7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use DBM::Deep::Engine;
40 use vars qw( $VERSION );
41 $VERSION = q(0.99_01);
44 # Set to 4 and 'N' for 32-bit offset tags (default). Theoretical limit of 4 GB per file.
45 # (Perl must be compiled with largefile support for files > 2 GB)
47 # Set to 8 and 'Q' for 64-bit offsets. Theoretical limit of 16 XB per file.
48 # (Perl must be compiled with largefile and 64-bit long support)
54 # Set to 4 and 'N' for 32-bit data length prefixes. Limit of 4 GB for each key/value.
55 # Upgrading this is possible (see above) but probably not necessary. If you need
56 # more than 4 GB for a single key or value, this module is really not for you :-)
58 #my $DATA_LENGTH_SIZE = 4;
59 #my $DATA_LENGTH_PACK = 'N';
60 our ($LONG_SIZE, $LONG_PACK, $DATA_LENGTH_SIZE, $DATA_LENGTH_PACK);
63 # Maximum number of buckets per list before another level of indexing is done.
64 # Increase this value for slightly greater speed, but larger database files.
65 # DO NOT decrease this value below 16, due to risk of recursive reindex overrun.
67 our $MAX_BUCKETS = 16;
70 # Better not adjust anything below here, unless you're me :-)
74 # Setup digest function for keys
76 our ($DIGEST_FUNC, $HASH_SIZE);
77 #my $DIGEST_FUNC = \&Digest::MD5::md5;
80 # Precalculate index and bucket sizes based on values above.
83 our ($INDEX_SIZE, $BUCKET_SIZE, $BUCKET_LIST_SIZE);
90 # Setup file and tag signatures. These should never change.
92 sub SIG_FILE () { 'DPDB' }
93 sub SIG_HASH () { 'H' }
94 sub SIG_ARRAY () { 'A' }
95 sub SIG_SCALAR () { 'S' }
96 sub SIG_NULL () { 'N' }
97 sub SIG_DATA () { 'D' }
98 sub SIG_INDEX () { 'I' }
99 sub SIG_BLIST () { 'B' }
100 sub SIG_SIZE () { 1 }
103 # Setup constants for users to pass to new()
105 sub TYPE_HASH () { SIG_HASH }
106 sub TYPE_ARRAY () { SIG_ARRAY }
107 sub TYPE_SCALAR () { SIG_SCALAR }
113 if (scalar(@_) > 1) {
115 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
119 elsif ( ref $_[0] ) {
120 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
121 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
126 $args = { file => shift };
134 # Class constructor method for Perl OO interface.
135 # Calls tie() and returns blessed reference to tied hash or array,
136 # providing a hybrid OO/tie interface.
139 my $args = $class->_get_args( @_ );
142 # Check if we want a tied hash or array.
145 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
146 $class = 'DBM::Deep::Array';
147 require DBM::Deep::Array;
148 tie @$self, $class, %$args;
151 $class = 'DBM::Deep::Hash';
152 require DBM::Deep::Hash;
153 tie %$self, $class, %$args;
156 return bless $self, $class;
161 # Setup $self and bless into this class.
166 # These are the defaults to be optionally overridden below
169 base_offset => length(SIG_FILE),
170 engine => 'DBM::Deep::Engine',
173 foreach my $param ( keys %$self ) {
174 next unless exists $args->{$param};
175 $self->{$param} = delete $args->{$param}
178 # locking implicitly enables autoflush
179 if ($args->{locking}) { $args->{autoflush} = 1; }
181 $self->{root} = exists $args->{root}
183 : DBM::Deep::_::Root->new( $args );
185 if (!defined($self->_fh)) { $self->{engine}->open( $self ); }
192 require DBM::Deep::Hash;
193 return DBM::Deep::Hash->TIEHASH( @_ );
198 require DBM::Deep::Array;
199 return DBM::Deep::Array->TIEARRAY( @_ );
202 #XXX Unneeded now ...
206 sub _get_bucket_value {
208 # Fetch single value given tag and MD5 digested key.
211 my ($tag, $md5) = @_;
212 my $keys = $tag->{content};
217 # Iterate through buckets, looking for a key match
220 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
221 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
222 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
226 # Hit end of list, no match
231 if ( $md5 ne $key ) {
236 # Found match -- seek to offset and read signature
239 seek($fh, $subloc + $self->_root->{file_offset}, SEEK_SET);
240 read( $fh, $signature, SIG_SIZE);
243 # If value is a hash or array, return new DBM::Deep object with correct offset
245 if (($signature eq TYPE_HASH) || ($signature eq TYPE_ARRAY)) {
246 my $obj = DBM::Deep->new(
248 base_offset => $subloc,
252 if ($self->_root->{autobless}) {
254 # Skip over value and plain key to see if object needs
257 seek($fh, $DATA_LENGTH_SIZE + $INDEX_SIZE, SEEK_CUR);
260 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
261 if ($size) { seek($fh, $size, SEEK_CUR); }
264 read( $fh, $bless_bit, 1);
265 if (ord($bless_bit)) {
267 # Yes, object needs to be re-blessed
270 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
271 if ($size) { read( $fh, $class_name, $size); }
272 if ($class_name) { $obj = bless( $obj, $class_name ); }
280 # Otherwise return actual value
282 elsif ($signature eq SIG_DATA) {
285 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
286 if ($size) { read( $fh, $value, $size); }
291 # Key exists, but content is null
301 # Delete single key/value pair given tag and MD5 digested key.
