12 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
13 |
use Carp qw/confess/; |
use Carp qw/confess/; |
14 |
use IO::File; |
use IO::File; |
15 |
|
use Getopt::Long; |
16 |
|
|
17 |
|
my $debug = 0; |
18 |
|
my $fuse_debug = 0; |
19 |
|
my $stats = 1; |
20 |
|
|
21 |
|
GetOptions( |
22 |
|
'debug+' => \$debug, |
23 |
|
'fuse-debug+' => \$fuse_debug, |
24 |
|
'stats!' => \$stats, |
25 |
|
); |
26 |
|
|
27 |
my $mount = { |
my $mount = { |
28 |
from => '/tmp/comp', |
from => '/tmp/comp', |
30 |
tmp => '/dev/shm/comp', |
tmp => '/dev/shm/comp', |
31 |
}; |
}; |
32 |
|
|
|
my $debug = shift @ARGV; |
|
|
|
|
33 |
my $skip_extensions_regex = qr/\.(?:sw[a-z]|gif|png|jpeg|jpg|avi|rar|zip|bz2|gz|tgz|avi|mpeg|mpg|tmp|temp)$/i; |
my $skip_extensions_regex = qr/\.(?:sw[a-z]|gif|png|jpeg|jpg|avi|rar|zip|bz2|gz|tgz|avi|mpeg|mpg|tmp|temp)$/i; |
34 |
|
|
35 |
# don't compress files smaller than this |
# don't compress files smaller than this |
44 |
|
|
45 |
my $pending; |
my $pending; |
46 |
|
|
47 |
|
sub real_name { |
48 |
|
my ( $dir, $name ) = @_; |
49 |
|
if ( -e "$dir/${name}.gz" ) { |
50 |
|
confess "ASSERT: unexpected $dir/$name exists" if -e "$dir/$name"; |
51 |
|
return "${name}.gz"; |
52 |
|
} |
53 |
|
return $name; |
54 |
|
} |
55 |
|
|
56 |
sub fixup { |
sub fixup { |
57 |
my ( $path ) = @_; |
my ( $path ) = @_; |
58 |
my $full = $mount->{from} . '/' . $path; |
return $mount->{from} . '/' . real_name( $mount->{from}, $path ); |
|
if ( -e $full . '.gz' ) { |
|
|
return $full . '.gz'; |
|
|
} |
|
|
return $full; |
|
59 |
} |
} |
60 |
|
|
61 |
sub original_name { |
sub original_name { |
83 |
return unpack("L", $buff); |
return unpack("L", $buff); |
84 |
} |
} |
85 |
|
|
|
sub unlink_all { |
|
|
my $file = shift; |
|
|
warn "# unlink_all( $file )\n"; |
|
|
|
|
|
my $path = fixup( $file ); |
|
|
unlink $path || return 0; |
|
|
|
|
|
my $tmp = $mount->{tmp} . '/' . $file; |
|
|
unlink $tmp if ( -e $tmp ); |
|
|
|
|
|
delete( $pending->{$file} ); |
|
|
return 1; |
|
|
} |
|
|
|
|
86 |
sub x_getattr { |
sub x_getattr { |
87 |
my ($file) = fixup(shift); |
my ($file) = fixup(shift); |
88 |
my (@list) = lstat($file); |
my (@list) = lstat($file); |
124 |
undef $s; |
undef $s; |
125 |
} |
} |
126 |
|
|
127 |
|
sub tmp_path { |
128 |
|
my $file = shift; |
129 |
|
|
130 |
|
my $path = fixup( $file ); |
131 |
|
|
132 |
|
my $op = 'UNKNOWN'; |
133 |
|
|
134 |
|
if (defined( $pending->{$file} )) { |
135 |
|
$path = $pending->{$file}->{path} || confess "no path for $file in ",dump( $pending ); |
136 |
|
$op = 'opened'; |
137 |
|
} else { |
138 |
|
my $tmp = $mount->{tmp} . '/' . $file; |
139 |
|
if ( -e $tmp ) { |
140 |
|
$path = $tmp; |
141 |
|
$op = 'existing'; |
142 |
|
} elsif ( $path =~ m/\.gz$/ ) { |
143 |
|
my $dest_path = $tmp; |
144 |
|
$dest_path =~ s!/[^/]+$!!; #!vim-fix |
145 |
|
mkpath $dest_path unless -e $dest_path; |
146 |
|
if ( -s $path ) { |
147 |
|
file_copy( '<:gzip', $path, '>', $tmp ) |
