5 Usage: $0 [-vfpsmutb] [-D DPI] [-d depth] [-F format] pdffile outdir
6 Convert pdffile - a pdf of scanned facing pages - to a set of images in outdir,
7 each with only one page.
10 -d depth: Use given depth. Default is 1 for black and white, 8 for colour.
11 Unpaper can only handle up to 8.
12 -D dpi: Use given DPI. Colour scans on library scanners are usually 300, bw 400.
13 -F extension: use the output format signified by the extension. Default is png
14 -s: skip masking/trimming. Overrides -m.
15 -f: Forceably redo everything.
16 -p: Forceably redo pdf conversion. Implies options below, equivalent to -f.
17 -m: Forceably redo masking/trimming and other preprocessing. Impiles options below.
18 -u: Forceably redo unpaper processing. Implies option below. IGNORED
19 -t: Forceably redo final trimming and cleaning. IGNORED
20 -b: Optimize for BeBook One.
24 ## Setup and utilities ##
25 source $(dirname $0)/my_seq.sh
41 while getopts 'vd:D:sfpmucF:b' OPTION
45 convertflags="$convertflags -verbose"
46 unpaperflags="$unpaperflags -v --time"
54 f) forcepdf=1; forcemask=1; forceunpaper=1; forceclean=1
56 p) forcepdf=1; forcemask=1; forceunpaper=1; forceclean=1
58 m) forcemask=1; forceunpaper=1; forceclean=1
60 u) forceunpaper=1; forceclean=1
64 F) extension="$OPTARG"
73 shift $(($OPTIND - 1))
75 # check we have an input and output!
76 if [[ $# != 2 ]]; then
77 echo "Wrong number of parameters (2 required, $# given: [$@])" >&2
86 base=$(basename $1 .pdf)
91 # figure out the number of pages
92 dscname=$dir/${base}.dsc
93 pdf2dsc $file $dscname || exit 1
94 pages=$(awk '$1 ~ "%%Pages" {print $2}' $dscname)
95 echo "Got $pages page(s)."
98 # guess at DPI, unless it's been set
99 # the scans do not seem to accurately reflect their resolution.
100 if [[ "$dpi" = "" ]]; then
101 # %k gives the number of colours/shades in the image.
102 # 2 for b/w, and the scanner seems to default to 400 dpi for that
103 # 256 for greyscale, which has to be selected manually (so I'm assuming 300 for no good reason)
104 # 31564 for colour, which defaults to 300
105 if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
110 if [[ $verbose == 1 ]]; then
111 echo "DPI guessed at $dpi."
115 # guess at depth, unless it's been set
116 if [[ "$depth" = "" ]]; then
117 if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
122 if [[ $verbose == 1 ]]; then
123 echo "Depth guessed at $depth."
127 # process pages 1 by 1 to avoid convert gobbling all the memory
128 for pg in `$my_seq 1 $pages`; do
129 echo "Processing page $pg."
131 pgn=$(printf '%03d' $pg)
134 origpnm=$dir/pg-${pgn}.pnm
135 if [[ ! $([ -e $origpnm ]) || $forcepdf ]]; then
136 convert $convertflags -depth $depth -density $dpi $file[$(expr $pg - 1)] \
141 preppnm=$dir/pg-pp-${pgn}.pnm
142 if [[ ! $skipmask && ( ! $([ -e $preppnm ]) || $forcemask ) ]]; then
144 # ... downscale, blur,
145 convert $convertflags -resize 10% -depth 8 -blur 10 -median 2 $origpnm $dir/pg-mask-${pgn}.pnm ||exit 1
147 # ... get crop co-ords. They're off by ~2 as I don't know how to
148 # properly correct for the border.
149 cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 1000% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.pnm info: || exit 1)
151 # ... crop and despeckle? the final pre-prepared image
152 convert $convertflags -crop $cropcords $origpnm $preppnm || exit 1
153 elif [[ $skipmask ]]; then
157 # check it hasn't mostly disappeared - e.g. if the scan was all black
158 # (e.g. forgot to put the book down when you first hit scan)
159 if [[ $(convert $preppnm -format '%[fx:s.w*s.h>1000]' info:) = "0" ]]; then
160 [[ $verbose == 1 ]] && echo "Discarding pg ${pgn}: not enough remains after masking."
165 #names go a bit funny here
166 #also, ignore flags starting here
167 unppnm=$dir/upg-${pgn}-%01d.pnm
168 unpaper $unpaperflags --layout double --overwrite --no-blackfilter -ni 10 -op 2 $preppnm $unppnm || exit 1
170 #detect if the page is 2-up
171 for subpg in $($my_seq 1 2); do
172 echo Processing subpg ${subpg}.
173 if $(dirname $0)/detect2pages.sh $dir/upg-${pgn}-${subpg}.pnm ${pgn} ${subpg}; then
174 if [[ $verbose == 1 ]]; then
175 echo "Resplitting subpg ${subpg}."
177 unpaper $unpaperflags --pre-rotate 90 --layout double --overwrite --no-blackfilter -op 2 $dir/upg-${pgn}-${subpg}.pnm $dir/upg-${pgn}-${subpg}-%01d.pnm
179 cp $dir/upg-${pgn}-${subpg}.pnm $dir/upg-${pgn}-${subpg}-1.pnm
184 #final convert and clean w/ bebook optimisation
185 if [[ $bebook ]]; then
186 convert $convertflags -colorspace Gray -median 1 $dir/upg-${pgn}-1-?.pnm -trim -resize 1800x2400 $dir/final-${pgn}-1-%01d.${extension} || exit 1
187 convert $convertflags -colorspace Gray -median 1 $dir/upg-${pgn}-2-?.pnm -trim -resize 1800x2400 $dir/final-${pgn}-2-%01d.${extension} || exit 1
189 convert $convertflags $dir/upg-${pgn}-1-?.pnm $dir/final-${pgn}-1-%01d.${extension} || exit 1
190 convert $convertflags $dir/upg-${pgn}-2-?.pnm $dir/final-${pgn}-2-%01d.${extension} || exit 1
196 mv $dir/final-*.${extension} $dir/pages