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