1 package DBM::Deep::Engine;
5 use Fcntl qw( :DEFAULT :flock :seek );
8 # Set to 4 and 'N' for 32-bit offset tags (default). Theoretical limit of 4 GB per file.
9 # (Perl must be compiled with largefile support for files > 2 GB)
11 # Set to 8 and 'Q' for 64-bit offsets. Theoretical limit of 16 XB per file.
12 # (Perl must be compiled with largefile and 64-bit long support)
15 # Set to 4 and 'N' for 32-bit data length prefixes. Limit of 4 GB for each key/value.
16 # Upgrading this is possible (see above) but probably not necessary. If you need
17 # more than 4 GB for a single key or value, this module is really not for you :-)
19 our ($LONG_SIZE, $LONG_PACK, $DATA_LENGTH_SIZE, $DATA_LENGTH_PACK);
21 # Maximum number of buckets per list before another level of indexing is done.
22 # Increase this value for slightly greater speed, but larger database files.
23 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
25 our $MAX_BUCKETS = 16;
27 our ($INDEX_SIZE, $BUCKET_SIZE, $BUCKET_LIST_SIZE);
32 # Precalculate index, bucket and bucket list sizes
35 #XXX I don't like this ...
36 set_pack() unless defined $LONG_SIZE;
38 $INDEX_SIZE = 256 * $LONG_SIZE;
39 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
40 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
45 # Set pack/unpack modes (see file header for more)
47 my ($long_s, $long_p, $data_s, $data_p) = @_;
49 $LONG_SIZE = $long_s ? $long_s : 4;
50 $LONG_PACK = $long_p ? $long_p : 'N';
52 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
53 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
61 # Set key digest function (default is MD5)
63 my ($digest_func, $hash_size) = @_;
65 $self->{digest} = $digest_func ? $digest_func : \&Digest::MD5::md5;
67 $HASH_SIZE = $hash_size ? $hash_size : 16;
81 digest => \&Digest::MD5::md5,
86 #XXX Where does 256 come from?
87 $self->{index_size} = 256 * $self->{long_size};
88 $self->{bucket_size} = $self->{hash_size} + $self->{long_size};
89 $self->{bucket_list_size} = $self->{max_buckets} * $self->{bucket_size};
98 $self->open( $obj ) if !defined $obj->_fh;
100 #XXX Is {end} busted? Can't we just seek( $fh, 0, SEEK_END ) ?
101 unless ( $obj->_root->{inode} ) {
102 my @stats = stat($obj->_fh);
103 $obj->_root->{inode} = $stats[1];
104 $obj->_root->{end} = $stats[7];
112 # Open a fh to the database, create if nonexistent.
113 # Make sure file signature matches DBM::Deep spec.
118 if (defined($obj->_fh)) { $self->close( $obj ); }
121 local $SIG{'__DIE__'};
122 # Theoretically, adding O_BINARY should remove the need for the binmode
123 # Of course, testing it is going to be ... interesting.
124 my $flags = O_RDWR | O_CREAT | O_BINARY;
127 sysopen( $fh, $obj->_root->{file}, $flags )
129 $obj->_root->{fh} = $fh;
130 }; if ($@ ) { $obj->_throw_error( "Received error: $@\n" ); }
131 if (! defined($obj->_fh)) {
132 return $obj->_throw_error("Cannot sysopen file: " . $obj->_root->{file} . ": $!");
137 #XXX Can we remove this by using the right sysopen() flags?
