added DELETED pending log
[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                                 }
96                         } elsif ( ! /^(MOVED|RENAMED)\#/ ) {
97                                 warn "skip $_\n";
98                                 next;
99                         }
100
101                         my ( undef, $from, $to ) = split(/\#/,$_,3);
102
103                         my ( $from_path, $to_path ) = map {
104                                 my $tmp = $data;
105                                 $tmp->{file} = $from_dir . $_;
106                                 $self->blob_path($tmp);
107                         } ( $from, $to );
108
109                         if ( ! -e $from_path ) {
110                                 warn "SKIPPED $from_path: $!";
111                                 next;
112                         }
113
114                         warn "MV $from_path -> $to_path";
115                         mkbasedir $to_path;
116                         rename $from_path, $to_path;
117         
118                         my $md5 = $self->md5sum($data)->get( $from_dir . $from );
119                         if ( ! $md5 ) {
120                                 warn "ERROR: no md5sum $from_dir $from " unless $md5;
121                                 next;
122                         }
123
124                         $self->md5sum($data)->out( $from_dir . $from );
125                         $self->md5sum($data)->put( $from_dir . $to => $md5 );
126
127                         warn "$md5 moved to $from_dir $to";
128                 }
129
130                 return 0; # skip dedup
131         }
132
133         #return $file->{size} > 4096 ? 1 : 0; # FIXME
134         return 1; # dedup
135 }
136
137 # never called by rsync directly!
138 sub new_file {
139         my ( $self,$data ) = @_;
140 #       $self->file_set($data);
141 }
142
143 sub remove_file {
144         my ( $self, $data ) = @_;
145
146         my $md5 = $self->md5sum($data)->get( $data->{file} );
147         return unless $md5; # directories don't have md5sums
148         my $path = $self->{md5pool} . '/' . $md5;
149         my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,
150                 $atime,$mtime,$ctime,$blksize,$blocks)
151                         = stat($path);
152         if ( $nlink == 1 ) {
153                 my $id = getpwnam 'md5';
154                 chown $id,$gid, $path;
155                 warn "# chown $id $gid $path";
156         }
157
158         $self->md5sum($data)->out( $data->{file} );
159 }
160
161 sub make_dir {
162         my ( $self, $data ) = @_;
163
164 }
165
166 sub new_link {
167         my ( $self, $data ) = @_;
168
169         warn "# new_link ",dump $data;
170
171         if ( $data->{file} =~ m{^(.*/?)\.send/([^/]+)/(.+)$} ) {
172                 my ( $dir, $to, $name ) = ( $1, $2, $3 );
173                 my $path = $self->blob_path($data);
174                 my $link_to = readlink $path;
175                 warn "$link_to";
176                 if ( $link_to =~ s{^\Q/rsyncd-munged/\E}{/} ) {
177
178                         my $s = $path;
179                         $s =~ s{/[^/]+$}{}; # strip filename
180                         while ( $link_to =~ s{/../}{/} ) {
181                                 $s =~ s{/[^/]+$}{} || die "can't strip $s";
182                                 warn "## simplify $s $link_to\n";
183                         }
184                         $s .= $link_to;
185
186                         my $d = $self->blob_path({
187                                 pid => $data->{pid},
188                                 file => $name
189                         });
190
191                         # $name can contain directories so we must create them
192                         my $to_dir = $d;
193                         $to_dir =~ s{/[^/]+$}{};
194                         make_path $to_dir if ! -e $to_dir;
195
196                         if ( ! -e $s ) {
197                                 warn "ERROR: can't find source $s";
198                         } else {
199
200                                 warn "link $s -> $d\n";
201                                 link $s, $d;
202
203                                 my ($l,$f) = ($1,$2) if $s =~ m{users/([^/]+)/blob/(.+)};
204
205 #                               my $origin = $self->file_get({
206 #                                       login => $l,
207 #                                       file  => $f,
208 #                               });
209 #                               $self->new_file($origin);
210                                 warn "INFO: sent file ",dump($l,$f);
211                                 my $md5 = $self->md5sum($data)->get($s);
212                                 $self->md5sum({ login => $to })->put($d => $md5 );
213                                 # FIXME broken!
