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 -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
23 ## Setup and utilities ##
24 # Mac OS X doesn't have seq. It has jot instead.
27 if [[ $(uname) == "Darwin" ]]; then my_seq=$Darwin_seq;
28 else my_seq=$Linux_seq;
45 while getopts 'vd:sfpmucF:' OPTION
49 convertflags="$convertflags -verbose"
50 unpaperflags="$unpaperflags -v --time"
58 f) forcepdf=1; forcemask=1; forceunpaper=1; forceclean=1
60 p) forcepdf=1; forcemask=1; forceunpaper=1; forceclean=1
62 m) forcemask=1; forceunpaper=1; forceclean=1
64 u) forceunpaper=1; forceclean=1
68 F) extension="$OPTARG"
75 shift $(($OPTIND - 1))
77 # check we have an input and output!
78 if [[ $# != 2 ]]; then
79 echo "Wrong number of parameters (2 required, $# given: [$@])" >&2
88 base=$(basename $1 .pdf)
93 # figure out the number of pages
94 dscname=$dir/${base}.dsc
95 pdf2dsc $file $dscname || exit 1
96 pages=$(awk '$1 ~ "%%Pages" {print $2}' $dscname)
97 echo "Got $pages page(s)."
100 # guess at DPI, unless it's been set
101 # the scans do not seem to accurately reflect their resolution.
102 if [[ "$dpi" = "" ]]; then
103 # %k gives the number of colours/shades in the image.
104 # 2 for b/w, and the scanner seems to default to 400 dpi for that
105 # 256 for greyscale, which has to be selected manually (so I'm assuming 300 for no good reason)
106 # 31564 for colour, which defaults to 300
107 if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
112 if [[ $verbose == 1 ]]; then
113 echo "DPI guessed at $dpi."
117 # guess at depth, unless it's been set
118 if [[ "$depth" = "" ]]; then
119 if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
124 if [[ $verbose == 1 ]]; then
125 echo "Depth guessed at $depth."
129 # process pages 1 by 1 to avoid convert gobbling all the memory
130 for pg in `$my_seq 1 $pages`; do
131 echo "Processing page $pg."
133 pgn=$(printf '%03d' $pg)
136 origpnm=$dir/pg-${pgn}.pnm
137 if [[ ! $([ -e $origpnm ]) || $forcepdf ]]; then
138 convert $convertflags -depth $depth -density $dpi $file[$(expr $pg - 1)] \
143 preppnm=$dir/pg-pp-${pgn}.pnm
144 if [[ ! $skipmask && ( ! $([ -e $preppnm ]) || $forcemask ) ]]; then
146 # ... downscale, blur,
147 convert $convertflags -resize 25% -depth 8 -blur 10 $origpnm $dir/pg-mask-${pgn}.pnm ||exit 1
149 # ... get crop co-ords. They're off by ~2 as I don't know how to
150 # properly correct for the border.
151 cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 400% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.pnm info: || exit 1)
153 # ... crop and despeckle? the final pre-prepared image
154 convert $convertflags -crop $cropcords $origpnm $preppnm || exit 1
155 elif [[ $skipmask ]]; then
160 #names go a bit funny here
161 #also, ignore flags starting here
162 unppnm=$dir/upg-${pgn}-%01d.pnm
163 unpaper $unpaperflags --layout double --overwrite --no-blackfilter -ni 10 -op 2 $preppnm $unppnm || exit 1
165 # final convert and clean
166 convert $convertflags $dir/upg-${pgn}-?.pnm $dir/upg-${pgn}-%01d.${extension} || exit 1
172 mv $dir/upg-*.${extension} $dir/pages