138 # Maybe ... q.v. above
139 binmode $fh; # for win32
141 if ($obj->_root->{autoflush}) {
142 my $old = select $fh;
147 seek($fh, 0 + $obj->_root->{file_offset}, SEEK_SET);
150 my $bytes_read = read( $fh, $signature, length(DBM::Deep->SIG_FILE));
153 # File is empty -- write signature and master index
156 seek($fh, 0 + $obj->_root->{file_offset}, SEEK_SET);
157 print( $fh DBM::Deep->SIG_FILE);
158 $self->create_tag($obj, $obj->_base_offset, $obj->_type, chr(0) x $INDEX_SIZE);
160 my $plain_key = "[base]";
161 print( $fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
163 # Flush the filehandle
164 my $old_fh = select $fh;
165 my $old_af = $|; $| = 1; $| = $old_af;
172 # Check signature was valid
174 unless ($signature eq DBM::Deep->SIG_FILE) {
175 $self->close( $obj );
176 return $obj->_throw_error("Signature not found -- file is not a Deep DB");
180 # Get our type from master index signature
182 my $tag = $self->load_tag($obj, $obj->_base_offset);
184 #XXX We probably also want to store the hash algorithm name and not assume anything
185 #XXX The cool thing would be to allow a different hashing algorithm at every level
188 return $obj->_throw_error("Corrupted file, no master index record");
190 if ($obj->{type} ne $tag->{signature}) {
191 return $obj->_throw_error("File type mismatch");
201 if ( my $fh = $obj->_root->{fh} ) {
204 $obj->_root->{fh} = undef;
211 # Given offset, signature and content, create tag and write to disk
214 my ($obj, $offset, $sig, $content) = @_;
215 my $size = length($content);
219 seek($fh, $offset + $obj->_root->{file_offset}, SEEK_SET);
220 print( $fh $sig . pack($DATA_LENGTH_PACK, $size) . $content );
222 if ($offset == $obj->_root->{end}) {
223 $obj->_root->{end} += DBM::Deep->SIG_SIZE + $DATA_LENGTH_SIZE + $size;
229 offset => $offset + DBM::Deep->SIG_SIZE + $DATA_LENGTH_SIZE,
236 # Given offset, load single tag and return signature, size and data
239 my ($obj, $offset) = @_;
243 seek($fh, $offset + $obj->_root->{file_offset}, SEEK_SET);
244 if (eof $fh) { return undef; }
247 read( $fh, $b, DBM::Deep->SIG_SIZE + $DATA_LENGTH_SIZE );
248 my ($sig, $size) = unpack( "A $DATA_LENGTH_PACK", $b );
251 read( $fh, $buffer, $size);
256 offset => $offset + DBM::Deep->SIG_SIZE + $DATA_LENGTH_SIZE,
263 # Given index tag, lookup single entry in index and return .
266 my ($obj, $tag, $index) = @_;
268 my $location = unpack($LONG_PACK, substr($tag->{content}, $index * $LONG_SIZE, $LONG_SIZE) );
269 if (!$location) { return; }
271 return $self->load_tag( $obj, $location );
276 # Adds one key/value pair to bucket list, given offset, MD5 digest of key,
277 # plain (undigested) key and value.
280 my ($obj, $tag, $md5, $plain_key, $value) = @_;
281 my $keys = $tag->{content};
285 my $root = $obj->_root;
287 my $is_dbm_deep = eval { local $SIG{'__DIE__'}; $value->isa( 'DBM::Deep' ) };
288 my $internal_ref = $is_dbm_deep && ($value->_root eq $root);
293 # Iterate through buckets, seeing if this is a new entry or a replace.
295 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
296 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
299 # Found empty bucket (end of list). Populate and exit loop.
303 $location = $internal_ref
304 ? $value->_base_offset
307 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
308 print( $fh $md5 . pack($LONG_PACK, $location) );
312 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
315 # Found existing bucket with same key. Replace with new value.
320 $location = $value->_base_offset;
321 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
322 print( $fh $md5 . pack($LONG_PACK, $location) );
326 seek($fh, $subloc + DBM::Deep->SIG_SIZE + $root->{file_offset}, SEEK_SET);
328 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
331 # If value is a hash, array, or raw value with equal or less size, we can
332 # reuse the same content area of the database. Otherwise, we have to create
333 # a new content area at the EOF.
336 my $r = Scalar::Util::reftype( $value ) || '';
337 if ( $r eq 'HASH' || $r eq 'ARRAY' ) {
338 $actual_length = $INDEX_SIZE;
340 # if autobless is enabled, must also take into consideration
341 # the class name, as it is stored along with key/value.
342 if ( $root->{autobless} ) {
343 my $value_class = Scalar::Util::blessed($value);
344 if ( defined $value_class && !$value->isa('DBM::Deep') ) {
345 $actual_length += length($value_class);
349 else { $actual_length = length($value); }
351 if ($actual_length <= $size) {
355 $location = $root->{end};
356 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $HASH_SIZE + $root->{file_offset}, SEEK_SET);
357 print( $fh pack($LONG_PACK, $location) );
365 # If this is an internal reference, return now.
