| 1 |
#!/usr/bin/perl |
| 2 |
use strict; |
| 3 |
|
| 4 |
use lib qw[/home/httpd/html/www/markup/html/whatpm |
| 5 |
/home/wakaba/work/manakai2/lib |
| 6 |
../lib |
| 7 |
]; |
| 8 |
|
| 9 |
my $log_uri; |
| 10 |
|
| 11 |
use Getopt::Long; |
| 12 |
GetOptions ( |
| 13 |
'log-uri=s' => \$log_uri, |
| 14 |
); |
| 15 |
die unless defined $log_uri; |
| 16 |
|
| 17 |
my $diff_command = 'diff'; |
| 18 |
my $diff_options = ['--rcs', '-a']; |
| 19 |
|
| 20 |
require Message::DOM::DOMImplementation; |
| 21 |
my $dom = 'Message::DOM::DOMImplementation'; |
| 22 |
|
| 23 |
my $ent = get_remote_entity ($log_uri); |
| 24 |
unless (defined $ent->{s}) { |
| 25 |
die "<$log_uri>: $ent->{error_status_text}"; |
| 26 |
} |
| 27 |
|
| 28 |
require RCSFormat; |
| 29 |
|
| 30 |
my $rcs = new RCSFormat; |
| 31 |
$rcs->{admin}->{strict} = 1; |
| 32 |
$rcs->{admin}->{comment} = '# '; |
| 33 |
$rcs->{admin}->{expand} = 'b'; |
| 34 |
|
| 35 |
my $text = {}; |
| 36 |
my $text_from = {}; |
| 37 |
|
| 38 |
#if ($ent->{s} =~ m!<title>CVS log for ([^<>]+)</title>!gc) { |
| 39 |
# $info->{file_path} = htunescape ($1); |
| 40 |
#} |
| 41 |
|
| 42 |
my $branches = {}; |
| 43 |
my $branch_rev = {}; |
| 44 |
if ($ent->{s} =~ m[View only Branch]gc) { |
| 45 |
while ($ent->{s} =~ m[<option value="([^"]+)"|</select]gc) { |
| 46 |
last unless defined $1; |
| 47 |
$branches->{$1} = 1; |
| 48 |
} |
| 49 |
} |
| 50 |
|
| 51 |
pos ($ent->{s}) = 0; |
| 52 |
|
| 53 |
REV: while ($ent->{s} =~ m!<hr size=1 noshade>!gc) { |
| 54 |
my $rev = ''; |
| 55 |
if ($ent->{s} =~ m!Revision!gc and |
| 56 |
$ent->{s} =~ m!<a href="([^"]+)"[^<>]*><b>([0-9.]+)</b></a>!gc) { |
| 57 |
my $uri = $dom->create_uri_reference ($1) |
| 58 |
->get_absolute_reference ($log_uri)->uri_reference; |
| 59 |
$rev = $2; |
| 60 |
|
| 61 |
my $revent = get_remote_entity ($uri); |
| 62 |
if (defined $revent->{s}) { |
| 63 |
$text->{$rev} = $revent->{s}; |
| 64 |
} |
| 65 |
|
| 66 |
$rcs->{admin}->{head} ||= $rev; |
| 67 |
} |
| 68 |
|
| 69 |
if ($ent->{s} =~ m!<i>\w+ (\w+)\s*(\d+)\s*(\d+):(\d+):(\d+) (\d+) UTC</i> \([^()]+\) by <i>([^<>]+)</i>!gc) { |
| 70 |
$rcs->{delta}->{$rev}->{date} = sprintf '%02d.%02d.%02d.%02d.%02d.%02d', |
| 71 |
$6, { |
| 72 |
Jan => '01', Feb => '02', Mar => '03', Apr => '04', |
| 73 |
May => '05', Jun => '06', Jul => '07', Aug => '08', |
| 74 |
Sep => '09', Oct => '10', Nov => '11', Dec => '12', |
| 75 |
}->{$1}, $2, $3, $4, $5; |
| 76 |
$rcs->{delta}->{$rev}->{author} = $7; |
