1 package CloudStore::Store;
10 use File::Path qw(make_path);
12 use Digest::MD5 qw(md5_base64);
13 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};
35 my ( $self,$data ) = @_;
39 my ( $self,$data ) = @_;
44 $dir =~ s{/[^/]+$}{}; # strip filename
45 mkdir $dir unless -e $dir;
49 my ( $self,$data ) = @_;
51 if ( $data->{file} =~ m{^(.*/)?.sync/send/([^/]+)$} ) {
53 warn "SEND $2 from $from_dir\n";
55 open(my $send, '<', $self->blob_path($data) );
59 my ( $to, $file ) = split(/\s+/,$_,2);
60 my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
64 $from->{file} = $from_dir . $file;
65 my $from_path = $self->blob_path($from);
67 if ( ! -r $from_path ) {
68 warn "ERROR: $from_path: $!";
72 my $to_path = "$dir/received/$file";
75 warn "SEND $from_path -> $to_path\n";
76 unlink $to_path if -e $to_path; # FIXME why we need this?
77 $sent_files->{$to} += link $from_path, $to_path;
79 my $md5 = $self->md5sum($data)->get( $from_dir . $file ) || warn "no md5 for $from_dir$file";
80 $self->md5sum({login => $to})->put( "/received/$file" => $md5 );
83 warn "SENT ",dump $sent_files;
85 return 0; # skip dedup
86 } elsif ( $data->{file} =~ m{^(.*/)?.sync/pending/([^/]+)$} ) {
88 warn "PENDIG $2 from $from_dir";
89 open(my $pend, '<', $self->blob_path($data) );
93 if ( m/^DELETED\#(.+)$/ ) {
94 my $path = $self->blob_path($data => $from_dir . $1 );
97 -d $path ? rmdir $path : unlink $path || warn "ERROR: unlink $path $!";
100 warn "MISSING $path to unlink";
103 } elsif ( ! /^(MOVED|RENAMED)\#/ ) {
108 my ( undef, $from, $to ) = split(/\#/,$_,3);
110 my ( $from_path, $to_path ) = map {
112 $tmp->{file} = $from_dir . $_;
113 $self->blob_path($tmp);
116 if ( ! -e $from_path ) {
117 warn "SKIPPED $from_path: $!";
121 warn "MV $from_path -> $to_path";
123 rename $from_path, $to_path;
125 my $md5 = $self->md5sum($data)->get( $from_dir . $from );
127 warn "ERROR: no md5sum $from_dir $from " unless $md5;
131 $self->md5sum($data)->out( $from_dir . $from );
132 $self->md5sum($data)->put( $from_dir . $to => $md5 );
134 warn "$md5 moved to $from_dir $to";
137 return 0; # skip dedup
140 #return $file->{size} > 4096 ? 1 : 0; # FIXME
144 # never called by rsync directly!
146 my ( $self,$data ) = @_;
147 # $self->file_set($data);
151 my ( $self, $data ) = @_;
153 my $md5 = $self->md5sum($data)->get( $data->{file} );
154 return unless $md5; # directories don't have md5sums
155 my $path = $self->{md5pool} . '/' . $md5;
156 my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,
157 $atime,$mtime,$ctime,$blksize,$blocks)
160 my $id = getpwnam 'md5';
161 chown $id,$gid, $path;
162 warn "# chown $id $gid $path";
165 $self->md5sum($data)->out( $data->{file} );
169 my ( $self, $data ) = @_;
174 my ( $self, $data ) = @_;
176 warn "# new_link ",dump $data;
178 if ( $data->{file} =~ m{^(.*/?)\.send/([^/]+)/(.+)$} ) {
179 my ( $dir, $to, $name ) = ( $1, $2, $3 );
180 my $path = $self->blob_path($data);
181 my $link_to = readlink $path;
183 if ( $link_to =~ s{^\Q/rsyncd-munged/\E}{/} ) {
186 $s =~ s{/[^/]+$}{}; # strip filename
187 while ( $link_to =~ s{/../}{/} ) {
188 $s =~ s{/[^/]+$}{} || die "can't strip $s";
189 warn "## simplify $s $link_to\n";
193 my $d = $self->blob_path({
198 # $name can contain directories so we must create them
202 warn "ERROR: can't find source $s";
205 warn "link $s -> $d\n";
208 my ($l,$f) = ($1,$2) if $s =~ m{users/([^/]+)/blob/(.+)};
210 # my $origin = $self->file_get({
214 # $self->new_file($origin);
215 warn "INFO: sent file ",dump($l,$f);
216 my $md5 = $self->md5sum($data)->get($s);
217 $self->md5sum({ login => $to })->put($d => $md5 );
223 warn "ERROR: can't SEND To:$to Name:$name Link:$link_to";
229 my ( $self, $data ) = @_;
231 my $login = $data->{login} || confess "missing login in ",dump $data;
233 return $self->{md5sum}->{$login} if exists $self->{md5sum}->{$login};
235 my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
238 my $md5_path = "$dir/.md5";
240 my $db = TokyoCabinet::HDB->new();
241 $db->open($md5_path, $db->OWRITER | $db->OCREAT)
242 or die "can't open $md5_path: ",$db->errmsg( $db->ecode );
244 warn "open $md5_path";
246 $self->{md5sum}->{$login} = $db;
251 my ( $self, $pid, $login ) = @_;
254 my ( undef, undef, $uid, $gid, undef, undef, $email, $dir, $shell ) =
257 $self->{pid}->{$pid} = {
270 my ( $self, $pid ) = @_;
271 foreach my $login ( keys %{ $self->{md5sum} } ) {
272 # FIXME only login's?