148 |
|
} else { |
149 |
|
confess "ASSERT: filesystem corruption, $path is zero size in ",dump( $pending ); |
150 |
|
} |
151 |
|
$path = $tmp; |
152 |
|
$op = 'created'; |
153 |
|
} |
154 |
|
confess "ASSERT: path shouldn't exist for $file in ", dump( $pending ) if defined( $pending->{$file}->{path} ); |
155 |
|
confess "ASSERT: open shouldn't exist for $file in ", dump( $pending ) if defined( $pending->{$file}->{open} ); |
156 |
|
$pending->{$file}->{path} = $path; |
157 |
|
$pending->{$file}->{open} = 0; # not really opened, just uncompressed |
158 |
|
warn "## tmp_file( $file ) $op $path [", -s $path, "]\n" if $debug; |
159 |
|
} |
160 |
|
return $path; |
161 |
|
} |
162 |
|
|
163 |
|
sub compress_file2path { |
164 |
|
my ( $file, $path ) = @_; |
165 |
|
|
166 |
|
my $dest = fixup( $file ); |
167 |
|
|
168 |
|
if ( defined($pending->{$file}) ) { |
169 |
|
my $pending_path = $pending->{$file}->{path} || confess "no path for $file in ",dump( $pending ); |
170 |
|
|
171 |
|
if ( $pending->{$file}->{open} > 1 ) { |
172 |
|
warn "$file used ", $pending->{$file}->{open}, " times, delaying compression\n"; |
173 |
|
return; |
174 |
|
} elsif ( ! $path ) { |
175 |
|
$path = $pending_path; |
176 |
|
} elsif ( $pending_path ne $path ) { |
177 |
|
confess "ASSERT: compressing into $path instead of $pending_path"; |
178 |
|
} |
179 |
|
} |
180 |
|
|
181 |
|
confess "need path" unless $path; |
182 |
|
|
183 |
|
# cleanup old compressed copy |
184 |
|
if ( $dest =~ /\.gz$/ ) { |
185 |
|
warn "## remove old $dest\n"; |
186 |
|
unlink $dest || confess "can't remove $dest: $!"; |
187 |
|
$dest =~ s/\.gz$//; |
188 |
|
confess "ASSERT: uncompressed $dest shouldn't exist!" if -e $dest; |
189 |
|
} |
190 |
|
|
191 |
|
if ( $path =~ $skip_extensions_regex ) { |
192 |
|
warn "$path [",-s $path,"] skipped compression\n"; |
193 |
|
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
194 |
|
} elsif ( -s $path < $min_compress_size ) { |
195 |
|
warn "$path [",-s $path,"] uncompressed, too small\n"; |
196 |
|
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
197 |
|
} else { |
198 |
|
warn "$path [",-s $path,"] compressing\n"; |
199 |
|
|
200 |
|
my $comp = $dest . '.gz'; |
201 |
|
file_copy( '<', $path, '>:gzip', $comp ); |
202 |
|
|
203 |
|
my ( $size_path, $size_comp ) = ( -s $path, -s $comp ); |
204 |
|
|
205 |
|
if ( $size_path <= $size_comp ) { |
206 |
|
warn ">>> $size_path <= $size_comp leaving uncompressed $dest\n"; |
207 |
|
unlink $comp || confess "can't remove: $comp: $!"; |
208 |
|
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
209 |
|
} else { |
210 |
|
warn ">>> compressed $size_path -> $size_comp ",int(($size_comp * 100) / $size_path),"% $comp\n"; |
211 |
|
|
212 |
|
# FIXME add timeout to remove uncompressed version? |
213 |
|
unlink $path || confess "can't remove $path: $!"; |
214 |
|
|
215 |
|
if ( -e $dest ) { |
216 |
|
warn "## cleanup uncompressed $dest\n" if $debug; |
