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