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 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 -s: skip masking/trimming. Overrides -m.
14 -f: Forceably redo everything.
15 -p: Forceably redo pdf conversion. Implies options below, equivalent to -f.
16 -m: Forceably redo masking/trimming and other preprocessing. Impiles options below.
17 -u: Forceably redo unpaper processing. Implies option below. IGNORED
18 -t: Forceably redo final trimming and cleaning. IGNORED
22 ## Setup and utilities ##
23 # Mac OS X doesn't have seq. It has jot instead.
26 if [[ $(uname) == "Darwin" ]]; then my_seq=$Darwin_seq;
27 else my_seq=$Linux_seq;
43 while getopts 'vd:sfpmuc' OPTION
47 convertflags="$convertflags -verbose"
48 unpaperflags="$unpaperflags -v --time"
56 f) forcepdf=1; forcemask=1; forceunpaper=1; forceclean=1
58 p) forcepdf=1; forcemask=1; forceunpaper=1; forceclean=1
60 m) forcemask=1; forceunpaper=1; forceclean=1
62 u) forceunpaper=1; forceclean=1
74 shift $(($OPTIND - 1))
76 # check we have an input and output!
77 if [[ $# != 2 ]]; then
78 echo "Wrong number of parameters (2 required, $# given: [$@])" >&2
87 base=$(basename $1 .pdf)
92 # figure out the number of pages
93 dscname=$dir/${base}.dsc
94 pdf2dsc $file $dscname || exit 1
95 pages=$(awk '$1 ~ "%%Pages" {print $2}' $dscname)
96 echo "Got $pages page(s)."
99 # guess at DPI, unless it's been set
100 # the scans do not seem to accurately reflect their resolution.
101 if [[ "$dpi" = "" ]]; then
102 # %k gives the number of colours/shades in the image.
103 # 2 for b/w, and the scanner seems to default to 400 dpi for that
104 # 256 for greyscale, which has to be selected manually (so I'm assuming 300 for no good reason)
105 # 31564 for colour, which defaults to 300
106 if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
111 if [[ $verbose == 1 ]]; then
112 echo "DPI guessed at $dpi."
116 # guess at depth, unless it's been set
117 if [[ "$depth" = "" ]]; then
118 if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
123 if [[ $verbose == 1 ]]; then
124 echo "Depth guessed at $depth."
128 # process pages 1 by 1 to avoid convert gobbling all the memory
129 for pg in `$my_seq 1 $pages`; do
130 echo "Processing page $pg."
132 pgn=$(printf '%03d' $pg)
135 origpnm=$dir/pg-${pgn}.pnm
136 if [[ ! $([ -e $origpnm ]) || $forcepdf ]]; then
137 convert $convertflags -depth $depth -density $dpi $file[$(expr $pg - 1)] \
142 preppnm=$dir/pg-pp-${pgn}.pnm
143 if [[ ! $skipmask && ( ! $([ -e $preppnm ]) || $forcemask ) ]]; then
145 # ... downscale, blur,
146 convert $convertflags -resize 25% -depth 8 -blur 10 $origpnm $dir/pg-mask-${pgn}.pnm ||exit 1
148 # ... get crop co-ords. They're off by ~2 as I don't know how to
149 # properly correct for the border.
150 cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 400% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.pnm info: || exit 1)
152 # ... crop and despeckle? the final pre-prepared image
153 convert $convertflags -crop $cropcords $origpnm $preppnm || exit 1
154 elif [[ $skipmask ]]; then
159 #names go a bit funny here
160 #also, ignore flags starting here
161 unppnm=$dir/upg-${pgn}-%01d.pnm
162 unpaper $unpaperflags --layout double --overwrite --no-blackfilter -ni 10 -op 2 $preppnm $unppnm || exit 1
164 # final convert and clean
165 convert $convertflags $dir/upg-${pgn}-?.pnm $dir/upg-${pgn}-%01d.tiff || exit 1
171 mv $dir/upg-*.tiff $dir/pages