#! /bin/bash
# Save this script as FindBWcomics.sh
# It can run on Windows if you have cygwin and ImageMagick installed
# This script scans a directory for all the cbr/cbz files it can find
# The for each file it extracts one image and determines if the image is in colors
# If not, it makes a copy to the directory given as 2nd parameter of the script
# I use this script to find B&W comic books for my eBook reader
# This script doesn't change the files in the source directory

# Note that there are many ways that this script can fail:
# - if the pages alternate between color and B&W, results will be random
# - if the archive contains other types of files
# - if the rar archive contains several different directories
# - if the inner pathname contains problematic characters such as [ ] { } ( ) / \ ' "...
# Many conversions will fail for this last reason

# (c) Guillaume Dargaud 2010 - www.gdargaud.net
# Free use and distribution
# History: 20100518 - 2nd version that can handle arbitrary file names

# You can change the following 2 parameters
export Limit=1000		# If fewer than those many colors, then we consider it BW. Default: 256
export Temp=/tmp		# Temporary directory
export DO=ln			# Use ln or cp for a link or a copy

if [ $# -ne 2 ]; 
then
	echo 1>&2 "Usage: $0 ScanDir DestDir"
	echo 1>&2 "Find all cbr/cbz comic book files and copy those in BW to DestDir"
	exit 127
fi

SourceDir=$1			# Directory to scan
export DestDir=$2		# Copy there if the file contains a BW imagemkdir $Temp/$$			# Temp storage

# Make a copy of the cbr or cbz file passed as argument if it contains mostly BW images
function CopyIfBW {
	shopt -s nocasematch
	function clean { rm -rf $Temp/$$; echo; exit; }
	mkdir -p $Temp/$$			# Temp storage
	
	# We don't trust the file extension, let's do a 'file' on it.
	Type=$( file -b "$1" )
	Type=${Type:0:3}
				
	case $Type in 
		Zip )	unzip -qq "$1" -d $Temp/$$
			;;
		RAR )	Source=$(readlink -f "$1")		# Convert to absolute path
				pushd $Temp/$$ >/dev/null
				unrar e -ep -inul -c- "$Source"
				#unrar-free -x --extract-no-paths "$Source"
				popd >/dev/null
				# Important note: sometimes unrar hangs. I just issue "killall unrar" and the script carries on
			;;
		* ) 	echo "ERROR: $1 is neither ZIP nor RAR but "$(file -b "$1") 
				return;
			;;
	esac

	chmod -R +w $Temp/$$
	# Pic a random files
	Third=$(( $(find $Temp/$$ -type f | wc -l )/3+3 ))
	File="$( find $Temp/$$ -type f | sort | head -$Third | tail -1 )"

	# Try various methods to determine colors:
	Class=$(convert "$File" -format "%[colorspace]" info:)
	if [[ x$Class == xGray ]]; then $DO -vi "$1" "$DestDir/"; clean; fi
	# Also usable, but probably redundant:
	# convert logo: -format "%[colorspace]" info:
	# Gray
	# convert logo: -type grayscale -format "%[channels]" info:
	# gray
	# convert logo: -format "%r" info:
	# PseudoClassGray

#	Colors=$( convert "$( find $Temp/$$ -type f | sort | head -$Half | tail -1 )" -format "%k" info:- )
	Colors=$(convert "$File" -format "%k" info:)
#echo "$Colors colors"
	if [[ x$Colors == x ]]; then echo "ERROR: $Type/$IT \| $1 \| $Image"; clean; fi
	if (( $Colors <= $Limit )); then $DO -vi "$1" "$DestDir/"; clean; fi
	# We assume the file is in color: Nothing to do
	rm -rf $Temp/$$
}
export -f CopyIfBW

find $SourceDir -type f \( -iname \*.cbz -o -iname \*.cbr \) -exec bash -c 'CopyIfBW "{}"' \;