Properly handle non-ascii characters for all languages enum keys

Author: abnegate

src/SDK/Language.php

@@ -86,24 +86,42 @@ public function getFilters(): array
 
     protected function toPascalCase(string $value): string
     {
-        return ucfirst($this->toCamelCase($value));
+        return \ucfirst($this->toCamelCase($value));
     }
 
     protected function toCamelCase($str): string
     {
-        $str = preg_replace('/[^a-z0-9' . implode("", []) . ']+/i', ' ', $str);
-        $str = trim($str);
-        $str = ucwords($str);
-        $str = str_replace(" ", "", $str);
-        return lcfirst($str);
+        // Normalize the string to decompose accented characters
+        $str = \Normalizer::normalize($str, \Normalizer::FORM_D);
+
+        // Remove accents and other residual non-ASCII characters
+        $str = \preg_replace('/\p{M}/u', '', $str);
+
+        $str = \preg_replace('/[^a-zA-Z0-9]+/', ' ', $str);
+        $str = \trim($str);
+        $str = \ucwords($str);
+        $str = \str_replace(' ', '', $str);
+        $str = \lcfirst($str);
+
+        return $str;
     }
 
     protected function toSnakeCase($str): string
     {
-        $str = \preg_replace('/([a-z])([A-Z])/', '$1 $2', $str);
-        $str = \explode(' ', $str);
-        $str = \implode('_', $str);
-        return \strtolower($str);
+        // Normalize the string to decompose accented characters
+        $str = \Normalizer::normalize($str, \Normalizer::FORM_D);
+
+        // Remove accents and other residual non-ASCII characters
+        $str = \preg_replace('/\p{M}/u', '', $str);
+
+        // Remove apostrophes before replacing non-word characters with underscores
+        $str = \str_replace("'", '', $str);
+        $str = \preg_replace('/[^a-zA-Z0-9]+/', '_', $str);
+        $str = \preg_replace('/_+/', '_', $str);
+        $str = \trim($str, '_');
+        $str = \strtolower($str);
+
+        return $str;
     }
 
     protected function toUpperSnakeCase($str): string

src/SDK/Language/Kotlin.php

@@ -443,7 +443,6 @@ public function getFilters(): array
                 if (isset($this->getIdentifierOverrides()[$value])) {
                     $value = $this->getIdentifierOverrides()[$value];
                 }
-                $value = \preg_replace('/[^a-zA-Z0-9]/', '', $value);
                 return $this->toUpperSnakeCase($value);
             }),
         ];

src/SDK/Language/Swift.php

@@ -471,7 +471,6 @@ public function getFilters(): array
                 if (isset($this->getIdentifierOverrides()[$value])) {
                     $value = $this->getIdentifierOverrides()[$value];
                 }
-                $value = \preg_replace('/[^a-zA-Z0-9]/', '', $value);
                 return $this->toCamelCase($value);
             }),
         ];

templates/deno/src/enums/enum.ts.twig

@@ -1,6 +1,6 @@
 export enum {{ enum.name | caseUcfirst | overrideIdentifier }} {
     {%~ for value in enum.enum %}
     {%~ set key = enum.keys is empty ? value : enum.keys[loop.index0] %}
-    {{ key | caseUcfirst | replace({'-': ''}) }} = '{{ value }}',
+    {{ key | caseEnumKey }} = '{{ value }}',
     {%~ endfor %}
 }

templates/node/lib/enums/enum.js.twig

@@ -1,7 +1,7 @@
 const {{ enum.name | caseUcfirst | overrideIdentifier }} = {
 {% for value in enum.enum %}
 {% set key = enum.keys is empty ? value : enum.keys[loop.index0] %}
-    {{ key | replace({'-': '',' ': ''}) | caseEnumKey }}: '{{value}}' {% if not loop.last %},{% endif %}
+    {{ key | caseEnumKey }}: '{{value}}' {% if not loop.last %},{% endif %}
 
 {% endfor %}
 }

templates/python/package/enums/enum.py.twig

@@ -3,5 +3,5 @@ from enum import Enum
 class {{ enum.name | caseUcfirst | overrideIdentifier }}(Enum):
 {% for value in enum.enum %}
 {% set key = enum.keys is empty ? value : enum.keys[loop.index0] %}
-    {{ key | caseUpper | replace({'-': '',' ':'_', '(': '', ')': '', '.': ''})}} = "{{ value }}"
+    {{ key | caseEnumKey }} = "{{ value }}"
 {% endfor %}

templates/ruby/lib/container/enums/enum.rb.twig

@@ -3,7 +3,7 @@ module {{ spec.title | caseUcfirst }}
         module {{ enum.name | caseUcfirst | overrideIdentifier }}
             {%~ for value in enum.enum %}
             {%~ set key = enum.keys is empty ? value : enum.keys[loop.index0] %}
-            {{ key | caseUpper | replace({'-': '_', ' ':'_', '(': '', ')': '', '.': ''}) }} = '{{ value }}'
+            {{ key | caseEnumKey }} = '{{ value }}'
             {%~ endfor %}
         end
     end