1 |
#!/opt/bin/perl |
2 |
|
3 |
BEGIN { require "common.pl" } |
4 |
our ($TXN, %stat, $db_env, $TPB, $db_info, $TODAY); |
5 |
|
6 |
use common::sense; |
7 |
|
8 |
our $NUM_DOWNLOADERS = 2; # minimum 2 |
9 |
|
10 |
our $db_http = table "http"; |
11 |
our $db_torrent = table "torrent"; |
12 |
|
13 |
our $TODO; |
14 |
|
15 |
############################################################################# |
16 |
|
17 |
my $pool = new Coro::Channel 10000; |
18 |
my @pool = map { |
19 |
async { |
20 |
while (my $args = $pool->get) { |
21 |
my $cb = pop @$args; |
22 |
$cb->(@$args); |
23 |
} |
24 |
} |
25 |
} 1..1; |
26 |
|
27 |
sub pool(&@) { |
28 |
$pool->put ([@_]); |
29 |
} |
30 |
|
31 |
sub http_extract($) { |
32 |
$_[0] =~ / id='NumComments'>(\d+)</ |
33 |
or die "no numcomments in <$_[0]>\n"; |
34 |
$1 |
35 |
} |
36 |
|
37 |
my $finish = new Coro::Channel 100; |
38 |
my @finishers = async { |
39 |
{ |
40 |
my $txn; |
41 |
$Coro::current->swap_sv (\$TXN, \$txn); |
42 |
} |
43 |
|
44 |
while (my $job = $finish->get) { |
45 |
# $TXN = $db_env->txn_begin; |
46 |
|
47 |
my ($i, $html, $torrent) = @$job; |
48 |
|
49 |
$html =~ s/^.*?Details for this torrent//s; |
50 |
$html =~ s/<div id="foot".*$//s; |
51 |
|
52 |
my $info = iget $i; |
53 |
|
54 |
my $old = get $db_http, $i; |
55 |
|
56 |
unless ($!) { |
57 |
my $old_html = dic_decompress 0, $old; |
58 |
|
59 |
if ((http_extract $html) ne eval { http_extract $old_html }) { |
60 |
++$stat{change}; |
61 |
$info->[1] = ($info->[1] >> 1) || 1; |
62 |
} else { |
63 |
++$stat{same}; |
64 |
$info->[1] *= 2; |
65 |
$info->[1] = 365 if $info->[1] > 365; |
66 |
} |
67 |
} |
68 |
|
69 |
$stat{data} += length $html; |
70 |
$html = dic_compress 0, $html if length $html; |
71 |
$stat{cmpr} += length $html; |
72 |
|
73 |
$stat{torrent} += length $torrent; |
74 |
|
75 |
put $db_torrent, $i, $torrent if defined $torrent; |
76 |
put $db_http, $i, $html; |
77 |
iput $i, [$TODAY, $info->[1]]; # check every day |
78 |
|
79 |
# db_txn_finish $TXN; |
80 |
# die "transaction failed" if $!; |
81 |
|
82 |
remove $TODO $i; |
83 |
} |
84 |
}; |
85 |
|
86 |
############################################################################# |
87 |
|
88 |
my $download = new Coro::Channel 200; |
89 |
my @downloaders = map { |
90 |
async { |
91 |
while (my $i = $download->get) { |
92 |
my ($html, $hdr) = GET "https://$TPB/torrent/$i"; |
93 |
|
94 |
if ($hdr->{Status} == 200) { |
95 |
BDB::db_exists $db_torrent, undef, $i, 0; |
96 |
if ($!) { |
97 |
if ($html =~ m%<a href="([^"]+)" title="Torrent File"%) { |
98 |
my $url = $1; |
99 |
$url =~ s/ /%20/g; |
100 |
$url =~ s%^//%https://%; |
101 |
$url =~ s%^/+%https://$TPB/%; |
102 |
my ($torrent, $hdr) = GET $url; |
103 |
|
104 |
if ($hdr->{Status} == 200) { |
105 |
$finish->put ([$i, $html, $torrent]); |
106 |
} elsif ($hdr->{Status} == 404) { |
107 |
++$stat{notorrent}; |
108 |
$finish->put ([$i, $html, undef]); |
