Narrow preg_match/preg_match_all subject string type when match is truthy

src/Type/Php/PregMatchTypeSpecifyingExtension.php

@@ -2,6 +2,8 @@
 
 namespace PHPStan\Type\Php;
 
+use PhpParser\Node\Expr;
+use PhpParser\Node\Expr\ArrayDimFetch;
 use PhpParser\Node\Expr\FuncCall;
 use PHPStan\Analyser\Scope;
 use PHPStan\Analyser\SpecifiedTypes;
@@ -41,13 +43,34 @@
 	{
 		$args = $node->getArgs();
 		$patternArg = $args[0] ?? null;
+		$subjectArg = $args[1] ?? null;
 		$matchesArg = $args[2] ?? null;
 		$flagsArg = $args[3] ?? null;
 
+		if ($patternArg === null) {
+			return new SpecifiedTypes();
+		}
+
+		$subjectTypes = new SpecifiedTypes();
 		if (
-			$patternArg === null || $matchesArg === null
+			$subjectArg !== null
+			&& $context->true()
+			&& $scope->getType($subjectArg->value)->isString()->yes()
+			&& !$this->isSubExprOfMatchesArg($subjectArg->value, $matchesArg !== null ? $matchesArg->value : null)
 		) {
-			return new SpecifiedTypes();
+			$subjectType = $this->regexShapeMatcher->matchSubjectExpr($patternArg->value, $scope);
+			if ($subjectType !== null) {
+				$subjectTypes = $this->typeSpecifier->create(
+					$subjectArg->value,
+					$subjectType,
+					$context,
+					$scope,
+				)->setRootExpr($node);
+			}
+		}
+
+		if ($matchesArg === null) {
+			return $subjectTypes;
 		}
 
 		$flagsType = null;
@@ -69,7 +92,7 @@
 			$matchedType = $this->regexShapeMatcher->matchAllExpr($patternArg->value, $flagsType, $wasMatched, $scope);
 		}
 		if ($matchedType === null) {
-			return new SpecifiedTypes();
+			return $subjectTypes;
 		}
 
 		$overwrite = false;
@@ -88,7 +111,21 @@
 			$types = $types->setAlwaysOverwriteTypes();
 		}
 
-		return $types;
+		return $types->unionWith($subjectTypes);
+	}
+
+	private function isSubExprOfMatchesArg(Expr $subject, ?Expr $matchesVar): bool
+	{
+		if ($matchesVar === null) {
+			return false;
+		}
+		$rootVar = $subject;
+		while ($rootVar instanceof ArrayDimFetch) {
+			$rootVar = $rootVar->var;
+		}
+		return $rootVar instanceof Expr\Variable
+			&& $matchesVar instanceof Expr\Variable
+			&& $rootVar->name === $matchesVar->name;
 	}
 
 }

src/Type/Php/RegexArrayShapeMatcher.php

@@ -61,6 +61,27 @@ public function matchExpr(Expr $patternExpr, ?Type $flagsType, TrinaryLogic $was
 		return $this->matchPatternType($this->getPatternType($patternExpr, $scope), $flagsType, $wasMatched, false);
 	}
 
+	public function matchSubjectExpr(Expr $patternExpr, Scope $scope): ?Type
+	{
+		$patternType = $this->getPatternType($patternExpr, $scope);
+		$constantStrings = $patternType->getConstantStrings();
+		if (count($constantStrings) === 0) {
+			return null;
+		}
+
+		$subjectTypes = [];
+		foreach ($constantStrings as $constantString) {
+			$astWalkResult = $this->regexGroupParser->parseGroups($constantString->getValue());
+			if ($astWalkResult === null) {
+				return null;
+			}
+
+			$subjectTypes[] = $astWalkResult->getSubjectBaseType();
+		}
+
+		return TypeCombinator::union(...$subjectTypes);
+	}
+
 	private function matchPatternType(Type $patternType, ?Type $flagsType, TrinaryLogic $wasMatched, bool $matchesAll): ?Type
 	{
 		if ($wasMatched->no()) {

tests/PHPStan/Analyser/nsrt/bug-14710.php

@@ -0,0 +1,110 @@
+<?php declare(strict_types = 1);
+
+namespace Bug14710;
+
+use function PHPStan\Testing\assertType;
+
+function ternaryPregMatch(string $x): void {
+	(preg_match('/^(a|b|c)$/', $x)) ?
+		assertType('non-falsy-string', $x)
+		: assertType('string', $x);
+}
+
+function ifPregMatch(string $x): void {
+	if (preg_match('/^(a|b|c)$/', $x)) {
+		assertType('non-falsy-string', $x);
+	} else {
+		assertType('string', $x);
+	}
+}
+
+function ternaryPregMatchWithMatches(string $x): void {
+	(preg_match('/^(a|b|c)$/', $x, $matches)) ?
+		assertType('non-falsy-string', $x)
+		: assertType('string', $x);
+}
+
+function ifPregMatchWithMatches(string $x): void {
+	if (preg_match('/^(a|b|c)$/', $x, $matches)) {
+		assertType('non-falsy-string', $x);
+	} else {
+		assertType('string', $x);
+	}
+}
+
+function pregMatchNonEmpty(string $x): void {
+	if (preg_match('/foo/', $x)) {
+		assertType('non-falsy-string', $x);
+	}
+}
+
+function pregMatchNoNarrow(string $x): void {
+	if (preg_match('/^$/', $x)) {
+		assertType('string', $x);
+	}
+}
+
+function pregMatchAll(string $x): void {
+	if (preg_match_all('/^(a|b|c)$/', $x)) {
+		assertType('non-falsy-string', $x);
+	} else {
+		assertType('string', $x);
+	}
+}
+
+function negatedPregMatch(string $x): void {
+	if (!preg_match('/^(a|b|c)$/', $x)) {
+		assertType('string', $x);
+	} else {
+		assertType('non-falsy-string', $x);
+	}
+}
+
+function pregMatchCompare(string $x): void {
+	if (preg_match('/^(a|b|c)$/', $x) === 1) {
+		assertType('non-falsy-string', $x);
+	}
+}
+
+function pregMatchNotIdentical(string $x): void {
+	if (preg_match('#ExtensionInterface$#', $x) !== 1) {
+		return;
+	}
+	assertType('non-falsy-string', $x);
+}
+
+function pregMatchNotEqual(string $x): void {
+	if (preg_match('#ExtensionInterface$#', $x) != 1) {
+		return;
+	}
+	assertType('non-falsy-string', $x);
+}
+
+function pregMatchWithNonConstantPattern(string $pattern, string $x): void {
+	if (preg_match($pattern, $x)) {
+		assertType('string', $x);
+	}
+}
+
+function pregMatchSubjectSharesVarWithMatches(): void {
+	$matches = ['', '', 'foo'];
+	if (preg_match('/^(a|b|c)$/', $matches[2], $matches)) {
+		assertType("array{non-falsy-string, 'a'|'b'|'c'}", $matches);
+	}
+}
+
+function pregMatchNullableSubject(?string $x): void {
+	// a null subject is coerced to '' which cannot match a non-empty pattern, so null is removed
+	if (preg_match('/^(a|b|c)$/', $x)) {
+		assertType('string|null', $x); // could be non-falsy-string
+	} else {
+		assertType('string|null', $x);
+	}
+}
+
+function pregMatchIntStringSubject(int|string $x): void {
+	// an int subject can be coerced to a matching string, so narrowing it away would be unsound
+	if (preg_match('/^(a|b|c)$/', $x)) {
+		assertType('int|string', $x);
+	}
+}