feature #39699 [String] Made AsciiSlugger fallback to parent locale's symbolsMap (jontjs)

nicolas-grekas · nicolas-grekas · commit e60721014b01 · 2021-01-26T10:34:31.000+01:00
This PR was merged into the 5.3-dev branch. Discussion ---------- [String] Made AsciiSlugger fallback to parent locale's symbolsMap | Q | A | ------------- | --- | Branch? | 5.x | Bug fix? | no | New feature? | yes | Deprecations? | no | Tickets | Fix #39178 | License | MIT | Doc PR | symfony/symfony-docs#14776 The Slugger already performed a fallback from (e.g.) 'en_GB' to (e.g.) 'en' for the transliterator, this PR adds similar behaviour for the symbols map. Commits ------- 916a8cf [String] Make AsciiSlugger fallback to parent locale's symbolsMap
diff --git a/src/Symfony/Component/String/CHANGELOG.md b/src/Symfony/Component/String/CHANGELOG.md
@@ -1,6 +1,11 @@
 CHANGELOG
 =========
 
+5.3
+---
+
+ * Made `AsciiSlugger` fallback to parent locale's symbolsMap
+
 5.2.0
 -----
 
diff --git a/src/Symfony/Component/String/Slugger/AsciiSlugger.php b/src/Symfony/Component/String/Slugger/AsciiSlugger.php
@@ -111,6 +111,8 @@ public function slug(string $string, string $separator = '-', string $locale = n
         }
 
         if ($this->symbolsMap instanceof \Closure) {
+            // If the symbols map is passed as a closure, there is no need to fallback to the parent locale
+            // as the closure can just provide substitutions for all locales of interest.
             $symbolsMap = $this->symbolsMap;
             array_unshift($transliterator, static function ($s) use ($symbolsMap, $locale) {
                 return $symbolsMap($s, $locale);
@@ -119,9 +121,20 @@ public function slug(string $string, string $separator = '-', string $locale = n
 
         $unicodeString = (new UnicodeString($string))->ascii($transliterator);
 
-        if (\is_array($this->symbolsMap) && isset($this->symbolsMap[$locale])) {
-            foreach ($this->symbolsMap[$locale] as $char => $replace) {
-                $unicodeString = $unicodeString->replace($char, ' '.$replace.' ');
+        if (\is_array($this->symbolsMap)) {
+            $map = null;
+            if (isset($this->symbolsMap[$locale])) {
+                $map = $this->symbolsMap[$locale];
+            } else {
+                $parent = self::getParentLocale($locale);
+                if ($parent && isset($this->symbolsMap[$parent])) {
+                    $map = $this->symbolsMap[$parent];
+                }
+            }
+            if ($map) {
+                foreach ($map as $char => $replace) {
+                    $unicodeString = $unicodeString->replace($char, ' '.$replace.' ');
+                }
             }
         }
 
@@ -143,17 +156,28 @@ private function createTransliterator(string $locale): ?\Transliterator
         }
 
         // Locale not supported and no parent, fallback to any-latin
-        if (false === $str = strrchr($locale, '_')) {
+        if (!$parent = self::getParentLocale($locale)) {
             return $this->transliterators[$locale] = null;
         }
 
         // Try to use the parent locale (ie. try "de" for "de_AT") and cache both locales
-        $parent = substr($locale, 0, -\strlen($str));
-
         if ($id = self::LOCALE_TO_TRANSLITERATOR_ID[$parent] ?? null) {
             $transliterator = \Transliterator::create($id.'/BGN') ?? \Transliterator::create($id);
         }
 
         return $this->transliterators[$locale] = $this->transliterators[$parent] = $transliterator ?? null;
     }
+
+    private static function getParentLocale(?string $locale): ?string
+    {
+        if (!$locale) {
+            return null;
+        }
+        if (false === $str = strrchr($locale, '_')) {
+            // no parent locale
+            return null;
+        }
+
+        return substr($locale, 0, -\strlen($str));
+    }
 }
diff --git a/src/Symfony/Component/String/Tests/SluggerTest.php b/src/Symfony/Component/String/Tests/SluggerTest.php
@@ -65,6 +65,37 @@ public function testSlugCharReplacementLocaleMethod()
         $this->assertSame('yo_y_tu_a_esta_direccion_slug_en_senal_test_es', $slug);
     }
 
+    public function testSlugCharReplacementLocaleConstructWithoutSymbolsMap()
+    {
+        $slugger = new AsciiSlugger('en');
+        $slug = (string) $slugger->slug('you & me with this address slug@test.uk', '_');
+
+        $this->assertSame('you_and_me_with_this_address_slug_at_test_uk', $slug);
+    }
+
+    public function testSlugCharReplacementParentLocaleConstructWithoutSymbolsMap()
+    {
+        $slugger = new AsciiSlugger('en_GB');
+        $slug = (string) $slugger->slug('you & me with this address slug@test.uk', '_');
+
+        $this->assertSame('you_and_me_with_this_address_slug_at_test_uk', $slug);
+    }
+
+    public function testSlugCharReplacementParentLocaleConstruct()
+    {
+        $slugger = new AsciiSlugger('fr_FR', ['fr' => ['&' => 'et', '@' => 'chez']]);
+        $slug = (string) $slugger->slug('toi & moi avec cette adresse slug@test.fr', '_');
+
+        $this->assertSame('toi_et_moi_avec_cette_adresse_slug_chez_test_fr', $slug);
+    }
+
+    public function testSlugCharReplacementParentLocaleMethod()
+    {
+        $slugger = new AsciiSlugger(null, ['es' => ['&' => 'y', '@' => 'en senal']]);
+        $slug = (string) $slugger->slug('yo & tu a esta dirección slug@test.es', '_', 'es_ES');
+        $this->assertSame('yo_y_tu_a_esta_direccion_slug_en_senal_test_es', $slug);
+    }
+
     public function testSlugClosure()
     {
         $slugger = new AsciiSlugger(null, function ($s, $locale) {
