From 0c48381bdc0bfd199166b721493ae8eb43556335 Mon Sep 17 00:00:00 2001
From: "B. Watson" <urchlay@slackware.uk>
Date: Fri, 3 Feb 2023 21:15:02 -0500
Subject: sbopkglint: check PDF files in doc dir

---
 sbopkglint.d/10-docs.t.sh | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

(limited to 'sbopkglint.d')

diff --git a/sbopkglint.d/10-docs.t.sh b/sbopkglint.d/10-docs.t.sh
index 1e9e02a..c6710b7 100644
--- a/sbopkglint.d/10-docs.t.sh
+++ b/sbopkglint.d/10-docs.t.sh
@@ -18,6 +18,10 @@
 # 444), and check that any +x file looks like some kind of script
 # (perl/shell/python/etc).
 
+# also, any PDF files found in the doc dir are checked with pdfinfo,
+# to make sure they're really PDFs and don't contain errors that
+# prevent them being displayed.
+
 DOCDIR=usr/doc/$PRGNAM-$VERSION
 
 # existence of the doc dir was already checked by a previous test,
@@ -50,6 +54,23 @@ if [ -d "$DOCDIR" ]; then
 			fi
 		done
 	fi
+
+	# PDF filenames often have spaces/punctuation, don't use find.
+	# This is not ideal (only goes 3 levels deep), but it's easy.
+	for i in $DOCDIR/*.pdf $DOCDIR/*/*.pdf $DOCDIR/*/*/*.pdf \
+	         $DOCDIR/*.PDF $DOCDIR/*/*.PDF $DOCDIR/*/*/*.PDF
+	do
+		if [ -e "$i" ]; then
+			# pdfinfo non-fatal errors don't result in non-zero exit status,
+			# but generally they do mean un-renderable PDFs, so check its stderr
+			# in addition to its exit status.
+			pdfinfo "$i" &> .tmp.pdfinfo
+			if [ "$?" != "0" ] ||  grep -q '^Syntax Error' .tmp.pdfinfo; then
+				warn "broken PDF file: $i"
+			fi
+			rm -f .tmp.pdfinfo
+		fi
+	done
 fi
 
 # allow /usr/doc/HTML to exist, though we don't check its contents here.
-- 
cgit v1.2.3