| 77 |
} |
| 78 |
|
| 79 |
$ent->{s} =~ m[(?=<(?>br|pre)>)]gc; |
| 80 |
|
| 81 |
if ($ent->{s} =~ m[\G<br>Branch:((?>(?!<(?>[bh]r|pre)).)+)]gcs) { |
| 82 |
my $b = $1; |
| 83 |
while ($b =~ m!><b>([^<>]+)</b></a>!gc) { |
| 84 |
my $branch = $1; |
| 85 |
if ($branch ne 'MAIN') { |
| 86 |
my $rev = $rev; |
| 87 |
$rev =~ s/\.(\d+)\.\d+\z//; |
| 88 |
push @{$rcs->{admin}->{symbols} ||= []}, [$branch => "$rev.0.$1"]; |
| 89 |
} |
| 90 |
} |
| 91 |
$ent->{s} =~ m[(?=<(?>br|pre)>)]gc; |
| 92 |
} |
| 93 |
|
| 94 |
if ($ent->{s} =~ m[\G<br>Branch point]gcs) { |
| 95 |
$ent->{s} =~ m[(?=<(?>br|pre)>)]gc; |
| 96 |
} |
| 97 |
|
| 98 |
if ($ent->{s} =~ m[\G<br>CVS Tags:((?>(?!<(?>[bh]r|pre)).)+)]gcs) { |
| 99 |
my $b = $1; |
| 100 |
while ($b =~ m!><b>([^<>]+)</b></a>!gc) { |
| 101 |
my $rev = $branches->{$1} ? "$rev.0.".((++$branch_rev->{$rev})*2) : $rev; |
| 102 |
if ($1 eq 'HEAD') { |
| 103 |
$rcs->{admin}->{head} = $rev; |
| 104 |
$rcs->{deltatext}->{$rev}->{text} = $text->{$rev}; |
| 105 |
} else { |
| 106 |
push @{$rcs->{admin}->{symbols} ||= []}, [$1 => $rev]; |
| 107 |
} |
| 108 |
} |
| 109 |
$ent->{s} =~ m[(?=<(?>br|pre)>)]gc; |
| 110 |
} |
| 111 |
|
| 112 |
if ($ent->{s} =~ m[\G<br>Branch point]gcs) { |
| 113 |
$ent->{s} =~ m[(?=<(?>br|pre)>)]gc; |
| 114 |
} |
| 115 |
|
| 116 |
if ($ent->{s} =~ m!\G<br>Changes since <b>([0-9.]+):!gc) { |
| 117 |
my $from_rev = $1; |
| 118 |
if ($rev =~ /\A[0-9]+\.[0-9]+\z/) { |
| 119 |
$rcs->{delta}->{$rev}->{next} = $from_rev; |
| 120 |
$text_from->{$from_rev} = $rev; |
| 121 |
} elsif ($from_rev =~ /\A[0-9]+\.[0-9]+\z/) { |
| 122 |
push @{$rcs->{delta}->{$from_rev}->{branches} ||= []}, $rev; |
| 123 |
$text_from->{$rev} = $from_rev; |
| 124 |
} else { |
| 125 |
$rcs->{delta}->{$from_rev}->{next} = $rev; |
| 126 |
$text_from->{$rev} = $from_rev; |
| 127 |
} |
| 128 |
} |
| 129 |
|
| 130 |
if ($ent->{s} =~ m!<pre>(.*?)</pre>!gcs) { |
| 131 |
$rcs->{deltatext}->{$rev}->{log} = htunescape ($1); |
| 132 |
} |
| 133 |
|
| 134 |
$rcs->{delta}->{$rev}->{state} = 'Exp'; |
| 135 |
} # REV |
| 136 |
|
| 137 |
delete $text_from->{$rcs->{admin}->{head}}; |
| 138 |
|
| 139 |
my %symbol; |
| 140 |
for (@{$rcs->{admin}->{symbols} || []}) { |
| 141 |
$symbol{$_->[0]} = $_->[1]; |
| 142 |
} |
| 143 |
$rcs->{admin}->{symbols} = [map {[$_ => $symbol{$_}]} keys %symbol]; |
| 144 |
|
| 145 |
|
| 146 |
require File::Temp; |
| 147 |
require IPC::Open2; |
| 148 |
while (keys %$text_from) { |
| 149 |
