From: Daniel Axtens <dja@ucc.gu.uwa.edu.au>
Date: Sat, 6 Aug 2011 04:59:28 +0000 (+0800)
Subject: Don't redo binarise if not needed.
X-Git-Url: https://git.ucc.asn.au/?a=commitdiff_plain;h=88f2331a5a94726bdf7c2b28f11c203a3f9ae15a;p=dja%2Fscandal.git

Don't redo binarise if not needed.
---

diff --git a/scan2pages.sh b/scan2pages.sh
index 069b421..cbfe52f 100755
--- a/scan2pages.sh
+++ b/scan2pages.sh
@@ -128,9 +128,18 @@ for scanpgnum in `$my_seq 1 $pages`; do
 done;
 
 # do ocr binarise
-[ -e $outdir/scanpgs ] && rm -r $outdir/scanpgs
-ocropus book2pages $outdir/scanpgs $outdir/scanpg-*.png || exit 1
+if [ -e $outdir/scanpgs ]; then
+	#assume (for the sake of speed in repeated runs) that
+	# if the last file exists, the process completed successfully
+	if [ -e $outdir/scanpgs/$(printf '%04d' $pages).bin.png ]; then
+		[[ $verbose ]] && echo Binarisation already complete
+	else
+		rm -r $outdir/scanpgs
+	fi
+fi
+[ -e $outdir/scanpgs ] || ocropus book2pages $outdir/scanpgs $outdir/scanpg-*.png || exit 1
 
+# process each binarised scan page
 for scanpgnum in  `$my_seq 1 $pages`; do
 
 	binscanpg=$outdir/scanpgs/$(printf '%04d' $scanpgnum).bin.png