Refactor RegexGroupParser - extract RegexAlternation class

conf/config.neon

@@ -1506,10 +1506,10 @@ services:
 		class: PHPStan\Type\Php\RegexArrayShapeMatcher
 
 	-
-		class: PHPStan\Type\Php\RegexGroupParser
+		class: PHPStan\Type\Regex\RegexGroupParser
 
 	-
-		class: PHPStan\Type\Php\RegexExpressionHelper
+		class: PHPStan\Type\Regex\RegexExpressionHelper
 
 	-
 		class: PHPStan\Type\Php\ReflectionClassConstructorThrowTypeExtension

src/Rules/Regexp/RegularExpressionPatternRule.php

@@ -9,7 +9,7 @@
 use PHPStan\Analyser\Scope;
 use PHPStan\Rules\Rule;
 use PHPStan\Rules\RuleErrorBuilder;
-use PHPStan\Type\Php\RegexExpressionHelper;
+use PHPStan\Type\Regex\RegexExpressionHelper;
 use function in_array;
 use function sprintf;
 use function str_starts_with;

src/Rules/Regexp/RegularExpressionQuotingRule.php

@@ -15,7 +15,7 @@
 use PHPStan\Rules\Rule;
 use PHPStan\Rules\RuleErrorBuilder;
 use PHPStan\ShouldNotHappenException;
-use PHPStan\Type\Php\RegexExpressionHelper;
+use PHPStan\Type\Regex\RegexExpressionHelper;
 use function array_filter;
 use function array_merge;
 use function array_values;

src/Type/Php/RegexArrayShapeMatcher.php

@@ -14,10 +14,13 @@
 use PHPStan\Type\Constant\ConstantStringType;
 use PHPStan\Type\IntegerRangeType;
 use PHPStan\Type\IntegerType;
+use PHPStan\Type\Regex\RegexAlternation;
+use PHPStan\Type\Regex\RegexCapturingGroup;
+use PHPStan\Type\Regex\RegexExpressionHelper;
+use PHPStan\Type\Regex\RegexGroupParser;
 use PHPStan\Type\StringType;
 use PHPStan\Type\Type;
 use PHPStan\Type\TypeCombinator;
-use function array_key_exists;
 use function array_reverse;
 use function count;
 use function in_array;
@@ -117,7 +120,7 @@ private function matchRegex(string $regex, ?int $flags, TrinaryLogic $wasMatched
 			// regex could not be parsed by Hoa/Regex
 			return null;
 		}
-		[$groupList, $groupCombinations, $markVerbs] = $parseResult;
+		[$groupList, $markVerbs] = $parseResult;
 
 		$trailingOptionals = 0;
 		foreach (array_reverse($groupList) as $captureGroup) {
@@ -128,7 +131,7 @@ private function matchRegex(string $regex, ?int $flags, TrinaryLogic $wasMatched
 		}
 
 		$onlyOptionalTopLevelGroup = $this->getOnlyOptionalTopLevelGroup($groupList);
-		$onlyTopLevelAlternationId = $this->getOnlyTopLevelAlternationId($groupList);
+		$onlyTopLevelAlternation = $this->getOnlyTopLevelAlternation($groupList);
 
 		if (
 			!$matchesAll
@@ -160,12 +163,11 @@ private function matchRegex(string $regex, ?int $flags, TrinaryLogic $wasMatched
 		} elseif (
 			!$matchesAll
 			&& $wasMatched->yes()
-			&& $onlyTopLevelAlternationId !== null
-			&& array_key_exists($onlyTopLevelAlternationId, $groupCombinations)
+			&& $onlyTopLevelAlternation !== null
 		) {
 			$combiTypes = [];
 			$isOptionalAlternation = false;
-			foreach ($groupCombinations[$onlyTopLevelAlternationId] as $groupCombo) {
+			foreach ($onlyTopLevelAlternation->getGroupCombinations() as $groupCombo) {
 				$comboList = $groupList;
 
 				$beforeCurrentCombo = true;
@@ -176,7 +178,10 @@ private function matchRegex(string $regex, ?int $flags, TrinaryLogic $wasMatched
 						$beforeCurrentCombo = false;
 					} elseif ($beforeCurrentCombo && !$group->resetsGroupCounter()) {
 						$group->forceNonOptional();
-					} elseif ($group->getAlternationId() === $onlyTopLevelAlternationId && !$this->containsUnmatchedAsNull($flags ?? 0, $matchesAll)) {
+					} elseif (
+						$group->getAlternationId() === $onlyTopLevelAlternation->getId()
+						&& !$this->containsUnmatchedAsNull($flags ?? 0, $matchesAll)
+					) {
 						unset($comboList[$groupId]);
 					}
 				}
@@ -243,9 +248,9 @@ private function getOnlyOptionalTopLevelGroup(array $captureGroups): ?RegexCaptu
 	/**
 	 * @param array<int, RegexCapturingGroup> $captureGroups
 	 */
-	private function getOnlyTopLevelAlternationId(array $captureGroups): ?int
+	private function getOnlyTopLevelAlternation(array $captureGroups): ?RegexAlternation
 	{
-		$alternationId = null;
+		$alternation = null;
 		foreach ($captureGroups as $captureGroup) {
 			if (!$captureGroup->isTopLevel()) {
 				continue;
@@ -255,14 +260,14 @@ private function getOnlyTopLevelAlternationId(array $captureGroups): ?int
 				return null;
 			}
 