366 # No need to write value or plain key
373 # If bucket didn't fit into list, split into a new index level
376 seek($fh, $tag->{ref_loc} + $root->{file_offset}, SEEK_SET);
377 print( $fh pack($LONG_PACK, $root->{end}) );
379 my $index_tag = $self->create_tag($obj, $root->{end}, DBM::Deep->SIG_INDEX, chr(0) x $INDEX_SIZE);
382 $keys .= $md5 . pack($LONG_PACK, 0);
384 for (my $i=0; $i<=$MAX_BUCKETS; $i++) {
385 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
387 my $old_subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) +
388 $HASH_SIZE, $LONG_SIZE));
389 my $num = ord(substr($key, $tag->{ch} + 1, 1));
391 if ($offsets[$num]) {
392 my $offset = $offsets[$num] + DBM::Deep->SIG_SIZE + $DATA_LENGTH_SIZE;
393 seek($fh, $offset + $root->{file_offset}, SEEK_SET);
395 read( $fh, $subkeys, $BUCKET_LIST_SIZE);
397 for (my $k=0; $k<$MAX_BUCKETS; $k++) {
398 my $subloc = unpack($LONG_PACK, substr($subkeys, ($k * $BUCKET_SIZE) +
399 $HASH_SIZE, $LONG_SIZE));
401 seek($fh, $offset + ($k * $BUCKET_SIZE) + $root->{file_offset}, SEEK_SET);
402 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
408 $offsets[$num] = $root->{end};
409 seek($fh, $index_tag->{offset} + ($num * $LONG_SIZE) + $root->{file_offset}, SEEK_SET);
410 print( $fh pack($LONG_PACK, $root->{end}) );
412 my $blist_tag = $self->create_tag($obj, $root->{end}, DBM::Deep->SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
414 seek($fh, $blist_tag->{offset} + $root->{file_offset}, SEEK_SET);
415 print( $fh $key . pack($LONG_PACK, $old_subloc || $root->{end}) );
420 $location ||= $root->{end};
421 } # re-index bucket list
424 # Seek to content area and store signature, value and plaintext key
428 seek($fh, $location + $root->{file_offset}, SEEK_SET);
431 # Write signature based on content type, set content length and write actual value.
433 my $r = Scalar::Util::reftype($value) || '';
435 print( $fh DBM::Deep->TYPE_HASH );
436 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
437 $content_length = $INDEX_SIZE;
439 elsif ($r eq 'ARRAY') {
440 print( $fh DBM::Deep->TYPE_ARRAY );
441 print( $fh pack($DATA_LENGTH_PACK, $INDEX_SIZE) . chr(0) x $INDEX_SIZE );
442 $content_length = $INDEX_SIZE;
444 elsif (!defined($value)) {
445 print( $fh DBM::Deep->SIG_NULL );
446 print( $fh pack($DATA_LENGTH_PACK, 0) );
450 print( $fh DBM::Deep->SIG_DATA );
451 print( $fh pack($DATA_LENGTH_PACK, length($value)) . $value );
452 $content_length = length($value);
456 # Plain key is stored AFTER value, as keys are typically fetched less often.
458 print( $fh pack($DATA_LENGTH_PACK, length($plain_key)) . $plain_key );
461 # If value is blessed, preserve class name
463 if ( $root->{autobless} ) {
464 my $value_class = Scalar::Util::blessed($value);
465 if ( defined $value_class && $value_class ne 'DBM::Deep' ) {
467 # Blessed ref -- will restore later
470 print( $fh pack($DATA_LENGTH_PACK, length($value_class)) . $value_class );
471 $content_length += 1;
472 $content_length += $DATA_LENGTH_SIZE + length($value_class);
476 $content_length += 1;
481 # If this is a new content area, advance EOF counter
483 if ($location == $root->{end}) {
484 $root->{end} += DBM::Deep->SIG_SIZE;
485 $root->{end} += $DATA_LENGTH_SIZE + $content_length;
486 $root->{end} += $DATA_LENGTH_SIZE + length($plain_key);
490 # If content is a hash or array, create new child DBM::Deep object and
491 # pass each key or element to it.
494 my $branch = DBM::Deep->new(
495 type => DBM::Deep->TYPE_HASH,
496 base_offset => $location,
499 foreach my $key (keys %{$value}) {
500 $branch->STORE( $key, $value->{$key} );
503 elsif ($r eq 'ARRAY') {
504 my $branch = DBM::Deep->new(
505 type => DBM::Deep->TYPE_ARRAY,
506 base_offset => $location,
510 foreach my $element (@{$value}) {
511 $branch->STORE( $index, $element );
519 return $obj->_throw_error("Fatal error: indexing failed -- possibly due to corruption in file");
522 sub get_bucket_value {
524 # Fetch single value given tag and MD5 digested key.
