From eea6e429dd86b4cd0d23808348548c7173e78698 Mon Sep 17 00:00:00 2001
From: Vitor Mattos <1079143+vitormattos@users.noreply.github.com>
Date: Fri, 24 Apr 2026 00:37:23 -0300
Subject: [PATCH] fix: support multi-space object headers

Signed-off-by: Vitor Mattos <1079143+vitormattos@users.noreply.github.com>
---
 .../PullRequestNearbyObjectHeaderOffset.pdf   | Bin 0 -> 2540 bytes
 .../PdfParser/RawData/RawDataParser.php       |  27 +++++++++++++++---
 .../Integration/RawData/RawDataParserTest.php |   3 ++
 3 files changed, 26 insertions(+), 4 deletions(-)
 create mode 100644 samples/bugs/rawdata/PullRequestNearbyObjectHeaderOffset.pdf

diff --git a/samples/bugs/rawdata/PullRequestNearbyObjectHeaderOffset.pdf b/samples/bugs/rawdata/PullRequestNearbyObjectHeaderOffset.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..950fb8f57140569b5abc17603fda6c0b86085e29
GIT binary patch
literal 2540
zcmbtWU2fwx5N^>o0lfp^MSui_Oj5QKNpQ4??HF4$-ULaC76|%4OSGe1TNFsTQF4~v
zsSmwDdyx($S(Y8O*rIwcs87R>-wfx+C%4m?P5VRZ<j=o;|HGmbA%eu>lQkMyc$Tg;
zmk7_+O!Fy^MZzt7&9jw$LcSG6meP{5jaB^i%~$YzZSgFrAy^p}I%xAfyyP6=IoB*<
znpyaMyU-={^Za^G><2CC(fcB2@n^2lg0Ir-qSr%da?$(Aos)S!;ZNy@SIO^RKR(2-
zpW`6uy^kzwRD{JQ-@q}b*sL=Z78gARa(D=91LIy~p^;3M;oWpr+rZmJ?@8-C#Q6F7
zx&J)q3%SBH2m(wT>^L@9*y<(Itgtin&QMewQ?6o}<~kJ_GMsE7w%Q1-4$qS%Gm=}}
zEKN5hyZMGQh0~;z1ca4!m@}y(I>PN&IIsC0i)<<6hGj92#*JAUHR`aD(~|ceBsY39
z)h}b$kHyBw!L&bUi4AY1NVYMT(ZK$|)76t!rw{Wp1ZjnKpAhYz_etMHr>z(Om92$X
zD?@x%y0GW^KjPu)yTunQ2k~Sl9_+#iq^}n80d#ceufKkg{A0HAvKo+2p@1TX5Z@+_
z?f4IrgoK3D{|<f$8Q~+#X7^kq>GI`EcDF45A$nChK*mQF9JnhHcxg%uiuR%>%NnxY
z)^bNF-?9~Sgj*(grjY}k<IchpDe|!>&?A9&H*k?Zbbhg*>XVdp;CEb!t&F)s&~c!p
z;EBjI&?yAH$PfZnHxuV=FEIj?g?~sBg&qgS9*&xbEy$s*os=AGP4$|iQ(oje);vKQ
zt`zWu=w|T=5R`2<3ut>ROVEmET(YvXYcvxYDGv1sH=3Zu3sPye0=nm?RC8NFv*P`;
zCMIqblUvDuMM1T$)RC6D;NQeauC{9(B4{G(ZeIV87k}ghcyFTL4xZm}NSY~aS_op@
z%;#Oql^gq8Iqx9WB2QyPXg96bMO%5Xr}a8$<2BnVjy#*%{u}CDTr2-8>3cf2gARHW
zzC@ZUxX~@#UWif!eOE}kJi*q4+9vFFD?G!tcGrwZ3EOLrt{JS~M1FX1bUE*CwtAgr
zCUbYWS`*dzQW90`y$c7Uxf-Pt7^7>kGU*0&u&clwJF3Or<UG5K%OtL91SWc_p-{}O
zYxV-;3gP=QjdumT;?PfER{yB@j?I*CHMkK>^~2;UBwtzt(V<1^hdRh}-GPgi-_{rn
zJlJL)lr<QDZ!qG)<sI4i0W^an?A(K#Il??2((@Q2(3y@f*SK(m4MFoU=C}~ULpwq!
z+{`fsUR7A*FCozT8api^k+>}XR^6ws+{6lhKShrta^-l=l~0GRcNt7B2h<ySlfi5}
z9(Y&8Cp1E%=@lJZ2GsG!#HH?RJekZOv_pSJhOn2C2=4MuKx1k;Ce*dm<*hiB`qs(G
I)y>TM8|$;wYXATM