304 my ($tag, $md5) = @_;
305 my $keys = $tag->{content};
310 # Iterate through buckets, looking for a key match
313 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
314 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
315 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
319 # Hit end of list, no match
324 if ( $md5 ne $key ) {
329 # Matched key -- delete bucket and return
331 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $self->_root->{file_offset}, SEEK_SET);
332 print( $fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
333 print( $fh chr(0) x $BUCKET_SIZE );
343 # Check existence of single key given tag and MD5 digested key.
346 my ($tag, $md5) = @_;
347 my $keys = $tag->{content};
350 # Iterate through buckets, looking for a key match
353 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
354 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
355 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
359 # Hit end of list, no match
364 if ( $md5 ne $key ) {
369 # Matched key -- return true
377 sub _find_bucket_list {
379 # Locate offset for bucket list, given digested key
385 # Locate offset for bucket list using digest index system
388 my $tag = $self->{engine}->load_tag($self, $self->_base_offset);
389 if (!$tag) { return; }
391 while ($tag->{signature} ne SIG_BLIST) {
392 $tag = $self->{engine}->index_lookup($self, $tag, ord(substr($md5, $ch, 1)));
393 if (!$tag) { return; }
400 sub _traverse_index {
402 # Scan index and recursively step into deeper levels, looking for next key.
404 my ($self, $offset, $ch, $force_return_next) = @_;
405 $force_return_next = undef unless $force_return_next;
407 my $tag = $self->{engine}->load_tag($self, $offset );
411 if ($tag->{signature} ne SIG_BLIST) {
412 my $content = $tag->{content};
414 if ($self->{return_next}) { $start = 0; }
415 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
417 for (my $index = $start; $index < 256; $index++) {
418 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
420 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
421 if (defined($result)) { return $result; }
425 $self->{return_next} = 1;
428 elsif ($tag->{signature} eq SIG_BLIST) {
429 my $keys = $tag->{content};
430 if ($force_return_next) { $self->{return_next} = 1; }
433 # Iterate through buckets, looking for a key match
435 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
436 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
437 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
441 # End of bucket list -- return to outer loop
443 $self->{return_next} = 1;
446 elsif ($key eq $self->{prev_md5}) {
448 # Located previous key -- return next one found
450 $self->{return_next} = 1;
453 elsif ($self->{return_next}) {
455 # Seek to bucket location and skip over signature
457 seek($fh, $subloc + SIG_SIZE + $self->_root->{file_offset}, SEEK_SET);
460 # Skip over value to get to plain key
463 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
464 if ($size) { seek($fh, $size, SEEK_CUR); }
467 # Read in plain key and return as scalar
470 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
471 if ($size) { read( $fh, $plain_key, $size); }
477 $self->{return_next} = 1;
478 } # tag is a bucket list
485 # Locate next key, given digested previous one
487 my $self = $_[0]->_get_self;
489 $self->{prev_md5} = $_[1] ? $_[1] : undef;
490 $self->{return_next} = 0;
493 # If the previous key was not specifed, start at the top and
494 # return the first one found.
496 if (!$self->{prev_md5}) {
497 $self->{prev_md5} = chr(0) x $HASH_SIZE;
498 $self->{return_next} = 1;
501 return $self->_traverse_index( $self->_base_offset, 0 );
506 # If db locking is set, flock() the db file. If called multiple
507 # times before unlock(), then the same number of unlocks() must
508 # be called before the lock is released.
510 my $self = $_[0]->_get_self;
512 $type = LOCK_EX unless defined $type;
514 if (!defined($self->_fh)) { return; }
516 if ($self->_root->{locking}) {
517 if (!$self->_root->{locked}) {
518 flock($self->_fh, $type);
520 # refresh end counter in case file has changed size
521 my @stats = stat($self->_root->{file});
522 $self->_root->{end} = $stats[7];
524 # double-check file inode, in case another process
525 # has optimize()d our file while we were waiting.
526 if ($stats[1] != $self->_root->{inode}) {
527 $self->{engine}->open( $self ); # re-open
528 flock($self->_fh, $type); # re-lock
529 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
532 $self->_root->{locked}++;
542 # If db locking is set, unlock the db file. See note in lock()
543 # regarding calling lock() multiple times.
545 my $self = $_[0]->_get_self;
547 if (!defined($self->_fh)) { return; }
549 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
550 $self->_root->{locked}--;
551 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
560 my $self = shift->_get_self;
561 my ($spot, $value) = @_;
566 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
567 my $type = $value->_type;
568 ${$spot} = $type eq TYPE_HASH ? {} : [];
569 $value->_copy_node( ${$spot} );
572 my $r = Scalar::Util::reftype( $value );
573 my $c = Scalar::Util::blessed( $value );
574 if ( $r eq 'ARRAY' ) {
575 ${$spot} = [ @{$value} ];
578 ${$spot} = { %{$value} };
580 ${$spot} = bless ${$spot}, $c
589 # Copy single level of keys or elements to new DB handle.
