RegexArrayShapeMatcher - enforce list-type when no named captures

src/Type/Php/RegexArrayShapeMatcher.php

@@ -8,7 +8,6 @@
 use PHPStan\TrinaryLogic;
 use PHPStan\Type\Accessory\AccessoryArrayListType;
 use PHPStan\Type\ArrayType;
-use PHPStan\Type\Constant\ConstantArrayType;
 use PHPStan\Type\Constant\ConstantArrayTypeBuilder;
 use PHPStan\Type\Constant\ConstantIntegerType;
 use PHPStan\Type\Constant\ConstantStringType;
@@ -62,7 +61,7 @@ public function matchExpr(Expr $patternExpr, ?Type $flagsType, TrinaryLogic $was
 	private function matchPatternType(Type $patternType, ?Type $flagsType, TrinaryLogic $wasMatched, bool $matchesAll): ?Type
 	{
 		if ($wasMatched->no()) {
-			return new ConstantArrayType([], []);
+			return ConstantArrayTypeBuilder::createEmpty()->getArray();
 		}
 
 		$constantStrings = $patternType->getConstantStrings();
@@ -146,8 +145,11 @@ private function matchRegex(string $regex, ?int $flags, TrinaryLogic $wasMatched
 
 			if (!$this->containsUnmatchedAsNull($flags, $matchesAll)) {
 				// positive match has a subject but not any capturing group
+				$builder = ConstantArrayTypeBuilder::createEmpty();
+				$builder->setOffsetValueType(new ConstantIntegerType(0), $this->createSubjectValueType($subjectBaseType, $flags, $matchesAll));
+
 				$combiType = TypeCombinator::union(
-					new ConstantArrayType([new ConstantIntegerType(0)], [$this->createSubjectValueType($subjectBaseType, $flags, $matchesAll)], [1], [], TrinaryLogic::createYes()),
+					$builder->getArray(),
 					$combiType,
 				);
 			}
@@ -206,7 +208,10 @@ private function matchRegex(string $regex, ?int $flags, TrinaryLogic $wasMatched
 				)
 			) {
 				// positive match has a subject but not any capturing group
-				$combiTypes[] = new ConstantArrayType([new ConstantIntegerType(0)], [$this->createSubjectValueType($subjectBaseType, $flags, $matchesAll)], [1], [], TrinaryLogic::createYes());
+				$builder = ConstantArrayTypeBuilder::createEmpty();
+				$builder->setOffsetValueType(new ConstantIntegerType(0), $this->createSubjectValueType($subjectBaseType, $flags, $matchesAll));
+
+				$combiTypes[] = $builder->getArray();
 			}
 
 			return TypeCombinator::union(...$combiTypes);
@@ -238,6 +243,7 @@ private function buildArrayType(
 		bool $matchesAll,
 	): Type
 	{
+		$forceList = count($markVerbs) === 0;
 		$builder = ConstantArrayTypeBuilder::createEmpty();
 
 		// first item in matches contains the overall match.
@@ -256,6 +262,8 @@ private function buildArrayType(
 			$optional = $this->isGroupOptional($captureGroup, $wasMatched, $flags, $isTrailingOptional, $matchesAll);
 
 			if ($captureGroup->isNamed()) {
+				$forceList = false;
+
 				$builder->setOffsetValueType(
 					$this->getKeyType($captureGroup->getName()),
 					$groupValueType,
@@ -288,13 +296,17 @@ private function buildArrayType(
 			$arrayType = TypeCombinator::intersect(new ArrayType(new IntegerType(), $builder->getArray()), new AccessoryArrayListType());
 			if (!$wasMatched->yes()) {
 				$arrayType = TypeCombinator::union(
-					new ConstantArrayType([], []),
+					ConstantArrayTypeBuilder::createEmpty()->getArray(),
 					$arrayType,
 				);
 			}
 			return $arrayType;
 		}
 