-			if ($alternationId === null) {
-				$alternationId = $captureGroup->getAlternationId();
-			} elseif ($alternationId !== $captureGroup->getAlternationId()) {
+			if ($alternation === null) {
+				$alternation = $captureGroup->getAlternation();
+			} elseif ($alternation->getId() !== $captureGroup->getAlternation()->getId()) {
 				return null;
 			}
 		}
 
-		return $alternationId;
+		return $alternation;
 	}
 
 	/**

src/Type/Regex/RegexAlternation.php

@@ -0,0 +1,39 @@
+<?php declare(strict_types = 1);
+
+namespace PHPStan\Type\Regex;
+
+use function array_key_exists;
+
+final class RegexAlternation
+{
+
+	/** @var array<int, list<int>> */
+	private array $groupCombinations = [];
+
+	public function __construct(private int $alternationId)
+	{
+	}
+
+	public function getId(): int
+	{
+		return $this->alternationId;
+	}
+
+	public function pushGroup(int $combinationIndex, RegexCapturingGroup $group): void
+	{
+		if (!array_key_exists($combinationIndex, $this->groupCombinations)) {
+			$this->groupCombinations[$combinationIndex] = [];
+		}
+
+		$this->groupCombinations[$combinationIndex][] = $group->getId();
+	}
+
+	/**
+	 * @return array<int, list<int>>
+	 */
+	public function getGroupCombinations(): array
+	{
+		return $this->groupCombinations;
+	}
+
+}

src/Type/Php/RegexCapturingGroup.php → src/Type/Regex/RegexCapturingGroup.php

@@ -1,6 +1,6 @@
 <?php declare(strict_types = 1);
 
-namespace PHPStan\Type\Php;
+namespace PHPStan\Type\Regex;
 
 use PHPStan\Type\Type;
 
@@ -12,7 +12,7 @@ class RegexCapturingGroup
 	public function __construct(
 		private int $id,
 		private ?string $name,
-		private ?int $alternationId,
+		private ?RegexAlternation $alternation,
 		private bool $inOptionalQuantification,
 		private RegexCapturingGroup|RegexNonCapturingGroup|null $parent,
 		private Type $type,
@@ -40,15 +40,27 @@ public function resetsGroupCounter(): bool
 		return $this->parent instanceof RegexNonCapturingGroup && $this->parent->resetsGroupCounter();
 	}
 
-	/** @phpstan-assert-if-true !null $this->getAlternationId() */
+	/**
+	 * @phpstan-assert-if-true !null $this->getAlternationId()
+	 * @phpstan-assert-if-true !null $this->getAlternation()
+	 */
 	public function inAlternation(): bool
 	{
-		return $this->alternationId !== null;
+		return $this->alternation !== null;
+	}
+
+	public function getAlternation(): ?RegexAlternation
+	{
+		return $this->alternation;
 	}
 
 	public function getAlternationId(): ?int
 	{
-		return $this->alternationId;
+		if ($this->alternation === null) {
+			return null;
+		}
+
+		return $this->alternation->getId();
 	}
 
 	public function isOptional(): bool

src/Type/Php/RegexExpressionHelper.php → src/Type/Regex/RegexExpressionHelper.php

@@ -1,6 +1,6 @@
 <?php declare(strict_types = 1);
 
-namespace PHPStan\Type\Php;
+namespace PHPStan\Type\Regex;
 
 use PhpParser\Node\Expr;
 use PhpParser\Node\Expr\BinaryOp\Concat;

src/Type/Php/RegexGroupParser.php → src/Type/Regex/RegexGroupParser.php

@@ -1,6 +1,6 @@
 <?php declare(strict_types = 1);
 
-namespace PHPStan\Type\Php;
+namespace PHPStan\Type\Regex;
 
 use Hoa\Compiler\Llk\Llk;
 use Hoa\Compiler\Llk\Parser;
@@ -20,7 +20,6 @@
 use PHPStan\Type\StringType;
 use PHPStan\Type\Type;
 use PHPStan\Type\TypeCombinator;
-use function array_key_exists;
 use function count;
 use function in_array;
 use function is_int;
@@ -45,7 +44,7 @@ public function __construct(
 	}
 
 	/**
-	 * @return array{array<int, RegexCapturingGroup>, array<int, array<int, int[]>>, list<string>}|null
+	 * @return array{array<int, RegexCapturingGroup>, list<string>}|null
 	 */
 	public function parseGroups(string $regex): ?array
 	{
@@ -75,44 +74,40 @@ public function parseGroups(string $regex): ?array
 		}
 
