Refactor deserializer to avoid integer conversion

Author: marcospassos

src/Atn/ATNDeserializer.php

@@ -68,9 +68,14 @@ public function __construct(?ATNDeserializationOptions $options = null)
         $this->deserializationOptions = $options ?? ATNDeserializationOptions::defaultOptions();
     }
 
-    public function deserialize(string $data) : ATN
+    /**
+     * @param array<int> $data
+     */
+    public function deserialize(array $data) : ATN
     {
-        $this->reset($data);
+        $this->data = $data;
+        $this->pos = 0;
+
         $this->checkVersion();
         $atn = $this->readATN();
         $this->readStates($atn);
@@ -104,22 +109,6 @@ public function deserialize(string $data) : ATN
         return $atn;
     }
 
-    private function reset(string $data) : void
-    {
-        $characters = \preg_split('//u', $data, -1, \PREG_SPLIT_NO_EMPTY);
-
-        if ($characters === false) {
-            return;
-        }
-
-        for ($i = 0, $length = \count($characters); $i < $length; $i++) {
-            $this->data[] = StringUtils::codePoint($characters[$i]);
-        }
-
-        $this->pos = 0;
-    }
-
-
     private function checkVersion() : void
     {
         $version = $this->readInt();

src/Parser.php

@@ -32,10 +32,8 @@ abstract class Parser extends Recognizer
      * {@see ATN} with bypass alternatives.
      *
      * @see ATNDeserializationOptions::isGenerateRuleBypassTransitions()
-     *
-     * @var array<string, ATN>
      */
-    private static $bypassAltsAtnCache = [];
+    private static ?ATN $bypassAltsAtnCache = null;
 
     /**
      * The error handling strategy for the parser. The default value is a new
@@ -394,17 +392,14 @@ public function setTokenFactory(TokenFactory $factory) : void
      */
     public function getATNWithBypassAlts() : ATN
     {
-        $serializedAtn = $this->getSerializedATN();
-        $result = self::$bypassAltsAtnCache[$serializedAtn] ?? null;
-
-        if ($result === null) {
+        if (self::$bypassAltsAtnCache === null) {
             $deserializationOptions = new ATNDeserializationOptions();
             $deserializationOptions->setGenerateRuleBypassTransitions(true);
-            $result = (new ATNDeserializer($deserializationOptions))->deserialize($serializedAtn);
-            self::$bypassAltsAtnCache[$serializedAtn] = $result;
+            self::$bypassAltsAtnCache = (new ATNDeserializer($deserializationOptions))
+                ->deserialize($this->getSerializedATN());
         }
 
-        return $result;
+        return self::$bypassAltsAtnCache;
     }
 
     public function getErrorHandler() : ANTLRErrorStrategy

src/Recognizer.php

@@ -113,8 +113,10 @@ public function getTokenType(string $tokenName) : int
      * If this recognizer was generated, it will have a serialized ATN
      * representation of the grammar. For interpreters, we don't know
      * their serialized ATN despite having created the interpreter from it.
+     *
+     * @return array<int>
      */
-    public function getSerializedATN() : string
+    public function getSerializedATN() : array
     {
         throw new \InvalidArgumentException('there is no serialized ATN');
     }