From 53538eb33efc11fb1f81cef9729aed284d0a1dff Mon Sep 17 00:00:00 2001
From: Brian Huisman <bhuisman@greywyvern.com>
Date: Thu, 24 Aug 2023 10:14:51 -0400
Subject: [PATCH] Check for wrong line-endings when getting xref (#635)

If we didn't find the `xref` command at the offset specified, then replace Windows `\r\n` line endings with Unix style `\n` and try again. If it succeeds, then edit the line-endings and proceed as normal. Otherwise continue on to the `decodeXrefStream()` method.

Fixes parsing of existing test suite file **/samples/bugs/Issue95_ANSI.pdf** the test for which would normally be passed over because of the `@group linux-only` flag. Remove this flag, as all assertions in the `testDecodeText()` function now resolve as true in any environment.
---
 src/Smalot/PdfParser/RawData/RawDataParser.php | 17 +++++++++++++++--
 tests/PHPUnit/Integration/FontTest.php         |  3 ---
 2 files changed, 15 insertions(+), 5 deletions(-)

diff --git a/src/Smalot/PdfParser/RawData/RawDataParser.php b/src/Smalot/PdfParser/RawData/RawDataParser.php
index 1a4583c0..ec8f600b 100644
--- a/src/Smalot/PdfParser/RawData/RawDataParser.php
+++ b/src/Smalot/PdfParser/RawData/RawDataParser.php
@@ -901,8 +901,15 @@ protected function getXrefData(string $pdfData, int $offset = 0, array $xref = [
             // Cross-Reference
             $xref = $this->decodeXref($pdfData, $startxref, $xref);
         } else {
-            // Cross-Reference Stream
-            $xref = $this->decodeXrefStream($pdfData, $startxref, $xref);
+            // Check if the $pdfData might have the wrong line-endings
+            $pdfDataUnix = str_replace("\r\n", "\n", $pdfData);
+            if ($startxref < \strlen($pdfDataUnix) && strpos($pdfDataUnix, 'xref', $startxref) == $startxref) {
+                // Return Unix-line-ending flag
+                $xref = ['Unix' => true];
+            } else {
+                // Cross-Reference Stream
+                $xref = $this->decodeXrefStream($pdfData, $startxref, $xref);
+            }
         }
         if (empty($xref)) {
             throw new \Exception('Unable to find xref');
@@ -937,6 +944,12 @@ public function parseData(string $data): array
         // get xref and trailer data
         $xref = $this->getXrefData($pdfData);
 
+        // If we found Unix line-endings
+        if (isset($xref['Unix'])) {
+            $pdfData = str_replace("\r\n", "\n", $pdfData);
+            $xref = $this->getXrefData($pdfData);
+        }
+
         // parse all document objects
         $objects = [];
         foreach ($xref['xref'] as $obj => $offset) {
diff --git a/tests/PHPUnit/Integration/FontTest.php b/tests/PHPUnit/Integration/FontTest.php
index b07bbf76..599a4203 100644
--- a/tests/PHPUnit/Integration/FontTest.php
+++ b/tests/PHPUnit/Integration/FontTest.php
@@ -294,9 +294,6 @@ public function testDecodeUnicode(): void
         $this->assertEquals('AB', Font::decodeUnicode("\xFE\xFF\x00A\x00B"));
     }
 
-    /**
-     * @group linux-only
-     */
     public function testDecodeText(): void
     {
         $filename = $this->rootDir.'/samples/Document1_pdfcreator_nocompressed.pdf';