my $rev = each %$text_from; |
| 150 |
my $any; |
| 151 |
if (defined $text->{$text_from->{$rev}}) { |
| 152 |
$any = 1; |
| 153 |
my $from_file = new File::Temp; |
| 154 |
binmode $from_file; |
| 155 |
$from_file->print ($text->{$text_from->{$rev}}); |
| 156 |
my $to_file = new File::Temp; |
| 157 |
binmode $to_file; |
| 158 |
$to_file->print ($text->{$rev}); |
| 159 |
my (undef, $diff_filemame) = File::Temp::tempfile |
| 160 |
('DIFFXXXX', DIR => File::Temp::tempdir (), OPEN => 0); |
| 161 |
IPC::Open2::open2 (my $diffin, my $diffout, |
| 162 |
$diff_command, @{$diff_options}, |
| 163 |
$from_file->filename => $to_file->filename) |
| 164 |
or die "$0: $diff_command: $!"; |
| 165 |
binmode $diffin; |
| 166 |
local $/ = undef; |
| 167 |
$rcs->{deltatext}->{$rev}->{text} = <$diffin>; |
| 168 |
delete $text_from->{$rev}; |
| 169 |
} |
| 170 |
last unless $any; |
| 171 |
} |
| 172 |
|
| 173 |
print $rcs->stringify; |
| 174 |
|
| 175 |
sub htunescape ($) { |
| 176 |
my $s = shift; |
| 177 |
$s =~ s!<[^<>]+>!!g; |
| 178 |
$s =~ s/</</g; |
| 179 |
$s =~ s/>/>/g; |
| 180 |
$s =~ s/&/&/g; |
| 181 |
return $s; |
| 182 |
} # htunescape |
| 183 |
|
| 184 |
sub get_remote_entity ($) { |
| 185 |
my $request_uri = $_[0]; |
| 186 |
my $r = {}; |
| 187 |
|
| 188 |
my $uri = $dom->create_uri_reference ($request_uri); |
| 189 |
unless ({ |
| 190 |
http => 1, |
| 191 |
}->{lc $uri->uri_scheme}) { |
| 192 |
return {uri => $request_uri, request_uri => $request_uri, |
| 193 |
error_status_text => 'URI scheme not allowed'}; |
| 194 |
} |
| 195 |
|
| 196 |
require Message::Util::HostPermit; |
| 197 |
my $host_permit = new Message::Util::HostPermit; |
| 198 |
$host_permit->add_rule (<<EOH); |
| 199 |
Allow host=suika port=80 |
| 200 |
Deny host=suika |
| 201 |
Allow host=suika.fam.cx port=80 |
| 202 |
Deny host=suika.fam.cx |
| 203 |
Deny host=localhost |
| 204 |
Deny host=*.localdomain |
| 205 |
Deny ipv4=0.0.0.0/8 |
| 206 |
Deny ipv4=10.0.0.0/8 |
| 207 |
Deny ipv4=127.0.0.0/8 |
| 208 |
Deny ipv4=169.254.0.0/16 |
| 209 |
Deny ipv4=172.0.0.0/11 |
| 210 |
Deny ipv4=192.0.2.0/24 |
| 211 |
Deny ipv4=192.88.99.0/24 |
| 212 |
Deny ipv4=192.168.0.0/16 |
| 213 |
Deny ipv4=198.18.0.0/15 |
| 214 |
Deny ipv4=224.0.0.0/4 |
| 215 |
Deny ipv4=255.255.255.255/32 |
| 216 |
Deny ipv6=0::0/0 |
| 217 |
Allow host=* |
| 218 |
EOH |
| 219 |
unless ($host_permit->check ($uri->uri_host, $uri->uri_port || 80)) { |
| 220 |
return {uri => $request_uri, request_uri => $request_uri, |
