ocrwm

#!/bin/bash
set -u

########################################
 # Set these variables
########################################
# Set the default text for the watermark
label='Copyright ©2013 KZ Gedenkstätte Neuengamme'

# Download a free font from Google Fonts
# http://www.google.com/fonts#UsePlace:use/Collection:Gudea
font='/Users/aes9h/Downloads/fonts/Gudea/Gudea-Regular.ttf' 


########################################
 # No need to edit past this point
########################################
CONV="$(which convert)"
COMP="$(which composite)"
IDEN="$(which identify)"
TESS="$(which tesseract)"

# Set colors for echo
ESQ="\x1b["
red=$ESQ'1;31;40m'
blue=$ESQ'1;34;40m'
yellow=$ESQ'33;40m'
green=$ESQ'32;40m'
magenta=$ESQ'35;40m'
reset=$(tput sgr0)

USAGE='ocrwm [-copw] [-l "text"] [/path/to/folder]'

# Get the options
while getopts "cl:opw" options; do

    case $options in
        c ) NOCOPY="true";;
        l ) LABEL="$OPTARG";;
        o ) OCR="true";;
        p ) PDF="true";;
        w ) WATERMARK="true";;
        \? ) echo -e "$USAGE"
             exit 1;;
        * ) echo -e "$USAGE"
            exit 1;;
    esac
done
shift $((OPTIND - 1))


# Set the watermark text
if [[ -z "${LABEL:-}" ]]; then
    setlabel="$label"
else
    setlabel="$LABEL"
fi

# If font not set above, set variable to empty
if [[ -f "$font" ]]; then
    setfont="-font $font"
else
    setfont=''
fi


# Get the working directory
if [[ -z ${1:-} ]]; then
    path=$(pwd)
else
    cd "$1"
    path=$(pwd)
fi


if [[ "true" != "${NOCOPY:-}" ]]; then
    if [[ ! -d copies ]]; then
       mkdir -p copies
    fi
else
    echo -e "${red}Please note, OCR and Watermark assumes that a copy has been made in the png format and is available in the 'copies' directory.${reset}"
fi

if [[ "true" == "${WATERMARK:-}" ]]; then
    if [[ ! -d marked ]]; then
        mkdir -p marked 
    fi
fi

if [[ "true" == "${OCR:-}" ]]; then
    if [[ ! -d ocr ]]; then
        mkdir -p ocr 
    fi
fi


# Use find piped to while with IFS to account for names with spaces and other
# characters that would break in a for loop. Putting IFS in a while loop means
# we don't have to worry about it being unset/reset.
find "$path" -type f -depth 1 \
    \( -iname \*.jpg -o -iname \*.jpeg \
    -o -iname \*.png -o -iname \*.gif \
    -o -iname "\*.tif*" \) \
    -print0 | while IFS= read -r -d '' file 
    do
        # while parameters -r = do not read slashes in file names specially
        #               -d '' = NUL delimeted

    # ## deletes longest match from front of variable
    # % deletes shortest match from end of variable
    fullname="${file##*/}"
    extension="${fullname##*.}"
    filename="${fullname%.*}"
    newname="${filename}.png"

    if [[ -f "$file" ]]; then
       ########################################################################
        # create a copy in png form in the copy directory
       ########################################################################
       if [[ "true" == "${NOCOPY:-}" ]]; then
           echo -e "${red}Skipping making a copy of $fullname.${reset}"
       else
           echo -e "Making copy of ${blue}$fullname${reset}."
           if [[ "png" == "$extension" ]]; then
               cp "$path/$fullname" "$path/copies/"
           else
               $CONV "$path/$fullname" "$path/copies/$newname"
           fi
       fi

       ######################################################################## 
        # OCR the png and put in ocr directory
       ######################################################################## 
       if [[ "true" == "${OCR:-}" ]]; then
           echo -e "Running OCR on ${yellow}$newname${reset}."
           #command  #input    #output(.txt) #language=German
           $TESS "$path/copies/$newname" "$path/ocr/$filename" -l deu 
       fi

       ######################################################################## 
        # Add watermark
       ######################################################################## 
       if [[ "true" == "${WATERMARK:-}" ]]; then
           echo -e "Putting water mark on ${green}$newname${reset}."
            width=$($IDEN -format %w "$path/copies/$newname") 
            # height=$($IDEN -format %h "$path/copies/$newname") 
            # w=$((width * 30))
            s=$((width/2))
            # h=$((s+$s))
            # p=$((height/35))


            # Add black text in gold box under image
            # convert \
            #     $1 \
            #     -background Gold \
            #     -pointsize $s \
            #     -font $font \
            #     -gravity south \
            #     -splice 0x$h \
            #     -annotate +0+22 "$label" \
            #     marked/$1

            # $CONV -background '#0008' -fill white -gravity center -size $w \
            #     caption:"$label" \
            #     dragon.gif +swap -gravity south -composite  anno_caption.jpg

            # Overlay white text on dark grey transparent background
            $CONV \
                -background '#00000080' \
                -fill white \
                -size "$s" \
                $setfont \
                label:"$setlabel" \
                miff:- | \
                    $COMP \
                    -gravity south \
                    -geometry +0+3 \
                    - "$path/copies/$newname" "$path/marked/$newname"
        fi # end water mark section

    fi
    echo

done


######################################################################## 
# Compile into one pdf
######################################################################## 
if [[ "true" == "${PDF:-}" ]]; then
   echo -e "${magenta}Consolidating images into one pdf.${reset}"
   $CONV "$path/copies/*.png" "$path/combined.pdf"
fi