590 # Recurse for nested structures
592 my $self = shift->_get_self;
595 if ($self->_type eq TYPE_HASH) {
596 my $key = $self->first_key();
598 my $value = $self->get($key);
599 $self->_copy_value( \$db_temp->{$key}, $value );
600 $key = $self->next_key($key);
604 my $length = $self->length();
605 for (my $index = 0; $index < $length; $index++) {
606 my $value = $self->get($index);
607 $self->_copy_value( \$db_temp->[$index], $value );
616 # Recursively export into standard Perl hashes and arrays.
618 my $self = $_[0]->_get_self;
621 if ($self->_type eq TYPE_HASH) { $temp = {}; }
622 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
625 $self->_copy_node( $temp );
633 # Recursively import Perl hash/array structure
635 #XXX This use of ref() seems to be ok
636 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
638 my $self = $_[0]->_get_self;
641 #XXX This use of ref() seems to be ok
644 # struct is not a reference, so just import based on our type
648 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
649 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
652 my $r = Scalar::Util::reftype($struct) || '';
653 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
654 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
656 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
657 $self->push( @$struct );
660 return $self->_throw_error("Cannot import: type mismatch");
668 # Rebuild entire database into new file, then move
669 # it back on top of original.
671 my $self = $_[0]->_get_self;
673 #XXX Need to create a new test for this
674 # if ($self->_root->{links} > 1) {
675 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
678 my $db_temp = DBM::Deep->new(
679 file => $self->_root->{file} . '.tmp',
683 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
687 $self->_copy_node( $db_temp );
691 # Attempt to copy user, group and permissions over to new file
693 my @stats = stat($self->_fh);
694 my $perms = $stats[2] & 07777;
697 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
698 chmod( $perms, $self->_root->{file} . '.tmp' );
700 # q.v. perlport for more information on this variable
701 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
703 # Potential race condition when optmizing on Win32 with locking.
704 # The Windows filesystem requires that the filehandle be closed
705 # before it is overwritten with rename(). This could be redone
709 $self->{engine}->close( $self );
712 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
713 unlink $self->_root->{file} . '.tmp';
715 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
719 $self->{engine}->close( $self );
720 $self->{engine}->open( $self );
727 # Make copy of object and return
729 my $self = $_[0]->_get_self;
731 return DBM::Deep->new(
732 type => $self->_type,
733 base_offset => $self->_base_offset,
739 my %is_legal_filter = map {
742 store_key store_value
743 fetch_key fetch_value
748 # Setup filter function for storing or fetching the key or value
750 my $self = $_[0]->_get_self;
752 my $func = $_[2] ? $_[2] : undef;
754 if ( $is_legal_filter{$type} ) {
755 $self->_root->{"filter_$type"} = $func;
769 # Get access to the root structure
771 my $self = $_[0]->_get_self;
772 return $self->{root};
777 # Get access to the raw fh
779 #XXX It will be useful, though, when we split out HASH and ARRAY
780 my $self = $_[0]->_get_self;
781 return $self->_root->{fh};
786 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
788 my $self = $_[0]->_get_self;
789 return $self->{type};
794 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
796 my $self = $_[0]->_get_self;
797 return $self->{base_offset};
805 die "DBM::Deep: $_[1]\n";
810 # Precalculate index, bucket and bucket list sizes
813 #XXX I don't like this ...
814 set_pack() unless defined $LONG_SIZE;
816 $INDEX_SIZE = 256 * $LONG_SIZE;
817 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
818 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
823 # Set pack/unpack modes (see file header for more)
825 my ($long_s, $long_p, $data_s, $data_p) = @_;
827 $LONG_SIZE = $long_s ? $long_s : 4;
828 $LONG_PACK = $long_p ? $long_p : 'N';
830 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
831 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
838 # Set key digest function (default is MD5)
840 my ($digest_func, $hash_size) = @_;
842 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
843 $HASH_SIZE = $hash_size ? $hash_size : 16;
850 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
855 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
859 # tie() methods (hashes and arrays)
864 # Store single hash key/value or array element in database.
866 my $self = $_[0]->_get_self;
869 # User may be storing a hash, in which case we do not want it run
870 # through the filtering system
871 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
872 ? $self->_root->{filter_store_value}->($_[2])
875 my $md5 = $DIGEST_FUNC->($key);
877 unless ( _is_writable( $self->_fh ) ) {
878 $self->_throw_error( 'Cannot write to a readonly filehandle' );
882 # Request exclusive lock for writing
884 $self->lock( LOCK_EX );
889 # Locate offset for bucket list using digest index system
891 my $tag = $self->{engine}->load_tag($self, $self->_base_offset);
893 $tag = $self->{engine}->create_tag($self, $self->_base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
897 while ($tag->{signature} ne SIG_BLIST) {
898 my $num = ord(substr($md5, $ch, 1));
900 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
901 my $new_tag = $self->{engine}->index_lookup($self, $tag, $num);
904 seek($fh, $ref_loc + $self->_root->{file_offset}, SEEK_SET);
905 print( $fh pack($LONG_PACK, $self->_root->{end}) );
907 $tag = $self->{engine}->create_tag($self, $self->_root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
909 $tag->{ref_loc} = $ref_loc;
917 $tag->{ref_loc} = $ref_loc;
924 # Add key/value to bucket list
926 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
935 # Fetch single value or element given plain key or array index
937 my $self = shift->_get_self;
940 my $md5 = $DIGEST_FUNC->($key);
943 # Request shared lock for reading
945 $self->lock( LOCK_SH );
947 my $tag = $self->_find_bucket_list( $md5 );
954 # Get value from bucket list
956 my $result = $self->_get_bucket_value( $tag, $md5 );
960 #XXX What is ref() checking here?
