sane-scan-pdf/scan

#!/bin/bash

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

DEVICE=fujitsu
OUTPUT=scan.pdf
OUTPUTARR=()
USEOUTPUT=0
USEARRAY=0
APPEND=0
RESOLUTION=300
MODE=Lineart
MODE_CHANGED=0
MODE_HW_DEFAULT=0
SCRIPT="$DIR/scan_perpage"
DUPLEX=0
UNPAPER=0
SEARCHABLE=0
LANGUAGE=eng
MAXPAGE=
TRUNCPAGE=0
HELP=0
SIZE=
DEFAULTSIZE=1
PGHEIGHT=
PGHEIGHTIN=
PGWIDTH=
PGWIDTHIN=
CROP=0
DESKEW=0
DRIVER_OPTION=
VERBOSE=0
SKIP_EMPTY_PAGES=0

TMP_DIR=$(mktemp -d -p "" scan.XXXXXXXXXX)
cleanup()
{
  rm -rf "$TMP_DIR"
}
trap cleanup EXIT

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

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

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

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

  --mode-hw-default) MODE_HW_DEFAULT=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 ;;

  --no-default-size) DEFAULTSIZE=0 ;;

  --crop) CROP=1 ;;

  --deskew) DESKEW=1 ;;

  --unpaper) UNPAPER=1 ;;

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

  --language) shift; LANGUAGE=$1 ;;

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

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

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

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

  -xo|--driver-options) shift; DRIVER_OPTION=$1;;

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

  esac
  shift   # next option
done

if [[ $HELP == 1 ]]; then
  echo "$(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. Use --mode-hw-default to not set any mode"
  echo " --mode-hw-default"
  echo "   Do not set the mode explicitly, use the hardware default"
  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) -- truncation happens after --skip-empty-pages"
  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 " -xo, --driver-options"
  echo "  Send additional options to the scanner driver e.g."
  echo "  -xo \"--whatever bar --frobnitz baz\""
  echo " --no-default-size"
  echo "   Disable default page size, useful if driver does not support page size/location arguments"
  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 == 1 && $USEOUTPUT == 1 ]]; then
  echo >&2 "Use one of -o or -l. Aborting."
  exit 1
fi

if [[ $MODE_CHANGED == 1 && $MODE_HW_DEFAULT == 1 ]]; then
  echo >&2 "Use one of -m/--mode or --mode-hardware-default. Aborting."
  exit 1
fi

if [[ $MODE_CHANGED == 0 && $MODE_HW_DEFAULT == 0 ]]; then
  echo >&2 'Warning: neither the -m/--mode nor --mode-hw-default argument was specified. The current default is "Lineart", however -m/--mode is a SANE device-specific switch. Therefore, in a future version when --mode is not specified, `'"`basename "$0"`""'"' will defer to the device-specific default. If you wish to continue using "Lineart" as your scan mode, either verify that is the default for your scanner (scanadf --help -d <device>), or specify --mode Lineart explicitly on the command line.'
fi

if [[ $USEOUTPUT == 1 && "$OUTPUT" == "" ]]; then
  echo >&2 "Output file must be specified. Aborting."
  exit 1
fi

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

if [[ $USEARRAY == 1 && ${#OUTPUTARR[@]} == 0 ]]; then
  echo >&2 "At least one file must be specified with -l. Aborting."
  exit 1
fi

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

if [[ $USEARRAY == 1 ]]; then
  OUTPUT=("${OUTPUTARR[@]}")
fi

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

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

PS2PDF_OPTS=

if [[ $CROP != 1 && $SIZE == "" && $DEFAULTSIZE == 1 ]]; then
  # Default to Letter size, but only if crop is not specified and this feature is not disabled
  SIZE=Letter
fi

case "$SIZE" in

  Letter) PGHEIGHT=279.4; PGWIDTH=215.9 ;;

  Legal) PGHEIGHT=355.6; PGWIDTH=215.9 ;;

  A4) PGHEIGHT=297; PGWIDTH=210 ;;

esac

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

if [[ $CROP != 1 && "$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 --ald=yes"
  # In duplex mode, the driver's buffer for the back side image will be larger than necessary, oh well
  # http://sane.10972.n7.nabble.com/Fujitsu-backend-and-iX500-scanning-page-longer-than-14-Inches-td19303.html
  PGHEIGHT="--page-height 9999 -y 9999"
  PGWIDTH="--page-width 9999 -x 9999"
  PS2PDF_OPTS="-dEPSCrop"
fi

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

if [[ $SKIP_EMPTY_PAGES == 1 && ! -x "$(command -v bc)" ]]; then
  echo >&2 'Warning: `--skip-empty-pages` specified, but `bc` not available. Disabling empty page detection.'
  SKIP_EMPTY_PAGES=0
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 "cleanup; rm -rf $LOCKFILE" EXIT
  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
if [[ $MODE_HW_DEFAULT == 1 ]]; then
  MODE=
else
  MODE="--mode '$MODE'"
fi
eval scanadf -d \'"$DEVICE"\' $MAXPAGE $PGHEIGHT $PGWIDTH -S $SCRIPT --script-wait --resolution $RESOLUTION $MODE $DESKEW $CROP $DRIVER_OPTION $SOURCE -o $TMP_DIR/scan-%04d

shopt -s extglob nullglob
pdffiles=($TMP_DIR/scan-[0-9]*.pdf)
numscans=${#pdffiles[@]}
if [[ $numscans > 0 ]]; then
  echo "Processing $numscans pages"
  if [[ $numscans > $TRUNCPAGE && $TRUNCPAGE > 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 > 1 && $USEARRAY == 1 ]]; then
    output_count=${#OUTPUT[@]}
    echo "Naming $numscans pdfs based on output list of $output_count names..."
    index=0
    while (( $index < $output_count && $numscans > $index )); do
      let "scanno = $index + 1"
      if [[ -f "${OUTPUT[$index]}" ]]; then
        mv "${OUTPUT[$index]}" "${OUTPUT[$index]}.orig"
        if [[ $APPEND == 1 ]]; then
          pdffiles=()
          if [[ -f "${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 > 1 || $APPEND == 1 ]]; then
    echo "Concatenating pdfs..."
    if [[ -f "$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