5 Usage: $0 [-vfpsmut] [-d depth] 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. Unpaper can only handle up to 8.
11 -D dpi: Use given DPI. Colour scans on library scanners are usually 300, bw 400.
12 -s: skip masking/trimming. Overrides -m.
13 -f: Forceably redo everything.
14 -p: Forceably redo pdf conversion. Implies options below, equivalent to -f.
15 -m: Forceably redo masking/trimming and other preprocessing. Impiles options below.
16 -u: Forceably redo unpaper processing. Implies option below. IGNORED
17 -t: Forceably redo final trimming and cleaning. IGNORED
21 ## Setup and utilities ##
22 # Mac OS X doesn't have seq. It has jot instead.
25 if [[ $(uname) == "Darwin" ]]; then my_seq=$Darwin_seq;
26 else my_seq=$Linux_seq;
41 while getopts 'vd:sfpmuc' OPTION
45 convertflags="$convertflags -verbose"
46 unpaperflags="$unpaperflags -v --time"
48 d) depthflags="-depth $OPTARG"
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
72 shift $(($OPTIND - 1))
74 # check we have an input and output!
75 if [[ $# != 2 ]]; then
76 echo "Wrong number of parameters (2 required, $# given: [$@])" >&2
85 base=$(basename $1 .pdf)
90 # figure out the number of pages
91 dscname=$dir/${base}.dsc
92 pdf2dsc $file $dscname || exit 1
93 pages=$(awk '$1 ~ "%%Pages" {print $2}' $dscname)
94 echo "Got $pages page(s)."
97 # guess at DPI, unless it's been set
98 # the scans do not seem to accurately reflect their resolution.
99 if [[ "$dpi" = "" ]]; then
100 # %k gives the number of colours/shades in the image.
101 # 2 for b/w, and the scanner seems to default to 400 dpi for that
102 # 256 for greyscale, which has to be selected manually (so I'm assuming 300 for no good reason)
103 # 31564 for colour, which defaults to 300
104 if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
109 if [[ $verbose == 1 ]]; then
110 echo "DPI guessed at $dpi."
114 # process pages 1 by 1 to avoid convert gobbling all the memory
115 for pg in `$my_seq 1 $pages`; do
116 echo "Processing page $pg."
118 pgn=$(printf '%03d' $pg)
121 origpnm=$dir/pg-${pgn}.pnm
122 if [[ ! $([ -e $origpnm ]) || $forcepdf ]]; then
123 convert $convertflags $depthflags -density $dpi $file[$(expr $pg - 1)] \
128 preppnm=$dir/pg-pp-${pgn}.pnm
129 if [[ ! $skipmask && ( ! $([ -e $preppnm ]) || $forcemask ) ]]; then
131 # ... downscale, blur,
132 convert $convertflags -resize 25% -depth 8 -blur 10 $origpnm $dir/pg-mask-${pgn}.pnm ||exit 1
134 # ... get crop co-ords. They're off by ~2 as I don't know how to
135 # properly correct for the border.
136 cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 400% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.pnm info: || exit 1)
138 # ... crop and despeckle? the final pre-prepared image
139 convert $convertflags -crop $cropcords $origpnm $preppnm || exit 1
140 elif [[ $skipmask ]]; then
145 #names go a bit funny here
146 #also, ignore flags starting here
147 unppnm=$dir/upg-${pgn}-%01d.pnm
148 unpaper $unpaperflags --layout double --overwrite --no-blackfilter -ni 10 -op 2 $preppnm $unppnm || exit 1
150 # final convert and clean
151 convert $convertflags $dir/upg-${pgn}-?.pnm $dir/upg-${pgn}-%01d.tiff || exit 1
157 mv $dir/upg-*.tiff $dir/pages