+		if ($forceList) {
+			return TypeCombinator::intersect($builder->getArray(), new AccessoryArrayListType());
+		}
+
 		return $builder->getArray();
 	}
 

tests/PHPStan/Analyser/nsrt/bug-11311.php

@@ -191,12 +191,12 @@ function (string $s): void {
 
 function (string $s): void {
 	preg_match('/%a(\d*)/', $s, $matches, PREG_UNMATCHED_AS_NULL);
-	assertType("array{0?: string, 1?: ''|numeric-string|null}", $matches); // could be array{0?: string, 1?: ''|numeric-string}
+	assertType("list{0?: string, 1?: ''|numeric-string|null}", $matches); // could be array{0?: string, 1?: ''|numeric-string}
 };
 
 function (string $s): void {
 	preg_match('/%a(\d*)?/', $s, $matches, PREG_UNMATCHED_AS_NULL);
-	assertType("array{0?: string, 1?: ''|numeric-string|null}", $matches); // could be array{0?: string, 1?: ''|numeric-string}
+	assertType("list{0?: string, 1?: ''|numeric-string|null}", $matches); // could be array{0?: string, 1?: ''|numeric-string}
 };
 
 function (string $s): void {
@@ -222,5 +222,5 @@ function (string $s): void {
 
 function (string $s): void {
 	preg_match('~a|(\d)|(\s)~', $s, $matches, PREG_UNMATCHED_AS_NULL);
-	assertType("array{0?: string, 1?: numeric-string|null, 2?: non-empty-string|null}", $matches);
+	assertType("list{0?: string, 1?: numeric-string|null, 2?: non-empty-string|null}", $matches);
 };

tests/PHPStan/Analyser/nsrt/bug-11580.php

@@ -26,7 +26,7 @@ public function bad2(string $in): void
 	public function bad3(string $in): void
 	{
 		$result = preg_match('~^/xxx/([\w\-]+)/?([\w\-]+)?/?$~', $in, $matches);
-		assertType('array{0?: string, 1?: non-empty-string, 2?: non-empty-string}', $matches);
+		assertType('list{0?: string, 1?: non-empty-string, 2?: non-empty-string}', $matches);
 		if ($result) {
 			assertType('array{0: non-falsy-string, 1: non-empty-string, 2?: non-empty-string}', $matches);
 		}

tests/PHPStan/Analyser/nsrt/preg_match_shapes.php

@@ -300,7 +300,7 @@ function (string $size): void {
 	if (preg_match('~^a\.(b)?(c)?d~', $size, $matches) !== 1) {
 		throw new InvalidArgumentException(sprintf('Invalid size "%s"', $size));
 	}
-	assertType("array{0: non-falsy-string, 1?: ''|'b', 2?: 'c'}", $matches);
+	assertType("list{0: non-falsy-string, 1?: ''|'b', 2?: 'c'}", $matches);
 };
 
 function (string $size): void {
@@ -525,15 +525,15 @@ function bug11323(string $s): void {
 
 function (string $s): void {
 	preg_match('/%a(\d*)/', $s, $matches);
-	assertType("array{0?: string, 1?: ''|numeric-string}", $matches);
+	assertType("list{0?: string, 1?: ''|numeric-string}", $matches);
 };
 
 class Bug11376
 {
 	public function test(string $str): void
 	{
 		preg_match('~^(?:(\w+)::)?(\w+)$~', $str, $matches);
-		assertType('array{0?: string, 1?: string, 2?: non-empty-string}', $matches);
+		assertType('list{0?: string, 1?: string, 2?: non-empty-string}', $matches);
 	}
 
 	public function test2(string $str): void
@@ -564,7 +564,7 @@ function (string $s): void {
 	}
 
 	if (preg_match($p, $s, $matches)) {
-		assertType("array{0: non-falsy-string, 1: 'x'|'£'|numeric-string, 2?: ''|numeric-string, 3?: 'x'}", $matches);
+		assertType("list{0: non-falsy-string, 1: 'x'|'£'|numeric-string, 2?: ''|numeric-string, 3?: 'x'}", $matches);
 	}
 };
 
@@ -730,7 +730,7 @@ function (string $s): void {
 
 function (string $s): void {
 	preg_match('~a|(\d)|(\s)~', $s, $matches);
-	assertType("array{0?: string, 1?: '', 2?: non-empty-string}|array{0?: string, 1?: numeric-string}", $matches);
+	assertType("list{0?: string, 1?: '', 2?: non-empty-string}|list{0?: string, 1?: numeric-string}", $matches);
 };
 
 function bug11490 (string $expression): void {
@@ -762,13 +762,13 @@ function bug11604 (string $string): void {
 		return;
 	}
 
