pdf-wget.sh

#!/bin/sh
LINKURL="http://techinfo.honda.com/rjanisis/pubs/OM/AC9595/AC9595MAINIDX.pdf"
BASEURL="http://techinfo.honda.com/rjanisis/pubs/OM/AC9595/"
mkdir download-pdf 2> /dev/null
cd download-pdf
wget "$LINKURL"
mkdir PDF 2> /dev/null
while `ls *.pdf 2>&1 >/dev/null`; do
	PDFs_INDONE=`wc -l DONE.list | cut -d' ' -f1`
	PDFs_INDIR=`ls *.pdf|wc -l`
	echo "$PDFs_INDONE $PDFs_INDIR"
	if [ "$PDFs_INDONE" = "$PDFs_INDIR" ]; then
		break;
	fi
	for PDF in `ls *.pdf`; do
		if grep "$PDF" DONE.list 2>&1 > /dev/null; then
			continue
		fi
		pdftohtml -i "$PDF"
		BASENAME=`basename "$PDF" .pdf`
		wget -c -B "$BASEURL" --mirror -nd --force-html -A pdf -i "$BASENAME"'s.html' 2>&1 /dev/null
		echo "$PDF" >> DONE.list
	done
done