X-Git-Url: https://git.ucc.asn.au/?p=dja%2Fscandal.git;a=blobdiff_plain;f=scan2pages.sh;h=ff8db05fc4e839aaeb3e250f423946892c25c2a8;hp=662d87749a541ea3142f7aceafa0df76ad03903b;hb=850e3073ddd9b0dea8ecaf97ebc14430ff1c3698;hpb=eab68b895041b14605257517b214d14474037d4b

diff --git a/scan2pages.sh b/scan2pages.sh
index 662d877..ff8db05 100755
--- a/scan2pages.sh
+++ b/scan2pages.sh
@@ -7,7 +7,7 @@ Convert pdffile - a pdf of scanned facing pages - to a set of images in outdir,
 each with only one page.
 OPTIONS:
 	-v: Be verbose.
-	-d depth: Use given depth. Default is 1 for black and white, 8 for colour.
+	-d depth: Use given depth. Default is 1 for black and white, 8 for colour/gray.
 		Unpaper can only handle up to 8.
 	-D dpi: Use given DPI. Colour scans on library scanners are usually 300, bw 400.
 	-F extension: use the output format signified by the extension. Default is png
@@ -15,20 +15,14 @@ OPTIONS:
 	-f: Forceably redo everything.
 	-p: Forceably redo pdf conversion. Implies options below, equivalent to -f.
 	-m: Forceably redo masking/trimming and other preprocessing. Impiles options below.
-	-u: Forceably redo unpaper processing. Implies option below. IGNORED
+	-u: Forceably redo unpaper processing. Implies option below.
 	-t: Forceably redo final trimming and cleaning. IGNORED
 	-b: Optimize for BeBook One.
 __EOF__
 }
 
 ## Setup and utilities ##
-# Mac OS X doesn't have seq. It has jot instead.
-Linux_seq="seq"
-Darwin_seq="jot -"
-if [[ $(uname) == "Darwin" ]]; then my_seq=$Darwin_seq;
-else my_seq=$Linux_seq;
-fi;
-
+source $(dirname $0)/my_seq.sh
 
 ## Process flags ##
 forcepdf=
@@ -137,44 +131,68 @@ for pg in `$my_seq 1 $pages`; do
 	pgn=$(printf '%03d' $pg)
 	
 	# convert from pdf
-	origpnm=$dir/pg-${pgn}.pnm
-	if [[ ! $([ -e $origpnm ]) || $forcepdf ]]; then
+	origfile=$dir/pg-${pgn}.png
+	if [ ! -e $origfile ] || [[ $forcepdf ]]; then
 		convert $convertflags -depth $depth -density $dpi $file[$(expr $pg - 1)] \
-			$origpnm || exit 1
+			$origfile || exit 1
 	fi;
 	
 	# preprocess it!
 	preppnm=$dir/pg-pp-${pgn}.pnm
-	if [[ ! $skipmask && ( ! $([ -e $preppnm ]) || $forcemask ) ]]; then
+	if [[ ! $skipmask ]] && ( [ ! -e $preppnm ] || [[ $forcemask ]] ); then
 		# create mask: 
 		# ... downscale, blur,
-		convert $convertflags -resize 25% -depth 8 -blur 10 $origpnm $dir/pg-mask-${pgn}.pnm ||exit 1
+		convert $convertflags -resize 10% -depth 8 -blur 10 -median 2 $origfile $dir/pg-mask-${pgn}.png ||exit 1
 
 		# ... get crop co-ords. They're off by ~2 as I don't know how to
 		# properly correct for the border.
-		cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 400% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.pnm info: || exit 1)
+		cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 1000% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.png info: || exit 1)
 		
 		# ... crop and despeckle? the final pre-prepared image
-		convert $convertflags -crop $cropcords $origpnm $preppnm || exit 1
+		convert $convertflags -crop $cropcords $origfile $preppnm || exit 1
 	elif [[ $skipmask ]]; then
-		cp $origpnm $preppnm
+		cp $origfile $preppnm
+	fi;
+	
+	# check it hasn't mostly disappeared - e.g. if the scan was all black
+	# (e.g. forgot to put the book down when you first hit scan)
+	if [[ $(convert  $preppnm -format '%[fx:s.w*s.h>1000]' info:) = "0" ]]; then
+		[[ $verbose == 1 ]] && echo "Discarding pg ${pgn}: not enough remains after masking."
+		continue; 
 	fi;
 	
 	#unpaper it
 	#names go a bit funny here
-	#also, ignore flags starting here
-	unppnm=$dir/upg-${pgn}-%01d.pnm
-	unpaper $unpaperflags --layout double --overwrite --no-blackfilter -ni 10 -op 2 $preppnm $unppnm || exit 1
+	unppnmbase=$dir/upg-${pgn}
+	if [ ! -e ${unppnmbase}-1.pnm ] || [ ! -e ${unppnmbase}-2.pnm ] || [[ $forceunpaper ]]; then
+		unpaper $unpaperflags --layout double --overwrite -ni 10 -op 2 $preppnm ${unppnmbase}-%01d.pnm || exit 1
+	fi;
+	
+	#detect if the page is 2-up
+	for subpg in $($my_seq 1 2); do
+		echo Processing subpg ${subpg}.
+		if $(dirname $0)/detect2pages.sh $dir/upg-${pgn}-${subpg}.pnm ${pgn} ${subpg}; then
+			if [[ $verbose == 1 ]]; then
+				echo "Resplitting subpg ${subpg}."
+			fi
+			unpaper $unpaperflags --pre-rotate 90 --layout double --overwrite -op 2 $dir/upg-${pgn}-${subpg}.pnm $dir/upg-${pgn}-${subpg}-%01d.pnm
+		else
+			cp $dir/upg-${pgn}-${subpg}.pnm $dir/upg-${pgn}-${subpg}-1.pnm
+		fi;
+
+	done;
 
 	#final convert and clean w/ bebook optimisation
 	if [[ $bebook ]]; then
-		convert $convertflags $dir/upg-${pgn}-?.pnm -depth 4 -resize 600x800 $dir/final-${pgn}-%01d.${extension} || exit 1
+		convert $convertflags -colorspace Gray -median 1 $dir/upg-${pgn}-1-?.pnm -trim -resize 1800x2400 $dir/final-${pgn}-1-%01d.${extension} || exit 1
+		convert $convertflags -colorspace Gray -median 1 $dir/upg-${pgn}-2-?.pnm -trim -resize 1800x2400 $dir/final-${pgn}-2-%01d.${extension} || exit 1
 	else
-		convert $convertflags $dir/upg-${pgn}-?.pnm $dir/final-${pgn}-%01d.${extension} || exit 1
+		convert $convertflags $dir/upg-${pgn}-1-?.pnm $dir/final-${pgn}-1-%01d.${extension} || exit 1
+		convert $convertflags $dir/upg-${pgn}-2-?.pnm $dir/final-${pgn}-2-%01d.${extension} || exit 1
 	fi
 	
 done
 
-mkdir $dir/pages
+mkdir -p $dir/pages
 mv $dir/final-*.${extension} $dir/pages