961 #YYY Filters only apply on scalar values, so the ref check is making
962 #YYY sure the fetched bucket is a scalar, not a child hash or array.
963 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
964 ? $self->_root->{filter_fetch_value}->($result)
970 # Delete single key/value pair or element given plain key or array index
972 my $self = $_[0]->_get_self;
975 my $md5 = $DIGEST_FUNC->($key);
978 # Request exclusive lock for writing
980 $self->lock( LOCK_EX );
982 my $tag = $self->_find_bucket_list( $md5 );
991 my $value = $self->_get_bucket_value( $tag, $md5 );
992 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
993 $value = $self->_root->{filter_fetch_value}->($value);
996 my $result = $self->_delete_bucket( $tag, $md5 );
999 # If this object is an array and the key deleted was on the end of the stack,
1000 # decrement the length variable.
1010 # Check if a single key or element exists given plain key or array index
1012 my $self = $_[0]->_get_self;
1015 my $md5 = $DIGEST_FUNC->($key);
1018 # Request shared lock for reading
1020 $self->lock( LOCK_SH );
1022 my $tag = $self->_find_bucket_list( $md5 );
1025 # For some reason, the built-in exists() function returns '' for false
1033 # Check if bucket exists and return 1 or ''
1035 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
1044 # Clear all keys from hash, or all elements from array.
1046 my $self = $_[0]->_get_self;
1049 # Request exclusive lock for writing
1051 $self->lock( LOCK_EX );
1053 my $fh = $self->_fh;
1055 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
1061 $self->{engine}->create_tag($self, $self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
1069 # Public method aliases
1071 sub put { (shift)->STORE( @_ ) }
1072 sub store { (shift)->STORE( @_ ) }
1073 sub get { (shift)->FETCH( @_ ) }
1074 sub fetch { (shift)->FETCH( @_ ) }
1075 sub delete { (shift)->DELETE( @_ ) }
1076 sub exists { (shift)->EXISTS( @_ ) }
1077 sub clear { (shift)->CLEAR( @_ ) }
1079 package DBM::Deep::_::Root;
1093 filter_store_key => undef,
1094 filter_store_value => undef,
1095 filter_fetch_key => undef,
1096 filter_fetch_value => undef,
1102 if ( $self->{fh} && !$self->{file_offset} ) {
1103 $self->{file_offset} = tell( $self->{fh} );
1111 return unless $self;
1113 close $self->{fh} if $self->{fh};
1124 DBM::Deep - A pure perl multi-level hash/array DBM
1129 my $db = DBM::Deep->new( "foo.db" );
1131 $db->{key} = 'value'; # tie() style
1134 $db->put('key' => 'value'); # OO style
1135 print $db->get('key');
1137 # true multi-level support
1138 $db->{my_complex} = [
1139 'hello', { perl => 'rules' },
1145 A unique flat-file database module, written in pure perl. True
1146 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1147 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1148 handle millions of keys and unlimited hash levels without significant
1149 slow-down. Written from the ground-up in pure perl -- this is NOT a
1150 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1151 Mac OS X and Windows.
1155 Hopefully you are using Perl's excellent CPAN module, which will download
1156 and install the module for you. If not, get the tarball, and run these
1168 Construction can be done OO-style (which is the recommended way), or using
1169 Perl's tie() function. Both are examined here.
1171 =head2 OO CONSTRUCTION
1173 The recommended way to construct a DBM::Deep object is to use the new()
1174 method, which gets you a blessed, tied hash or array reference.
1176 my $db = DBM::Deep->new( "foo.db" );
1178 This opens a new database handle, mapped to the file "foo.db". If this
1179 file does not exist, it will automatically be created. DB files are
1180 opened in "r+" (read/write) mode, and the type of object returned is a
1181 hash, unless otherwise specified (see L<OPTIONS> below).
1183 You can pass a number of options to the constructor to specify things like
1184 locking, autoflush, etc. This is done by passing an inline hash:
1186 my $db = DBM::Deep->new(
1192 Notice that the filename is now specified I<inside> the hash with
1193 the "file" parameter, as opposed to being the sole argument to the
1194 constructor. This is required if any options are specified.
1195 See L<OPTIONS> below for the complete list.
1199 You can also start with an array instead of a hash. For this, you must
1200 specify the C<type> parameter:
1202 my $db = DBM::Deep->new(
1204 type => DBM::Deep->TYPE_ARRAY
1207 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1208 a new DB file. If you create a DBM::Deep object with an existing file, the
1209 C<type> will be loaded from the file header, and an error will be thrown if
1210 the wrong type is passed in.
1212 =head2 TIE CONSTRUCTION
1214 Alternately, you can create a DBM::Deep handle by using Perl's built-in
1215 tie() function. The object returned from tie() can be used to call methods,
1216 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
1217 file (as expected with most tie'd objects).
1220 my $db = tie %hash, "DBM::Deep", "foo.db";
1223 my $db = tie @array, "DBM::Deep", "bar.db";
1225 As with the OO constructor, you can replace the DB filename parameter with
1226 a hash containing one or more options (see L<OPTIONS> just below for the
1229 tie %hash, "DBM::Deep", {
1237 There are a number of options that can be passed in when constructing your
1238 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1244 Filename of the DB file to link the handle to. You can pass a full absolute
1245 filesystem path, partial path, or a plain filename if the file is in the
1246 current working directory. This is a required parameter (though q.v. fh).
