X-Git-Url: https://git.ucc.asn.au/?p=dja%2Fscandal.git;a=blobdiff_plain;f=scan2pages.sh;h=025205df3ed758deace151548991dbeeba3005d9;hp=12f56709a31e87e8a9ab0cdf3788e85dda4f8671;hb=5a22e3f88799a29482d078ba4e8ca76c5f4af3cc;hpb=9c757d3063751e5f907c86241edfc1efa59bc6f4

diff --git a/scan2pages.sh b/scan2pages.sh
index 12f5670..025205d 100755
--- a/scan2pages.sh
+++ b/scan2pages.sh
@@ -1,42 +1,53 @@
 #!/bin/bash
 
-UNPAPER_PATH="/Users/dja/Applications/unpaper-0.3/"
-PATH=$UNPAPER_PATH:$PATH
-
 function usage {
 cat >&2 << __EOF__ 
-Usage: $0 [-vfpsmut] [-d depth] pdffile outdir
+Usage: $0 [-vfpsmutb] [-D DPI] [-d depth] [-F format] pdffile outdir
 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. Unpaper can only handle up to 8.
+	-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
 	-s: skip masking/trimming. Overrides -m.
 	-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 ##
+source $(dirname $0)/my_seq.sh
 
-# process for -f flag to forceably redo all conversions
+## Process flags ##
 forcepdf=
 forcemask=
 forceunpaper=
 forceclean=
 convertflags=
 unpaperflags=
-depthflags="-depth 1"
+depthset=
+depth=
+dpi=
 skipmask=
-while getopts 'vd:sfpmuc' OPTION
+verbose=
+extension="png"
+bebook=
+while getopts 'vd:D:sfpmucF:b' OPTION
 do
 	case $OPTION in
-	v)  convertflags="$convertflags -verbose"
+	v)	verbose=1
+		convertflags="$convertflags -verbose"
 		unpaperflags="$unpaperflags -v --time"
 		;;
-	d)	depthflags="-depth $OPTARG"
+	d)	depth="$OPTARG"
+		;;
+	D)	dpi=$OPTARG
 		;;
 	s)	skipmask=1;
 		;;
@@ -50,9 +61,10 @@ do
 		;;
 	t)	forceclean=1
 		;;
-#	b)	bflag=1
-#		bval="$OPTARG"
-#		;;
+	F)	extension="$OPTARG"
+		;;
+	b)	bebook=1
+		;;
 	?)	usage
 		exit 2
 		;;
@@ -83,28 +95,58 @@ pages=$(awk '$1 ~ "%%Pages" {print $2}' $dscname)
 echo "Got $pages page(s)."
 rm $dscname
 
+# guess at DPI, unless it's been set
+# the scans do not seem to accurately reflect their resolution.
+if [[ "$dpi" = "" ]]; then
+	# %k gives the number of colours/shades in the image.
+	# 2 for b/w, and the scanner seems to default to 400 dpi for that
+	# 256 for greyscale, which has to be selected manually (so I'm assuming 300 for no good reason)
+	# 31564 for colour, which defaults to 300
+	if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
+		dpi=400;
+	else
+		dpi=300
+	fi
+	if [[ $verbose == 1 ]]; then
+		echo "DPI guessed at $dpi."
+	fi
+fi
+
+# guess at depth, unless it's been set
+if [[ "$depth" = "" ]]; then
+	if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
+		depth=1;
+	else
+		depth=8
+	fi
+	if [[ $verbose == 1 ]]; then
+		echo "Depth guessed at $depth."
+	fi
+fi
+
 # process pages 1 by 1 to avoid convert gobbling all the memory
-for pg in `jot - 1 $pages`; do
+for pg in `$my_seq 1 $pages`; do
 	echo "Processing page $pg."
 	
 	pgn=$(printf '%03d' $pg)
 	
 	# convert from pdf
 	origpnm=$dir/pg-${pgn}.pnm
-	if [[ ! $([ -e $origpnm ]) || $forcepdf ]]; then
-		convert $convertflags $depthflags -density 300 $file[$(expr $pg - 1)] \
+	if [ ! -e $origpnm ] || [[ $forcepdf ]]; then
+		convert $convertflags -depth $depth -density $dpi $file[$(expr $pg - 1)] \
 			$origpnm || 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 $origpnm $dir/pg-mask-${pgn}.pnm ||exit 1
 
-		# ... get crop co-ords
-		cropcords=$(convert -resize 400% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.pnm info: || 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 1000% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.pnm info: || exit 1)
 		
 		# ... crop and despeckle? the final pre-prepared image
 		convert $convertflags -crop $cropcords $origpnm $preppnm || exit 1
@@ -112,18 +154,45 @@ for pg in `jot - 1 $pages`; do
 		cp $origpnm $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
-
-	# final convert and clean
-	convert $convertflags $dir/upg-${pgn}-?.pnm $dir/upg-${pgn}-%01d.tiff || exit 1
+	unppnmbase=$dir/upg-${pgn}
+	if [ ! -e ${unppnmbase}-1.pnm ] || [ ! -e ${unppnmbase}-2.pnm ] || [[ $forceunpaper ]]; then
+		unpaper $unpaperflags --layout double --overwrite --no-blackfilter -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 --no-blackfilter -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 -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}-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
-mv $dir/upg-*.tiff $dir/pages
+mkdir -p $dir/pages
+mv $dir/final-*.${extension} $dir/pages