Validate target-language in XLF files

src/Parser/XliffParser.php

@@ -16,6 +16,9 @@
 use InvalidArgumentException;
 use SimpleXMLElement;
 
+use function preg_match;
+use function strtolower;
+
 /**
  * XliffParser.
  *
@@ -112,14 +115,58 @@ public static function getSupportedFileExtensions(): array
 
     public function getLanguage(): string
     {
-        if (preg_match(
-            '/^([a-z]{2})\./i',
-            $this->getFileName(),
-            $matches,
-        )) {
-            return $matches[1];
+        return $this->getLanguageFromFileName() ?? $this->getSourceLanguage();
+    }
+
+    public function isVersion2(): bool
+    {
+        return $this->isVersion2;
+    }
+
+    /**
+     * Extracts the expected locale from the filename, supporting both
+     * prefix convention (de.locallang.xlf, TYPO3 style) and
+     * suffix convention (messages.de.xlf, Symfony/Laravel style).
+     * Returns null if the filename carries no locale.
+     */
+    public function getLanguageFromFileName(): ?string
+    {
+        $fileName = $this->getFileName();
+
+        // Prefix convention: de.locallang.xlf, de_AT.locallang.xlf, de_DE.locallang.xlf
+        if (preg_match('/^([a-z]{2})(?:[-_][A-Z]{2})?\./i', $fileName, $matches)) {
+            return strtolower($matches[1]);
         }
 
+        // Suffix convention: messages.de.xlf, messages.de_AT.xlf, messages.de_DE.xlf
+        if (preg_match('/\.([a-z]{2})(?:[-_][A-Z]{2})?\.(?:xlf|xliff)$/i', $fileName, $matches)) {
+            return strtolower($matches[1]);
+        }
+
+        return null;
+    }
+
+    /**
+     * Returns the normalized target language declared in the XLIFF file, or null if not set.
+     * Region suffix is stripped to match getLanguageFromFileName() behavior (e.g. "de-AT" → "de").
+     * XLIFF 1.x: target-language attribute on <file>.
+     * XLIFF 2.x: trgLang attribute on <xliff>.
+     */
+    public function getTargetLanguage(): ?string
+    {
+        $lang = $this->isVersion2
+            ? (string) ($this->xml['trgLang'] ?? '')
+            : (string) ($this->xml->file['target-language'] ?? '');
+
+        if ('' === $lang) {
+            return null;
+        }
+
+        return strtolower((string) preg_replace('/[-_][^-_]+$/', '', $lang));
+    }
+
+    private function getSourceLanguage(): string
+    {
         if ($this->isVersion2) {
             return (string) ($this->xml['srcLang'] ?? '');
         }

src/Validator/XliffSchemaValidator.php

@@ -20,6 +20,7 @@
 use Symfony\Component\Translation\Util\XliffUtils;
 
 use function sprintf;
+use function strtolower;
 
 /**
  * XliffSchemaValidator.
@@ -31,40 +32,79 @@ class XliffSchemaValidator extends AbstractValidator implements ValidatorInterfa
 {
     public function processFile(ParserInterface $file): array
     {
-        try {
-            /*
-             * With XmlUtils::loadFile() we always get a strange symfony error related to global composer autoloading issue.
-             *      Call to undefined method Symfony\Component\Filesystem\Filesystem::readFile()
-             */
-            if (!file_exists($file->getFilePath())) {
-                $this->logger?->error('File does not exist: '.$file->getFileName());
-
-                return [];
-            }
+        /*
+         * With XmlUtils::loadFile() we always get a strange symfony error related to global composer autoloading issue.
+         *      Call to undefined method Symfony\Component\Filesystem\Filesystem::readFile()
+         */
+        if (!file_exists($file->getFilePath())) {
+            $this->logger?->error('File does not exist: '.$file->getFileName());
+
+            return [];
+        }
 
-            $fileContent = file_get_contents($file->getFilePath());
-            if (false === $fileContent) {
-                $this->logger?->error('Failed to read file: '.$file->getFileName());
+        $fileContent = file_get_contents($file->getFilePath());
+        if (false === $fileContent) {
+            $this->logger?->error('Failed to read file: '.$file->getFileName());
 
-                return [];
-            }
+            return [];
+        }
+
+        try {
             $dom = XmlUtils::parse($fileContent);
+        } catch (Exception $e) {
+            $this->logger?->error('Failed to parse XML: '.$e->getMessage());
+
+            return [];
+        }
+
+        // Schema validation — may throw for unsupported XLIFF versions (e.g. 2.x)
+        $errors = [];
+        try {
             $errors = XliffUtils::validateSchema($dom);
         } catch (Exception $e) {
             if (str_contains($e->getMessage(), 'No support implemented for loading XLIFF version')) {
                 $this->logger?->notice(sprintf('Skipping %s: %s', $this->getShortName(), $e->getMessage()));
             } else {
                 $this->logger?->error('Failed to validate XML schema: '.$e->getMessage());
             }
-
-            return [];
         }
 
-        if (!empty($errors)) {
+        // Additional check: if filename encodes a locale, verify it matches target-language in the file header
+        if (!$file instanceof XliffParser) {
             return $errors;
         }
 
