySupportedNegotiations();
+        }
+
+        if ($header === null || $header === '') {
+            return $supportedLocales[0];
+        }
+
+        $acceptable      = $this->parseHeader($header);
+        $fallbackLocales = [];
+
+        foreach ($acceptable as $accept) {
+            // if acceptable quality is zero, skip it.
+            if ($accept['q'] === 0.0) {
+                continue;
+            }
+
+            // if acceptable value is "anything", return the first available
+            if ($accept['value'] === '*') {
+                return $supportedLocales[0];
+            }
+
+            // look for exact match
+            if (in_array($accept['value'], $supportedLocales, true)) {
+                return $accept['value'];
+            }
+
+            // set a fallback locale
+            $fallbackLocales[] = strtok($accept['value'], '-');
+        }
+
+        foreach ($fallbackLocales as $fallbackLocale) {
+            // look for exact match
+            if (in_array($fallbackLocale, $supportedLocales, true)) {
+                return $fallbackLocale;
+            }
+
+            // look for regional locale match
+            foreach ($supportedLocales as $locale) {
+                if (str_starts_with($locale, $fallbackLocale . '-')) {
+                    return $locale;
+                }
+            }
+        }
+
+        return $supportedLocales[0];
+    }
+
     /**
      * Parses an Accept* header into it's multiple values.
      *

tests/system/HTTP/NegotiateTest.php

@@ -16,6 +16,7 @@
 use CodeIgniter\HTTP\Exceptions\HTTPException;
 use CodeIgniter\Test\CIUnitTestCase;
 use Config\App;
+use Config\Feature;
 use PHPUnit\Framework\Attributes\Group;
 
 /**
@@ -111,11 +112,23 @@ public function testNegotiatesEncodingBasics(): void
 
     public function testAcceptLanguageBasics(): void
     {
-        $this->request->setHeader('Accept-Language', 'da, en-gb;q=0.8, en;q=0.7');
+        $this->request->setHeader('Accept-Language', 'da, en-gb, en-us;q=0.8, en;q=0.7');
 
         $this->assertSame('da', $this->negotiate->language(['da', 'en']));
         $this->assertSame('en-gb', $this->negotiate->language(['en-gb', 'en']));
         $this->assertSame('en', $this->negotiate->language(['en']));
+
+        // Will find the first locale instead of "en-gb"
+        $this->assertSame('en-us', $this->negotiate->language(['en-us', 'en-gb', 'en']));
+        $this->assertSame('en', $this->negotiate->language(['en', 'en-us', 'en-gb']));
+
+        config(Feature::class)->strictLocaleNegotiation = true;
+
+        $this->assertSame('da', $this->negotiate->language(['da', 'en']));
+        $this->assertSame('en-gb', $this->negotiate->language(['en-gb', 'en']));
+        $this->assertSame('en', $this->negotiate->language(['en']));
+        $this->assertSame('en-gb', $this->negotiate->language(['en-us', 'en-gb', 'en']));
+        $this->assertSame('en-gb', $this->negotiate->language(['en', 'en-us', 'en-gb']));
     }
 
     /**
@@ -125,7 +138,19 @@ public function testAcceptLanguageMatchesBroadly(): void
     {
         $this->request->setHeader('Accept-Language', 'fr-FR,fr;q=0.9,en-US;q=0.8,en;q=0.7');
 
-        $this->assertSame('fr', $this->negotiate->language(['fr', 'en']));
+        $this->assertSame('fr', $this->negotiate->language(['fr', 'fr-FR', 'en']));
+        $this->assertSame('fr-FR', $this->negotiate->language(['fr-FR', 'fr', 'en']));
+        $this->assertSame('fr-BE', $this->negotiate->language(['fr-BE', 'fr', 'en']));
+        $this->assertSame('en', $this->negotiate->language(['en', 'en-US']));
+        $this->assertSame('fr-BE', $this->negotiate->language(['ru', 'en-GB', 'fr-BE']));
+
+        config(Feature::class)->strictLocaleNegotiation = true;
+
+        $this->assertSame('fr-FR', $this->negotiate->language(['fr', 'fr-FR', 'en']));
+        $this->assertSame('fr-FR', $this->negotiate->language(['fr-FR', 'fr', 'en']));
+        $this->assertSame('fr', $this->negotiate->language(['fr-BE', 'fr', 'en']));
+        $this->assertSame('en-US', $this->negotiate->language(['en', 'en-US']));
+        $this->assertSame('fr-BE', $this->negotiate->language(['ru', 'en-GB', 'fr-BE']));
     }
 
     public function testBestMatchEmpty(): void

user_guide_src/source/changelogs/v4.6.0.rst

@@ -239,6 +239,13 @@ Routing
 
 - Now you can specify multiple hostnames when restricting routes.
 
