1 package CloudStore::Store;
6 use base 'CloudStore::MD5sum';
11 use File::Path qw(make_path);
13 use Digest::MD5 qw(md5_base64);
14 use Data::Dump qw(dump);
20 my ($class,$group) = @_;
23 api => CloudStore::API->new( $group ),
27 $self->{md5pool} = $self->{api}->{md5}->{dir};
29 warn "# new ",dump $self if $ENV{DEBUG};
34 sub api { $_[0]->{api} }
38 $dir =~ s{/[^/]+$}{}; # strip filename
39 mkdir $dir unless -e $dir;
43 my ( $self,$data ) = @_;
47 if ( $data->{file} =~ m{^(.*/)?.sync/send/([^/]+)$} ) {
49 warn "SEND $2 from $from_dir\n";
51 open(my $send, '<', $self->blob_path($data) );
55 my ( $to, $file ) = split(/\s+/,$_,2);
56 my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
60 $from->{file} = $from_dir . $file;
61 my $from_path = $self->blob_path($from);
63 if ( ! -r $from_path ) {
64 warn "ERROR: $from_path: $!";
68 my $to_path = "$dir/received/$file";
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;
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 );
79 warn "SENT ",dump $sent_files;
81 return 0; # skip dedup
82 } elsif ( $data->{file} =~ m{^(.*/)?.sync/pending/([^/]+)$} ) {
84 warn "PENDIG $2 from $from_dir";
85 open(my $pend, '<', $self->blob_path($data) );
89 if ( m/^DELETED\#(.+)$/ ) {
90 my $path = $self->blob_path($data => $from_dir . $1 );
93 -d $path ? rmdir $path : unlink $path || warn "ERROR: unlink $path $!";
96 warn "MISSING $path to unlink";
99 } elsif ( ! /^(MOVED|RENAMED)\#/ ) {
104 my ( undef, $from, $to ) = split(/\#/,$_,3);
106 my ( $from_path, $to_path ) = map {
108 $tmp->{file} = $from_dir . $_;
109 $self->blob_path($tmp);
112 if ( ! -e $from_path ) {
113 warn "SKIPPED $from_path: $!";
117 warn "MV $from_path -> $to_path";
119 rename $from_path, $to_path;
121 my $md5 = $self->md5sum($data)->get( $from_dir . $from );
123 warn "ERROR: no md5sum $from_dir $from " unless $md5;
127 $self->md5sum($data)->out( $from_dir . $from );
128 $self->md5sum($data)->put( $from_dir . $to => $md5 );
129 $self->md5sum_close($data);
131 warn "$md5 moved to $from_dir $to";
134 return 0; # skip dedup
139 if ( $data->{file} =~ m{^(.*/)?.sync/} ) {
140 # ignore .sync/ files from client
144 #return $file->{size} > 4096 ? 1 : 0; # FIXME
148 # never called by rsync directly!
150 my ( $self,$data ) = @_;
151 # $self->file_set($data);
154 # client doesn't issue --delete
156 my ( $self, $data ) = @_;
158 my $md5 = $self->md5sum($data)->get( $data->{file} );
159 return unless $md5; # directories don't have md5sums
160 my $path = $self->{md5pool} . '/' . $md5;
161 my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,
162 $atime,$mtime,$ctime,$blksize,$blocks)
165 my $user = $self->{api}->user_info($data->{login});
167 if ( $nlink == 1 && $uid == $user->{uid} ) {
168 $self->append( $user, 'removed', -$size, $uid, $data->{file} );
169 my $id = getpwnam 'md5';
170 chown $id,$gid, $path;
171 warn "# chown $id $gid $path";
174 $self->md5sum($data)->out( $data->{file} );
175 $self->md5sum_close($data);
177 $self->api->append_meta('md5sum', $user, 'delete', $data->{file} );
181 my ( $self, $data ) = @_;
186 my ( $self, $data ) = @_;
188 warn "# new_link ",dump $data;
192 if ( $data->{file} =~ m{^(.*/?)\.send/([^/]+)/(.+)$} ) {
193 my ( $dir, $to, $name ) = ( $1, $2, $3 );
194 my $path = $self->blob_path($data);
195 my $link_to = readlink $path;
197 if ( $link_to =~ s{^\Q/rsyncd-munged/\E}{/} ) {
200 $s =~ s{/[^/]+$}{}; # strip filename
201 while ( $link_to =~ s{/../}{/} ) {
202 $s =~ s{/[^/]+$}{} || die "can't strip $s";
203 warn "## simplify $s $link_to\n";
207 my $d = $self->blob_path({
212 # $name can contain directories so we must create them
216 warn "ERROR: can't find source $s";
219 warn "link $s -> $d\n";
222 my ($l,$f) = ($1,$2) if $s =~ m{users/([^/]+)/blob/(.+)};