literal 0
HcmV?d00001

diff --git a/src/Smalot/PdfParser/RawData/RawDataParser.php b/src/Smalot/PdfParser/RawData/RawDataParser.php
index 6b1b7ea5a..7133b62ba 100644
--- a/src/Smalot/PdfParser/RawData/RawDataParser.php
+++ b/src/Smalot/PdfParser/RawData/RawDataParser.php
@@ -534,7 +534,7 @@ protected function decodeXrefStream(string $pdfData, int $startxref, array $xref
     protected function getObjectHeaderPattern(array $objRefs): string
     {
         // consider all whitespace character (PDF specifications)
-        return '/'.$objRefs[0].$this->config->getPdfWhitespacesRegex().$objRefs[1].$this->config->getPdfWhitespacesRegex().'obj/';
+        return '/'.$objRefs[0].$this->config->getPdfWhitespacesRegex().'+'.$objRefs[1].$this->config->getPdfWhitespacesRegex().'+obj/';
     }
 
     protected function getObjectHeaderLen(array $objRefs): int
@@ -567,6 +567,7 @@ protected function getIndirectObject(string $pdfData, array $xref, string $objRe
             throw new \Exception('Invalid object reference for $obj.');
         }
 
+        $objHeaderPattern = $this->getObjectHeaderPattern($objRefArr);
         $objHeaderLen = $this->getObjectHeaderLen($objRefArr);
 
         /*
@@ -576,9 +577,27 @@ protected function getIndirectObject(string $pdfData, array $xref, string $objRe
         $offset += strspn($pdfData, $this->config->getPdfWhitespaces(), $offset);
         // ignore leading zeros for object number
         $offset += strspn($pdfData, '0', $offset);
-        if (0 == preg_match($this->getObjectHeaderPattern($objRefArr), substr($pdfData, $offset, $objHeaderLen))) {
-            // an indirect reference to an undefined object shall be considered a reference to the null object
-            return ['null', 'null', $offset];
+        if (0 == preg_match($objHeaderPattern, substr($pdfData, $offset, 33), $headerMatches)) {
+            // Some malformed files have slightly inaccurate xref offsets.
+            // Try to recover by locating the expected object header nearby.
+            $searchStart = max(0, $offset - 64);
+            $searchLen = 192;
+            if (
+                preg_match(
+                    $objHeaderPattern,
+                    substr($pdfData, $searchStart, $searchLen),
+                    $headerMatches,
+                    \PREG_OFFSET_CAPTURE
+                ) > 0
+            ) {
+                $offset = $searchStart + $headerMatches[0][1];
+                $objHeaderLen = \strlen($headerMatches[0][0]);
+            } else {
+                // an indirect reference to an undefined object shall be considered a reference to the null object
+                return ['null', 'null', $offset];
+            }
+        } else {
+            $objHeaderLen = \strlen($headerMatches[0]);
         }
 
         /*
diff --git a/tests/PHPUnit/Integration/RawData/RawDataParserTest.php b/tests/PHPUnit/Integration/RawData/RawDataParserTest.php
index c13759770..ac3a52c75 100644
--- a/tests/PHPUnit/Integration/RawData/RawDataParserTest.php
+++ b/tests/PHPUnit/Integration/RawData/RawDataParserTest.php
@@ -334,6 +334,9 @@ public static function provideRawDataRegressionFixtures(): iterable
         yield 'pr799 xref subsection with multiple spaces' => [
             'rawdata/PullRequestXrefSubsectionMultipleSpaces.pdf',
         ];
+        yield 'pr800 object header with multiple spaces (nearby xref offset)' => [
+            'rawdata/PullRequestNearbyObjectHeaderOffset.pdf',
+        ];
     }
 
     /**