Don't redo binarise if not needed.

[dja/scandal.git] / scan2pages.sh

diff --git a/scan2pages.sh b/scan2pages.sh
index dbd66ee..cbfe52f 100755 (executable)
--- 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' $scanpgnum).bin.png
@@ -162,7 +171,11 @@ for scanpgnum in  `$my_seq 1 $pages`; do
        #unpaper it
        physpgbase=$outdir/physpg-${scanpgnum}
        if [ ! -e ${physpgbase}-1.pnm ] || [ ! -e ${physpgbase}-2.pnm ]; then
-               unpaper $unpaperflags --layout double --overwrite -ni 10 -op 2 $cleanscanpg ${physpgbase}-%01d.pnm || exit 1
+               # most of unpaper's processing is redundant given ocropus, and somewhat too agressive
+               unpaper $unpaperflags --layout double --overwrite -op 2 \
+                       --no-processing \
+                       $cleanscanpg ${physpgbase}-%01d.pnm || exit 1
+
        fi;
        
        for physpgnum in $($my_seq 1 2); do