[psr1] Fix XML declarations misidentified as PHP short tags

src/Analyser/FileAnalysisProvider.php

@@ -4,6 +4,7 @@
 
 namespace Boundwize\StructArmed\Analyser;
 
+use Boundwize\StructArmed\Util\InlineHtmlOpeningTagMatcher;
 use Boundwize\StructArmed\Util\Path;
 use PhpParser\Error;
 use PhpParser\Node;
@@ -33,7 +34,6 @@
 use function token_get_all;
 use function trim;
 
-use const PREG_OFFSET_CAPTURE;
 use const T_INLINE_HTML;
 use const T_OPEN_TAG;
 use const T_OPEN_TAG_WITH_ECHO;
@@ -221,12 +221,12 @@ private function invalidPhpTagLineForToken(int $id, string $text, int $tokenLine
 
         if (
             $id !== T_INLINE_HTML
-            || preg_match('/<\?(?!php(?:\s|$)|=)/', $text, $matches, PREG_OFFSET_CAPTURE) !== 1
+            || ($tagOffset = InlineHtmlOpeningTagMatcher::invalidInlineHtmlTagOffset($text)) === null
         ) {
             return null;
         }
 
-        return $tokenLine + substr_count(substr($text, 0, $matches[0][1]), "\n");
+        return $tokenLine + substr_count(substr($text, 0, $tagOffset), "\n");
     }
 
     private function contents(string $file): string

src/Rule/Rules/File/Psr1PhpTagsRule.php

@@ -9,6 +9,7 @@
 use Boundwize\StructArmed\Rule\FileAnalysisRuleInterface;
 use Boundwize\StructArmed\Rule\FixableInterface;
 use Boundwize\StructArmed\Rule\RuleViolation;
+use Boundwize\StructArmed\Util\InlineHtmlOpeningTagMatcher;
 
 use function file_get_contents;
 use function file_put_contents;
@@ -21,7 +22,6 @@
 use function substr_count;
 use function token_get_all;
 
-use const PREG_OFFSET_CAPTURE;
 use const T_INLINE_HTML;
 use const T_OPEN_TAG;
 
@@ -150,9 +150,8 @@ private function replacementForInvalidTag(int $id, string $text, int $tokenLine,
 
         $searchOffset = 0;
 
-        while (preg_match('/<\?(?!php(?:\s|$)|=)/', $text, $matches, PREG_OFFSET_CAPTURE, $searchOffset) === 1) {
-            $tagOffset = $matches[0][1];
-            $tagLine   = $tokenLine + substr_count(substr($text, 0, $tagOffset), "\n");
+        while (($tagOffset = InlineHtmlOpeningTagMatcher::invalidInlineHtmlTagOffset($text, $searchOffset)) !== null) {
+            $tagLine = $tokenLine + substr_count(substr($text, 0, $tagOffset), "\n");
 
             if ($tagLine === $targetLine) {
                 return [

src/Util/InlineHtmlOpeningTagMatcher.php

@@ -0,0 +1,41 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Boundwize\StructArmed\Util;
+
+use function array_map;
+use function implode;
+use function preg_match;
+use function preg_quote;
+use function sprintf;
+
+use const PREG_OFFSET_CAPTURE;
+
+final class InlineHtmlOpeningTagMatcher
+{
+    /** @var list<string> */
+    private const ALLOWED_TAGS = [
+        'php',
+        'xml',
+        'xml-stylesheet',
+    ];
+
+    public static function invalidInlineHtmlTagOffset(string $text, int $offset = 0): ?int
+    {
+        $allowedTags = implode('|', array_map(
+            static fn (string $target): string => preg_quote($target, '/'),
+            self::ALLOWED_TAGS,
+        ));
+        $pattern     = sprintf(
+            '/<\?(?!(?:%s)(?:\s|$|\?>)|=)/',
+            $allowedTags,
+        );
+
+        if (preg_match($pattern, $text, $matches, PREG_OFFSET_CAPTURE, $offset) !== 1) {
+            return null;
+        }
+
+        return $matches[0][1];
+    }
+}

tests/Analyser/FileAnalysisProviderTest.php

@@ -6,6 +6,7 @@
 
 use Boundwize\StructArmed\Analyser\FileAnalysis;
 use Boundwize\StructArmed\Analyser\FileAnalysisProvider;
+use Boundwize\StructArmed\Util\InlineHtmlOpeningTagMatcher;
 use PHPUnit\Framework\Attributes\CoversClass;
 use PHPUnit\Framework\Attributes\DataProvider;
 use PHPUnit\Framework\TestCase;
@@ -14,6 +15,7 @@
 