224 # my $origin = $self->file_get({
228 # $self->new_file($origin);
229 warn "INFO: sent file ",dump($l,$f);
230 my $md5 = $self->md5sum($data)->get($s);
231 $self->md5sum({ login => $to })->put($d => $md5 );
237 warn "ERROR: can't SEND To:$to Name:$name Link:$link_to";
246 my ( $self, $pid, $login ) = @_;
249 $self->{pid}->{$pid} = $self->{api}->user_info($login);
255 my ( $self, $pid ) = @_;
257 $self->md5sum_close( $self->{pid}->{$pid} );
259 delete $self->{pid}->{$pid};
264 my ( $self, $data ) = @_;
265 if ( $data =~ m/\[(\d+)\] rsync \w+ (\S+) \w+ (\S+)/ ) {
266 my ( $pid, $module, $login ) = ( $1, $2, $3 );
267 $self->init_pid_login( $pid, $login );
268 } elsif ( $data =~ m/\[(\d+)\] sent \S+ bytes\s+received \S+ bytes/ ) {
270 $self->cleanup_pid( $pid );
272 warn "## rsync_log $data";
277 my ( $self, $data, $path ) = @_;
278 my $blob = $self->{pid}->{ $data->{pid} }->{dir};
280 warn "ERROR: $data->{pid} not found, possible restart?";
281 $self->init_pid_login( $data->{pid}, $data->{login} );
282 $blob = $self->{pid}->{ $data->{pid} }->{dir} || die "no dir for ", dump( $self->{pid}->{ $data->{pid} } );
284 $blob .= '/' . ( defined $path ? $path : $data->{file} );
290 my ( $self,$data ) = @_;
292 my $path = $self->blob_path($data);
294 if ( $data->{itemize} =~ m/^[c>]([fdL])/ ) { # received change/create
297 if ( $type eq 'f' ) {
298 $self->modify_file( $data ) && # selective dedup
299 $self->dedup( $data, $path );
300 } elsif ( $type eq 'd' ) {
301 $self->make_dir( $data );
302 } elsif ( $type eq 'L' ) {
303 $self->new_link( $data );
305 die "unknown type $type ", dump $data;
307 } elsif ( $data->{itemize} =~ m/\*deleting/ ) {
308 $self->removed_file($data);
310 warn "IGNORED ",dump($data) if $ENV{DEBUG};
317 $self->{api}->append( @_ );
321 my ( $self, $data ) = @_;
323 my $pool = $self->{md5pool} || die "no md5pool in ",dump $self;
324 mkdir $pool unless -e $pool;
326 my $md5 = $data->{md5} || die "no md5 in ",dump $data;
327 my $path = $self->blob_path($data);
330 warn "ERROR missing path $path";
334 my $pool_md5 = "$pool/$md5";
336 if ( -e $pool_md5 ) {
337 warn "dedup hit $md5 $path\n";
339 my ($pool_uid,$pool_size) = (stat($pool_md5))[4,7];
340 my $user = $self->{api}->user_info( $data->{login} );
342 if ( $pool_uid != $user->{uid} ) {
343 if ( $pool_uid != $self->{api}->{md5}->{uid} ) {
344 chown $self->{api}->{md5}->{uid}, $self->{api}->{md5}->{gid}, $pool_md5;
345 chmod oct("0444"), $pool_md5;
346 my $steal_user = $self->{api}->user_info( $pool_uid );
347 $self->append( $steal_user, 'dedup-steal', $pool_size, $pool_uid, $data->{file} );
349 $self->append( $user, 'dedup', $pool_size, $pool_uid, $data->{file} );
352 my $dedup = $path . '.dedup';
353 rename $path, $dedup;
354 link "$pool/$md5", $path;
357 link $path, "$pool/$md5";
358 warn "dedup +++ $md5 $path";
361 $self->md5sum($data)->put( $data->{file} => $md5 );
362 $self->md5sum_close($data);
364 $self->api->append_meta('md5sum', $data->{login}, $md5, $data->{file} );
367 my $empty_md5 = " " x 32;
370 my ( $self, $data, $path ) = @_;
372 if ( $data->{file} =~ /^(.+\/)?md5sum$/ ) {
375 warn "IMPORT ", $data->{file}, "\n";
376 open(my $md5sum, '<', $path);
379 my ( $md5, $file ) = split(/\s+/,$_,2);
380 if ( ! -e "$self->{md5pool}/$md5" ) {
381 warn "MISSING $md5 $file\n";
385 login => $data->{login},
390 my $new_path = $self->blob_path($new);
391 if ( ! -e $new_path ) {
392 $self->{api}->send_file( 'md5' => $md5, $data->{login}, "$dir$file" );
393 $self->md5pool( $new );
395 $self->md5pool( $new );
398 warn "INFO imported $imported files from ",dump($data);
400 return; # don't put md5sum files into pool
403 if ( $data->{md5} ne $empty_md5 ) {
404 $self->md5pool( $data );
406 warn "empty md5", dump $data;