Fixed importing of links in none PdfDocEncoding

Setasign · Sep 25, 2024 · fbd3570 · fbd3570
1 parent 04065f3
commit fbd3570
Show file tree

Hide file tree

Showing 5 changed files with 47 additions and 1 deletion.
diff --git a/src/FpdfTrait.php b/src/FpdfTrait.php
@@ -140,8 +140,8 @@ protected function _putlinks($n)
             $rect = sprintf('%.2F %.2F %.2F %.2F', $pl[0], $pl[1], $pl[0] + $pl[2], $pl[1] - $pl[3]);
             $this->_put('<</Type /Annot /Subtype /Link /Rect [' . $rect . ']', false);
             if (is_string($pl[4])) {
-                $this->_put('/A <</S /URI /URI ' . $this->_textstring($pl[4]) . '>>');
                 if (isset($pl['importedLink'])) {
+                    $this->_put('/A <</S /URI /URI (' . $this->_escape($pl[4]) . ')>>');
                     $values = $pl['importedLink']['pdfObject']->value;
 
                     foreach ($values as $name => $entry) {
@@ -158,6 +158,7 @@ protected function _putlinks($n)
                         $this->_put($s);
                     }
                 } else {
+                    $this->_put('/A <</S /URI /URI ' . $this->_textstring($pl[4]) . '>>');
                     $this->_put('/Border [0 0 0]', false);
                 }
                 $this->_put('>>');

diff --git a/src/PdfReader/Page.php b/src/PdfReader/Page.php
@@ -281,6 +281,8 @@ public function getContentStream()
      * All coordinates are normalized in view to rotation and translation of the boundary-box, so that their
      * origin is lower-left.
      *
+     * The URI is the binary value of the PDF string object. It can be in PdfDocEncoding or in UTF-16BE encoding.
+     *
      * @return array
      */
     public function getExternalLinks($box = PageBoundaries::CROP_BOX)

diff --git a/tests/_files/pdfs/links/tuto6.pdf b/tests/_files/pdfs/links/tuto6.pdf
diff --git a/tests/functional/LinkHandling/AbstractTest.php b/tests/functional/LinkHandling/AbstractTest.php
@@ -741,4 +741,31 @@ public function testImportOfSpecialPageBoundaries()
         $reader = new PdfReader(new PdfParser(StreamReader::createByString($pdfString)));
         $this->compareExpectedLinks(1, $expectedLinks, $reader);
     }
+
+    public function testLinkInUtf16Encoding()
+    {
+        $pdf = $this->getInstance();
+        $pdf->AddPage();
+        // This file has its link in UTF-16BE saved.
+        $pdf->setSourceFile(__DIR__ . '/../../_files/pdfs/links/tuto6.pdf');
+        $tplId = $pdf->importPage(2, PageBoundaries::CROP_BOX, true, true);
+        $pdf->useTemplate($tplId);
+        $pdfString = $this->save($pdf);
+//        file_put_contents(__DIR__ . '/test.pdf', $pdfString);
+
+        $expectedLinks = [
+            [
+                // the strings are in UTF-16BE: http://pdf.wtf/ümlaut
+                'uri' => "\xFE\xFF\x00h\x00t\x00t\x00p\x00:\x00/\x00/\x00p\x00d\x00f\x00.\x00w\x00t\x00f\x00/\x00\xfc\x00m\x00l\x00a\x00u\x00t",
+                'rect' => [28.35, 749.82, 113.39, 807.87],
+            ],
+            [
+                'uri' => "\xFE\xFF\x00h\x00t\x00t\x00p\x00:\x00/\x00/\x00p\x00d\x00f\x00.\x00w\x00t\x00f\x00/\x00\xfc\x00m\x00l\x00a\x00u\x00t",
+                'rect' => [387.18, 756.93, 468.87, 770.93],
+            ],
+        ];
+
+        $reader = new PdfReader(new PdfParser(StreamReader::createByString($pdfString)));
+        $this->compareExpectedLinks(1, $expectedLinks, $reader);
+    }
 }
diff --git a/tests/functional/PdfReader/PageTest.php b/tests/functional/PdfReader/PageTest.php
@@ -92,6 +92,22 @@ public function getExternalLinksProvider()
                 [
                     1 => []
                 ]
+            ],
+            [
+                __DIR__ . '/../../_files/pdfs/links/tuto6.pdf',
+                [
+                    2 => [
+                        [
+                            // the strings are in UTF-16BE: http://pdf.wtf/ümlaut
+                            'uri' => "\xFE\xFF\x00h\x00t\x00t\x00p\x00:\x00/\x00/\x00p\x00d\x00f\x00.\x00w\x00t\x00f\x00/\x00\xfc\x00m\x00l\x00a\x00u\x00t",
+                            'rect' => new Rectangle(28.35, 807.87, 113.39, 749.82)
+                        ],
+                        [
+                            'uri' => "\xFE\xFF\x00h\x00t\x00t\x00p\x00:\x00/\x00/\x00p\x00d\x00f\x00.\x00w\x00t\x00f\x00/\x00\xfc\x00m\x00l\x00a\x00u\x00t",
+                            'rect' => new Rectangle(387.18, 770.93, 468.87, 756.93)
+                        ],
+                    ]
+                ]
             ]
         ];
     }