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 # Mac OS X doesn't have seq. It has jot instead.
28 if [[ $(uname) == "Darwin" ]]; then my_seq=$Darwin_seq;
29 else my_seq=$Linux_seq;
47 while getopts 'vd:D:sfpmucF:b' OPTION
51 convertflags="$convertflags -verbose"
52 unpaperflags="$unpaperflags -v --time"
60 f) forcepdf=1; forcemask=1; forceunpaper=1; forceclean=1
62 p) forcepdf=1; forcemask=1; forceunpaper=1; forceclean=1
64 m) forcemask=1; forceunpaper=1; forceclean=1
66 u) forceunpaper=1; forceclean=1
70 F) extension="$OPTARG"
79 shift $(($OPTIND - 1))
81 # check we have an input and output!
82 if [[ $# != 2 ]]; then
83 echo "Wrong number of parameters (2 required, $# given: [$@])" >&2
92 base=$(basename $1 .pdf)
97 # figure out the number of pages
98 dscname=$dir/${base}.dsc
99 pdf2dsc $file $dscname || exit 1
100 pages=$(awk '$1 ~ "%%Pages" {print $2}' $dscname)
101 echo "Got $pages page(s)."
104 # guess at DPI, unless it's been set
105 # the scans do not seem to accurately reflect their resolution.
106 if [[ "$dpi" = "" ]]; then
107 # %k gives the number of colours/shades in the image.
108 # 2 for b/w, and the scanner seems to default to 400 dpi for that
109 # 256 for greyscale, which has to be selected manually (so I'm assuming 300 for no good reason)
110 # 31564 for colour, which defaults to 300
111 if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
116 if [[ $verbose == 1 ]]; then
117 echo "DPI guessed at $dpi."
121 # guess at depth, unless it's been set
122 if [[ "$depth" = "" ]]; then
123 if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
128 if [[ $verbose == 1 ]]; then
129 echo "Depth guessed at $depth."
133 # process pages 1 by 1 to avoid convert gobbling all the memory
134 for pg in `$my_seq 1 $pages`; do
135 echo "Processing page $pg."
137 pgn=$(printf '%03d' $pg)
140 origpnm=$dir/pg-${pgn}.pnm
141 if [[ ! $([ -e $origpnm ]) || $forcepdf ]]; then
142 convert $convertflags -depth $depth -density $dpi $file[$(expr $pg - 1)] \
147 preppnm=$dir/pg-pp-${pgn}.pnm
148 if [[ ! $skipmask && ( ! $([ -e $preppnm ]) || $forcemask ) ]]; then
150 # ... downscale, blur,
151 convert $convertflags -resize 25% -depth 8 -blur 10 $origpnm $dir/pg-mask-${pgn}.pnm ||exit 1
153 # ... get crop co-ords. They're off by ~2 as I don't know how to
154 # properly correct for the border.
155 cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 400% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.pnm info: || exit 1)
157 # ... crop and despeckle? the final pre-prepared image
158 convert $convertflags -crop $cropcords $origpnm $preppnm || exit 1
159 elif [[ $skipmask ]]; then
164 #names go a bit funny here
165 #also, ignore flags starting here
166 unppnm=$dir/upg-${pgn}-%01d.pnm
167 unpaper $unpaperflags --layout double --overwrite --no-blackfilter -ni 10 -op 2 $preppnm $unppnm || exit 1
169 #final convert and clean w/ bebook optimisation
170 if [[ $bebook ]]; then
171 convert $convertflags $dir/upg-${pgn}-?.pnm -depth 4 -resize 600x800 $dir/final-${pgn}-%01d.${extension} || exit 1
173 convert $convertflags $dir/upg-${pgn}-?.pnm $dir/final-${pgn}-%01d.${extension} || exit 1
179 mv $dir/final-*.${extension} $dir/pages