From 7e1e281af322d7764445aa66970413965d50e533 Mon Sep 17 00:00:00 2001
From: David Huggins-Daines <dhd@ecolingui.ca>
Date: Wed, 27 Nov 2024 10:40:59 -0500
Subject: [PATCH] fix: ensure that bytes is actually bytes (fixes: #1059)

---
 pdfminer/pdfdevice.py                      |   4 ++--
 samples/contrib/issue-1059-cmap-decode.pdf | Bin 0 -> 3785 bytes
 tests/test_tools_pdf2txt.py                |   6 ++++++
 3 files changed, 8 insertions(+), 2 deletions(-)
 create mode 100644 samples/contrib/issue-1059-cmap-decode.pdf

diff --git a/pdfminer/pdfdevice.py b/pdfminer/pdfdevice.py
index 2374601c..184922f9 100644
--- a/pdfminer/pdfdevice.py
+++ b/pdfminer/pdfdevice.py
@@ -167,7 +167,7 @@ def render_string_horizontal(
             if isinstance(obj, (int, float)):
                 x -= obj * dxscale
                 needcharspace = True
-            else:
+            elif isinstance(obj, bytes):
                 for cid in font.decode(obj):
                     if needcharspace:
                         x += charspace
@@ -207,7 +207,7 @@ def render_string_vertical(
             if isinstance(obj, (int, float)):
                 y -= obj * dxscale
                 needcharspace = True
-            else:
+            elif isinstance(obj, bytes):
                 for cid in font.decode(obj):
                     if needcharspace:
                         y += charspace
diff --git a/samples/contrib/issue-1059-cmap-decode.pdf b/samples/contrib/issue-1059-cmap-decode.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..452973df63a8c203031ad9106e24c70172eba900
GIT binary patch
literal 3785
zcmb7H*>c-96n*Dc;KdU;9gzTcW;~j)Eh$l3*N!AN4<0|z5+t#eNQH!E;{I3r0r|4t
zi;c9L#pc))4|or7a5s`CH^XtCxruY~_1~}mISwO~R1beSXJ_6`zUB?_VPNXbANSn5
z%tgMf)`Nk=%Ox5)cvD+nR)<m$=KWSIo4Wu>_lhN_>2UC#<&=V#q9IJJa#>gVi|WfA
z1(O)YlFYyNuK2RZpN-z+H88chDQ~Jnz2FU~9#^HnA4V8XGabO;O)HTLPH3yBBPi5@
zdXptgS%}9|yT8cXI*0x8c0AVOcm0UQr0>Tm!HX>N?=3gcHgLm^rH02SjX)rrkOvp~
zq)~!HiMf{|QY*kkhA7kbV2_N#N%0tN8crJdh<t+rL;<2!W2W_h!+z)GT5L!dQE*Ko
zYM$>L?{YGnl!YMPO<gT!T#!{!E^F;=^1#<ciLqN23!#x5TkP^ZLGtXe5qvi(R~0!s
zBi<AUF!dwpDSs00x?b|SDA%NSdpSEL-t4g7Z}|>GOUYnBmV5;hUFG{f@*O9hr1edl
z`oL)|FhwCBkinFgMYZJ3K40)UFV~!$Q8)(VYz)VMKnfq-&lKxnwb<k}SO=paSW_A>
z{0T-Qg?xoz3cMFcG{$)Wg$hXm4I+gkfd;Wcl0bt*AxWS?s*ohmAd~8}4%CV6v`$zY
zYD|6UbLkq~a#3;{mVLFy-N}E<JRMB9k~J&KLKomg-f%1!j1~U=GG^Y8Hw&=MOMxTJ
z9puKh%$`ISY-}X4oThE4v4o~i;tKT$<azaeGK9c-CYrKxaq+s?i>ihx&ryiLs)E+D
z|G<m&Mvyo#5``>0Z@kUd4T&ZF0{R5O;+#r|Nz>ih1o~I8I4{>*>;*GO%|8j2sqSyM
z-&O#~oE;v7Y<$W)m|?zZA<}7|2-Y)nn0s$jLNY-f+uig(ST@q2^HjHGff+}z=sGKQ
zIuVqv?TzOLv8n3bWOUo``qaCGf)qtn4xx|rhVOjpQ}~N$<|mZ(Y5IoJH>ciJwfwOG
z)ba8F1<|`I7IoEBEAfupKmiIeuLwK}PQ7^{w!C*yRe!V7uCz|8*R6u3K5Ki$XXmry
zCKzo>`sxDOH(Bf)=ErUnlO%BclmHAu7rL2(TLUxdrYQlaj1Am4pWZs9gOgHHubqpz
zBU^ju#+lE^d?g=Pup)Fm#Y|(IXufmqyo>Aix3haP|J51IosZwwmql*S>I2u}3`U=Z
zMi5iR2EIn0Vz&BJPx-k%KMq~i(U&k6o;OyXf<}-wHt-etH1nyx@!;qBUJ)EdQ#Vx6
zf!#yb?1wCYbbyB)Vum<vvHybAb=VmTTu3*YUK-o{VgYm0uuU+FU*RFMl%L>Km|`e8
zc!@|Adn-~Wl@)2Q6{+nFevL>?eJV11fk>SItw@zwtw@cHC}VH%D?}RVQ<2dNM6yn$
zvLL~hVk$wyjJZ?>8Z6EA#_?(B8q*!1xZ8s?G}g-2pp6s5G9<y|ie>u3<yK#^Kz22N
zXk*a^+8V45ORA11bt;G%DWNq+Qwxkz!!^?6R9hq5nk%Um0vo*1?G=(|Jmf>Ifg)?J
zj4j->eRlQs2MyDD0UW7jlpZnH#v{sC8i5xjR$fIS>NGyq$ajO*wV6hii3CeT0Paw;
zXy@n^G%`HKd^ZjyQ|(@1Bx@`?hmELeBwNSwL6hAmlL8IX248ONyMFxWeB(9jOU+l{
zZx%TCTB;5bM-e2$ir6ve`|=hk+p!oP#dvha7=yR2qcJ~%7meew^!a=Nz7QVS1>raB
z;^$-0bFTaxIhv1!@T@)_i;~WKj;Qlu%WM4ZoE86Y@QWuu8{{hyyiQgXd|bji0K9F%
U*GXOr>AxThGw0-FbUk+d2iZM|8UO$Q

literal 0
HcmV?d00001

diff --git a/tests/test_tools_pdf2txt.py b/tests/test_tools_pdf2txt.py
index e80a5e69..e1ae7a01 100644
--- a/tests/test_tools_pdf2txt.py
+++ b/tests/test_tools_pdf2txt.py
@@ -68,6 +68,12 @@ def test_contrib_issue_350(self):
         """
         run("contrib/issue-00352-asw-oct96-p41.pdf")
 
+    def test_contrib_issue_1059_textseq(self):
+        """Ensure that CMaps are robust to non-strings in text
+        sequences
+        (https://github.com/pdfminer/pdfminer.six/issues/1059)."""
+        run("contrib/issue-1059-cmap-decode.pdf")
+
     def test_scancode_patchelf(self):
         """Regression test for https://github.com/euske/pdfminer/issues/96"""
         run("scancode/patchelf.pdf")