don't try to rename non-existing files
[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 warn $_;
89                         if ( ! /^(MOVED|RENAMED)\#/ ) {
90                                 warn "skip $_\n";
91                                 next;
92                         }
93
94                         my ( undef, $from, $to ) = split(/\#/,$_,3);
95
96                         my ( $from_path, $to_path ) = map {
97                                 my $tmp = $data;
98                                 $tmp->{file} = $from_dir . $_;
99                                 $self->blob_path($tmp);
100                         } ( $from, $to );
101
102                         if ( ! -e $from_path ) {
103                                 warn "SKIPPED $from_path: $!";
104                                 next;
105                         }
106
107                         warn "MV $from_path -> $to_path";
108                         mkbasedir $to_path;
109                         rename $from_path, $to_path;
110         
111                         my $md5 = $self->md5sum($data)->get( $from_dir . $from );
112                         die "no md5sum $from_dir $from " unless $md5;
113
114                         $self->md5sum($data)->out( $from_dir . $from );
115                         $self->md5sum($data)->put( $from_dir . $to => $md5 );
116
117                         warn "$md5 moved to $from_dir $to";
118                 }
119
120                 return 0; # skip dedup
121         }
122
123         #return $file->{size} > 4096 ? 1 : 0; # FIXME
124         return 1; # dedup
125 }
126
127 # never called by rsync directly!
128 sub new_file {
129         my ( $self,$data ) = @_;
130 #       $self->file_set($data);
131 }
132
133 sub remove_file {
134         my ( $self, $data ) = @_;
135
136         my $md5 = $self->md5sum($data)->get( $data->{file} );
137         return unless $md5; # directories don't have md5sums
138         my $path = $self->{md5pool} . '/' . $md5;
139         my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,
140                 $atime,$mtime,$ctime,$blksize,$blocks)
141                         = stat($path);
142         if ( $nlink == 1 ) {
143                 my $id = getpwnam 'md5';
144                 chown $id,$gid, $path;
145                 warn "# chown $id $gid $path";
146         }
147
148         $self->md5sum($data)->out( $data->{file} );
149 }
150
151 sub make_dir {
152         my ( $self, $data ) = @_;
153
154 }
155
156 sub new_link {
157         my ( $self, $data ) = @_;
158
159         warn "# new_link ",dump $data;
160
161         if ( $data->{file} =~ m{^(.*/?)\.send/([^/]+)/(.+)$} ) {
162                 my ( $dir, $to, $name ) = ( $1, $2, $3 );
163                 my $path = $self->blob_path($data);
164                 my $link_to = readlink $path;
165                 warn "$link_to";
166                 if ( $link_to =~ s{^\Q/rsyncd-munged/\E}{/} ) {
167
168                         my $s = $path;
169                         $s =~ s{/[^/]+$}{}; # strip filename
170                         while ( $link_to =~ s{/../}{/} ) {
171                                 $s =~ s{/[^/]+$}{} || die "can't strip $s";
172                                 warn "## simplify $s $link_to\n";
173                         }
174                         $s .= $link_to;
175
176                         my $d = $self->blob_path({
177                                 pid => $data->{pid},
178                                 file => $name
179                         });
180
181                         # $name can contain directories so we must create them
182                         my $to_dir = $d;
183                         $to_dir =~ s{/[^/]+$}{};
184                         make_path $to_dir if ! -e $to_dir;
185
186                         if ( ! -e $s ) {
187                                 warn "ERROR: can't find source $s";
188                         } else {
189
190                                 warn "link $s -> $d\n";
191                                 link $s, $d;
192
193                                 my ($l,$f) = ($1,$2) if $s =~ m{users/([^/]+)/blob/(.+)};
194
195 #                               my $origin = $self->file_get({
196 #                                       login => $l,
197 #                                       file  => $f,
198 #                               });
199 #                               $self->new_file($origin);
200                                 warn "INFO: sent file ",dump($l,$f);
201                                 my $md5 = $self->md5sum($data)->get($s);
202                                 $self->md5sum({ login => $to })->put($d => $md5 );
203                                 # FIXME broken!