1250 If you want, you can pass in the fh instead of the file. This is most useful for doing
1253 my $db = DBM::Deep->new( { fh => \*DATA } );
1255 You are responsible for making sure that the fh has been opened appropriately for your
1256 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
1257 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
1258 needs to read from the fh.
1262 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
1263 not need to set this. However, it's there if you want it.
1265 If you pass in fh and do not set this, it will be set appropriately.
1269 This parameter specifies what type of object to create, a hash or array. Use
1270 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1271 This only takes effect when beginning a new file. This is an optional
1272 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1276 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1277 function to lock the database in exclusive mode for writes, and shared mode for
1278 reads. Pass any true value to enable. This affects the base DB handle I<and
1279 any child hashes or arrays> that use the same DB file. This is an optional
1280 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1284 Specifies whether autoflush is to be enabled on the underlying filehandle.
1285 This obviously slows down write operations, but is required if you may have
1286 multiple processes accessing the same DB file (also consider enable I<locking>).
1287 Pass any true value to enable. This is an optional parameter, and defaults to 0
1292 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1293 restore them when fetched. This is an B<experimental> feature, and does have
1294 side-effects. Basically, when hashes are re-blessed into their original
1295 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1296 able to call any DBM::Deep methods on them. You have been warned.
1297 This is an optional parameter, and defaults to 0 (disabled).
1301 See L<FILTERS> below.
1305 Setting I<debug> mode will make all errors non-fatal, dump them out to
1306 STDERR, and continue on. This is for debugging purposes only, and probably
1307 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1309 B<NOTE>: This parameter is considered deprecated and should not be used anymore.
1313 =head1 TIE INTERFACE
1315 With DBM::Deep you can access your databases using Perl's standard hash/array
1316 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
1317 treat them as such. DBM::Deep will intercept all reads/writes and direct them
1318 to the right place -- the DB file. This has nothing to do with the
1319 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
1320 using regular hashes and arrays, rather than calling functions like C<get()>
1321 and C<put()> (although those work too). It is entirely up to you how to want
1322 to access your databases.
1326 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1327 or even nested hashes (or arrays) using standard Perl syntax:
1329 my $db = DBM::Deep->new( "foo.db" );
1331 $db->{mykey} = "myvalue";
1333 $db->{myhash}->{subkey} = "subvalue";
1335 print $db->{myhash}->{subkey} . "\n";
1337 You can even step through hash keys using the normal Perl C<keys()> function:
1339 foreach my $key (keys %$db) {
1340 print "$key: " . $db->{$key} . "\n";
1343 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1344 pushes them onto an array, all before the loop even begins. If you have an
1345 extra large hash, this may exhaust Perl's memory. Instead, consider using
1346 Perl's C<each()> function, which pulls keys/values one at a time, using very
1349 while (my ($key, $value) = each %$db) {
1350 print "$key: $value\n";
1353 Please note that when using C<each()>, you should always pass a direct
1354 hash reference, not a lookup. Meaning, you should B<never> do this:
1357 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1359 This causes an infinite loop, because for each iteration, Perl is calling
1360 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1361 it effectively keeps returning the first key over and over again. Instead,
1362 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1366 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1367 reference. This includes inserting, removing and manipulating elements,
1368 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1369 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1370 or simply be a nested array reference inside a hash. Example:
1372 my $db = DBM::Deep->new(
1373 file => "foo-array.db",
1374 type => DBM::Deep->TYPE_ARRAY
1378 push @$db, "bar", "baz";
1379 unshift @$db, "bah";
1381 my $last_elem = pop @$db; # baz
1382 my $first_elem = shift @$db; # bah
1383 my $second_elem = $db->[1]; # bar
1385 my $num_elements = scalar @$db;
1389 In addition to the I<tie()> interface, you can also use a standard OO interface
1390 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1391 array) has its own methods, but both types share the following common methods:
1392 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1396 =item * new() / clone()
1398 These are the constructor and copy-functions.
1400 =item * put() / store()
1402 Stores a new hash key/value pair, or sets an array element value. Takes two
1403 arguments, the hash key or array index, and the new value. The value can be
1404 a scalar, hash ref or array ref. Returns true on success, false on failure.
1406 $db->put("foo", "bar"); # for hashes
1407 $db->put(1, "bar"); # for arrays
1409 =item * get() / fetch()
1411 Fetches the value of a hash key or array element. Takes one argument: the hash
1412 key or array index. Returns a scalar, hash ref or array ref, depending on the
1415 my $value = $db->get("foo"); # for hashes
1416 my $value = $db->get(1); # for arrays
1420 Checks if a hash key or array index exists. Takes one argument: the hash key
1421 or array index. Returns true if it exists, false if not.
1423 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1424 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1428 Deletes one hash key/value pair or array element. Takes one argument: the hash
1429 key or array index. Returns true on success, false if not found. For arrays,
1430 the remaining elements located after the deleted element are NOT moved over.
1431 The deleted element is essentially just undefined, which is exactly how Perl's
1432 internal arrays work. Please note that the space occupied by the deleted
1433 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1434 below for details and workarounds.