217 |
|
unlink $dest || confess "can't remove $dest: $!"; |
218 |
|
} |
219 |
|
} |
220 |
|
|
221 |
|
} |
222 |
|
} |
223 |
|
|
224 |
sub x_open { |
sub x_open { |
225 |
my ($file) = shift; |
my ($file) = shift; |
226 |
my ($mode) = shift; |
my ($mode) = shift; |
236 |
return 0; |
return 0; |
237 |
} |
} |
238 |
|
|
|
$pending->{$file}->{open}++; |
|
|
|
|
239 |
my $mode_desc = { |
my $mode_desc = { |
240 |
rdonly => $mode && O_RDONLY, |
rdonly => $mode && O_RDONLY, |
241 |
rdwr => $mode && O_RDWR, |
rdwr => $mode && O_RDWR, |
243 |
create => $mode && O_CREAT, |
create => $mode && O_CREAT, |
244 |
trunc => $mode && O_TRUNC, |
trunc => $mode && O_TRUNC, |
245 |
}; |
}; |
246 |
my $path = fixup($file); |
|
247 |
|
my $path = tmp_path( $file ); |
248 |
|
|
249 |
warn "## open( $file, $mode ) pending: ", $pending->{$file}->{open}, " mode $mode: ", dump( $mode_desc )," $path [", -s $path, "]\n" if $debug; |
warn "## open( $file, $mode ) pending: ", $pending->{$file}->{open}, " mode $mode: ", dump( $mode_desc )," $path [", -s $path, "]\n" if $debug; |
250 |
|
|
251 |
my $fh; |
my $fh; |
252 |
|
|
253 |
my $tmp = $mount->{tmp} . '/' . $file; |
if ( sysopen($fh , $path, $mode) ) { |
254 |
if ( -e $tmp ) { |
close($fh) || confess "can't close $path: $!"; |
255 |
$path = $tmp; |
warn "<<< sysopen $path [", -e $path ? -s $path : 'new' , "]\n"; |
256 |
} elsif ( $path =~ m/\.gz$/ ) { |
$pending->{$file}->{open}++; |
257 |
my $dest_path = $tmp; |
return 0; |
258 |
$dest_path =~ s!/[^/]+$!!; #!vim-fix |
} else { |
259 |
mkpath $dest_path unless -e $dest_path; |
warn "ERROR: can't open $path -- $!"; |
260 |
file_copy( '<:gzip', $path, '>', $tmp ); |
return -$!; |
|
$path = $tmp; |
|
261 |
} |
} |
|
warn "<<< open abs path: $path [", -e $path ? -s $path : 'new' , "]\n"; |
|
|
return -$! unless sysopen($fh , $path, $mode); |
|
|
close($fh); |
|
262 |
|
|
|
$pending->{$file}->{path} = $path; |
|
|
return 0; |
|
263 |
} |
} |
264 |
|
|
265 |
sub x_read { |
sub x_read { |
276 |
|
|
277 |
if(seek($fh,$off,SEEK_SET)) { |
if(seek($fh,$off,SEEK_SET)) { |
278 |
read($fh,$rv,$bufsize); |
read($fh,$rv,$bufsize); |
279 |
|
$pending->{$file}->{read} += length($rv) if $stats; |
280 |
} |
} |
281 |
|
|
282 |
return $rv; |
return $rv; |
284 |
|
|
285 |
sub x_write { |
sub x_write { |
286 |
my ($file,$buf,$off) = @_; |
my ($file,$buf,$off) = @_; |
287 |
$pending->{$file}->{write}++; |
|
288 |
my $rv; |
my $rv; |
289 |
my $path = fixup($file); |
my $path = fixup($file); |
290 |
|
|
292 |
|
|
293 |
return -ENOENT() unless -e $path; |
return -ENOENT() unless -e $path; |
294 |
|
|
295 |
|
$path = $pending->{$file}->{path} || confess "no path for $file in ", dump( $pending ); |
296 |
|
confess "write into non-existant $path for $file" unless -e $path; |
297 |
|
|
298 |
my $fh = new IO::File; |
my $fh = new IO::File; |
299 |
return -ENOSYS() unless open($fh,'+<',$pending->{$file}->{path}); |
