1 |
#!/bin/sh |
#!/bin/sh |
2 |
|
|
3 |
#export http_proxy=http://proxy.pliva.hr:8080 |
# proxy settings (same as in firebird) |
4 |
|
fping -q proxy && export http_proxy=http://proxy:8080 |
5 |
|
if [ -z "$http_proxy" ] ; then |
6 |
|
fping -q proxy.lan && export http_proxy=http://proxy.lan:8080 |
7 |
|
fi |
8 |
|
# user agent (same as in firebird) |
9 |
|
ua="Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040506 Firefox/0.8" |
10 |
|
cookie_file="cookie.txt" |
11 |
|
|
12 |
|
# wait between pages |
13 |
|
export wait=120 |
14 |
|
|
15 |
|
if [ ! -z "$http_proxy" ] ; then |
16 |
|
echo "Using proxy $http_proxy" |
17 |
|
fi |
18 |
|
|
19 |
|
if [ -z "$1" ] ; then |
20 |
|
echo "Usage: $0 ISBN" |
21 |
|
exit 1 |
22 |
|
fi |
23 |
|
|
24 |
|
if [ -e orig ] ; then |
25 |
|
echo "orig directory found. Resume download? [Y/n]" |
26 |
|
read ans |
27 |
|
if [ "$ans" = "n" ] ; then |
28 |
|
exit 1; |
29 |
|
fi |
30 |
|
mv orig/* . |
31 |
|
rm -Rf orig |
32 |
|
|
33 |
|
grep -l 'promo.asp' * | xargs -i rm {} |
34 |
|
grep -l 'This is only a preview of the full book' * | xargs -i rm {} |
35 |
|
fi |
36 |
|
|
37 |
#isbn="0-201-41975-0" |
isbn=`echo $1 | sed 's/-//g' | tr '[a-z]' '[A-Z]'` |
|
isbn="0-672-32240-4" |
|
38 |
|
|
39 |
wait=10 |
function mirror() { |
40 |
|
|
41 |
isbn2=`echo $isbn | sed 's/-//g'` |
url="$1" |
42 |
|
|
43 |
function mirror() { |
file=`echo $url | sed -e s,http://[^?]*?,index.html?, -e s,#.*$,, -e s,/,%2F,g` |
44 |
wget -p -nH -nc -k \ |
if [ -e "$file" ] ; then |
45 |
--random-wait --wait=$wait -t 0 \ |
# echo "skip $url" |
46 |
--load-cookies=/home/dpavlin/.phoenix/default/g6b45nt6.slt/cookies.txt \ |
echo -n "." |
47 |
-U "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031206 Firebird/0.7" \ |
return |
48 |
$1 |
fi |
49 |
|
|
50 |
|
cookie="" |
51 |
|
if echo $url | grep '/index' >/dev/null ; then |
52 |
|
echo -n "no login (index) " |
53 |
|
elif echo $url | grep 'mode=toc' >/dev/null ; then |
54 |
|
echo -n "no login (toc) " |
55 |
|
else |
56 |
|
if [ ! -e $cookie_file ] ; then |
57 |
|
echo "cookies file $cookie_file doesn't exits! Please create it." |
58 |
|
echo "It should be in format:" |
59 |
|
echo "Cookie: Site=UICode=&Portal=oreilly&GUID=..." |
60 |
|
exit 1 |
61 |
|
fi |
62 |
|
read cookie < $cookie_file |
63 |
|
if [ -z "$cookie" ] ; then |
64 |
|
echo "Empty cookie file $cookie_file !" |
65 |
|
exit 1 |
66 |
|
fi |
67 |
|
fi |
68 |
|
|
69 |
|
if [ -z "$cookie" ] ; then |
70 |
|
echo "$url [no cookie]" |
71 |
|
else |
72 |
|
echo "$url [with cookie]" |
73 |
|
fi |
74 |
|
|
75 |
|
wget -q -p -nH -nc -k -t 1 -U "$ua" --cookies=off --header="$cookie" $url |
76 |
|
perl -e '$t=rand($ENV{wait} || 120);print "sleep for $t sec.\n"; sleep($t);' |
77 |
|
|
|
# -D safari.oreilly.com \ |
|
|
# -A 0-201-41975-0 \ |
|
78 |
} |
} |
79 |
|
|
80 |
function geturl() { |
function geturl() { |
81 |
hindent -s $1 | grep $2 | grep -i href | grep mode=[st][eo]c | \ |
hindent -s $1 | grep -i href | grep mode=[st][eo]c | \ |
82 |
sed -e 's/^.*<a.*href="//i' \ |
sed -e 's/^.*<a.*href="//i' \ |
83 |
-e 's/".*//' -e 's/amp;//g' \ |
-e 's/".*//' -e 's/amp;//g' \ |
84 |
-e 's,^[^\?]*\?,http://safari.oreilly.com/,' \ |
-e 's,^[^\?]*\?,http://safari.oreilly.com/,' \ |
