bash-common/chepdftk.sh

#!/bin/bash
# Umbrella function to simplify use of pdftk
# Created April 20, 2010
# Taha Ahmed, taha@chepec.se
# **************************
# Usage:
#			ARG 1: filename of pdf, e.g., 'oxides.pdf'
#			ARG 2: specify operation,
#					 - 'dump_data' --> writes metadata to $reportfile in current directory
#			ARG 3: used when updating metadata, specifies the input text file

# $*	Stores all the arguments that were entered on the command line ($1 $2 ...).
# $@	Stores all the arguments that were entered on the command line, individually quoted ("$1" "$2" ...).
# -o is logical OR - can be evaluated within single brackets
# -a is logical AND - can be evaluated within single brackets

#############################################################
## Original pdftk syntax
## *********************
## Dump metadata to file
#= pdftk <in.pdf> dump_data output <$reportfile>
## Update metadata from file
#= pdftk <in.pdf> update_info <$reportfile> output <out.pdf>
#############################################################

# Perhaps good to save original PDF

# Could be nice to be able to input metadata updates directly at the bash prompt, something like
# First, script displays current fields
# echo Which field do you want to update/create?
# echo <Display choice as a list>
# read choice newfieldinfo
# do the whole pdftk update stuff
# voila!

#set -x # for debugging

NAME=$(basename $0)
echo "***********************************************************"
echo "This is $NAME. Write $NAME -h for help."

# Set defaults
DUMPMETADATA=yes # default action
UPDATEMETADATA=no

# Set filetypes
pdfext=".pdf"
metadataext=".pdfmeta"

usage () {
	cat <<EOF
NAME
	$NAME - Wraps pdftk in a personalised run environment.

SYNOPSIS
	$NAME [OPTION], or
	$NAME [FILENAME.PDF]

DESCRIPTION
	The following options can be used:
	
	-h, --help
		Displays this help text.
		
	If no option nor filename is specified, the script defaults to
	dumping metadata information for all PDF files present in the 
	current directory into *.pdfmeta files, one for each PDF.
	If no PDF files are present in the current directory, the script
	exits without taking any further action.

DETAILS


EXAMPLE
	Extract metadata for all PDFs in directory and dump into separate files
	$NAME -a
	
	Default usage - dump PDF metadata in <filename>.pdfmeta
	$NAME <filename>.pdf

AUTHOR
	Taha Ahmed <taha.ahmed at mkem.uu.se>

EOF
}

OPTIONS=$@

for OPTION in "$@"; do
	case "$OPTION" in
		-h | --help )
			usage
			exit 0
			;;
		-a | --all )
			# Perform default action on all PDFs in current directory
			
			;;
			* ) # no options, just run with default settings
			;;
	esac
done

FILES=$OPTIONS

# Check if no input files were given
if [ ! -n "$FILES" ]; then
	# Perhaps do default action for all pdf files in current directory?
	echo -e "\nError: There is no input files! Usage is:\n"
	usage
	exit 1
fi

# Check file types (pre-assumes a file extension exists)
for FILE in "$FILES"; do
	PDFFILE=$(echo $FILE | grep '.[Pp][Dd][Ff]$')
	TEXFILE=$(echo $FILE | grep '.[Tt][Ee][Xx]$')    
	if [ ! -n "$PDFFILE" -a ! -n "$TEXFILE" ]; then
		echo "$FILE is not a supported file type!"
		echo "It should be one of: .pdf or .tex"
		exit 1
	fi
	NEWFILES="$NEWFILES $FILE"
	unset PDFFILE TEXFILE
done
FILES=$NEWFILES

# Execution
for FILE in "$FILES"; do
	# Metadata filename is pdf basename plus .pdfmeta extension
	METADATAFILE=${FILE%.*}$metadataext # metadata filename

	if [ "$DUMPMETADATA" = "yes" ]; then
		#echo "pdftk $FILE dump_data output $METADATAFILE"
		
		## It seems pdftk has problems dealing with spaces in filenames or paths.
		## Possible work-arounds:
			## Escape spaces with back-slashes?
			## Check for spaces in filename, if present, copy to temporary file without spaces,
			## then run call pdftk, then rename again...
		
		pdftk $FILE dump_data output $METADATAFILE
	fi
done

exit 0

### STUFF BELOW IS OLD
reportfile=./chepdftk-report.txt

# Check number of args supplied
if [ $# -eq 0 ]; then
	# Zero arguments
	echo No arguments given.
	exit
else
	# Some arguments were passed.
	if [ $# -eq 1 ]; then
		# One argument was passed
		echo One argument.
	else
		# More than one argument
		if [ $# -eq 2 ]; then
			# Two arguments were passed
			echo Two arguments.
			if [ $2 == "dump_data" ]; then
				pdftk $1 $2 output $reportfile
			fi
		else
			# More than two arguments
			if [ $# -eq 3 ]; then
				# Three arguments were passed
				echo Three arguments.
			else
				# More than three arguments
				echo More than three arguments. Too many.
			fi
		fi
	fi
fi