 #[CoversClass(FileAnalysis::class)]
 #[CoversClass(FileAnalysisProvider::class)]
+#[CoversClass(InlineHtmlOpeningTagMatcher::class)]
 final class FileAnalysisProviderTest extends TestCase
 {
     public function testAnalysesPsr1FactsAndCachesThemByFile(): void
@@ -92,8 +94,12 @@ public static function lightweightAnalysisProvider(): iterable
         yield 'invalid UTF-8' => ["<?php echo \"\xB1\";", false, false, null];
         yield 'short tag' => ['<? echo "short";', false, true, 1];
         yield 'uppercase tag' => ['<?PHP echo "upper";', false, true, 1];
+        yield 'PHP-like inline HTML' => ['<?php?>', false, true, null];
         yield 'valid tag' => ['<?php echo "valid";', false, true, null];
         yield 'echo tag' => ['<?= "echo";', false, true, null];
+        yield 'XML declaration' => ['<?xml version="1.0"?>', false, true, null];
+        yield 'XML stylesheet' => ['<?xml-stylesheet href="style.xsl"?>', false, true, null];
+        yield 'arbitrary XML processing instruction' => ['<?xml-custom value="x"?>', false, true, 1];
         yield 'plain text' => ['plain text', false, true, null];
     }
 

tests/Rule/File/Psr1PhpTagsRuleTest.php

@@ -11,6 +11,7 @@
 use Boundwize\StructArmed\Rule\Rules\File\PhpFileFinder;
 use Boundwize\StructArmed\Rule\Rules\File\Psr1PhpTagsRule;
 use Boundwize\StructArmed\Rule\RuleViolation;
+use Boundwize\StructArmed\Util\InlineHtmlOpeningTagMatcher;
 use PHPUnit\Framework\Attributes\CoversClass;
 use PHPUnit\Framework\TestCase;
 
@@ -31,6 +32,7 @@
 
 #[CoversClass(PhpFileFinder::class)]
 #[CoversClass(Psr1PhpTagsRule::class)]
+#[CoversClass(InlineHtmlOpeningTagMatcher::class)]
 final class Psr1PhpTagsRuleTest extends TestCase
 {
     public function testViolatesShortOpenTag(): void
@@ -51,6 +53,44 @@ public function testViolatesShortOpenTag(): void
         }
     }
 
+    public function testViolatesAssignmentUsingShortOpenTag(): void
+    {
+        $basePath = $this->makeTempDir();
+
+        try {
+            mkdir($basePath . '/src');
+            file_put_contents($basePath . '/src/Foo.php', '<? $value = 1; ?>');
+
+            $violations = (new Psr1PhpTagsRule(['src/']))->evaluateProjectAll($basePath, Architecture::define());
+
+            $this->assertCount(1, $violations);
+            $this->assertSame(1, $violations[0]->line);
+        } finally {
+            unlink($basePath . '/src/Foo.php');
+            rmdir($basePath . '/src');
+            rmdir($basePath);
+        }
+    }
+
+    public function testViolatesCommentPrefixedShortOpenTag(): void
+    {
+        $basePath = $this->makeTempDir();
+
+        try {
+            mkdir($basePath . '/src');
+            file_put_contents($basePath . '/src/Foo.php', "<? /* comment */ echo 'x'; ?>");
+
+            $violations = (new Psr1PhpTagsRule(['src/']))->evaluateProjectAll($basePath, Architecture::define());
+
+            $this->assertCount(1, $violations);
+            $this->assertSame(1, $violations[0]->line);
+        } finally {
+            unlink($basePath . '/src/Foo.php');
+            rmdir($basePath . '/src');
+            rmdir($basePath);
+        }
+    }
+
     public function testPassesLongAndEchoTags(): void
     {
         $basePath = $this->makeTempDir();
@@ -69,6 +109,42 @@ public function testPassesLongAndEchoTags(): void
         }
     }
 
+    public function testPassesXmlProcessingInstructionsInInlineHtml(): void
+    {
+        $basePath = $this->makeTempDir();
+
+        try {
+            mkdir($basePath . '/src');
+            $contents = <<<'PHP'
+                <?php
+
+                header('Content-Type: application/xml');
+
+                ?>
+                <?xml version="1.0" encoding="UTF-8"?>
+                <?xml-stylesheet type="text/xsl" href="sitemap.xsl"?>
+                <urlset />
+                PHP;
+            file_put_contents($basePath . '/src/sitemap.php', $contents);
+
+            $psr1PhpTagsRule = new Psr1PhpTagsRule(['src/']);
+            $violations      = $psr1PhpTagsRule->evaluateProjectAll($basePath, Architecture::define());
+
+            $this->assertSame([], $violations);
+            $this->assertFalse($psr1PhpTagsRule->fix(new RuleViolation(
+                message:   'File must use only valid PHP tags',
+                file:      $basePath . '/src/sitemap.php',
+                line:      6,
+                className: '',
+            )));
+            $this->assertSame($contents, file_get_contents($basePath . '/src/sitemap.php'));
+        } finally {
+            unlink($basePath . '/src/sitemap.php');
+            rmdir($basePath . '/src');
+            rmdir($basePath);
+        }
+    }
+
     public function testIgnoresPhpTagsInsideStrings(): void
     {
         $basePath = $this->makeTempDir();