1436 $db->delete("foo"); # for hashes
1437 $db->delete(1); # for arrays
1441 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1442 value. Please note that the space occupied by the deleted keys/values or
1443 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1444 details and workarounds.
1446 $db->clear(); # hashes or arrays
1448 =item * lock() / unlock()
1454 Recover lost disk space.
1456 =item * import() / export()
1458 Data going in and out.
1460 =item * set_digest() / set_pack() / set_filter()
1462 q.v. adjusting the interal parameters.
1468 For hashes, DBM::Deep supports all the common methods described above, and the
1469 following additional methods: C<first_key()> and C<next_key()>.
1475 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1476 fetched in an undefined order (which appears random). Takes no arguments,
1477 returns the key as a scalar value.
1479 my $key = $db->first_key();
1483 Returns the "next" key in the hash, given the previous one as the sole argument.
1484 Returns undef if there are no more keys to be fetched.
1486 $key = $db->next_key($key);
1490 Here are some examples of using hashes:
1492 my $db = DBM::Deep->new( "foo.db" );
1494 $db->put("foo", "bar");
1495 print "foo: " . $db->get("foo") . "\n";
1497 $db->put("baz", {}); # new child hash ref
1498 $db->get("baz")->put("buz", "biz");
1499 print "buz: " . $db->get("baz")->get("buz") . "\n";
1501 my $key = $db->first_key();
1503 print "$key: " . $db->get($key) . "\n";
1504 $key = $db->next_key($key);
1507 if ($db->exists("foo")) { $db->delete("foo"); }
1511 For arrays, DBM::Deep supports all the common methods described above, and the
1512 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1513 C<unshift()> and C<splice()>.
1519 Returns the number of elements in the array. Takes no arguments.
1521 my $len = $db->length();
1525 Adds one or more elements onto the end of the array. Accepts scalars, hash
1526 refs or array refs. No return value.
1528 $db->push("foo", "bar", {});
1532 Fetches the last element in the array, and deletes it. Takes no arguments.
1533 Returns undef if array is empty. Returns the element value.
1535 my $elem = $db->pop();
1539 Fetches the first element in the array, deletes it, then shifts all the
1540 remaining elements over to take up the space. Returns the element value. This
1541 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1544 my $elem = $db->shift();
1548 Inserts one or more elements onto the beginning of the array, shifting all
1549 existing elements over to make room. Accepts scalars, hash refs or array refs.
1550 No return value. This method is not recommended with large arrays -- see
1551 <LARGE ARRAYS> below for details.
1553 $db->unshift("foo", "bar", {});
1557 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1558 -f splice> for usage -- it is too complicated to document here. This method is
1559 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1563 Here are some examples of using arrays:
1565 my $db = DBM::Deep->new(
1567 type => DBM::Deep->TYPE_ARRAY
1570 $db->push("bar", "baz");
1571 $db->unshift("foo");
1574 my $len = $db->length();
1575 print "length: $len\n"; # 4
1577 for (my $k=0; $k<$len; $k++) {
1578 print "$k: " . $db->get($k) . "\n";
1581 $db->splice(1, 2, "biz", "baf");
1583 while (my $elem = shift @$db) {
1584 print "shifted: $elem\n";
1589 Enable automatic file locking by passing a true value to the C<locking>
1590 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1592 my $db = DBM::Deep->new(
1597 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1598 mode for writes, and shared mode for reads. This is required if you have
1599 multiple processes accessing the same database file, to avoid file corruption.
1600 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1601 NFS> below for more.
1603 =head2 EXPLICIT LOCKING
1605 You can explicitly lock a database, so it remains locked for multiple
1606 transactions. This is done by calling the C<lock()> method, and passing an
1607 optional lock mode argument (defaults to exclusive mode). This is particularly
1608 useful for things like counters, where the current value needs to be fetched,
1609 then incremented, then stored again.
1612 my $counter = $db->get("counter");
1614 $db->put("counter", $counter);
1623 You can pass C<lock()> an optional argument, which specifies which mode to use
1624 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1625 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1626 same as the constants defined in Perl's C<Fcntl> module.
1628 $db->lock( DBM::Deep->LOCK_SH );
1632 =head1 IMPORTING/EXPORTING
1634 You can import existing complex structures by calling the C<import()> method,
1635 and export an entire database into an in-memory structure using the C<export()>
1636 method. Both are examined here.
1640 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1641 walking the structure and adding keys/elements to the database as you go,
1642 simply pass a reference to the C<import()> method. This recursively adds
1643 everything to an existing DBM::Deep object for you. Here is an example:
1648 array1 => [ "elem0", "elem1", "elem2" ],
1650 subkey1 => "subvalue1",
1651 subkey2 => "subvalue2"
1655 my $db = DBM::Deep->new( "foo.db" );
1656 $db->import( $struct );
1658 print $db->{key1} . "\n"; # prints "value1"
1660 This recursively imports the entire C<$struct> object into C<$db>, including
1661 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1662 keys are merged with the existing ones, replacing if they already exist.
1663 The C<import()> method can be called on any database level (not just the base
1664 level), and works with both hash and array DB types.