109 |
|
110 |
#warn "https://$TPB/torrent/$i $url\n"; |
111 |
#remove $TODO $i; |
112 |
} else { |
113 |
warn "$url https://$TPB/torrent/$i: status $hdr->{Status}\n"; |
114 |
} |
115 |
} else { |
116 |
++$stat{notorrentlink}; |
117 |
$finish->put ([$i, $html, undef]); |
118 |
} |
119 |
} else { |
120 |
$finish->put ([$i, $html, undef]); |
121 |
} |
122 |
} elsif ($hdr->{Status} == 404) { |
123 |
remove $TODO $i; |
124 |
del $db_info, $i; |
125 |
} else { |
126 |
warn "https://$TPB/torrent/$i: status $hdr->{Status}\n"; |
127 |
} |
128 |
} |
129 |
} |
130 |
} 1..$NUM_DOWNLOADERS; |
131 |
|
132 |
############################################################################# |
133 |
|
134 |
if (0) { |
135 |
#sput "last_recent", 7100000; |
136 |
sput "last_recent", 3210000; |
137 |
} |
138 |
|
139 |
if (0) { |
140 |
db_env_txn_checkpoint $db_env; |
141 |
db_compact $db_info; |
142 |
db_env_txn_checkpoint $db_env; |
143 |
return; |
144 |
} |
145 |
|
146 |
async { |
147 |
$TODO = new Set::IntSpan sget "todo"; |
148 |
|
149 |
if (1) { |
150 |
my $last = sget "last_recent"; |
151 |
|
152 |
my ($html, $hdr) = GET "https://$TPB/recent"; |
153 |
|
154 |
$html =~ m%<div class="detName">\s*<a href="/torrent/(\d+)/% |
155 |
or die "https://$TPB/recent: unable to parse recent torrent ($html)"; |
156 |
|
157 |
my $next = $1; |
158 |
#$next = 7043798;#d# |
159 |
|
160 |
$TODO->U ([[$last, $next]]); |
161 |
|
162 |
sput todo => "$TODO"; |
163 |
sput last_recent => $next; |
164 |
} |
165 |
|
166 |
my $sputter = AE::timer 10, 10, sub { |
167 |
async_pool { |
168 |
sput todo => "$TODO"; |
169 |
}; |
170 |
}; |
171 |
|
172 |
# new torrents |
173 |
if (1) { |
174 |
for (spans $TODO) { |
175 |
for my $i ($_->[0] .. $_->[1]) { |
176 |
$stat{stat} = $i; |
177 |
BDB::db_exists $db_http, undef, $i, 0; |
178 |
unless ($!) { |
179 |
BDB::db_exists $db_torrent, undef, $i, 0; |
180 |
unless ($!) { |
181 |
# already have both |
182 |
remove $TODO $i; |
183 |
next; |
184 |
} |
185 |
} |
186 |
|
187 |
iput $i, [0, 1]; |
188 |
# $download->put ($i);#d# |
189 |
} |
190 |
} |
191 |
} |
192 |
|
193 |
# refresh |
194 |
if (1) { |
195 |
my $c = $db_info->cursor (undef, 0); |
196 |
my ($k, $v); |
197 |
db_c_get $c, $k, $v, BDB::SET_RANGE; |
198 |
my @i; |
199 |
|
200 |
until ($!) { |
201 |
my $info = decode_cbor $v; |
202 |
push @i, $k |
203 |
# $download->put ($k) |
204 |
if $info->[0] + $info->[1] <= $TODAY; |
205 |
db_c_get $c, $k, $v, BDB::NEXT; |
206 |
#d# last if @i > 200000;#d# |
207 |
} |
208 |
|
209 |
undef $c; |
210 |
|
211 |
$download->put ($_) |
212 |
for @i; |
213 |
} |
214 |
|
215 |
$download->shutdown; |
216 |
$_->join |
217 |
for @downloaders; |
218 |
|
219 |
$finish->shutdown; |
220 |
$_->join for @finishers; |
221 |
|
222 |
$pool->shutdown; |
223 |
$_->join for @pool; |
224 |
|
225 |
sput todo => "$TODO"; |
226 |
|
227 |
EV::unloop; |
228 |
}; |
229 |
|
230 |
EV::loop; |
231 |
|