ignore all .sync files from client
[cloudstore.git] / lib / CloudStore / Store.pm
1 package CloudStore::Store;
2 use warnings;
3 use strict;
4
5 use lib 'lib';
6 use base 'CloudStore::MD5sum';
7 use CloudStore::API;
8
9 use autodie;
10 use JSON::XS;
11 use File::Path qw(make_path);
12 use File::Slurp qw();
13 use Digest::MD5 qw(md5_base64);
14 use Data::Dump qw(dump);
15 use Carp qw(confess);
16
17 use WarnColor;
18
19 sub new {
20         my ($class,$group) = @_;
21
22         my $self = {
23                 api => CloudStore::API->new( $group ),
24         };
25         bless $self, $class;
26
27         $self->{md5pool} = $self->{api}->{md5}->{dir};
28
29         warn "# new ",dump $self if $ENV{DEBUG};
30
31         return $self;
32 }
33
34 sub mkbasedir {
35         my $dir = shift;
36         $dir =~ s{/[^/]+$}{}; # strip filename
37         mkdir $dir unless -e $dir;
38 }
39
40 sub modify_file {
41         my ( $self,$data ) = @_;
42
43 =for removed
44
45         if ( $data->{file} =~ m{^(.*/)?.sync/send/([^/]+)$} ) {
46                 my $from_dir = $1;
47                 warn "SEND $2 from $from_dir\n";
48                 my $sent_files;
49                 open(my $send, '<', $self->blob_path($data) );
50                 while(<$send>) {
51                         s/[\n\r]+$//;
52
53                         my ( $to, $file ) = split(/\s+/,$_,2);
54                         my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
55                                 getpwnam $to;
56
57                         my $from = $data;
58                         $from->{file} = $from_dir . $file;
59                         my $from_path = $self->blob_path($from);
60
61                         if ( ! -r $from_path ) {
62                                 warn "ERROR: $from_path: $!";
63                                 next;
64                         }
65
66                         my $to_path = "$dir/received/$file";
67                         mkbasedir $to_path;
68
69                         warn "SEND $from_path -> $to_path\n";
70                         unlink $to_path if -e $to_path; # FIXME why we need this?
71                         $sent_files->{$to} += link $from_path, $to_path;
72                         # FIXME cross-shard
73                         my $md5 = $self->md5sum($data)->get( $from_dir . $file ) || warn "no md5 for $from_dir$file";
74                         $self->md5sum({login => $to})->put( "/received/$file" => $md5 );
75                 }
76
77                 warn "SENT ",dump $sent_files;
78
79                 return 0; # skip dedup
80         } elsif ( $data->{file} =~ m{^(.*/)?.sync/pending/([^/]+)$} ) {
81                 my $from_dir = $1;
82                 warn "PENDIG $2 from $from_dir";
83                 open(my $pend, '<', $self->blob_path($data) );
84                 while(<$pend>) {
85                         s/[\n\r]+$//;
86
87                         if ( m/^DELETED\#(.+)$/ ) {
88                                 my $path = $self->blob_path($data => $from_dir . $1 );
89                                 if ( -e $path ) {
90                                         warn "UNLINK $path";
91                                         -d $path ? rmdir $path : unlink $path || warn "ERROR: unlink $path $!";
92                                         next;
93                                 } else {
94                                         warn "MISSING $path to unlink";
95                                         next;
96                                 }
97                         } elsif ( ! /^(MOVED|RENAMED)\#/ ) {
98                                 warn "skip $_\n";
99                                 next;
100                         }
101
102                         my ( undef, $from, $to ) = split(/\#/,$_,3);
103
104                         my ( $from_path, $to_path ) = map {
105                                 my $tmp = $data;
106                                 $tmp->{file} = $from_dir . $_;
107                                 $self->blob_path($tmp);
108                         } ( $from, $to );
109
110                         if ( ! -e $from_path ) {
111                                 warn "SKIPPED $from_path: $!";
112                                 next;
113                         }
114
115                         warn "MV $from_path -> $to_path";
116                         mkbasedir $to_path;
117                         rename $from_path, $to_path;
118         
119                         my $md5 = $self->md5sum($data)->get( $from_dir . $from );
120                         if ( ! $md5 ) {
121                                 warn "ERROR: no md5sum $from_dir $from " unless $md5;
122                                 next;
123                         }
124
125                         $self->md5sum($data)->out( $from_dir . $from );
126                         $self->md5sum($data)->put( $from_dir . $to => $md5 );
127
128                         warn "$md5 moved to $from_dir $to";
129                 }
130
131                 return 0; # skip dedup
132         }
133
134 =cut
135
136         if ( $data->{file} =~ m{^(.*/)?.sync/} ) {
137                 # ignore .sync/ files from client
138                 return 0;
139         }
140
141         #return $file->{size} > 4096 ? 1 : 0; # FIXME
142         return 1; # dedup
143 }
144
145 # never called by rsync directly!