527 my ($obj, $tag, $md5) = @_;
528 my $keys = $tag->{content};
533 # Iterate through buckets, looking for a key match
536 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
537 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
538 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
542 # Hit end of list, no match
547 if ( $md5 ne $key ) {
552 # Found match -- seek to offset and read signature
555 seek($fh, $subloc + $obj->_root->{file_offset}, SEEK_SET);
556 read( $fh, $signature, DBM::Deep->SIG_SIZE);
559 # If value is a hash or array, return new DBM::Deep object with correct offset
561 if (($signature eq DBM::Deep->TYPE_HASH) || ($signature eq DBM::Deep->TYPE_ARRAY)) {
562 my $obj = DBM::Deep->new(
564 base_offset => $subloc,
568 if ($obj->_root->{autobless}) {
570 # Skip over value and plain key to see if object needs
573 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
576 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
577 if ($size) { seek($fh, $size, SEEK_CUR); }
580 read( $fh, $bless_bit, 1);
581 if (ord($bless_bit)) {
583 # Yes, object needs to be re-blessed
586 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
587 if ($size) { read( $fh, $class_name, $size); }
588 if ($class_name) { $obj = bless( $obj, $class_name ); }
596 # Otherwise return actual value
598 elsif ($signature eq DBM::Deep->SIG_DATA) {
601 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
602 if ($size) { read( $fh, $value, $size); }
607 # Key exists, but content is null
617 # Delete single key/value pair given tag and MD5 digested key.
620 my ($obj, $tag, $md5) = @_;
621 my $keys = $tag->{content};
626 # Iterate through buckets, looking for a key match
629 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
630 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
631 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
635 # Hit end of list, no match
640 if ( $md5 ne $key ) {
645 # Matched key -- delete bucket and return
647 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $obj->_root->{file_offset}, SEEK_SET);
648 print( $fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
649 print( $fh chr(0) x $BUCKET_SIZE );
659 # Check existence of single key given tag and MD5 digested key.
662 my ($obj, $tag, $md5) = @_;
663 my $keys = $tag->{content};
666 # Iterate through buckets, looking for a key match
669 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
670 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
671 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
675 # Hit end of list, no match
680 if ( $md5 ne $key ) {
685 # Matched key -- return true
693 sub find_bucket_list {
695 # Locate offset for bucket list, given digested key
698 my ($obj, $md5) = @_;
701 # Locate offset for bucket list using digest index system
704 my $tag = $self->load_tag($obj, $obj->_base_offset);
705 if (!$tag) { return; }
707 while ($tag->{signature} ne DBM::Deep->SIG_BLIST) {
708 $tag = $self->index_lookup($obj, $tag, ord(substr($md5, $ch, 1)));
709 if (!$tag) { return; }
718 # Scan index and recursively step into deeper levels, looking for next key.
721 my ($obj, $offset, $ch, $force_return_next) = @_;
722 $force_return_next = undef unless $force_return_next;
724 my $tag = $self->load_tag($obj, $offset );
728 if ($tag->{signature} ne DBM::Deep->SIG_BLIST) {
729 my $content = $tag->{content};
731 if ($obj->{return_next}) { $start = 0; }
732 else { $start = ord(substr($obj->{prev_md5}, $ch, 1)); }
734 for (my $index = $start; $index < 256; $index++) {
735 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
737 my $result = $self->traverse_index( $obj, $subloc, $ch + 1, $force_return_next );
738 if (defined($result)) { return $result; }
742 $obj->{return_next} = 1;
745 elsif ($tag->{signature} eq DBM::Deep->SIG_BLIST) {
746 my $keys = $tag->{content};
747 if ($force_return_next) { $obj->{return_next} = 1; }
750 # Iterate through buckets, looking for a key match
752 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
753 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
754 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
758 # End of bucket list -- return to outer loop
760 $obj->{return_next} = 1;
763 elsif ($key eq $obj->{prev_md5}) {
765 # Located previous key -- return next one found
767 $obj->{return_next} = 1;
770 elsif ($obj->{return_next}) {
772 # Seek to bucket location and skip over signature
774 seek($fh, $subloc + DBM::Deep->SIG_SIZE + $obj->_root->{file_offset}, SEEK_SET);
777 # Skip over value to get to plain key
780 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
781 if ($size) { seek($fh, $size, SEEK_CUR); }
784 # Read in plain key and return as scalar
787 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
788 if ($size) { read( $fh, $plain_key, $size); }
794 $obj->{return_next} = 1;
795 } # tag is a bucket list
802 # Locate next key, given digested previous one
807 $obj->{prev_md5} = $_[1] ? $_[1] : undef;
808 $obj->{return_next} = 0;
811 # If the previous key was not specifed, start at the top and
812 # return the first one found.
814 if (!$obj->{prev_md5}) {
815 $obj->{prev_md5} = chr(0) x $HASH_SIZE;
816 $obj->{return_next} = 1;
819 return $self->traverse_index( $obj, $obj->_base_offset, 0 );