return -ENOSYS() unless open($fh,'+<',$path); |
300 |
if($rv = seek( $fh ,$off,SEEK_SET)) { |
if($rv = seek( $fh ,$off,SEEK_SET)) { |
301 |
$rv = print( $fh $buf ); |
$rv = print( $fh $buf ); |
302 |
warn "## write ", $pending->{$file}->{path}, " $off ",length( $buf ), "\n" if $debug; |
my $size = length($buf); |
303 |
|
warn "## write $path offset $off [$size]\n" if $debug; |
304 |
|
$pending->{$file}->{write} += $size; |
305 |
} |
} |
306 |
$rv = -ENOSYS() unless $rv; |
$rv = -ENOSYS() unless $rv; |
307 |
close($fh); |
close($fh) || warn "can't close $path: $!"; |
308 |
return length($buf); |
return length($buf); |
309 |
} |
} |
310 |
|
|
311 |
sub err { return (-shift || -$!) } |
sub err { return (-shift || -$!) } |
312 |
|
|
313 |
sub x_readlink { return readlink(fixup(shift)); } |
sub x_readlink { return readlink(fixup(shift)); } |
314 |
sub x_unlink { return unlink_all( shift ) ? 0 : -$! } |
|
315 |
|
sub x_unlink { |
316 |
|
my $file = shift; |
317 |
|
my $path = fixup( $file ); |
318 |
|
|
319 |
|
if ( $file =~ m#\Q/.fuse_hidden\E# ) { |
320 |
|
return unlink $path ? 0 : -$1; |
321 |
|
} |
322 |
|
|
323 |
|
warn "# unlink( $file )\n"; |
324 |
|
|
325 |
sub x_symlink { return symlink(shift,fixup(shift)) ? 0 : -$!; } |
unlink $path || return 0; |
326 |
|
|
327 |
|
my $tmp = $mount->{tmp} . '/' . $file; |
328 |
|
unlink $tmp if ( -e $tmp ); |
329 |
|
|
330 |
|
delete( $pending->{$file} ); |
331 |
|
return 0; |
332 |
|
} |
333 |
|
|
334 |
|
sub x_symlink { |
335 |
|
my ($from,$to) = @_; |
336 |
|
|
337 |
|
my $from_path = $from; #fixup( $from ); |
338 |
|
my $to_path = fixup( $to ); |
339 |
|
|
340 |
|
my $rv = symlink( $from_path, $to_path ) ? 0 : -$!; |
341 |
|
warn "# symlink( $from_path -> $to_path ) = $rv\n" if $debug; |
342 |
|
|
343 |
|
my $tmp = $mount->{tmp} . '/' . $from; |
344 |
|
if ( -e $tmp ) { |
345 |
|
my $tmp_to = $mount->{$tmp} . '/' . $to; |
346 |
|
symlink( $tmp, $tmp_to ) || confess "can't symlink $tmp -> $tmp_to: $!"; |
347 |
|
} |
348 |
|
return $rv; |
349 |
|
} |
350 |
|
|
351 |
|
sub x_link { |
352 |
|
my ($from,$to) = @_; |
353 |
|
|
354 |
|
my $from_path = fixup($from); |
355 |
|
my $to_path = fixup($to); |
356 |
|
$to_path .= '.gz' if ( $from_path =~ m/\.gz$/ && $to_path !~ m/\.gz$/ ); |
357 |
|
|
358 |
|
my $rv = link( $from_path, $to_path ) ? 0 : -$!; |
359 |
|
|
360 |
|
warn "# link( $from_path -> $to_path ) = $rv\n" if $debug; |
361 |
|
|
362 |
|
return $rv; |
363 |
|
} |
364 |
|
|
365 |
sub x_rename { |
sub x_rename { |
366 |
my ($old,$new) = @_; |
my ($old,$new) = @_; |
373 |
|
|
374 |
my $tmp = $mount->{tmp} . '/' . $old; |
my $tmp = $mount->{tmp} . '/' . $old; |
375 |
if ( -e $tmp ) { |
if ( -e $tmp ) { |
376 |
my $new_tmp = $mount->{tmp} . '/' . $new; |
if ( $new =~ m#\Q/.fuse_hidden\E# ) { |
377 |
rename $tmp, $new_tmp || confess "can't rename $tmp -> $new_tmp : $!"; |
unlink $tmp || confess "can't unlink $tmp for $new"; |
378 |
|