-        return [];
+        $expectedLanguage = $file->getLanguageFromFileName();
+        if (null !== $expectedLanguage) {
+            $targetLang = $file->getTargetLanguage();
+            $isVersion2 = $file->isVersion2();
+            $attribute = $isVersion2 ? 'trgLang' : 'target-language';
+            $element = $isVersion2 ? '<xliff>' : '<file>';
+
+            if (null === $targetLang) {
+                $errors[] = [
+                    'message' => sprintf(
+                        'Missing "%s" attribute on %s node; expected "%s" based on filename',
+                        $attribute,
+                        $element,
+                        $expectedLanguage,
+                    ),
+                    'level' => 'ERROR',
+                ];
+            } elseif (strtolower($targetLang) !== $expectedLanguage) {
+                $errors[] = [
+                    'message' => sprintf(
+                        '"%s" attribute "%s" does not match filename language "%s"',
+                        $attribute,
+                        $targetLang,
+                        $expectedLanguage,
+                    ),
+                    'level' => 'ERROR',
+                ];
+            }
+        }
+
+        return $errors;
     }
 
     public function formatIssueMessage(Issue $issue, string $prefix = ''): string

tests/src/Parser/XliffParserTest.php

@@ -411,6 +411,114 @@ public function testGetLanguageFromSrcLangAttributeXliff2(): void
         $this->assertSame('en', $parser->getLanguage());
     }
 
+    public function testGetLanguageFromFileNamePrefixConvention(): void
+    {
+        $parser = new XliffParser($this->prefixedXliffFile); // de.messages.xlf
+        $this->assertSame('de', $parser->getLanguageFromFileName());
+    }
+
+    public function testGetLanguageFromFileNameSuffixConvention(): void
+    {
+        $parser = new XliffParser($this->targetLanguageXliffFile); // messages.de.xlf
+        $this->assertSame('de', $parser->getLanguageFromFileName());
+    }
+
+    public function testGetLanguageFromFileNamePrefixConventionWithRegion(): void
+    {
+        $file = $this->tempDir.'/de_DE.locallang.xlf';
+        file_put_contents($file, $this->prefixedXliffContent);
+
+        $parser = new XliffParser($file);
+        $this->assertSame('de', $parser->getLanguageFromFileName());
+    }
+
+    public function testGetLanguageFromFileNameSuffixConventionWithRegion(): void
+    {
+        $file = $this->tempDir.'/messages.de_DE.xlf';
+        file_put_contents($file, $this->targetLanguageXliffContent);
+
+        $parser = new XliffParser($file);
+        $this->assertSame('de', $parser->getLanguageFromFileName());
+    }
+
+    public function testGetLanguageFromFileNameReturnsNullForSourceFile(): void
+    {
+        $parser = new XliffParser($this->validXliffFile); // messages.xlf — no locale
+        $this->assertNull($parser->getLanguageFromFileName());
+    }
+
+    public function testGetLanguageFromFileNameReturnsNullForXliff2SourceFile(): void
+    {
+        $parser = new XliffParser($this->xliff2File); // messages_v2.xlf — no locale
+        $this->assertNull($parser->getLanguageFromFileName());
+    }
+
+    public function testGetTargetLanguageReturnsValueWhenSet(): void
+    {
+        $parser = new XliffParser($this->targetLanguageXliffFile); // target-language="de"
+        $this->assertSame('de', $parser->getTargetLanguage());
+    }
+
+    public function testGetTargetLanguageReturnsNullWhenNotSet(): void
+    {
+        $parser = new XliffParser($this->validXliffFile); // no target-language attribute
+        $this->assertNull($parser->getTargetLanguage());
+    }
+
+    public function testGetTargetLanguageXliff2ReturnsValueWhenSet(): void
+    {
+        $parser = new XliffParser($this->xliff2TargetFile); // trgLang="de"
+        $this->assertSame('de', $parser->getTargetLanguage());
+    }
+
+    public function testGetTargetLanguageXliff2ReturnsNullWhenNotSet(): void
+    {
+        $parser = new XliffParser($this->xliff2File); // no trgLang
+        $this->assertNull($parser->getTargetLanguage());
+    }
+
+    public function testGetTargetLanguageStripsRegionSuffixWithHyphen(): void
+    {
+        $file = $this->tempDir.'/region_hyphen.xlf';
+        file_put_contents($file, <<<'EOT'
+<?xml version="1.0" encoding="utf-8"?>
+<xliff xmlns="urn:oasis:names:tc:xliff:document:1.2" version="1.2">
+  <file source-language="en" target-language="de-AT" datatype="plaintext" original="region_hyphen.xlf">
+    <body><trans-unit id="k"><source>x</source></trans-unit></body>
+  </file>
+</xliff>
+EOT);
+        $this->assertSame('de', (new XliffParser($file))->getTargetLanguage());
+    }
+
+    public function testGetTargetLanguageStripsRegionSuffixWithUnderscore(): void
+    {
+        $file = $this->tempDir.'/region_underscore.xlf';
+        file_put_contents($file, <<<'EOT'
+<?xml version="1.0" encoding="utf-8"?>
+<xliff xmlns="urn:oasis:names:tc:xliff:document:1.2" version="1.2">
+  <file source-language="en" target-language="de_AT" datatype="plaintext" original="region_underscore.xlf">
+    <body><trans-unit id="k"><source>x</source></trans-unit></body>
+  </file>
+</xliff>
+EOT);
+        $this->assertSame('de', (new XliffParser($file))->getTargetLanguage());
+    }
+
+    public function testGetTargetLanguageXliff2StripsRegionSuffix(): void
+    {
+        $file = $this->tempDir.'/region_v2.xlf';
+        file_put_contents($file, <<<'EOT'
+<?xml version="1.0" encoding="utf-8"?>
+<xliff xmlns="urn:oasis:names:tc:xliff:document:2.0" version="2.0" srcLang="en" trgLang="de-AT">
+  <file id="messages">
+    <unit id="k"><segment><source>x</source></segment></unit>
+  </file>
+</xliff>
+EOT);
+        $this->assertSame('de', (new XliffParser($file))->getTargetLanguage());
+    }
+
     public function testGetContentByKeyFallsBackToSourceWhenTargetIsEmptyXliff2(): void
     {
         $fallbackFile = $this->tempDir.'/v2_fallback.xlf';