85 |
-e 's/#$//' \ |
-e 's/#.*$//' \ |
86 |
-e 's/srchText=//' | \ |
-e 's/\&srchText=//' \ |
87 |
grep -v open=false | \ |
-e 's/open=false/open=true/' | \ |
88 |
grep -v 'view=[A-Z].*%2F[^i]' | \ |
grep '&s=1&b=1&f=1&t=1&c=1&u=1&r=&o=1&n=1&d=1&p=1&a=0' | \ |
89 |
grep -v 'view=[A-Z].*/[^i]' | \ |
grep -i "xmlid=[0-9A-Za-z\-][0-9A-Za-z\-]*" | \ |
90 |
grep '&s=1&b=1&f=1&t=1&c=1&u=1&r=&o=1' | \ |
grep -vi "xmlid=[0-9A-Za-z\-]*/[0-9][0-9][0-9][0-9][0-9][0-9][0-9]" | \ |
91 |
sort -u >> in |
sort -u >> in |
92 |
} |
} |
93 |
|
|
94 |
echo > in |
function uniqurl() { |
95 |
#mirror "http://safari.oreilly.com/?XmlId=$isbn" |
mv in in.tmp |
96 |
|
grep -v 'view=[A-Z]' in.tmp | grep 'a=0$' | sort -u > in |
97 |
|
grep 'view=[A-Z].*/index' in.tmp | grep 'a=0$' | sort -u >> in |
98 |
|
links=`wc -l in | cut -d" " -f1` |
99 |
|
echo "found $links unique links" |
100 |
|
} |
101 |
|
|
102 |
|
function mirror_in() { |
103 |
|
cat in | while read url ; do |
104 |
|
mirror "$url" |
105 |
|
#sleep $wait |
106 |
|
|
107 |
|
if grep 'promo.asp' `ls -t index.html* | head -3` >/dev/null ; then |
108 |
|
echo "WARNING: safari seems to logout you as user. Aborting." |
109 |
|
exit 1 |
110 |
|
fi |
111 |
|
|
112 |
|
if grep -i '>Account locked<' `ls -t index.html* | head -3` >/dev/null ; then |
113 |
|
echo "WARNING: your safari account is locked. Aborting." |
114 |
|
exit 1 |
115 |
|
fi |
116 |
|
|
117 |
|
if grep -i 'session disabled' `ls -t index.html* | head -3` >/dev/null ; then |
118 |
|
echo "WARNING: your safari session is disabled. Aborting." |
119 |
|
exit 1 |
120 |
|
fi |
121 |
|
|
122 |
|
if grep -i 'This is only a preview of the full book' `ls -t index.html* | head -3` >/dev/null ; then |
123 |
|
echo "WARNING: you didn't add this book to your bookshelf!" |
124 |
|
exit 1 |
125 |
|
fi |
126 |
|
done |
127 |
|
echo |
128 |
|
} |
129 |
|
|
130 |
|
echo -n > in |
131 |
|
mirror "http://safari.oreilly.com/$isbn" |
132 |
|
echo |
133 |
|
|
134 |
echo "extract URLs from first page..." |
file=`ls *index.html?XmlId=*` |
135 |
geturl "index.html?XmlId=$isbn" $isbn |
isbn=`echo $file | cut -d= -f2` |
136 |
|
echo "extract URLs from first page $file... [$isbn]" |
137 |
|
geturl $file $isbn |
138 |
|
uniqurl |
139 |
|
|
140 |
mirror "-i in" |
mirror_in |
141 |
|
|
142 |
echo -n "extracting URLs [1]" |
echo -n "extracting URLs [1]" |
143 |
ls index.html* | while read file ; do |
ls index.html* | while read file ; do |
146 |
done |
done |
147 |
echo |
echo |
148 |
|
|
149 |
sort -u in > in2 |
uniqurl |
150 |
|
|
151 |
mirror "-i in2" |
mirror_in |
152 |
|
|
153 |
echo > in |
echo -n > in |
154 |
echo -n "extracting URLs [2]" |
echo -n "extracting URLs [2]" |
155 |
ls index.html* | while read file ; do |
ls index.html* | while read file ; do |
156 |
echo -n "." |
echo -n "." |
157 |
geturl $file $isbn |
geturl $file $isbn |
158 |
done |
done |
159 |
|
|
160 |
sort -u in > in2 |
uniqurl |
161 |
|
|
162 |
|
mirror_in |
163 |
|
|
164 |
|
# convert links in html |
165 |
|
bn=`basename $0` |
166 |
|
dir=`echo $0 | sed "s/$bn$//"` |
167 |
|
ls index.html* | xargs -i $dir/filter.pl {} |
168 |
|
mkdir orig |
169 |
|
mv index.html* orig/ |
170 |
|
|
|
mirror "-i in2" |
|