OBSOLETE. SEE http://gist.github.com/749471

OBSOLETE

#!/bin/sh
#
# Copyright (c) 2010 Shinnosuke Suzuki <sasasin@sasasin.net>
# You can distribute this file under the GPL.
#
# Modified by TADA Tadashi <t@tdtds.jp>
#
FILE=$1
if [ -z $FILE ]; then
	echo usage: pdf2kindle PDFfile [top bottom left right]
	exit
elif [ ! -e $FILE ]; then
	echo "$FILE is not found."
	exit
fi
echo $(basename $FILE .pdf)

TOP=$(($2+10))
BOTTOM=$(($3+10))
LEFT=$(($4+10))
RIGHT=$(($5+10))

# 画像補正の並列数
# 負荷がキツイなら下げる
PARA=2

TMPDIR=./$(basename $FILE .pdf)
mkdir -p $TMPDIR/ 2>/dev/null
mkdir $TMPDIR/pgm 2>/dev/null
mkdir $TMPDIR/png 2>/dev/null
mkdir $TMPDIR/pdf 2>/dev/null
mkdir $TMPDIR/pgm_err 2>/dev/null

touch $TMPDIR/crop.txt
echo "TOP=$TOP" >> $TMPDIR/crop.txt
echo "BOTTOM=$BOTTOM" >> $TMPDIR/crop.txt
echo "LEFT=$LEFT" >> $TMPDIR/crop.txt
echo "RIGHT=$RIGHT" >> $TMPDIR/crop.txt

pdf2pgm() {
	echo -n "extracting image files from source PDF.."
	pdftoppm -gray $FILE $TMPDIR/pgm/pdf2mobi
	echo '.'
}

pgm2png() {
echo -n "cropping image files"
rm -f $TMPDIR/convert_error.log
touch $TMPDIR/convert_error.log
i=0
ls $TMPDIR/pgm/*.pgm \
| while read F
do
	echo -n '.'
	(
		# 上下左右の裁断と自動トリム。
		# TODO:実行速度が遅いので、どうにかしてOpenCV化したい
		convert \
			$TMPDIR/pgm/$(basename $F) \
			-level '0%,100%' \
			-chop ${LEFT}x${TOP} -flip -flop \
			-chop ${RIGHT}x${BOTTOM} -flip -flop \
			-fuzz 50% -trim \
			-quality 0 \
			$TMPDIR/png/$(basename $F .pgm).png \
			;
	) &
	# 同時に$PARA並行までに制限
	i=$(($i+1))
	if [ $i -eq $PARA ]; then
		i=0
		wait
   fi

done 2>> $TMPDIR/convert_error.log
wait
echo '.'

TO=$TMPDIR/pgm_err
echo -n "getting error information as $TO..."
cat $TMPDIR/convert_error.log \
| tr '`' '\n' \
| tr "'" '\n' \
| grep -E '\.pgm$' \
| while read F
do
	ERROR_PPM=$F
	ERROR_PNG=$(echo $F | sed -e 's/\/pgm\//\/png\//g' | sed -e 's/\.pgm$/\.png/g')
	cp $ERROR_PPM $ERROR_PNG $TO/
done
if [ 0 -ne $(find $TO -type f | wc -l) ];then
	echo "-> if error files not empty, try to decrease -fuzz vlue."
	find $TO
else
	echo "-> no error."
fi

}

dumpmetadata() {
	if [ ! -e $TMPDIR/pdf/metadata.txt ]; then
		echo "extracting PDF metadata..."
		pdftk $FILE dump_data output $TMPDIR/pdf/metadata.txt
	fi
}

png2pdf() {
	echo -n "generationg destination PDF"
	i=0
	while [ $i != 10 ]; do
		echo -n "."
		if [ 0 -ne $(ls $TMPDIR/png/pdf2mobi-${i}??.png 2>/dev/null | wc -l) ];then
			convert $TMPDIR/png/pdf2mobi-${i}??.png $TMPDIR/pdf/$i.pdf
		fi
		i=$(($i+1))
	done
	rm -f $TMPDIR/$(basename $FILE .pdf).out.tmp.pdf
	pdftk $TMPDIR/pdf/*.pdf \
		cat output $TMPDIR/$(basename $FILE .pdf).out.tmp.pdf
	pdftk $TMPDIR/$(basename $FILE .pdf).out.tmp.pdf \
		update_info $TMPDIR/pdf/metadata.txt \
		output $TMPDIR/$(basename $FILE .pdf).out.pdf
	mv $TMPDIR/$(basename $FILE .pdf).out.pdf $TMPDIR/../
	echo '.'
	ls -l $TMPDIR/../$(basename $FILE .pdf).out.pdf
}

dumpmetadata
pdf2pgm
pgm2png
png2pdf

このスクリプトは古くなっており、すでにメンテされていません。
最新のKindlize手法はRakefileによって実現されています。以下を参照して下さい。

自炊PDFをKindle3向けにトリミングするRakefile
https://gist.github.com/749471