--- get_book.sh	2003/12/14 22:27:18	1.2
+++ get_book.sh	2004/02/15 11:40:43	1.7
@@ -1,27 +1,66 @@
 #!/bin/sh
 
-#export http_proxy=http://proxy:8080
+# proxy settings (same as in firebird)
+fping proxy && export http_proxy=http://proxy:8080
+# user agent (same as in firebird)
+ua="Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031225 Firebird/0.7"
+
+# wait between pages
+export wait=120
+
 
 if [ -z "$1" ] ; then
 	echo "Usage: $0 ISBN"
 	exit 1
 fi
 
-isbn=$1;
+if [ -e orig ] ; then
+	echo "orig directory found. Resume download? [Y/n]"
+	read ans
+	if [ "$ans" = "n" ] ; then
+		exit 1;
+	fi
+	mv orig/* .
+	rm -Rf orig
 
-wait=10
+	grep -l 'promo.asp' * | xargs -i rm {}
+	grep -l 'This is only a preview of the full book' * | xargs -i rm {}
+fi
+
+isbn=$1;
 
 isbn2=`echo $isbn | sed 's/-//g'`
 
 function mirror() {
+
+	url="$1"
+
+	file=`echo $url | sed -e s,http://[^?]*?,index.html?, -e s,#.*$,, -e s,/,%2F,g`
+	if [ -e "$file" ] ; then
+#		echo "skip $url"
+		return
+	fi
+
+	if echo $url | grep '/index' >/dev/null ; then
+		cookies=""
+		echo -n "no login (index) "
+	elif echo $url | grep 'mode=toc' >/dev/null ; then
+		cookies=""
+		echo -n "no login (toc) "
+	else
+		cookies="--load-cookies=/home/dpavlin/.phoenix/default/g6b45nt6.slt/cookies.txt"
+		echo -n "with login "
+	fi
+	echo $url
+
 	wget -p -nH -nc -k \
-		--random-wait --wait=$wait -t 0 \
-		--load-cookies=/home/dpavlin/.phoenix/default/g6b45nt6.slt/cookies.txt \
-		-U "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031206 Firebird/0.7" \
-		$1
+		-t 1 -U "$ua" \
+		$cookies $url
 
 #	-D safari.oreilly.com \
 #	-A 0-201-41975-0 \
+
+	perl -e '$t=rand($ENV{wait} || 30);print "sleep for $t sec.\n"; sleep($t);'
 }
 
 function geturl() {
@@ -34,6 +73,7 @@
 		-e 's/open=false/open=true/' | \
 		grep '&s=1&b=1&f=1&t=1&c=1&u=1&r=&o=1' | \
 		grep $2 | \
+		grep -v "$2/[0-9][0-9][0-9][0-9][0-9][0-9][0-9]" | \
 		sort -u >> in
 }
 
@@ -43,14 +83,41 @@
 	grep 'view=[A-Z].*/index' in.tmp | sort -u >> in
 }
 
-echo > in
+function mirror_in() {
+	cat in | while read url ; do
+		mirror "$url"
+		#sleep $wait
+
+		if grep 'promo.asp' `ls -t index.html* | head -3` >/dev/null ; then
+			echo "WARNING: safari seems to logout you as user. Aborting."
+			exit 1
+		fi
+
+		if grep -i '>Account locked<' `ls -t index.html* | head -3` >/dev/null ; then
+			echo "WARNING: your safari account is locked. Aborting."
+			exit 1
+		fi
+
+		if grep -i 'session disabled' `ls -t index.html* | head -3` >/dev/null ; then
+			echo "WARNING: your safari session is disabled. Aborting."
+			exit 1
+		fi
+
+		if grep -i 'This is only a preview of the full book' `ls -t index.html* | head -3` >/dev/null ; then
+			echo "WARNING: you didn't add this book to your bookshelf!"
+			exit 1
+		fi
+	done
+}
+
+echo -n > in
 mirror "http://safari.oreilly.com/?XmlId=$isbn"
 
 echo "extract URLs from first page..."
 geturl "index.html?XmlId=$isbn" $isbn
 uniqurl
 
-mirror "-i in"
+mirror_in
 
 echo -n "extracting URLs [1]"
 ls index.html* | while read file ; do
@@ -61,9 +128,9 @@
 
 uniqurl
 
-mirror "-i in"
+mirror_in
 
-echo > in
+echo -n > in
 echo -n "extracting URLs [2]"
 ls index.html* | while read file ; do
 	echo -n "."
@@ -72,7 +139,7 @@
 
 uniqurl
 
-mirror "-i in"
+mirror_in
 
 # convert links in html
 bn=`basename $0`