076cc12c1a6a267ce87170fd1ae69ee78817b2b3
[cloudstore.git] / lib / CloudStore / Store.pm
1 package CloudStore::Store;
2 use warnings;
3 use strict;
4
5 use autodie;
6 use JSON::XS;
7 use File::Path qw(make_path);
8 use File::Slurp qw();
9 use Digest::MD5 qw(md5_base64);
10 use Data::Dump qw(dump);
11 use Carp qw(confess);
12 use TokyoCabinet;
13
14 use WarnColor;
15
16 sub new {
17         my $class = shift;
18
19         my $self = {@_};
20         bless $self, $class;
21
22         die "no dir" unless $self->{dir};
23         $self->{md5pool} = $self->{dir} . '/md5';
24
25         warn "# new ",dump $self if $ENV{DEBUG};
26
27         return $self;
28 }
29
30 sub user_set {
31         my ( $self,$data ) = @_;
32 }
33
34 sub user_get {
35         my ( $self,$data ) = @_;
36 }
37
38 sub mkbasedir {
39         my $dir = shift;
40         $dir =~ s{/[^/]+$}{}; # strip filename
41         mkdir $dir unless -e $dir;
42 }
43
44 sub modify_file {
45         my ( $self,$data ) = @_;
46
47         if ( $data->{file} =~ m{^(.*/)?.sync/send/([^/]+)$} ) {
48                 my $from_dir = $1;
49                 warn "SEND $2 from $from_dir\n";
50                 my $sent_files;
51                 open(my $send, '<', $self->blob_path($data) );
52                 while(<$send>) {
53                         s/[\n\r]+$//;
54
55                         my ( $to, $file ) = split(/\s+/,$_,2);
56                         my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
57                                 getpwnam $to;
58
59                         my $from = $data;
60                         $from->{file} = $from_dir . $file;
61                         my $from_path = $self->blob_path($from);
62
63                         if ( ! -r $from_path ) {
64                                 warn "ERROR: $from_path: $!";
65                                 next;
66                         }
67
68                         my $to_path = "$dir/received/$file";
69                         mkbasedir $to_path;
70
71                         warn "SEND $from_path -> $to_path\n";
72                         unlink $to_path if -e $to_path; # FIXME why we need this?
73                         $sent_files->{$to} += link $from_path, $to_path;
74                         # FIXME cross-shard
75                         my $md5 = $self->md5sum($data)->get( $from_dir . $file ) || warn "no md5 for $from_dir$file";
76                         $self->md5sum({login => $to})->put( "/received/$file" => $md5 );
77                 }
78
79                 warn "SENT ",dump $sent_files;
80
81                 return 0; # skip dedup
82         } elsif ( $data->{file} =~ m{^(.*/)?.sync/pending/([^/]+)$} ) {
83                 my $from_dir = $1;
84                 warn "PENDIG $2 from $from_dir";
85                 open(my $pend, '<', $self->blob_path($data) );
86                 while(<$pend>) {
87                         s/[\n\r]+$//;
88
89                         if ( m/^DELETED\#(.+)$/ ) {
90                                 my $path = $from_dir . $1;
91                                 if ( -e $path ) {
92                                         unlink $path && warn "unlink $path\n";
93                                 } else {
94                                         warn "MISSING $path to unlink\n";
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         #return $file->{size} > 4096 ? 1 : 0; # FIXME
135         return 1; # dedup
136 }
137
138 # never called by rsync directly!