} else { |
379 |
|
my $new_tmp = $mount->{tmp} . '/' . $new; |
380 |
|
rename $tmp, $new_tmp || confess "can't rename $tmp -> $new_tmp : $!"; |
381 |
|
} |
382 |
} |
} |
383 |
|
|
384 |
if (defined( $pending->{$old} )) { |
if (defined( $pending->{$old} )) { |
389 |
$pending->{$new}->{path} = $path; |
$pending->{$new}->{path} = $path; |
390 |
|
|
391 |
delete( $pending->{$old} ); |
delete( $pending->{$old} ); |
392 |
|
warn "## tweaking pending to ", dump( $pending ) if $debug; |
393 |
} |
} |
394 |
|
|
395 |
return $err; |
return $err; |
396 |
} |
} |
|
sub x_link { return link(fixup(shift),fixup(shift)) ? 0 : -$! } |
|
397 |
|
|
398 |
sub x_chown { |
sub x_chown { |
399 |
my ($path) = fixup(shift); |
my ($file,$uid,$gid) = @_; |
400 |
|
my $path = fixup($file); |
401 |
print "nonexistent $path\n" unless -e $path; |
print "nonexistent $path\n" unless -e $path; |
|
my ($uid,$gid) = @_; |
|
402 |
# perl's chown() does not chown symlinks, it chowns the symlink's |
# perl's chown() does not chown symlinks, it chowns the symlink's |
403 |
# target. it fails when the link's target doesn't exist, because |
# target. it fails when the link's target doesn't exist, because |
404 |
# the stat64() syscall fails. |
# the stat64() syscall fails. |
405 |
# this causes error messages when unpacking symlinks in tarballs. |
# this causes error messages when unpacking symlinks in tarballs. |
406 |
my ($err) = syscall(&SYS_lchown,$path,$uid,$gid,$path) ? -$! : 0; |
my ($err) = syscall(&SYS_lchown,$path,$uid,$gid,$path) ? -$! : 0; |
407 |
|
|
408 |
|
my $tmp = $mount->{tmp} . '/' . $file; |
409 |
|
syscall(&SYS_lchown,$file,$uid,$gid,$path) if -e $tmp; |
410 |
|
|
411 |
return $err; |
return $err; |
412 |
} |
} |
413 |
|
|
420 |
|
|
421 |
sub x_truncate { |
sub x_truncate { |
422 |
my ( $file,$size ) = @_; |
my ( $file,$size ) = @_; |
423 |
my $path = fixup($file); |
|
424 |
|
#confess "no pending file $file to truncate in ", dump( $pending ) unless defined( $pending->{$file} ); |
425 |
|
|
426 |
|
my $path = tmp_path( $file ); |
427 |
my $rv = truncate( $path, $size ) ? 0 : -$! ; |
my $rv = truncate( $path, $size ) ? 0 : -$! ; |
428 |
if ( $path =~ m/\.gz$/ ) { |
warn "## truncate( $file $size ) $path [", -s $path, "] = $rv\n" if $debug; |
429 |
my $no_gz = $path; |
compress_file2path( $file, $path ); |
430 |
$no_gz =~ s/\.gz$//; |
|
|
rename $path, $no_gz || confess "can't rename $path -> $no_gz: $!"; |
|
|
} |
|
|
warn "## truncate( $file $size ) $path [", -s $path, "]\n"; |
|
|
$pending->{$file}->{write}++; |
|
431 |
return $rv; |
return $rv; |
432 |
} |
} |
433 |
|
|
434 |
sub x_utime { return utime($_[1],$_[2],fixup($_[0])) ? 0:-$!; } |
sub x_utime { return utime($_[1],$_[2],fixup($_[0])) ? 0:-$!; } |
435 |
|
|
436 |
sub x_mkdir { my ($name, $perm) = @_; return 0 if mkdir(fixup($name),$perm); return -$!; } |
sub x_mkdir { my ($name, $perm) = @_; return 0 if mkdir(fixup($name),$perm); return -$!; } |