273 $self->{md5sum}->{$login}->close;
274 warn "close md5sum $login";
276 delete $self->{md5sum};
278 delete $self->{pid}->{$pid};
283 my ( $self, $data ) = @_;
284 if ( $data =~ m/\[(\d+)\] rsync \w+ (\S+) \w+ (\S+)/ ) {
285 my ( $pid, $module, $login ) = ( $1, $2, $3 );
286 $self->init_pid_login( $pid, $login );
287 } elsif ( $data =~ m/\[(\d+)\] sent \S+ bytes\s+received \S+ bytes/ ) {
289 $self->cleanup_pid( $pid );
291 # warn "## rsync_log $data";
296 my ( $self, $data, $path ) = @_;
297 my $blob = $self->{pid}->{ $data->{pid} }->{dir} || die "no dir for $data->{pid} in ",dump( $self->{pid} );
298 $blob .= '/' . ( defined $path ? $path : $data->{file} );
304 my ( $self,$data ) = @_;
306 my $path = $self->blob_path($data);
308 if ( $data->{itemize} =~ m/^[c>]([fdL])/ ) { # received change/create
311 if ( $type eq 'f' ) {
312 $self->modify_file( $data ) && # selective dedup
313 $self->dedup( $data, $path );
314 } elsif ( $type eq 'd' ) {
315 $self->make_dir( $data );
316 } elsif ( $type eq 'L' ) {
317 $self->new_link( $data );
319 die "unknown type $type ", dump $data;
321 } elsif ( $data->{itemize} =~ m/\*deleting/ ) {
322 $self->remove_file($data);
324 warn "IGNORED ",dump($data) if $ENV{DEBUG};
330 my ( $self, $data ) = @_;
332 my $pool = $self->{md5pool} || die "no md5pool in ",dump $self;
333 mkdir $pool unless -e $pool;
335 my $md5 = $data->{md5} || die "no md5 in ",dump $data;
336 my $path = $self->blob_path($data);
338 if ( -e "$pool/$md5" ) {
339 warn "dedup hit $md5 $path\n";
340 my $dedup = $path . '.dedup';
341 rename $path, $dedup;
342 link "$pool/$md5", $path;
346 link $path, "$pool/$md5";
347 warn "dedup +++ $md5 $path";
350 $self->md5sum($data)->put( $data->{file} => $md5 );
353 my $empty_md5 = " " x 32;
356 my ( $self, $data, $path ) = @_;
358 if ( $data->{file} =~ /^(.+\/)?md5sum$/ ) {
361 warn "IMPORT ", $data->{file}, "\n";
362 open(my $md5sum, '<', $path);
365 my ( $md5, $file ) = split(/\s+/,$_,2);
366 if ( ! -e "$self->{md5path}/$md5" ) {
367 warn "MISSING $md5 $file\n";
375 my $new_path = $self->blob_path($new);
376 if ( ! -e $new_path ) {
377 # create path from md5sum file
379 $imported += link "$self->{md5path}/$md5", $new_path;
380 $self->new_file($new);
381 warn "import from $path ",dump($new);
382 $self->md5pool( $new );
384 $self->md5pool( $new );
387 print "INFO imported $imported files from ",dump($data);
390 if ( $data->{md5} ne $empty_md5 ) {
391 $self->md5sum($data)->put( $data->{file} => $data->{md5} );
392 $self->md5pool( $data );
394 warn "empty md5", dump $data;