+Negotiator
+==========
+
+- Added a feature flag ``Feature::$strictLocaleNegotiation`` to enable strict locale comparision.
+  Previously, response with language headers ``Accept-language: en-US,en-GB;q=0.9`` returned the first allowed language ``en`` could instead of the exact language ``en-US`` or ``en-GB``.
+  Set the value to ``true`` to enable comparison not only by language code ('en' - ISO 639-1) but also by regional code ('en-US' - ISO 639-1 plus ISO 3166-1 alpha).
+
 Testing
 =======
 

user_guide_src/source/incoming/content_negotiation.rst

@@ -102,6 +102,49 @@ and German you would do something like:
 In this example, 'en' would be returned as the current language. If no match is found, it will return the first element
 in the ``$supported`` array, so that should always be the preferred language.
 
+Strict Locale Negotiation
+-------------------------
+
+.. versionadded:: 4.6.0
+
+By default, locale is determined on a lossy comparison basis. So only the first part of the locale string is taken
+into account (language). This is usually sufficient. But sometimes we want to be able to distinguish between regional versions such as
+``en-US`` and ``en-GB`` to serve different content.
+
+For such cases, we have introduced a new setting that can be enabled via ``Config\Feature::$strictLocaleNegotiation``. This will ensure
+that the strict comparison will be made in the first place.
+
+.. note::
+
+    CodeIgniter comes with translations only for primary language tags ('en', 'fr', etc.). So if you enable this feature and your
+    settings in ``Config\App::$supportedLocales`` include regional language tags ('en-US', 'fr-FR', etc.), then keep in mind that
+    if you have your own translation files, you **must also change** the folder names for CodeIgniter's translation files to match
+    what you put in the ``$supportedLocales`` array.
+
+Now let's consider the below example. The browser's preferred language will be set as this::
+
+    GET /foo HTTP/1.1
+    Accept-Language: fr; q=1.0, en-GB; q=0.5
+
+In this example, the browser would prefer French, with a second choice of English (United Kingdom). Your website on another hand
+supports German and English (United States):
+
+.. literalinclude:: content_negotiation/008.php
+
+In this example, 'en-US' would be returned as the current language. If no match is found, it will return the first element
+in the ``$supported`` array. Here is how exactly the locale selection process works.
+
+Even though the 'fr' is preferred by the browser it is not in our ``$supported`` array. The same problem occurs with 'en-GB', but here
+we will be able to search for variants. First, we will fallback to the most general locale (in this case 'en') which again is not in our
+array. Then we will search for the regional locale 'en-'. And that's when our value from the ``$supported`` array will be matched.
+We will return 'en-US'.
+
+So the process of selecting a locale is as follows:
+
+#. strict match ('en-GB') - ISO 639-1 plus ISO 3166-1 alpha-2
+#. general locale match ('en') - ISO 639-1
+#. regional locale match ('en-') - ISO 639-1 plus "wildcard" for ISO 3166-1 alpha-2
+
 Encoding
 ========
 

user_guide_src/source/incoming/content_negotiation/008.php

@@ -0,0 +1,10 @@
+<?php
+
+$supported = [
+    'de',
+    'en-US',
+];
+
+$lang = $request->negotiate('language', $supported);
+// or
+$lang = $negotiate->language($supported);

user_guide_src/source/installation/upgrade_458.rst

@@ -52,4 +52,4 @@ All Changes
 This is a list of all files in the **project space** that received changes;
 many will be simple comments or formatting that have no effect on the runtime:
 
-- @TODO
+- @TODO

user_guide_src/source/installation/upgrade_460.rst

@@ -211,6 +211,7 @@ Config
 
 - app/Config/Feature.php
     - ``Config\Feature::$autoRoutesImproved`` has been changed to ``true``.
+    - ``Config\Feature::$strictLocaleNegotiation`` has been added.
 - app/Config/Routing.php
     - ``Config\Routing::$translateUriToCamelCase`` has been changed to ``true``.
 
@@ -220,4 +221,4 @@ All Changes
 This is a list of all files in the **project space** that received changes;
 many will be simple comments or formatting that have no effect on the runtime:
 
-- @TODO
+- app/Config/Feature.php