146 sub new_file {
147         my ( $self,$data ) = @_;
148 #       $self->file_set($data);
149 }
150
151 sub removed_file {
152         my ( $self, $data ) = @_;
153
154         my $md5 = $self->md5sum($data)->get( $data->{file} );
155         return unless $md5; # directories don't have md5sums
156         my $path = $self->{md5pool} . '/' . $md5;
157         my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,
158                 $atime,$mtime,$ctime,$blksize,$blocks)
159                         = stat($path);
160
161         my $user = $self->{api}->user_info($data->{login});
162
163         if ( $nlink == 1 && $uid == $user->{uid} ) {
164                 $self->append( $user, 'removed', -$size, $uid, $data->{file} );
165                 my $id = getpwnam 'md5';
166                 chown $id,$gid, $path;
167                 warn "# chown $id $gid $path";
168         }
169
170         $self->md5sum($data)->out( $data->{file} );
171 }
172
173 sub make_dir {
174         my ( $self, $data ) = @_;
175
176 }
177
178 sub new_link {
179         my ( $self, $data ) = @_;
180
181         warn "# new_link ",dump $data;
182
183         if ( $data->{file} =~ m{^(.*/?)\.send/([^/]+)/(.+)$} ) {
184                 my ( $dir, $to, $name ) = ( $1, $2, $3 );
185                 my $path = $self->blob_path($data);
186                 my $link_to = readlink $path;
187                 warn "$link_to";
188                 if ( $link_to =~ s{^\Q/rsyncd-munged/\E}{/} ) {
189
190                         my $s = $path;
191                         $s =~ s{/[^/]+$}{}; # strip filename
192                         while ( $link_to =~ s{/../}{/} ) {
193                                 $s =~ s{/[^/]+$}{} || die "can't strip $s";
194                                 warn "## simplify $s $link_to\n";
195                         }
196                         $s .= $link_to;
197
198                         my $d = $self->blob_path({
199                                 pid => $data->{pid},
200                                 file => $name
201                         });
202
203                         # $name can contain directories so we must create them
204                         mkbasedir $d;
205
206                         if ( ! -e $s ) {
207                                 warn "ERROR: can't find source $s";
208                         } else {
209
210                                 warn "link $s -> $d\n";
211                                 link $s, $d;
212
213                                 my ($l,$f) = ($1,$2) if $s =~ m{users/([^/]+)/blob/(.+)};
214
215 #                               my $origin = $self->file_get({
216 #                                       login => $l,
217 #                                       file  => $f,
218 #                               });
219 #                               $self->new_file($origin);
220                                 warn "INFO: sent file ",dump($l,$f);
221                                 my $md5 = $self->md5sum($data)->get($s);
222                                 $self->md5sum({ login => $to })->put($d => $md5 );
223                                 # FIXME broken!