-	assertType("array{0: non-empty-string, 1?: ''|'XX', 2?: 'YY'}", $matches);
+	assertType("list{0: non-empty-string, 1?: ''|'XX', 2?: 'YY'}", $matches);
 	// could be array{string, '', 'YY'}|array{string, 'XX'}|array{string}
 }
 
 function bug11604b (string $string): void {
 	if (preg_match('/(XX)|(YY)?(ZZ)/', $string, $matches)) {
-		assertType("array{0: non-empty-string, 1?: ''|'XX', 2?: ''|'YY', 3?: 'ZZ'}", $matches);
+		assertType("list{0: non-empty-string, 1?: ''|'XX', 2?: ''|'YY', 3?: 'ZZ'}", $matches);
 	}
 }
 
@@ -935,11 +935,11 @@ function bugEmptySubexpression (string $string): void {
 	}
 
 	if (preg_match('~((a)||(b))~', $string, $matches)) {
-		assertType("array{0: string, 1: ''|'a'|'b', 2?: ''|'a', 3?: 'b'}", $matches);
+		assertType("list{0: string, 1: ''|'a'|'b', 2?: ''|'a', 3?: 'b'}", $matches);
 	}
 
 	if (preg_match('~((a)|()|(b))~', $string, $matches)) {
-		assertType("array{0: string, 1: ''|'a'|'b', 2?: ''|'a', 3?: '', 4?: 'b'}", $matches);
+		assertType("list{0: string, 1: ''|'a'|'b', 2?: ''|'a', 3?: '', 4?: 'b'}", $matches);
 	}
 }
 
@@ -1010,3 +1010,8 @@ function bug12749f(string $str): void
 		assertType('array{non-empty-string}', $match); // could be numeric-string
 	}
 }
+
+function bug12397(string $string) : array {
+	$m = preg_match('#\b([A-Z]{2,})-(\d+)#', $string, $match);
+	assertType('list{0?: string, 1?: non-falsy-string, 2?: numeric-string}', $match);
+}

tests/PHPStan/Analyser/nsrt/preg_replace_callback_shapes.php

@@ -22,7 +22,7 @@ function (string $s): void {
 	preg_replace_callback(
 		'/(foo)?(bar)?(baz)?/',
 		function ($matches) {
-			assertType("array{0: array{string, int<-1, max>}, 1?: array{''|'foo', int<-1, max>}, 2?: array{''|'bar', int<-1, max>}, 3?: array{'baz', int<-1, max>}}", $matches);
+			assertType("list{0: array{string, int<-1, max>}, 1?: array{''|'foo', int<-1, max>}, 2?: array{''|'bar', int<-1, max>}, 3?: array{'baz', int<-1, max>}}", $matches);
 			return '';
 		},
 		$s,

tests/PHPStan/Rules/Arrays/NonexistentOffsetInArrayDimFetchRuleTest.php

@@ -900,4 +900,11 @@ public function testNarrowSuperglobals(): void
 		$this->analyse([__DIR__ . '/data/narrow-superglobal.php'], []);
 	}
 
+	public function testBug11602(): void
+	{
+		$this->reportPossiblyNonexistentGeneralArrayOffset = true;
+
+		$this->analyse([__DIR__ . '/data/bug-11602.php'], []);
+	}
+
 }

tests/PHPStan/Rules/Arrays/data/bug-11602.php

@@ -0,0 +1,23 @@
+<?php
+
+namespace Bug11602;
+
+class HelloWorld
+{
+	public function parseRef(string $coordinate, string $ref): string
+	{
+		if (preg_match('/^([A-Z]{1,3})([0-9]{1,7})(:([A-Z]{1,3})([0-9]{1,7}))?$/', $ref, $matches) !== 1) {
+			return $ref;
+		}
+		if (!isset($matches[3])) { // single cell, not range
+			return $coordinate;
+		}
+		$minRow = (int) $matches[2];
+		$maxRow = (int) $matches[5];
+		$rows = $maxRow - $minRow + 1;
+		$minCol = $matches[1];
+		$maxCol = $matches[4];
+
+		return "$minCol$minRow:$maxCol$maxRow";
+	}
+}