Always return UTF-8 from text string metadata

src/Document/Dictionary/DictionaryValue/TextString/TextStringValue.php

@@ -5,6 +5,7 @@
 
 use Override;
 use PrinsFrank\PdfParser\Document\Dictionary\DictionaryValue\DictionaryValue;
+use PrinsFrank\PdfParser\Document\Encoding\PDFDocEncoding;
 use PrinsFrank\PdfParser\Exception\ParseFailureException;
 
 /** @api */
@@ -15,42 +16,29 @@ public function __construct(
 
     /** @throws ParseFailureException */
     public function getText(): string {
-        if (str_starts_with($this->textStringValue, '(') && str_ends_with($this->textStringValue, ')')) {
-            $value = preg_replace_callback(
-                '/\\\\([0-7]{1,3})/',
-                fn(array $matches) => mb_chr((int) octdec($matches[1])),
-                substr($this->textStringValue, 1, -1),
+        if (str_starts_with($this->textStringValue, '/')) {
+            return preg_replace_callback(
+                '/#([0-9A-F]{2})/',
+                fn(array $matches) => chr((int) hexdec($matches[1])),
+                $this->textStringValue,
             ) ?? throw new ParseFailureException();
-
-            return str_replace(
-                ['\\\\', '\n', '\r', '\t', '\b', '\f', '\(', '\)'],
-                ['\\', "\n", "\r", "\t", "\x08", "\f", '(', ')'],
-                $value,
-            );
         }
 
-        if (str_starts_with($this->textStringValue, '<') && str_ends_with($this->textStringValue, '>')) {
-            $binaryValue = hex2bin(substr($this->textStringValue, 1, -1));
-            if ($binaryValue === false) {
-                throw new ParseFailureException('Invalid hex string');
-            }
+        $binaryValue = $this->getBinaryString();
 
-            if (str_starts_with($binaryValue, "\xFE\xFF")) {
-                return mb_convert_encoding(substr($binaryValue, 2), 'UTF-8', 'UTF-16BE');
-            }
+        if (str_starts_with($binaryValue, "\xFE\xFF")) {
+            return mb_convert_encoding(substr($binaryValue, 2), 'UTF-8', 'UTF-16BE');
+        }
 
-            return $binaryValue;
+        if (str_starts_with($binaryValue, "\xFF\xFE")) {
+            return mb_convert_encoding(substr($binaryValue, 2), 'UTF-8', 'UTF-16LE');
         }
 
-        if (str_starts_with($this->textStringValue, '/')) {
-            return preg_replace_callback(
-                '/#([0-9A-F]{2})/',
-                fn(array $matches) => chr((int) hexdec($matches[1])),
-                $this->textStringValue,
-            ) ?? throw new ParseFailureException();
+        if (str_starts_with($binaryValue, "\xEF\xBB\xBF")) {
+            return substr($binaryValue, 3);
         }
 
-        throw new ParseFailureException(sprintf('Unrecognized format %s', $this->textStringValue));
+        return PDFDocEncoding::textToUnicode($binaryValue);
     }
 
     public function getBinaryString(): string {

tests/Samples/files/issue-127/contents.yml

@@ -3,7 +3,7 @@ version: '2.0'
 userPassword: null
 ownerPassword: null
 fileEncryptionKey: null
-title: !!binary /v8AMQAuADIAIABUAGEAYgBlAGwAbABlAF8AMQBfAEsAQQBfADEAOQAtADAANwAwADUAOQ==
+title: '1.2 Tabelle_1_KA_19-07059'
 producer: 'Kofax Power PDF'
 author: 'Rehm, Martin (LSN)'
 creator: 'Kofax Power PDF'

tests/Samples/files/issue-235/contents.yml

@@ -6,7 +6,7 @@ fileEncryptionKey: null
 title: ''
 producer: 'Adobe PDF Library 20.1.50'
 author: null
-creator: !!binary QWNyb2JhdCBQREZNYWtlciAyMCBm/HIgV29yZA==
+creator: 'Acrobat PDFMaker 20 für Word'
 creationDate: 2025-04-10T10:26:21+02:00
 modificationDate: 2025-04-10T10:28:29+02:00
 pages:

tests/Samples/files/issue-291/contents.yml

@@ -4,9 +4,9 @@ userPassword: null
 ownerPassword: null
 fileEncryptionKey: null
 title: "The Netherlands courts have jurisdiction to hear a representative action concerning the alleged anticompetitive conduct of Apple in relation to its App Store aimed at the Netherlands market\r"
-producer: !!binary /v8ATQBpAGMAcgBvAHMAbwBmAHQArgAgAFcAbwByAGQAIAAyADAAMgAx
+producer: 'Microsoft® Word 2021'
 author: null
-creator: !!binary /v8ATQBpAGMAcgBvAHMAbwBmAHQArgAgAFcAbwByAGQAIAAyADAAMgAx
+creator: 'Microsoft® Word 2021'
 creationDate: 2025-12-01T16:55:05+01:00
 modificationDate: 2025-12-01T16:55:05+01:00
 pages:

tests/Unit/Document/Dictionary/DictionaryValue/TextString/TextStringValueTest.php

@@ -80,6 +80,56 @@ public function testGetTextWithOctalCharacters(): void {
         );
     }
 
+    /** @see 7.9.2.2 Text string type — UTF-16BE with a leading byte order mark */
+    public function testGetTextConvertsUTF16BEToUTF8(): void {
+        // "Tïtle" as UTF-16BE (FE FF BOM) written as a hex string
+        static::assertSame(
+            'Tïtle',
+            (new TextStringValue('<FEFF005400EF0074006C0065>'))->getText(),
+        );
+
+        // The same UTF-16BE bytes written as a literal string with octal escapes
+        static::assertSame(
+            'Tïtle',
+            (new TextStringValue("(\376\377\000T\000\357\000t\000l\000e)"))->getText(),
+        );
+    }
+
+    /** @see 7.9.2.2 Text string type — UTF-16LE with a leading byte order mark */
+    public function testGetTextConvertsUTF16LEToUTF8(): void {
+        static::assertSame(
+            'Tïtle',
+            (new TextStringValue('<FFFE5400EF0074006C006500>'))->getText(),
+        );
+    }
+
+    /** @see 7.9.2.2.1 Text string type — UTF-8 with a leading byte order mark (PDF 2.0) */
+    public function testGetTextStripsUTF8BOM(): void {
+        static::assertSame(
+            'Tïtle',
+            (new TextStringValue('<EFBBBF54C3AF746C65>'))->getText(),
+        );
+    }
+
+    /** @see 7.9.2.2 Text string type — PDFDocEncoding (no byte order mark) is normalized to valid UTF-8 */
+    public function testGetTextNormalizesPDFDocEncodingToUTF8(): void {
+        // 0xFC ("ü", shared with Latin-1) written as a literal octal escape and as a hex string
+        static::assertSame(
+            'für',
+            (new TextStringValue('(f\374r)'))->getText(),
+        );
+        static::assertSame(
+            'für',
+            (new TextStringValue('<66FC72>'))->getText(),
+        );
+
+        // 0x80 ("•") and 0xA0 ("€") sit in the range where PDFDocEncoding diverges from Latin-1
+        static::assertSame(
+            '•€',
+            (new TextStringValue('<80A0>'))->getText(),
+        );
+    }
+
     /** @see 7.3.5, table 4 */
     public function testGetTextLiteralNames(): void {
         static::assertSame(