 		$capturingGroups = [];
-		$groupCombinations = [];
 		$alternationId = -1;
 		$captureGroupId = 100;
 		$markVerbs = [];
 		$this->walkRegexAst(
 			$ast,
-			false,
+			null,
 			$alternationId,
 			0,
 			false,
 			null,
 			$captureGroupId,
 			$capturingGroups,
-			$groupCombinations,
 			$markVerbs,
 			$captureOnlyNamed,
 			false,
 			$modifiers,
 		);
 
-		return [$capturingGroups, $groupCombinations, $markVerbs];
+		return [$capturingGroups, $markVerbs];
 	}
 
 	/**
 	 * @param array<int, RegexCapturingGroup> $capturingGroups
-	 * @param array<int, array<int, int[]>> $groupCombinations
 	 * @param list<string> $markVerbs
 	 */
 	private function walkRegexAst(
 		TreeNode $ast,
-		bool $inAlternation,
+		?RegexAlternation $alternation,
 		int &$alternationId,
 		int $combinationIndex,
 		bool $inOptionalQuantification,
 		RegexCapturingGroup|RegexNonCapturingGroup|null $parentGroup,
 		int &$captureGroupId,
 		array &$capturingGroups,
-		array &$groupCombinations,
 		array &$markVerbs,
 		bool $captureOnlyNamed,
 		bool $repeatedMoreThanOnce,
@@ -124,7 +119,7 @@ private function walkRegexAst(
 			$group = new RegexCapturingGroup(
 				$captureGroupId++,
 				null,
-				$inAlternation ? $alternationId : null,
+				$alternation,
 				$inOptionalQuantification,
 				$parentGroup,
 				$this->createGroupType(
@@ -139,7 +134,7 @@ private function walkRegexAst(
 			$group = new RegexCapturingGroup(
 				$captureGroupId++,
 				$name,
-				$inAlternation ? $alternationId : null,
+				$alternation,
 				$inOptionalQuantification,
 				$parentGroup,
 				$this->createGroupType(
@@ -151,15 +146,15 @@ private function walkRegexAst(
 			$parentGroup = $group;
 		} elseif ($ast->getId() === '#noncapturing') {
 			$group = new RegexNonCapturingGroup(
-				$inAlternation ? $alternationId : null,
+				$alternation,
 				$inOptionalQuantification,
 				$parentGroup,
 				false,
 			);
 			$parentGroup = $group;
 		} elseif ($ast->getId() === '#noncapturingreset') {
 			$group = new RegexNonCapturingGroup(
-				$inAlternation ? $alternationId : null,
+				$alternation,
 				$inOptionalQuantification,
 				$parentGroup,
 				true,
@@ -182,7 +177,7 @@ private function walkRegexAst(
 
 		if ($ast->getId() === '#alternation') {
 			$alternationId++;
-			$inAlternation = true;
+			$alternation = new RegexAlternation($alternationId);
 		}
 
 		if ($ast->getId() === '#mark') {
@@ -196,26 +191,21 @@ private function walkRegexAst(
 		) {
 			$capturingGroups[$group->getId()] = $group;
 
-			if (!array_key_exists($alternationId, $groupCombinations)) {
-				$groupCombinations[$alternationId] = [];
-			}
-			if (!array_key_exists($combinationIndex, $groupCombinations[$alternationId])) {
-				$groupCombinations[$alternationId][$combinationIndex] = [];
+			if ($alternation !== null) {
+				$alternation->pushGroup($combinationIndex, $group);
 			}
-			$groupCombinations[$alternationId][$combinationIndex][] = $group->getId();
 		}
 
 		foreach ($ast->getChildren() as $child) {
 			$this->walkRegexAst(
 				$child,
-				$inAlternation,
+				$alternation,
 				$alternationId,
 				$combinationIndex,
 				$inOptionalQuantification,
 				$parentGroup,
 				$captureGroupId,
 				$capturingGroups,
-				$groupCombinations,
 				$markVerbs,
 				$captureOnlyNamed,
 				$repeatedMoreThanOnce,

src/Type/Php/RegexNonCapturingGroup.php → src/Type/Regex/RegexNonCapturingGroup.php

@@ -1,12 +1,12 @@
 <?php declare(strict_types = 1);
 
-namespace PHPStan\Type\Php;
+namespace PHPStan\Type\Regex;
 
 class RegexNonCapturingGroup
 {
 
 	public function __construct(
-		private ?int $alternationId,
+		private ?RegexAlternation $alternation,
 		private bool $inOptionalQuantification,
 		private RegexCapturingGroup|RegexNonCapturingGroup|null $parent,
 		private bool $resetGroupCounter,
@@ -17,12 +17,16 @@ public function __construct(
 	/** @phpstan-assert-if-true !null $this->getAlternationId() */
 	public function inAlternation(): bool
 	{
-		return $this->alternationId !== null;
+		return $this->alternation !== null;
 	}
 
 	public function getAlternationId(): ?int
 	{
-		return $this->alternationId;
+		if ($this->alternation === null) {
+			return null;
+		}
+
+		return $this->alternation->getId();
 	}
 
 	public function isOptional(): bool