1 |
#!/bin/sh |
#!/bin/sh |
2 |
|
|
3 |
# proxy settings (same as in firebird) |
# proxy settings (same as in firebird) |
4 |
fping proxy && export http_proxy=http://proxy:8080 |
fping -q proxy && export http_proxy=http://proxy:8080 |
5 |
|
if [ -z "$http_proxy" ] ; then |
6 |
|
fping -q proxy.lan && export http_proxy=http://proxy.lan:8080 |
7 |
|
fi |
8 |
# user agent (same as in firebird) |
# user agent (same as in firebird) |
9 |
ua="Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031225 Firebird/0.7" |
ua="Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040506 Firefox/0.8" |
10 |
|
cookie_file="cookie.txt" |
11 |
|
|
12 |
# wait between pages |
# wait between pages |
13 |
export wait=120 |
export wait=120 |
14 |
|
|
15 |
|
if [ ! -z "$http_proxy" ] ; then |
16 |
|
echo "Using proxy $http_proxy" |
17 |
|
fi |
18 |
|
|
19 |
if [ -z "$1" ] ; then |
if [ -z "$1" ] ; then |
20 |
echo "Usage: $0 ISBN" |
echo "Usage: $0 ISBN" |
34 |
grep -l 'This is only a preview of the full book' * | xargs -i rm {} |
grep -l 'This is only a preview of the full book' * | xargs -i rm {} |
35 |
fi |
fi |
36 |
|
|
37 |
isbn=$1; |
isbn=`echo $1 | sed 's/-//g' | tr '[a-z]' '[A-Z]'` |
|
|
|
|
isbn2=`echo $isbn | sed 's/-//g'` |
|
38 |
|
|
39 |
function mirror() { |
function mirror() { |
40 |
|
|
43 |
file=`echo $url | sed -e s,http://[^?]*?,index.html?, -e s,#.*$,, -e s,/,%2F,g` |
file=`echo $url | sed -e s,http://[^?]*?,index.html?, -e s,#.*$,, -e s,/,%2F,g` |
44 |
if [ -e "$file" ] ; then |
if [ -e "$file" ] ; then |
45 |
# echo "skip $url" |
# echo "skip $url" |
46 |
|
echo -n "." |
47 |
return |
return |
48 |
fi |
fi |
49 |
|
|
50 |
|
cookie="" |
51 |
if echo $url | grep '/index' >/dev/null ; then |
if echo $url | grep '/index' >/dev/null ; then |
|
cookies="" |
|
52 |
echo -n "no login (index) " |
echo -n "no login (index) " |
53 |
elif echo $url | grep 'mode=toc' >/dev/null ; then |
elif echo $url | grep 'mode=toc' >/dev/null ; then |
|
cookies="" |
|
54 |
echo -n "no login (toc) " |
echo -n "no login (toc) " |
55 |
else |
else |
56 |
cookies="--load-cookies=/home/dpavlin/.phoenix/default/g6b45nt6.slt/cookies.txt" |
if [ ! -e $cookie_file ] ; then |
57 |
echo -n "with login " |
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 |
fi |
|
echo $url |
|
68 |
|
|
69 |
wget -p -nH -nc -k \ |
if [ -z "$cookie" ] ; then |
70 |
-t 1 -U "$ua" \ |
echo "$url [no cookie]" |
71 |
$cookies $url |
else |
72 |
|
echo "$url [with cookie]" |
73 |
|
fi |
74 |
|
|
75 |
# -D safari.oreilly.com \ |
wget -q -p -nH -nc -k -t 1 -U "$ua" --cookies=off --header="$cookie" $url |
76 |
# -A 0-201-41975-0 \ |
perl -e '$t=rand($ENV{wait} || 120);print "sleep for $t sec.\n"; sleep($t);' |
77 |
|
|
|
perl -e '$t=rand($ENV{wait} || 30);print "sleep for $t sec.\n"; sleep($t);' |
|
78 |
} |
} |
79 |
|
|
80 |
function geturl() { |
function geturl() { |
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 |
-e 's/open=false/open=true/' | \ |
-e 's/open=false/open=true/' | \ |
88 |
grep '&s=1&b=1&f=1&t=1&c=1&u=1&r=&o=1' | \ |
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 $2 | \ |
grep -i "xmlid=[0-9A-Za-z\-][0-9A-Za-z\-]*" | \ |
90 |
grep -v "$2/[0-9][0-9][0-9][0-9][0-9][0-9][0-9]" | \ |
grep -vi "xmlid=[0-9A-Za-z\-]*/[0-9][0-9][0-9][0-9][0-9][0-9][0-9]" | \ |
91 |
sort -u >> in |
sort -u >> in |
92 |
} |
} |
93 |
|
|
94 |
function uniqurl() { |
function uniqurl() { |
95 |
mv in in.tmp |
mv in in.tmp |
96 |
grep -v 'view=[A-Z]' in.tmp | sort -u > in |
grep -v 'view=[A-Z]' in.tmp | grep 'a=0$' | sort -u > in |
97 |
grep 'view=[A-Z].*/index' in.tmp | sort -u >> in |
grep 'view=[A-Z].*/index' in.tmp | grep 'a=0$' | sort -u >> in |
98 |
|
links=`wc -l in | cut -d" " -f1` |
99 |
|
echo "found $links unique links" |
100 |
} |
} |
101 |
|
|
102 |
function mirror_in() { |
function mirror_in() { |
124 |
exit 1 |
exit 1 |
125 |
fi |
fi |
126 |
done |
done |
127 |
|
echo |
128 |
} |
} |
129 |
|
|
130 |
echo -n > in |
echo -n > in |
131 |
mirror "http://safari.oreilly.com/?XmlId=$isbn" |
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 |
uniqurl |
139 |
|
|
140 |
mirror_in |
mirror_in |