make-tifpdf: create a 600 dpi pdf with 1-bit-tiff images and with colored 300 dpi jpg front and back cover pages for use on bitsavers.org

make-tifpdf

#!/usr/bin/bash

# init 20230124

if [ $# -lt 1 ] ; then
	echo "make-tifpdf: create a 600 dpi pdf with 1-bit-tiff images for uploading scans to → bitsavers.org"
	echo "Usage: make-tifpdf <infile.pdf> [outfile.pdf]"
	echo
	exit
fi

# set to 0 to convert everything to 600dpi 1-bit tiff
# set to 1 to convert the front and back cover to colored jpg with coverresolution
coloredcover=1

# resolution in dpi for front and back cover jpg
coverresolution=300

# tumble source: https://github.com/brouhaha/tumble
# tumble is a utility by Eric Smith to construct PDF files from one or more 1-bit-tiff image files

start=$(date +%s.%N)

# Path to tumble program
tumble="/home/benutzer/src/tumble/tumble"

infile=$1

red="\e[1;31m"
yellow="\e[1;33m"
green="\e[1;32m"
cyan="\e[1;36m"
nocol="\e[0m"

# directory for resulting output files
tumbledir="tumble"
mkdir -p $tumbledir

# directory for processed input files
donedir="done"
mkdir -p $donedir

function ProgressBar {

	# https://raw.githubusercontent.com/fearside/ProgressBar/master/progressbar.sh
	# Author : Teddy Skarin

	# 1. Create ProgressBar function
	# 1.1 Input is currentState($1) and totalState($2)

	# Process data
	let _progress=(${1}*100/${2}*100)/100
	let _done=(${_progress}*4)/10
	let _left=40-$_done

	# Build progressbar string lengths
	_done=$(printf "%${_done}s")
	_left=$(printf "%${_left}s")

	# 1.2 Build progressbar strings and print the ProgressBar line
	# 1.2.1 Output example:
	# 1.2.1.1 Progress : [########################################] 100%
	printf "\rProgress : [${_done// /#}${_left// /-}] ${_progress}%%"

}

if [ $# -eq 1 ] ; then
	outfile=$(echo "$tumbledir/${infile%.*}--tumble--600dpi.pdf" | sed -e "s/--.00dpi--tumble--600dpi/--tumble--600dpi/")
else
	outfile=$2
fi

if [ -f $outfile ] ; then

	echo "${infile}:"
	echo -e "${green}As the corresponding output file already exists, processing is skipped.${nocol}"

	if [ -f $infile ] ; then
		echo "Thus we simply move the input file to"
		echo "→ $donedir/$infile.done"
		mv "$infile" "$donedir/$infile.done"
	fi

	echo
	exit
fi

tmpdir=$(mktemp -d -t make-tifpdf-XXXXXXXXXX)


numberofpages=$(pdfinfo "$infile" | grep "Pages" | awk '{print $2}')
echo -e "${cyan}Processing:"
echo $(du -h "$infile")
echo -e "$numberofpages PDF pages${nocol}"

lastbutonepage=$(echo "$numberofpages - 1" | bc)

# explode any page in the infile pdf to single 600 dpi ppm file
# each file: filesize 62 MB

progress=0

if [ $coloredcover -eq 1 ] ; then

	let endprogress=$numberofpages*2
	let halfprogress=$numberofpages

	# process colored front and back cover
	ProgressBar 1 $endprogress
	pdftoppm -f 1 -singlefile -jpeg -r $coverresolution "$infile" $tmpdir/frontcover

	pdftoppm -f 2 -l $lastbutonepage -r 600 "$infile" $tmpdir/p
	ProgressBar $halfprogress $endprogress

	pdftoppm -f $numberofpages -singlefile -jpeg -r $coverresolution "$infile" $tmpdir/backcover

else

	ProgressBar $halfprogress $endprogress
	pdftoppm -r 600 "$infile" $tmpdir/p

fi


# number of files = number of pages

ppmpages=$(ls $tmpdir | wc -l)
j=$halfprogress

for i in $tmpdir/p*.ppm; do

	let j=$j+1
	ProgressBar $j $endprogress

#	"-colors 2" sometimes produces all black tif pages for all white pages
#	do not use it
#	convert $i -density 600 +dither -colors 2 -colorspace gray -contrast-stretch 0 -depth 1 "${i%.*}--600dpi.tif"

#	http://www.fmwconcepts.com/imagemagick/2colorthresh/index.php
#	PURPOSE:
#	To automatically thresholds an image to binary (b/w) format using an adaptive spatial subdivision color reduction technique.
#
#	DESCRIPTION:
#	2COLORTHRESH automatically thresholds an image to binary (b/w) format using an adaptive spatial subdivision color reduction technique.
#	This is the -colors IM operator as implemented with slight modification from Anthony's Examples
#	at http://www.imagemagick.org/Usage/quantize/#two_color
#
#	For algorithm details, see http://www.imagemagick.org/script/quantize.php 
#
#	adapted to output create 1-bit tiff files in the desired resolution 

	convert $i -density 600 +dither -colorspace gray -contrast-stretch 0 -depth 1 "${i%.*}--600dpi.tif"

done

# Each uncompressed tiff file is now a 1-bit BlackWhite image with 600 dpi
# each file: filesize 2,7 MB

if [ $coloredcover -eq 1 ] ; then

	# process colored front and back cover
	$tumble $tmpdir/frontcover.jpg $(ls $tmpdir/*.tif) $tmpdir/backcover.jpg -o "$outfile" &>/dev/null

else

	$tumble $(ls $tmpdir/*.tif) -o "$outfile" &>/dev/null

fi

let j=$endprogress
ProgressBar $j $endprogress

end=$(date +%s.%N)
runtime=$(echo "scale=2;($end - $start)/1" | bc)
pagesperminute=$(echo "scale=2;60 * $numberofpages / $runtime" | bc)
secondsperpage=$(echo "scale=2;$runtime / $numberofpages" | bc)

# postprocess infile
# move to $donedir
mv "$infile" "$donedir/$infile.done"

echo
echo $(du -h "$donedir/$infile.done")
echo $(du -h "$outfile")
echo
echo -e "$yellow=== Summary ===$nocol"

if [ $coloredcover -eq 1 ] ; then
	echo "Colored front/back cover: yes"
else
	echo "Colored front/back cover: no"
fi

echo "$runtime seconds processing time"
echo "PDF: $numberofpages pages"
echo "$pagesperminute pages/minute"
echo "$secondsperpage seconds/page"
echo

# cleaning up

rm -rf $tmpdir