139 sub new_file {
140         my ( $self,$data ) = @_;
141 #       $self->file_set($data);
142 }
143
144 sub remove_file {
145         my ( $self, $data ) = @_;
146
147         my $md5 = $self->md5sum($data)->get( $data->{file} );
148         return unless $md5; # directories don't have md5sums
149         my $path = $self->{md5pool} . '/' . $md5;
150         my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,
151                 $atime,$mtime,$ctime,$blksize,$blocks)
152                         = stat($path);
153         if ( $nlink == 1 ) {
154                 my $id = getpwnam 'md5';
155                 chown $id,$gid, $path;
156                 warn "# chown $id $gid $path";
157         }
158
159         $self->md5sum($data)->out( $data->{file} );
160 }
161
162 sub make_dir {
163         my ( $self, $data ) = @_;
164
165 }
166
167 sub new_link {
168         my ( $self, $data ) = @_;
169
170         warn "# new_link ",dump $data;
171
172         if ( $data->{file} =~ m{^(.*/?)\.send/([^/]+)/(.+)$} ) {
173                 my ( $dir, $to, $name ) = ( $1, $2, $3 );
174                 my $path = $self->blob_path($data);
175                 my $link_to = readlink $path;
176                 warn "$link_to";
177                 if ( $link_to =~ s{^\Q/rsyncd-munged/\E}{/} ) {
178
179                         my $s = $path;
180                         $s =~ s{/[^/]+$}{}; # strip filename
181                         while ( $link_to =~ s{/../}{/} ) {
182                                 $s =~ s{/[^/]+$}{} || die "can't strip $s";
183                                 warn "## simplify $s $link_to\n";
184                         }
185                         $s .= $link_to;
186
187                         my $d = $self->blob_path({
188                                 pid => $data->{pid},
189                                 file => $name
190                         });
191
192                         # $name can contain directories so we must create them
193                         my $to_dir = $d;
194                         $to_dir =~ s{/[^/]+$}{};
195                         make_path $to_dir if ! -e $to_dir;
196
197                         if ( ! -e $s ) {
198                                 warn "ERROR: can't find source $s";
199                         } else {
200
201                                 warn "link $s -> $d\n";
202                                 link $s, $d;
203
204                                 my ($l,$f) = ($1,$2) if $s =~ m{users/([^/]+)/blob/(.+)};
205
206 #                               my $origin = $self->file_get({
207 #                                       login => $l,
208 #                                       file  => $f,
209 #                               });
210 #                               $self->new_file($origin);
211                                 warn "INFO: sent file ",dump($l,$f);
212                                 my $md5 = $self->md5sum($data)->get($s);
213                                 $self->md5sum({ login => $to })->put($d => $md5 );
214                                 # FIXME broken!
215                         }
216
217
218                 } else {
219                         warn "ERROR: can't SEND To:$to Name:$name Link:$link_to";
220                 }
221         }
222 }
223
224 sub md5sum {
225         my ( $self, $data ) = @_;
226
227         my $login = $data->{login} || confess "missing login in ",dump $data;
228
229         return $self->{md5sum}->{$login} if exists $self->{md5sum}->{$login};
230
231         my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
232                 getpwnam $login;
233
234         my $md5_path = "$dir/.md5";
235
236         my $db = TokyoCabinet::HDB->new();
237         $db->open($md5_path, $db->OWRITER | $db->OCREAT)
238         or die "can't open $md5_path: ",$db->errmsg( $db->ecode );
239
240         warn "open $md5_path";
241
242         $self->{md5sum}->{$login} = $db;
243         return $db;
244 }
245
246 sub rsync_log {
247         my ( $self, $data ) = @_;
248         if ( $data =~ m/\[(\d+)\] rsync \w+ (\S+) \w+ (\S+)/ ) {
249                 my ( $pid, $module, $login ) = ( $1, $2, $3 );
250
251                 $login =~ s/\@.+//;
252                 my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
253                         getpwnam $login;
254
255                 $self->{pid}->{$pid} = {
256                         login => $login,
257                         uid => $uid,
258                         gid => $gid,
259                         email => $email,
260                         dir => $dir,
261                         shell => $shell,
262                 };