1666 B<Note:> Make sure your existing structure has no circular references in it.
1667 These will cause an infinite loop when importing.
1671 Calling the C<export()> method on an existing DBM::Deep object will return
1672 a reference to a new in-memory copy of the database. The export is done
1673 recursively, so all nested hashes/arrays are all exported to standard Perl
1674 objects. Here is an example:
1676 my $db = DBM::Deep->new( "foo.db" );
1678 $db->{key1} = "value1";
1679 $db->{key2} = "value2";
1681 $db->{hash1}->{subkey1} = "subvalue1";
1682 $db->{hash1}->{subkey2} = "subvalue2";
1684 my $struct = $db->export();
1686 print $struct->{key1} . "\n"; # prints "value1"
1688 This makes a complete copy of the database in memory, and returns a reference
1689 to it. The C<export()> method can be called on any database level (not just
1690 the base level), and works with both hash and array DB types. Be careful of
1691 large databases -- you can store a lot more data in a DBM::Deep object than an
1692 in-memory Perl structure.
1694 B<Note:> Make sure your database has no circular references in it.
1695 These will cause an infinite loop when exporting.
1699 DBM::Deep has a number of hooks where you can specify your own Perl function
1700 to perform filtering on incoming or outgoing data. This is a perfect
1701 way to extend the engine, and implement things like real-time compression or
1702 encryption. Filtering applies to the base DB level, and all child hashes /
1703 arrays. Filter hooks can be specified when your DBM::Deep object is first
1704 constructed, or by calling the C<set_filter()> method at any time. There are
1705 four available filter hooks, described below:
1709 =item * filter_store_key
1711 This filter is called whenever a hash key is stored. It
1712 is passed the incoming key, and expected to return a transformed key.
1714 =item * filter_store_value
1716 This filter is called whenever a hash key or array element is stored. It
1717 is passed the incoming value, and expected to return a transformed value.
1719 =item * filter_fetch_key
1721 This filter is called whenever a hash key is fetched (i.e. via
1722 C<first_key()> or C<next_key()>). It is passed the transformed key,
1723 and expected to return the plain key.
1725 =item * filter_fetch_value
1727 This filter is called whenever a hash key or array element is fetched.
1728 It is passed the transformed value, and expected to return the plain value.
1732 Here are the two ways to setup a filter hook:
1734 my $db = DBM::Deep->new(
1736 filter_store_value => \&my_filter_store,
1737 filter_fetch_value => \&my_filter_fetch
1742 $db->set_filter( "filter_store_value", \&my_filter_store );
1743 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1745 Your filter function will be called only when dealing with SCALAR keys or
1746 values. When nested hashes and arrays are being stored/fetched, filtering
1747 is bypassed. Filters are called as static functions, passed a single SCALAR
1748 argument, and expected to return a single SCALAR value. If you want to
1749 remove a filter, set the function reference to C<undef>:
1751 $db->set_filter( "filter_store_value", undef );
1753 =head2 REAL-TIME ENCRYPTION EXAMPLE
1755 Here is a working example that uses the I<Crypt::Blowfish> module to
1756 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1757 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1758 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1761 use Crypt::Blowfish;
1764 my $cipher = Crypt::CBC->new({
1765 'key' => 'my secret key',
1766 'cipher' => 'Blowfish',
1768 'regenerate_key' => 0,
1769 'padding' => 'space',
1773 my $db = DBM::Deep->new(
1774 file => "foo-encrypt.db",
1775 filter_store_key => \&my_encrypt,
1776 filter_store_value => \&my_encrypt,
1777 filter_fetch_key => \&my_decrypt,
1778 filter_fetch_value => \&my_decrypt,
1781 $db->{key1} = "value1";
1782 $db->{key2} = "value2";
1783 print "key1: " . $db->{key1} . "\n";
1784 print "key2: " . $db->{key2} . "\n";
1790 return $cipher->encrypt( $_[0] );
1793 return $cipher->decrypt( $_[0] );
1796 =head2 REAL-TIME COMPRESSION EXAMPLE
1798 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1799 compression / decompression of keys & values with DBM::Deep Filters.
1800 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1801 more on I<Compress::Zlib>.
1806 my $db = DBM::Deep->new(
1807 file => "foo-compress.db",
1808 filter_store_key => \&my_compress,
1809 filter_store_value => \&my_compress,
1810 filter_fetch_key => \&my_decompress,
1811 filter_fetch_value => \&my_decompress,
1814 $db->{key1} = "value1";
1815 $db->{key2} = "value2";
1816 print "key1: " . $db->{key1} . "\n";
1817 print "key2: " . $db->{key2} . "\n";
1823 return Compress::Zlib::memGzip( $_[0] ) ;
1826 return Compress::Zlib::memGunzip( $_[0] ) ;
1829 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1830 actually numerical index numbers, and are not filtered.
1832 =head1 ERROR HANDLING
1834 Most DBM::Deep methods return a true value for success, and call die() on
1835 failure. You can wrap calls in an eval block to catch the die.
1837 my $db = DBM::Deep->new( "foo.db" ); # create hash
1838 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1840 print $@; # prints error message
1842 =head1 LARGEFILE SUPPORT
1844 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1845 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1846 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1847 by calling the static C<set_pack()> method before you do anything else.
1849 DBM::Deep::set_pack(8, 'Q');
1851 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1852 instead of 32-bit longs. After setting these values your DB files have a
1853 theoretical maximum size of 16 XB (exabytes).
1855 B<Note:> Changing these values will B<NOT> work for existing database files.
1856 Only change this for new files, and make sure it stays set consistently
1857 throughout the file's life. If you do set these values, you can no longer
1858 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1859 back to 32-bit mode.