448 |
|
|
449 |
sub x_release { |
sub x_release { |
450 |
my ( $file, $mode ) = @_; |
my ( $file, $mode ) = @_; |
451 |
|
|
452 |
if ( ! defined( $pending->{$file} ) ) { |
if ( ! defined( $pending->{$file} ) ) { |
453 |
warn "release $file, NO PENDING DATA\n"; |
warn "release $file, NO PENDING DATA\n"; |
454 |
return 0; |
return 0; |
455 |
} elsif ( ! defined( $pending->{$file}->{write} ) ) { |
} elsif ( ! defined( $pending->{$file}->{write} ) ) { |
456 |
warn "release $file, not written into\n"; |
warn "release $file, not written into\n"; |
457 |
} elsif ( defined( $pending->{$file}->{open} ) && $pending->{$file}->{open} == 1 ) { |
} elsif ( $file =~ m#\Q/.fuse_hidden\E# ) { |
458 |
my $path = $pending->{$file}->{path} || confess "no path for $file ? ", dump( $pending ); |
warn "release internal $file\n" if $debug; |
|
my $dest = fixup( $file ); |
|
|
|
|
|
# cleanup old compressed copy |
|
|
if ( $dest =~ /\.gz$/ ) { |
|
|
warn "## remove old $dest\n"; |
|
|
unlink $dest || confess "can't remove $dest: $!"; |
|
|
$dest =~ s/\.gz$//; |
|
|
} |
|
|
|
|
|
if ( $file =~ $skip_extensions_regex ) { |
|
|
warn "release $file [",-s $path,"] skipped compression\n"; |
|
|
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
|
|
} elsif ( -s $path < $min_compress_size ) { |
|
|
warn "release $file [",-s $path,"] uncompressed, too small\n"; |
|
|
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
|
|
} else { |
|
|
warn "release $file [",-s $path,"] compressing\n"; |
|
|
|
|
|
my $comp = $dest . '.gz'; |
|
|
file_copy( '<', $path, '>:gzip', $comp ); |
|
|
|
|
|
my ( $size_path, $size_comp ) = ( -s $path, -s $comp ); |
|
|
|
|
|
if ( $size_path <= $size_comp ) { |
|
|
warn ">>> $size_path <= $size_comp leaving uncompressed\n"; |
|
|
unlink $comp || warn "can't reamove: $comp: $!"; |
|
|
} else { |
|
|
warn ">>> compressed $size_path -> $size_comp ",int(($size_comp * 100) / $size_path),"%\n"; |
|
|
# FIXME add timeout to remove uncompressed version? |
|
|
unlink $path || warn "can't remove $path: $!"; |
|
|
} |
|
|
} |
|
459 |
} else { |
} else { |
460 |
warn "release $file, but still used ", $pending->{$file}->{open} , " times, delaying compression\n"; |
compress_file2path( $file ); |
461 |
} |
} |
462 |
|
|
463 |
$pending->{$file}->{open}--; |
$pending->{$file}->{open}--; |
464 |
if ( $pending->{$file}->{open} == 0 ) { |
if ( $pending->{$file}->{open} == 0 ) { |
465 |
warn "## cleanup pending $file [", -s fixup($file), "]\n" if $debug; |
warn "## cleanup pending $file [", -s fixup($file), "]\n" if $debug; |
466 |
delete( $pending->{$file} ); |
delete( $pending->{$file} ); |
467 |
} |
} |
468 |
|
|
469 |
return 0; |
return 0; |
470 |
} |
} |
471 |
|
|
493 |
statfs =>"main::x_statfs", |
statfs =>"main::x_statfs", |
494 |
release =>"main::x_release", |
release =>"main::x_release", |
495 |
# threaded=>1, |
# threaded=>1, |
496 |
# debug => 1, |
debug => $fuse_debug, |
497 |
); |
); |