BeBook One arrived. Optimisations

[dja/scandal.git] / scan2pages.sh

diff --git a/scan2pages.sh b/scan2pages.sh
index 45f6215..399e1c2 100755 (executable)
--- a/scan2pages.sh
+++ b/scan2pages.sh
@@ -2,18 +2,22 @@
 
 function usage {
 cat >&2 << __EOF__ 
-Usage: $0 [-vfpsmut] [-d depth] pdffile outdir
+Usage: $0 [-vfpsmutb] [-D DPI] [-d depth] [-F format] pdffile outdir
 Convert pdffile - a pdf of scanned facing pages - to a set of images in outdir,
 each with only one page.
 OPTIONS:
        -v: Be verbose.
-       -d depth: Use given depth. Default is 1. Unpaper can only handle up to 8.
+       -d depth: Use given depth. Default is 1 for black and white, 8 for colour.
+               Unpaper can only handle up to 8.
+       -D dpi: Use given DPI. Colour scans on library scanners are usually 300, bw 400.
+       -F extension: use the output format signified by the extension. Default is png
        -s: skip masking/trimming. Overrides -m.
        -f: Forceably redo everything.
        -p: Forceably redo pdf conversion. Implies options below, equivalent to -f.
        -m: Forceably redo masking/trimming and other preprocessing. Impiles options below.
        -u: Forceably redo unpaper processing. Implies option below. IGNORED
        -t: Forceably redo final trimming and cleaning. IGNORED
+       -b: Optimize for BeBook One.
 __EOF__
 }
 
@@ -33,15 +37,23 @@ forceunpaper=
 forceclean=
 convertflags=
 unpaperflags=
-depthflags="-depth 1"
+depthset=
+depth=
+dpi=
 skipmask=
-while getopts 'vd:sfpmuc' OPTION
+verbose=
+extension="png"
+bebook=
+while getopts 'vd:sfpmucF:' OPTION
 do
        case $OPTION in
-       v)  convertflags="$convertflags -verbose"
+       v)      verbose=1
+               convertflags="$convertflags -verbose"
                unpaperflags="$unpaperflags -v --time"
                ;;
-       d)      depthflags="-depth $OPTARG"
+       d)      depth="$OPTARG"
+               ;;
+       D)      dpi=$OPTARG
                ;;
        s)      skipmask=1;
                ;;
@@ -55,9 +67,10 @@ do
                ;;
        t)      forceclean=1
                ;;
-#      b)      bflag=1
-#              bval="$OPTARG"
-#              ;;
+       F)      extension="$OPTARG"
+               ;;
+       b)      bebook=1
+               ;;
        ?)      usage
                exit 2
                ;;
@@ -88,6 +101,35 @@ pages=$(awk '$1 ~ "%%Pages" {print $2}' $dscname)
 echo "Got $pages page(s)."
 rm $dscname
 
+# guess at DPI, unless it's been set
+# the scans do not seem to accurately reflect their resolution.
+if [[ "$dpi" = "" ]]; then
+       # %k gives the number of colours/shades in the image.
+       # 2 for b/w, and the scanner seems to default to 400 dpi for that
+       # 256 for greyscale, which has to be selected manually (so I'm assuming 300 for no good reason)
+       # 31564 for colour, which defaults to 300
+       if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
+               dpi=400;
+       else
+               dpi=300
+       fi
+       if [[ $verbose == 1 ]]; then
+               echo "DPI guessed at $dpi."
+       fi
+fi
+
+# guess at depth, unless it's been set
+if [[ "$depth" = "" ]]; then
+       if [[ $( convert $file[0] -format "%k" info: || exit 1 ) == 2 ]]; then
+               depth=1;
+       else
+               depth=8
+       fi
+       if [[ $verbose == 1 ]]; then
+               echo "Depth guessed at $depth."
+       fi
+fi
+
 # process pages 1 by 1 to avoid convert gobbling all the memory
 for pg in `$my_seq 1 $pages`; do
        echo "Processing page $pg."
@@ -97,7 +139,7 @@ for pg in `$my_seq 1 $pages`; do
        # convert from pdf
        origpnm=$dir/pg-${pgn}.pnm
        if [[ ! $([ -e $origpnm ]) || $forcepdf ]]; then
-               convert $convertflags $depthflags -density 300 $file[$(expr $pg - 1)] \
+               convert $convertflags -depth $depth -density $dpi $file[$(expr $pg - 1)] \
                        $origpnm || exit 1
        fi;
        
@@ -124,12 +166,15 @@ for pg in `$my_seq 1 $pages`; do
        unppnm=$dir/upg-${pgn}-%01d.pnm
        unpaper $unpaperflags --layout double --overwrite --no-blackfilter -ni 10 -op 2 $preppnm $unppnm || exit 1
 
-       # final convert and clean
-       convert $convertflags $dir/upg-${pgn}-?.pnm $dir/upg-${pgn}-%01d.tiff || exit 1
-       
+       #final convert and clean w/ bebook optimisation
+       if [[ $bebook ]]; then
+               convert $convertflags $dir/upg-${pgn}-?.pnm -depth 4 -resize 600x800 final-${pgn}-%01d.${extension} || exit 1
+       else
+               convert $convertflags $dir/upg-${pgn}-?.pnm $dir/final-${pgn}-%01d.${extension} || exit 1
+       fi
        
 done
 
 mkdir $dir/pages
-mv $dir/upg-*.tiff $dir/pages
+mv $dir/upg-*.${extension} $dir/pages