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