sane-scan-pdf/scan

#!/bin/bash

DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)

DEVICE=fujitsu
OUTPUT=scan.pdf
USEARRAY=0
APPEND=0
RESOLUTION=300
MODE=Lineart
SCRIPT="$DIR/scan_perpage"
DUPLEX=0
UNPAPER=0
SEARCHABLE=0
LANGUAGE=eng
MAXPAGE=
TRUNCPAGE=0
HELP=0
SIZE=Letter
PGHEIGHT=
PGHEIGHTIN=
PGWIDTH=
PGWIDTHIN=
CROP=0
DESKEW=0
VERBOSE=0
SKIP_EMPTY_PAGES=0

TMP_DIR=`mktemp -d`
trap "rm -rf $TMP_DIR" 0

# Parse command-line options
while [ $# -gt 0 ]; do
  case "$1" in

  -v|--verbose) VERBOSE=1 ;;

  -d|--duplex) DUPLEX=1 ;;

  -m|--mode) shift; MODE=$1 ;;

  -r|--resolution) shift; RESOLUTION=$1 ;;

  -a|--append) APPEND=1 ;;

  -e|--max) shift; MAXPAGE=$1 ;;

  -t|--truncate) shift; TRUNCPAGE=$1 ;;

  -h|--help) HELP=1 ;;

  -s|--size) shift; SIZE=$1 ;;

  -ph|--page-height) shift; PGHEIGHT=$1 ;;

  -pw|--page-width) shift; PGWIDTH=$1 ;;

  --crop) CROP=1 ;;

  --deskew) DESKEW=1 ;;

  --unpaper) UNPAPER=1 ;;

  --searchable|--ocr) SEARCHABLE=1 ;;

  --language) LANGUAGE=$1 ;;

  --skip-empty-pages) SKIP_EMPTY_PAGES=1 ;;

  -o|--output) shift; OUTPUT="$1" ;;

  -l|--outputlist) shift; USEARRAY=1; OUTPUT=(); OUTPUT+=("$1") ;;

  -x|--device) shift; DEVICE=$1;;

  *) if [ $USEARRAY = 1 ]; then OUTPUT+=("$1"); else echo >&2 "Unknown argument: $1"; exit 1; fi ;;

  esac
  shift   # next option
done

if [ $HELP -eq 1 ]; then
  echo "$(basename $0) [OPTIONS]... [OUTPUT]"
  echo ""
  echo "OPTIONS"
  echo " -v, --verbose"
  echo "   Verbose output (this will slow down the scan due to the need to prevent interleaved output)"
  echo " -d, --duplex"
  echo "   Duplex scanning"
  echo " -m, --mode"
  echo "   Mode e.g. Lineart (default), Halftone, Gray, Color, etc."
  echo " -r, --resolution"
  echo "   Resolution e.g 300 (default)"
  echo " -a, --append"
  echo "   Append output to existing scan"
  echo " -e, --max <pages>"
  echo "   Max number of pages e.g. 2 (default is all pages)"
  echo " -t, --truncate <pages>"
  echo "   Truncate number of pages from end e.g. 1 (default is none)"
  echo " -s, --size"
  echo "   Page Size as type e.g. Letter (default), Legal, A4, no effect if --crop is specified"
  echo " -ph, --page-height"
  echo "   Custom Page Height in mm"
  echo " -pw, --page-width"
  echo "   Custom Page Width in mm"
  echo " -x, --device"
  echo "  Override scanner device name, defaulting to `fujitsu`"
  echo " --crop"
  echo "   Crop to contents (driver must support this)"
  echo " --deskew"
  echo "   Run driver deskew (driver must support this)"
  echo " --unpaper"
  echo "   Run post-processing deskew and black edge detection (requires unpaper)"
  echo " --ocr"
  echo "   Run OCR to make the PDF searchable (requires tesseract)"
  echo " --language <lang>"
  echo "   which language to use for OCR"
  echo " --skip-empty-pages"
  echo "   remove empty pages from resulting PDF document (e.g. one sided doc in duplex mode)"
  echo ""
  echo "OUTPUT"
  echo " -o, --output <outputfile>"
  echo "   Output to named file default=scan.pdf"
  echo " -l, --outputlist <outputfile-1...outputfile-n> Output to named files for each scanned page, can be used with append"
  echo ""
  exit 0
fi

if [ $USEARRAY = 0 -a -f "$OUTPUT" -a ! $APPEND = 1 ]; then
  echo >&2 "Output file $OUTPUT already exists. Delete or specify -a. Aborting."
  exit 1
fi

if [ $USEARRAY = 1 -a ! $APPEND = 1 ]; then
  for o in "${OUTPUT[@]}"; do
    if [ -f "$o" ]; then
      echo >&2 "Output file $o already exists. Delete or specify -a. Aborting."
      exit 1
    fi
  done
fi

SOURCE=""
if [ $DUPLEX -eq 1 ]; then
  SOURCE="--source \"ADF Duplex\""
fi

if [ "$MAXPAGE" != "" ]; then
  MAXPAGE="-e $MAXPAGE"
fi

PS2PDF_OPTS=

# Specify non-letter sizes in mm
case "$SIZE" in

  Letter) PGHEIGHTIN=11; PGWIDTHIN=8.5 ;;

