From d2256338b82730fef647a3cd8b65bc9040a7d73e Mon Sep 17 00:00:00 2001
From: Andrew Murray <radarhere@users.noreply.github.com>
Date: Thu, 20 Apr 2023 23:15:20 +1000
Subject: [PATCH] Use later value for duplicate xref entries

---
 Tests/images/duplicate_xref_entry.pdf | Bin 0 -> 3326 bytes
 Tests/test_pdfparser.py               |   6 ++++++
 src/PIL/PdfParser.py                  |   9 +++------
 3 files changed, 9 insertions(+), 6 deletions(-)
 create mode 100644 Tests/images/duplicate_xref_entry.pdf

diff --git a/Tests/images/duplicate_xref_entry.pdf b/Tests/images/duplicate_xref_entry.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..f57a57d61c6ad0649448af316f689db4e7f1322e
GIT binary patch
literal 3326
zcmeH}S#Z-v7{?{cr{WAa2{dFvvpL)(9cpQ1`C?K!#KsU>hQvt=rPBu=n?xDMN@V3?
zrVr4YH+r{EK<~cMmR?V#rRN(xfxeX9?GyB-54hh-K4N>ADRh|WbXZv$>)UU4zfY_G
z_mg)x7QYVrWZsL?8cFITgHlUqSjlG91%yQ(Ju+loMBs-qnleu`UPdBPQ&R&2yfC&j
zLy!<n>dCA+!3)F536e(v=uhw)HjrEf+<1Ch<!*TKLR6j0=?L?}jY=kuZbK3shHkG6
z#I*x=Zcj~-w9X5ob1B_ed1x9|<b@qdMxUfpZ6V6Uq%mn+$?8NyLM%EKBWX1QUW1CB
zA0}E<ot{y11lA7IAhN19I+IL88Zk1jq)B4)ATQiRauYCP(2pUUybe!hnlGFve95fb
z8Xt-?EXyz~98BR0CdxSNc8A^ObT}L?my>t5G`bra+-sVeSG25c>$rMtTYEd|@%5sv
zb=~dleWCt!B9>*jqc?JWSQ_y8Wr<G0x?C=IgS)k{u~q79@02dO72ae#P8t-)b}<$Y
z%X!$sF{T6Jw6R8GijB2!R-4`7bny+~aAYN8VL8rX<!m;q6}<0&e#YvtHFe^R_U4hK
zqieQB+Oz+lvwQIH(N&4#XL|f8ZLf=8eO2q4Ht#jPef`(UfnX>cSs#sUif`UBv~_gs
z=392`ymePPL&hg2m8rZwH@~pBwENDx?!M>V`|jWOz=IDx{K%t^J^sX@C!c!ynP;DS
z{)HD`dilsJufF#B8*jaR?45Vtd;fzEKl=ESPe1$o#K}{qzxw){Z@>HghaZ1B`|~fq
z{`UJHf1dk`=EX9cnHF5l#A@>LKwcKBm9si%UaVySPR?Vsbz*zd#t}zywz*5%<7^q+
zfAH{8SGPZLW>rc%&adu~PkYbO)QrsjSz>!HDYJ57mAp<dZjOb`<2=j_%(?UX`VOop
zoa8}9#W%1hC@Peqpb{jn&S_~79aZD7rlv=U4q6hZTm?fNQIM2SRF2`LP|YAJE-F&`
z6Sk3zk{nbQVWbEL_L(OPjH)>uq>@K|!@nqas$i;ijjE_d?$oguT~S4{zapYpI=RtG
zD^+gN(?~Fbi>YXYuTMN!nR-movjp{>FEBbom^`ERL#(ubDQYeTWeoMj)=Q$~7iCGr
zQb3eLyTD_cnz>+Sxn3=5WSkdKh&SV;R}>5c{6RF1N;VvTSd5-r*%pz*wKj~!dOtA$
z8(1?|j6`Z}7)O?k)wQIOh1yHEIiadLqD43Xl~CDDAXH%}H?AW3e2hKK>q$*F^1xCg
z!h5}&w#U7fQ0ei1>{Y|1x3;JROyAZjw$z}gZa^JVs$e9QVPs^UK(M3LObXR5c!fz9
zM%5K2RkQYg%p@w5f6XK+u8Uz3)J$3f&5zJ3Ce`YmdR`_d;bp+P{8Yjv0*Rtjp^^mO
zz6Mh01;_vmQ(DPLs@Q?kB|H6Nwu&tXpTnjN$OAh;8!FQchh?a>)ix>UuOpchia?32
z_QN4*8Oe-J5r;x%Jj>Sz1nc~yaCq4obrduJP{JUV=_e4GM+WpvBpSO@!$c_(4wy{*
E8-kd<b^rhX

literal 0
HcmV?d00001

diff --git a/Tests/test_pdfparser.py b/Tests/test_pdfparser.py
index 43e244c7b..105a838d9 100644
--- a/Tests/test_pdfparser.py
+++ b/Tests/test_pdfparser.py
@@ -117,3 +117,9 @@ def test_pdf_repr():
     assert pdf_repr(b"a)/b\\(c") == rb"(a\)/b\\\(c)"
     assert pdf_repr([123, True, {"a": PdfName(b"b")}]) == b"[ 123 true <<\n/a /b\n>> ]"
     assert pdf_repr(PdfBinary(b"\x90\x1F\xA0")) == b"<901FA0>"
+
+
+def test_duplicate_xref_entry():
+    pdf = PdfParser("Tests/images/duplicate_xref_entry.pdf")
+    assert pdf.xref_table.existing_entries[6][0] == 1197
+    pdf.close()
diff --git a/src/PIL/PdfParser.py b/src/PIL/PdfParser.py
index 1b3cb52a2..dc1012f54 100644
--- a/src/PIL/PdfParser.py
+++ b/src/PIL/PdfParser.py
@@ -957,14 +957,11 @@ class PdfParser:
                 check_format_condition(m, "xref entry not found")
                 offset = m.end()
                 is_free = m.group(3) == b"f"
-                generation = int(m.group(2))
                 if not is_free:
+                    generation = int(m.group(2))
                     new_entry = (int(m.group(1)), generation)
-                    check_format_condition(
-                        i not in self.xref_table or self.xref_table[i] == new_entry,
-                        "xref entry duplicated (and not identical)",
-                    )
-                    self.xref_table[i] = new_entry
+                    if i not in self.xref_table:
+                        self.xref_table[i] = new_entry
         return offset
 
     def read_indirect(self, ref, max_nesting=-1):