From e7d0960e9c0a457227ccdca5dc5b5d2e6348b0fb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Gr=C3=A9goire=20Paris?= <postmaster@greg0ire.fr>
Date: Mon, 28 Nov 2022 21:10:53 +0100
Subject: [PATCH] Allow using enums as a token type

Enums allow to avoid having to come up with values to assign for each
name (unless you use a backed enum), and will provide more type safety.
---
 lib/Doctrine/Common/Lexer/AbstractLexer.php   | 18 +++++-
 lib/Doctrine/Common/Lexer/Token.php           |  3 +-
 .../Common/Lexer/AbstractLexerTest.php        |  9 +++
 tests/Doctrine/Common/Lexer/EnumLexer.php     | 58 +++++++++++++++++++
 tests/Doctrine/Common/Lexer/TokenType.php     | 12 ++++
 5 files changed, 96 insertions(+), 4 deletions(-)
 create mode 100644 tests/Doctrine/Common/Lexer/EnumLexer.php
 create mode 100644 tests/Doctrine/Common/Lexer/TokenType.php

diff --git a/lib/Doctrine/Common/Lexer/AbstractLexer.php b/lib/Doctrine/Common/Lexer/AbstractLexer.php
index 44ea16b..9ea73f3 100644
--- a/lib/Doctrine/Common/Lexer/AbstractLexer.php
+++ b/lib/Doctrine/Common/Lexer/AbstractLexer.php
@@ -5,8 +5,13 @@
 namespace Doctrine\Common\Lexer;
 
 use ReflectionClass;
+use UnitEnum;
 
+use function assert;
+use function get_class;
 use function implode;
+use function is_int;
+use function is_string;
 use function preg_split;
 use function sprintf;
 use function substr;
@@ -18,7 +23,7 @@
 /**
  * Base class for writing simple lexers, i.e. for creating small DSLs.
  *
- * @template T of string|int
+ * @template T of UnitEnum|string|int
  */
 abstract class AbstractLexer
 {
@@ -275,13 +280,18 @@ protected function scan($input)
     /**
      * Gets the literal for a given token.
      *
-     * @param int|string $token
+     * @param T $token
      *
      * @return int|string
      */
     public function getLiteral($token)
     {
-        $className = static::class;
+        if ($token instanceof UnitEnum) {
+            $className = get_class($token);
+        } else {
+            $className = static::class;
+        }
+
         $reflClass = new ReflectionClass($className);
         $constants = $reflClass->getConstants();
 
@@ -291,6 +301,8 @@ public function getLiteral($token)
             }
         }
 
+        assert(is_string($token) || is_int($token));
+
         return $token;
     }
 
diff --git a/lib/Doctrine/Common/Lexer/Token.php b/lib/Doctrine/Common/Lexer/Token.php
index 475afa1..4c6d53a 100644
--- a/lib/Doctrine/Common/Lexer/Token.php
+++ b/lib/Doctrine/Common/Lexer/Token.php
@@ -8,11 +8,12 @@
 use Countable;
 use Doctrine\Deprecations\Deprecation;
 use ReturnTypeWillChange;
+use UnitEnum;
 
 use function in_array;
 
 /**
- * @template T of string|int
+ * @template T of UnitEnum|string|int
  * @implements ArrayAccess<string,mixed>
  */
 final class Token implements ArrayAccess, Countable
diff --git a/tests/Doctrine/Common/Lexer/AbstractLexerTest.php b/tests/Doctrine/Common/Lexer/AbstractLexerTest.php
index 5a8ebac..c0c5bc6 100644
--- a/tests/Doctrine/Common/Lexer/AbstractLexerTest.php
+++ b/tests/Doctrine/Common/Lexer/AbstractLexerTest.php
@@ -236,6 +236,15 @@ public function testGetLiteral(): void
         $this->assertSame('fake_token', $this->concreteLexer->getLiteral('fake_token'));
     }
 
+    public function testGetLiteralWithEnumLexer(): void
+    {
+        $enumLexer = new EnumLexer();
+        $this->assertSame(
+            'Doctrine\Tests\Common\Lexer\TokenType::OPERATOR',
+            $enumLexer->getLiteral(TokenType::OPERATOR)
+        );
+    }
+
     public function testIsA(): void
     {
         $this->assertTrue($this->concreteLexer->isA(11, 'int'));
diff --git a/tests/Doctrine/Common/Lexer/EnumLexer.php b/tests/Doctrine/Common/Lexer/EnumLexer.php
new file mode 100644
index 0000000..c80aaae
--- /dev/null
+++ b/tests/Doctrine/Common/Lexer/EnumLexer.php
@@ -0,0 +1,58 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Doctrine\Tests\Common\Lexer;
+
+use Doctrine\Common\Lexer\AbstractLexer;
+
+use function in_array;
+use function is_numeric;
+use function is_string;
+
+/** @extends AbstractLexer<TokenType> */
+class EnumLexer extends AbstractLexer
+{
+    /**
+     * {@inheritDoc}
+     */
+    protected function getCatchablePatterns()
+    {
+        return [
+            '=|<|>',
+            '[a-z]+',
+            '\d+',
+        ];
+    }
+
+    /**
+     * {@inheritDoc}
+     */
+    protected function getNonCatchablePatterns()
+    {
+        return [
+            '\s+',
+            '(.)',
+        ];
+    }
+
+    /**
+     * {@inheritDoc}
+     */
+    protected function getType(&$value)
+    {
+        if (is_numeric($value)) {
+            $value = (int) $value;
+
+            return TokenType::INT;
+        }
+
+        if (in_array($value, ['=', '<', '>'])) {
+            return TokenType::OPERATOR;
+        }
+
+        if (is_string($value)) {
+            return TokenType::STRING;
+        }
+    }
+}
diff --git a/tests/Doctrine/Common/Lexer/TokenType.php b/tests/Doctrine/Common/Lexer/TokenType.php
new file mode 100644
index 0000000..d4a7d53
--- /dev/null
+++ b/tests/Doctrine/Common/Lexer/TokenType.php
@@ -0,0 +1,12 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Doctrine\Tests\Common\Lexer;
+
+enum TokenType
+{
+    case INT;
+    case OPERATOR;
+    case STRING;
+}