From 3ef10b1ea817033aef3cf0e1aad21561b15d2475 Mon Sep 17 00:00:00 2001
From: David Huggins-Daines <dhd@ecolingui.ca>
Date: Wed, 27 Nov 2024 11:28:41 -0500
Subject: [PATCH] Verify that we don't have pdfminer.six#1059 (and warn about
 it) (#23)

* test: verify that we don not have pdfminer.six#1059 (and warn about it)

* fix: warn rather than throwing TypeError on bogus text objects
---
 playa/cli.py                               |   7 +++++++
 playa/page.py                              |   4 +++-
 samples/contrib/issue-1059-cmap-decode.pdf | Bin 0 -> 3785 bytes
 3 files changed, 10 insertions(+), 1 deletion(-)
 create mode 100644 samples/contrib/issue-1059-cmap-decode.pdf

diff --git a/playa/cli.py b/playa/cli.py
index d5ba69ca..21da3ad9 100644
--- a/playa/cli.py
+++ b/playa/cli.py
@@ -3,6 +3,7 @@
 """
 
 import argparse
+import logging
 import csv
 from pathlib import Path
 
@@ -26,12 +27,18 @@ def make_argparse() -> argparse.ArgumentParser:
         choices=["screen", "page", "user"],
         default="screen",
     )
+    parser.add_argument(
+        "--debug",
+        help="Very verbose debugging output",
+        action="store_true",
+    )
     return parser
 
 
 def main() -> None:
     parser = make_argparse()
     args = parser.parse_args()
+    logging.basicConfig(level=logging.DEBUG if args.debug else logging.WARNING)
     writer = csv.DictWriter(args.outfile, fieldnames=playa.fieldnames)
     writer.writeheader()
     for path in args.pdfs:
diff --git a/playa/page.py b/playa/page.py
index 5a0473a7..61b71525 100644
--- a/playa/page.py
+++ b/playa/page.py
@@ -1640,6 +1640,7 @@ def render_string(
                 if isinstance(obj, str):
                     obj = make_compat_bytes(obj)
                 if not isinstance(obj, bytes):
+                    log.warning("Found non-string %r in text object", obj)
                     continue
                 for cid in self.textstate.font.decode(obj):
                     if needcharspace:
@@ -2333,7 +2334,8 @@ def do_TJ(self, strings: PDFObject) -> None:
         positioning"""
         args = list_value(strings)
         if not all(isinstance(s, (int, float, bytes)) for s in args):
-            raise TypeError("TJ takes only strings and numbers, not %r" % args)
+            log.warning("Found non-string in text object %r", args)
+            return
         self.textobj.append(make_txt("TJ", *args))
 
     def do_Tj(self, s: PDFObject) -> None:
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