224                         }
225
226
227                 } else {
228                         warn "ERROR: can't SEND To:$to Name:$name Link:$link_to";
229                 }
230         }
231 }
232
233 sub init_pid_login {
234         my ( $self, $pid, $login ) = @_;
235
236         $login =~ s/\@.+//;
237         $self->{pid}->{$pid} = $self->{api}->user_info($login);
238
239         warn "created $pid";
240 }
241
242 sub cleanup_pid {
243         my ( $self, $pid ) = @_;
244
245         $self->md5sum_close;
246
247         delete $self->{pid}->{$pid};
248         warn "removed $pid";
249 }
250
251 sub rsync_log {
252         my ( $self, $data ) = @_;
253         if ( $data =~ m/\[(\d+)\] rsync \w+ (\S+) \w+ (\S+)/ ) {
254                 my ( $pid, $module, $login ) = ( $1, $2, $3 );
255                 $self->init_pid_login( $pid, $login );
256         } elsif ( $data =~ m/\[(\d+)\] sent \S+ bytes\s+received \S+ bytes/ ) {
257                 my $pid = $1;
258                 $self->cleanup_pid( $pid );
259         } else {
260                 warn "## rsync_log $data";
261         }
262 }
263
264 sub blob_path {
265         my ( $self, $data, $path ) = @_;
266         my $blob = $self->{pid}->{ $data->{pid} }->{dir} || die "no dir for $data->{pid} in ",dump( $self->{pid} );
267         $blob .= '/' . ( defined $path ? $path : $data->{file} );
268         return $blob;
269 }
270
271
272 sub rsync_transfer {
273         my ( $self,$data ) = @_;
274
275         my $path = $self->blob_path($data);
276
277         if ( $data->{itemize} =~ m/^[c>]([fdL])/ ) { # received change/create
278                 my $type = $1;
279
280                 if ( $type eq 'f' ) {
281                         $self->modify_file( $data ) && # selective dedup
282                         $self->dedup( $data, $path );
283                 } elsif ( $type eq 'd' ) {
284                         $self->make_dir( $data );
285                 } elsif ( $type eq 'L' ) {
286                         $self->new_link( $data );
287                 } else {
288                         die "unknown type $type ", dump $data;
289                 }
290         } elsif ( $data->{itemize} =~ m/\*deleting/ ) {
291                 $self->removed_file($data);
292         } else {
293                 warn "IGNORED ",dump($data) if $ENV{DEBUG};
294         }
295         return $data;
296 }
297
298 sub append {
299         my $self = shift @_;
300         $self->{api}->append( @_ );
301 }
302
303 sub md5pool {
304         my ( $self, $data ) = @_;
305
306         my $pool = $self->{md5pool} || die "no md5pool in ",dump $self;
307         mkdir $pool unless -e $pool;
308
309         my $md5 = $data->{md5} || die "no md5 in ",dump $data;
310         my $path = $self->blob_path($data);
311
312         my $pool_md5 = "$pool/$md5";
313
314         if ( -e $pool_md5 ) {
315                 warn "dedup hit $md5 $path\n";
316
317                 my ($pool_uid,$pool_size) = (stat($pool_md5))[4,7];
318                 my $user = $self->{api}->user_info( $data->{login} );
319
320                 if ( $pool_uid != $self->{api}->{md5}->{uid} ) {
321                         chown $self->{api}->{md5}->{uid}, $self->{api}->{md5}->{gid}, $pool_md5;
322                         chmod oct("0444"), $pool_md5;
323                         my $steal_user = $self->{api}->user_info( $pool_uid );
324                         $self->append( $steal_user, 'dedup-steal', $pool_size, $pool_uid, $data->{file} );
325                 }
326                 $self->append( $user, 'dedup', $pool_size, $pool_uid, $data->{file} );
327
328                 my $dedup = $path . '.dedup';
329                 rename $path, $dedup;
330                 link "$pool/$md5", $path;
331                 unlink $dedup;
332         } else {
333                 link $path, "$pool/$md5";
334                 warn "dedup +++ $md5 $path";
335         }
336
337         $self->md5sum($data)->put( $data->{file} => $md5 );
338 }
339
340 my $empty_md5 = " " x 32;
341
342 sub dedup {
343         my ( $self, $data, $path ) = @_;
344
345         if ( $data->{file} =~ /^(.+\/)?md5sum$/ ) {
346                 my $dir = $1;
347                 my $imported = 0;
348                 warn "IMPORT ", $data->{file}, "\n";
349                 open(my $md5sum, '<', $path);
350                 while(<$md5sum>) {
351                         chomp;
352                         my ( $md5, $file ) = split(/\s+/,$_,2);
353                         if ( ! -e "$self->{md5path}/$md5" ) {
354                                 warn "MISSING $md5 $file\n";
355                                 next;
356                         }
357                         my $new = {
358                                 pid => $data->{pid},
359                                 file => "$dir$file",
360                                 md5 => $md5,
361                         };
362                         my $new_path = $self->blob_path($new);
363                         if ( ! -e $new_path ) {
364                                 $self->{api}->send_file( 'md5' => $md5, $data->{login}, "$dir$file" );
365                                 $self->md5pool( $new );
366                         } else {
367                                 $self->md5pool( $new );
368                         }
369                 }
370                 print "INFO imported $imported files from ",dump($data);
371         }
372
373         if ( $data->{md5} ne $empty_md5 ) {
374                 $self->md5sum($data)->put( $data->{file} => $data->{md5} );
375                 $self->md5pool( $data );
376         } else {
377                 warn "empty md5", dump $data;
378         }
379 }
380
381 1;