  Legal) PGHEIGHT=355.6; PGWIDTH=215.9 ;;

  A4) PGHEIGHT=297; PGWIDTH=210 ;;

esac

if [ $CROP != 1 -a "$PGHEIGHT" != "" ]; then
  PGHEIGHTIN=$(units --compact -1 "$PGHEIGHT mm" 'in')
  PGHEIGHT="--page-height $PGHEIGHT -y $PGHEIGHT"
  PS2PDF_OPTS="-dEPSCrop"
fi

if [ $CROP != 1 -a "$PGWIDTH" != "" ]; then
  PGWIDTHIN=$(units --compact -1 "$PGWIDTH mm" 'in')
  PGWIDTH="--page-width $PGWIDTH -x $PGWIDTH"
  PS2PDF_OPTS="-dEPSCrop"
fi

if [ $CROP = 1 ]; then
  CROP="--swcrop=yes"
  PGHEIGHT=
  PGWIDTH=
  PGHEIGHTIN=
  PGWIDTHIN=
  PS2PDF_OPTS="-dEPSCrop"
fi

if [ $DESKEW = 1 ]; then
  DESKEW="--swdeskew=yes"
fi

export VERBOSE
export UNPAPER
export SEARCHABLE
export LANGUAGE
export RESOLUTION
export PGWIDTHIN
export PGHEIGHTIN
export PS2PDF_OPTS
export SKIP_EMPTY_PAGES

if [ $VERBOSE = 1 ]; then
  LOCKFILE=$(mktemp)
  trap "rm -rf $LOCKFILE" 0
  export LOCKFILE
fi;


echo >&2 "Scanning..."
#eval strace -f -o /tmp/scan-trace.txt scanadf -d $DEVICE $MAXPAGE $PGHEIGHT $PGWIDTH -S $SCRIPT --script-wait --resolution $RESOLUTION --mode $MODE $DESKEW $CROP $SOURCE -o scan-%04d
eval scanadf -d "$DEVICE" $MAXPAGE $PGHEIGHT $PGWIDTH -S $SCRIPT --script-wait --resolution $RESOLUTION --mode $MODE $DESKEW $CROP $SOURCE -o $TMP_DIR/scan-%04d

shopt -s extglob nullglob
pdffiles=($TMP_DIR/scan-[0-9]*.pdf)
numscans=${#pdffiles[@]}
if [ $numscans -gt 0 ]; then
  echo "Processing $numscans pages"
  if [ $numscans -gt $TRUNCPAGE -a $TRUNCPAGE -gt 0 ]; then
    for x in ${pdffiles[@]:$numscans-$TRUNCPAGE:$TRUNCPAGE}; do rm "$x"; done;
    pdffiles=(${pdffiles[@]:0:$numscans-$TRUNCPAGE})
    echo "Truncated $TRUNCPAGE pages"
  fi
  if [ $numscans -gt 1 -a $USEARRAY = 1 ]; then
    echo "Naming pdfs based on output list..."
    output_count=${#OUTPUT[@]}
    index=0
    while [ "$index" -lt "$output_count" ]; do
      let "scanno = $index + 1"
      if [ -f "${OUTPUT[$index]}" ]; then
        mv "${OUTPUT[$index]}" "${OUTPUT[$index]}.orig"
        if [ $APPEND -eq 1 ]; then
          pdffiles=()
          if [ -f "${OUTPUT[$index]}.orig" ]; then
            pdffiles+=("${OUTPUT[$index]}.orig")
          fi
          pdffiles+=($TMP_DIR/scan-*(0)$scanno.pdf)
          pdfunite "${pdffiles[@]}" "${OUTPUT[$index]}" && rm $TMP_DIR/scan-*(0)$scanno.pdf
        else
          mv $TMP_DIR/scan-*(0)$scanno.pdf "${OUTPUT[$index]}"
        fi
      else
        mv $TMP_DIR/scan-*(0)$scanno.pdf "${OUTPUT[$index]}"
      fi
      let "index = $index + 1"
    done
  elif [ $numscans -gt 1 -o $APPEND -eq 1 ]; then
    echo "Concatenating pdfs..."
    if [ -f "$OUTPUT" ]; then
      mv "$OUTPUT" "${OUTPUT}.orig"
    fi
    pdffiles=()
    if [ -f "${OUTPUT}.orig" ]; then
      pdffiles+=("${OUTPUT}.orig")
    fi
    pdffiles+=($TMP_DIR/scan-[0-9]*.pdf)
    pdfunite "${pdffiles[@]}" "$OUTPUT" && rm $TMP_DIR/scan-[0-9]*.pdf
  else
    if [ $USEARRAY = 1 ]; then
      mv $TMP_DIR/scan-0*.pdf "${OUTPUT[0]}"
    else
      mv $TMP_DIR/scan-0*.pdf "$OUTPUT"
    fi
  fi
  echo ""
  echo "Done."
else
  echo "Found no scans."
fi