263
264                 warn "created $pid";
265
266         } elsif ( $data =~ m/\[(\d+)\] sent \S+ bytes\s+received \S+ bytes/ ) {
267                 my $pid = $1;
268
269                 foreach my $login ( keys %{ $self->{md5sum} } ) {
270                         $self->{md5sum}->{$login}->close;
271                         warn "close md5sum $login";
272                 }
273                 delete $self->{md5sum};
274
275                 delete $self->{pid}->{$pid};
276                 warn "removed $pid";
277 warn dump $self;
278
279         } else {
280 #               warn "## rsync_log $data";
281         }
282 }
283
284 sub blob_path {
285         my ( $self, $data ) = @_;
286         my $blob = $self->{pid}->{ $data->{pid} }->{dir} || die "no dir for $data->{pid} in ",dump( $self->{pid} );
287         $blob .= '/' . $data->{file};
288         return $blob;
289 }
290
291
292 sub rsync_transfer {
293         my ( $self,$data ) = @_;
294
295         my $path = $self->blob_path($data);
296
297         if ( $data->{itemize} =~ m/^[c>]([fdL])/ ) { # received change/create
298                 my $type = $1;
299
300                 if ( $type eq 'f' ) {
301                         $self->modify_file( $data ) && # selective dedup
302                         $self->dedup( $data, $path );
303                 } elsif ( $type eq 'd' ) {
304                         $self->make_dir( $data );
305                 } elsif ( $type eq 'L' ) {
306                         $self->new_link( $data );
307                 } else {
308                         die "unknown type $type ", dump $data;
309                 }
310         } elsif ( $data->{itemize} =~ m/\*deleting/ ) {
311                 $self->remove_file($data);
312         } else {
313                 warn "IGNORED ",dump($data) if $ENV{DEBUG};
314         }
315         return $data;
316 }
317
318 sub md5pool {
319         my ( $self, $data ) = @_;
320
321         my $pool = $self->{md5pool} || die "no md5pool in ",dump $self;
322         mkdir $pool unless -e $pool;
323
324         my $md5 = $data->{md5} || die "no md5 in ",dump $data;
325         my $path = $self->blob_path($data);
326
327         if ( -e "$pool/$md5" ) {
328                 warn "dedup hit $md5 $path\n";
329                 my $dedup = $path . '.dedup';
330                 rename $path, $dedup;
331                 link "$pool/$md5", $path;
332                 unlink $dedup;
333                 # FIXME fix perms?
334         } else {
335                 link $path, "$pool/$md5";
336                 warn "dedup +++ $md5 $path";
337         }
338
339         $self->md5sum($data)->put( $data->{file} => $md5 );
340 }
341
342 my $empty_md5 = " " x 32;
343
344 sub dedup {
345         my ( $self, $data, $path ) = @_;
346
347         if ( $data->{file} =~ /^(.+\/)?md5sum$/ ) {
348                 my $dir = $1;
349                 my $imported = 0;
350                 warn "IMPORT ", $data->{file}, "\n";
351                 open(my $md5sum, '<', $path);
352                 while(<$md5sum>) {
353                         chomp;
354                         my ( $md5, $file ) = split(/\s+/,$_,2);
355                         if ( ! -e "$self->{md5path}/$md5" ) {
356                                 warn "MISSING $md5 $file\n";
357                                 next;
358                         }
359                         my $new = {
360                                 pid => $data->{pid},
361                                 file => "$dir$file",
362                                 md5 => $md5,
363                         };
364                         my $new_path = $self->blob_path($new);
365                         if ( ! -e $new_path ) {
366                                 # create path from md5sum file
367                                 my $only_dir = $1 if $new =~ m{^(.+)/[^/]+$};
368                                 make_path $only_dir unless -d $only_dir;
369                                 $imported += link "$self->{md5path}/$md5", $new_path;
370                                 $self->new_file($new);
371                                 warn "import from $path ",dump($new);
372                                 $self->md5pool( $new );
373                         } else {
374                                 $self->md5pool( $new );
375                         }
376                 }
377                 print "INFO imported $imported files from ",dump($data);
378         }
379
380         if ( $data->{md5} ne $empty_md5 ) {
381                 $self->md5pool( $data );
382         } else {
383                 warn "empty md5", dump $data;
384         }
385 }
386
387 1;