214                         }
215
216
217                 } else {
218                         warn "ERROR: can't SEND To:$to Name:$name Link:$link_to";
219                 }
220         }
221 }
222
223 sub md5sum {
224         my ( $self, $data ) = @_;
225
226         my $login = $data->{login} || confess "missing login in ",dump $data;
227
228         return $self->{md5sum}->{$login} if exists $self->{md5sum}->{$login};
229
230         my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
231                 getpwnam $login;
232
233         my $md5_path = "$dir/.md5";
234
235         my $db = TokyoCabinet::HDB->new();
236         $db->open($md5_path, $db->OWRITER | $db->OCREAT)
237         or die "can't open $md5_path: ",$db->errmsg( $db->ecode );
238
239         warn "open $md5_path";
240
241         $self->{md5sum}->{$login} = $db;
242         return $db;
243 }
244
245 sub rsync_log {
246         my ( $self, $data ) = @_;
247         if ( $data =~ m/\[(\d+)\] rsync \w+ (\S+) \w+ (\S+)/ ) {
248                 my ( $pid, $module, $login ) = ( $1, $2, $3 );
249
250                 $login =~ s/\@.+//;
251                 my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
252                         getpwnam $login;
253
254                 $self->{pid}->{$pid} = {
255                         login => $login,
256                         uid => $uid,
257                         gid => $gid,
258                         email => $email,
259                         dir => $dir,
260                         shell => $shell,
261                 };
262
263                 warn "created $pid";
264
265         } elsif ( $data =~ m/\[(\d+)\] sent \S+ bytes\s+received \S+ bytes/ ) {
266                 my $pid = $1;
267
268                 foreach my $login ( keys %{ $self->{md5sum} } ) {
269                         $self->{md5sum}->{$login}->close;
270                         warn "close md5sum $login";
271                 }
272                 delete $self->{md5sum};
273
274                 delete $self->{pid}->{$pid};
275                 warn "removed $pid";
276 warn dump $self;
277
278         } else {
279 #               warn "## rsync_log $data";
280         }
281 }
282
283 sub blob_path {
284         my ( $self, $data ) = @_;
285         my $blob = $self->{pid}->{ $data->{pid} }->{dir} || die "no dir for $data->{pid} in ",dump( $self->{pid} );
286         $blob .= '/' . $data->{file};
287         return $blob;
288 }
289
290
291 sub rsync_transfer {
292         my ( $self,$data ) = @_;
293
294         my $path = $self->blob_path($data);
295
296         if ( $data->{itemize} =~ m/^[c>]([fdL])/ ) { # received change/create
297                 my $type = $1;
298
299                 if ( $type eq 'f' ) {
300                         $self->modify_file( $data ) && # selective dedup
301                         $self->dedup( $data, $path );
302                 } elsif ( $type eq 'd' ) {
303                         $self->make_dir( $data );
304                 } elsif ( $type eq 'L' ) {
305                         $self->new_link( $data );
306                 } else {
307                         die "unknown type $type ", dump $data;
308                 }
309         } elsif ( $data->{itemize} =~ m/\*deleting/ ) {
310                 $self->remove_file($data);
311         } else {
312                 warn "IGNORED ",dump($data) if $ENV{DEBUG};
313         }
314         return $data;
315 }
316
317 sub md5pool {
318         my ( $self, $data ) = @_;
319
320         my $pool = $self->{md5pool} || die "no md5pool in ",dump $self;
321         mkdir $pool unless -e $pool;
322
323         my $md5 = $data->{md5} || die "no md5 in ",dump $data;
324         my $path = $self->blob_path($data);
325
326         if ( -e "$pool/$md5" ) {
327                 warn "dedup hit $md5 $path\n";
328                 my $dedup = $path . '.dedup';
329                 rename $path, $dedup;
330                 link "$pool/$md5", $path;
331                 unlink $dedup;
332                 # FIXME fix perms?
333         } else {
334                 link $path, "$pool/$md5";
335                 warn "dedup +++ $md5 $path";
336         }
337
338         $self->md5sum($data)->put( $data->{file} => $md5 );
339 }
340
341 my $empty_md5 = " " x 32;
342
343 sub dedup {
344         my ( $self, $data, $path ) = @_;
345
346         if ( $data->{file} =~ /^(.+\/)?md5sum$/ ) {
347                 my $dir = $1;
348                 my $imported = 0;
349                 warn "IMPORT ", $data->{file}, "\n";
350                 open(my $md5sum, '<', $path);
351                 while(<$md5sum>) {
352                         chomp;
353                         my ( $md5, $file ) = split(/\s+/,$_,2);
354                         if ( ! -e "$self->{md5path}/$md5" ) {
355                                 warn "MISSING $md5 $file\n";
356                                 next;
357                         }
358                         my $new = {
359                                 pid => $data->{pid},
360                                 file => "$dir$file",
361                                 md5 => $md5,
362                         };
363                         my $new_path = $self->blob_path($new);
364                         if ( ! -e $new_path ) {
365                                 # create path from md5sum file
366                                 my $only_dir = $1 if $new =~ m{^(.+)/[^/]+$};
367                                 make_path $only_dir unless -d $only_dir;
368                                 $imported += link "$self->{md5path}/$md5", $new_path;
369                                 $self->new_file($new);
370                                 warn "import from $path ",dump($new);
371                                 $self->md5pool( $new );
372                         } else {
373                                 $self->md5pool( $new );
374                         }
375                 }
376                 print "INFO imported $imported files from ",dump($data);
377         }
378
379         if ( $data->{md5} ne $empty_md5 ) {
380                 $self->md5pool( $data );
381         } else {
382                 warn "empty md5", dump $data;
383         }
384 }
385
386 1;