| 221 |
error_status_text => 'Connection to the host is forbidden'}; |
| 222 |
} |
| 223 |
|
| 224 |
require LWP::UserAgent; |
| 225 |
my $ua = WDCC::LWPUA->new; |
| 226 |
$ua->{wdcc_dom} = $dom; |
| 227 |
$ua->{wdcc_host_permit} = $host_permit; |
| 228 |
$ua->agent ('Mozilla'); ## TODO: for now. |
| 229 |
$ua->parse_head (0); |
| 230 |
$ua->protocols_allowed ([qw/http/]); |
| 231 |
$ua->max_size (1000_000); |
| 232 |
my $req = HTTP::Request->new (GET => $request_uri); |
| 233 |
my $res = $ua->request ($req); |
| 234 |
## TODO: 401 sets |is_success| true. |
| 235 |
if ($res->is_success) { |
| 236 |
$r->{base_uri} = $res->base; ## NOTE: It does check |Content-Base|, |Content-Location|, and <base>. ## TODO: Use our own code! |
| 237 |
$r->{uri} = $res->request->uri; |
| 238 |
$r->{request_uri} = $request_uri; |
| 239 |
|
| 240 |
## TODO: More strict parsing... |
| 241 |
my $ct = $res->header ('Content-Type'); |
| 242 |
if (defined $ct and $ct =~ m#^([0-9A-Za-z._+-]+/[0-9A-Za-z._+-]+)#) { |
| 243 |
$r->{media_type} = lc $1; |
| 244 |
} |
| 245 |
if (defined $ct and $ct =~ /;\s*charset\s*=\s*"?(\S+)"?/i) { |
| 246 |
$r->{charset} = lc $1; |
| 247 |
$r->{charset} =~ tr/\\//d; |
| 248 |
} |
| 249 |
|
| 250 |
$r->{s} = ''.$res->content; |
| 251 |
} else { |
| 252 |
$r->{uri} = $res->request->uri; |
| 253 |
$r->{request_uri} = $request_uri; |
| 254 |
$r->{error_status_text} = $res->status_line; |
| 255 |
} |
| 256 |
|
| 257 |
$r->{header_field} = []; |
| 258 |
$res->scan (sub { |
| 259 |
push @{$r->{header_field}}, [$_[0], $_[1]]; |
| 260 |
}); |
| 261 |
$r->{header_status_code} = $res->code; |
| 262 |
$r->{header_status_text} = $res->message; |
| 263 |
|
| 264 |
return $r; |
| 265 |
} # get_remote_entity |
| 266 |
|
| 267 |
package WDCC::LWPUA; |
| 268 |
BEGIN { push our @ISA, 'LWP::UserAgent'; } |
| 269 |
|
| 270 |
sub redirect_ok { |
| 271 |
my $ua = shift; |
| 272 |
unless ($ua->SUPER::redirect_ok (@_)) { |
| 273 |
return 0; |
| 274 |
} |
| 275 |
|
| 276 |
my $uris = $_[1]->header ('Location'); |
| 277 |
return 0 unless $uris; |
| 278 |
my $uri = $ua->{wdcc_dom}->create_uri_reference ($uris); |
| 279 |
unless ({ |
| 280 |
http => 1, |
| 281 |
}->{lc $uri->uri_scheme}) { |
| 282 |
return 0; |
| 283 |
} |
| 284 |
unless ($ua->{wdcc_host_permit}->check ($uri->uri_host, $uri->uri_port || 80)) { |
| 285 |
return 0; |
| 286 |
} |
| 287 |
return 1; |
| 288 |
} # redirect_ok |
Parent Directory
| 
Revision Log