204                         }
205
206
207                 } else {
208                         warn "ERROR: can't SEND To:$to Name:$name Link:$link_to";
209                 }
210         }
211 }
212
213 sub md5sum {
214         my ( $self, $data ) = @_;
215
216         my $login = $data->{login} || confess "missing login in ",dump $data;
217
218         return $self->{md5sum}->{$login} if exists $self->{md5sum}->{$login};
219
220         my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
221                 getpwnam $login;
222
223         my $md5_path = "$dir/.md5";
224
225         my $db = TokyoCabinet::HDB->new();
226         $db->open($md5_path, $db->OWRITER | $db->OCREAT)
227         or die "can't open $md5_path: ",$db->errmsg( $db->ecode );
228
229         warn "open $md5_path";
230
231         $self->{md5sum}->{$login} = $db;
232         return $db;
233 }
234
235 sub rsync_log {
236         my ( $self, $data ) = @_;
237         if ( $data =~ m/\[(\d+)\] rsync \w+ (\S+) \w+ (\S+)/ ) {
238                 my ( $pid, $module, $login ) = ( $1, $2, $3 );
239
240                 $login =~ s/\@.+//;
241                 my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
242                         getpwnam $login;
243
244                 $self->{pid}->{$pid} = {
245                         login => $login,
246                         uid => $uid,
247                         gid => $gid,
248                         email => $email,
249                         dir => $dir,
250                         shell => $shell,
251                 };
252
253                 warn "created $pid";
254
255         } elsif ( $data =~ m/\[(\d+)\] sent \S+ bytes\s+received \S+ bytes/ ) {
256                 my $pid = $1;
257
258                 foreach my $login ( keys %{ $self->{md5sum} } ) {
259                         $self->{md5sum}->{$login}->close;
260                         warn "close md5sum $login";
261                 }
262                 delete $self->{md5sum};
263
264                 delete $self->{pid}->{$pid};
265                 warn "removed $pid";
266 warn dump $self;
267
268         } else {
269 #               warn "## rsync_log $data";
270         }
271 }
272
273 sub blob_path {
274         my ( $self, $data ) = @_;
275         my $blob = $self->{pid}->{ $data->{pid} }->{dir} || die "no dir for $data->{pid} in ",dump( $self->{pid} );
276         $blob .= '/' . $data->{file};
277         return $blob;
278 }
279
280
281 sub rsync_transfer {
282         my ( $self,$data ) = @_;
283
284         my $path = $self->blob_path($data);
285
286         if ( $data->{itemize} =~ m/^[c>]([fdL])/ ) { # received change/create
287                 my $type = $1;
288
289                 if ( $type eq 'f' ) {
290                         $self->modify_file( $data ) && # selective dedup
291                         $self->dedup( $data, $path );
292                 } elsif ( $type eq 'd' ) {
293                         $self->make_dir( $data );
294                 } elsif ( $type eq 'L' ) {
295                         $self->new_link( $data );
296                 } else {
297                         die "unknown type $type ", dump $data;
298                 }
299         } elsif ( $data->{itemize} =~ m/\*deleting/ ) {
300                 $self->remove_file($data);
301         } else {
302                 warn "IGNORED ",dump($data) if $ENV{DEBUG};
303         }
304         return $data;
305 }
306
307 sub md5pool {
308         my ( $self, $data ) = @_;
309
310         my $pool = $self->{md5pool} || die "no md5pool in ",dump $self;
311         mkdir $pool unless -e $pool;
312
313         my $md5 = $data->{md5} || die "no md5 in ",dump $data;
314         my $path = $self->blob_path($data);
315
316         if ( -e "$pool/$md5" ) {
317                 warn "dedup hit $md5 $path\n";
318                 my $dedup = $path . '.dedup';
319                 rename $path, $dedup;
320                 link "$pool/$md5", $path;
321                 unlink $dedup;
322                 # FIXME fix perms?
323         } else {
324                 link $path, "$pool/$md5";
325                 warn "dedup +++ $md5 $path";
326         }
327
328         $self->md5sum($data)->put( $data->{file} => $md5 );
329 }
330
331 my $empty_md5 = " " x 32;
332
333 sub dedup {
334         my ( $self, $data, $path ) = @_;
335
336         if ( $data->{file} =~ /^(.+\/)?md5sum$/ ) {
337                 my $dir = $1;
338                 my $imported = 0;
339                 warn "IMPORT ", $data->{file}, "\n";
340                 open(my $md5sum, '<', $path);
341                 while(<$md5sum>) {
342                         chomp;
343                         my ( $md5, $file ) = split(/\s+/,$_,2);
344                         if ( ! -e "$self->{md5path}/$md5" ) {
345                                 warn "MISSING $md5 $file\n";
346                                 next;
347                         }
348                         my $new = {
349                                 pid => $data->{pid},
350                                 file => "$dir$file",
351                                 md5 => $md5,
352                         };
353                         my $new_path = $self->blob_path($new);
354                         if ( ! -e $new_path ) {
355                                 # create path from md5sum file
356                                 my $only_dir = $1 if $new =~ m{^(.+)/[^/]+$};
357                                 make_path $only_dir unless -d $only_dir;
358                                 $imported += link "$self->{md5path}/$md5", $new_path;
359                                 $self->new_file($new);
360                                 warn "import from $path ",dump($new);
361                                 $self->md5pool( $new );
362                         } else {
363                                 $self->md5pool( $new );
364                         }
365                 }
366                 print "INFO imported $imported files from ",dump($data);
367         }
368
369         if ( $data->{md5} ne $empty_md5 ) {
370                 $self->md5pool( $data );
371         } else {
372                 warn "empty md5", dump $data;
373         }
374 }
375
376 1;