1861 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1862 only a 32-bit Perl. However, I have received user reports that this does
1865 =head1 LOW-LEVEL ACCESS
1867 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1868 you can call the C<_fh()> method, which returns the handle:
1870 my $fh = $db->_fh();
1872 This method can be called on the root level of the datbase, or any child
1873 hashes or arrays. All levels share a I<root> structure, which contains things
1874 like the filehandle, a reference counter, and all the options specified
1875 when you created the object. You can get access to this root structure by
1876 calling the C<root()> method.
1878 my $root = $db->_root();
1880 This is useful for changing options after the object has already been created,
1881 such as enabling/disabling locking, or debug modes. You can also
1882 store your own temporary user data in this structure (be wary of name
1883 collision), which is then accessible from any child hash or array.
1885 =head1 CUSTOM DIGEST ALGORITHM
1887 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1888 keys. However you can override this, and use another algorithm (such as SHA-256)
1889 or even write your own. But please note that DBM::Deep currently expects zero
1890 collisions, so your algorithm has to be I<perfect>, so to speak.
1891 Collision detection may be introduced in a later version.
1895 You can specify a custom digest algorithm by calling the static C<set_digest()>
1896 function, passing a reference to a subroutine, and the length of the algorithm's
1897 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1898 objects. Here is a working example that uses a 256-bit hash from the
1899 I<Digest::SHA256> module. Please see
1900 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1905 my $context = Digest::SHA256::new(256);
1907 DBM::Deep::set_digest( \&my_digest, 32 );
1909 my $db = DBM::Deep->new( "foo-sha.db" );
1911 $db->{key1} = "value1";
1912 $db->{key2} = "value2";
1913 print "key1: " . $db->{key1} . "\n";
1914 print "key2: " . $db->{key2} . "\n";
1920 return substr( $context->hash($_[0]), 0, 32 );
1923 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1924 of bytes you specify in the C<set_digest()> function (in this case 32).
1926 =head1 CIRCULAR REFERENCES
1928 DBM::Deep has B<experimental> support for circular references. Meaning you
1929 can have a nested hash key or array element that points to a parent object.
1930 This relationship is stored in the DB file, and is preserved between sessions.
1933 my $db = DBM::Deep->new( "foo.db" );
1936 $db->{circle} = $db; # ref to self
1938 print $db->{foo} . "\n"; # prints "foo"
1939 print $db->{circle}->{foo} . "\n"; # prints "foo" again
1941 One catch is, passing the object to a function that recursively walks the
1942 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1943 C<export()> methods) will result in an infinite loop. The other catch is,
1944 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
1945 or C<next_key()> methods), you will get the I<target object's key>, not the
1946 ref's key. This gets even more interesting with the above example, where
1947 the I<circle> key points to the base DB object, which technically doesn't
1948 have a key. So I made DBM::Deep return "[base]" as the key name in that
1951 =head1 CAVEATS / ISSUES / BUGS
1953 This section describes all the known issues with DBM::Deep. It you have found
1954 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1956 =head2 UNUSED SPACE RECOVERY
1958 One major caveat with DBM::Deep is that space occupied by existing keys and
1959 values is not recovered when they are deleted. Meaning if you keep deleting
1960 and adding new keys, your file will continuously grow. I am working on this,
1961 but in the meantime you can call the built-in C<optimize()> method from time to
1962 time (perhaps in a crontab or something) to recover all your unused space.
1964 $db->optimize(); # returns true on success
1966 This rebuilds the ENTIRE database into a new file, then moves it on top of
1967 the original. The new file will have no unused space, thus it will take up as
1968 little disk space as possible. Please note that this operation can take
1969 a long time for large files, and you need enough disk space to temporarily hold
1970 2 copies of your DB file. The temporary file is created in the same directory
1971 as the original, named with a ".tmp" extension, and is deleted when the
1972 operation completes. Oh, and if locking is enabled, the DB is automatically
1973 locked for the entire duration of the copy.
1975 B<WARNING:> Only call optimize() on the top-level node of the database, and
1976 make sure there are no child references lying around. DBM::Deep keeps a reference
1977 counter, and if it is greater than 1, optimize() will abort and return undef.
1979 =head2 AUTOVIVIFICATION
1981 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1982 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1983 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1984 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1985 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1988 $db->{foo}->{bar} = "hello";
1990 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1991 being an empty hash. Try this instead, which works fine:
1993 $db->{foo} = { bar => "hello" };
1995 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1996 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1997 Probably a bug in Perl.
1999 =head2 FILE CORRUPTION
2001 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
2002 for a 32-bit signature when opened, but other corruption in files can cause
2003 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
2004 stuck in an infinite loop depending on the level of corruption. File write
2005 operations are not checked for failure (for speed), so if you happen to run
2006 out of disk space, DBM::Deep will probably fail in a bad way. These things will
2007 be addressed in a later version of DBM::Deep.
2011 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
2012 filesystems, but will NOT protect you from file corruption over NFS. I've heard
2013 about setting up your NFS server with a locking daemon, then using lockf() to
2014 lock your files, but your mileage may vary there as well. From what I
2015 understand, there is no real way to do it. However, if you need access to the
2016 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
2017 lockf(), see the L<LOW-LEVEL ACCESS> section above.
2019 =head2 COPYING OBJECTS
2021 Beware of copying tied objects in Perl. Very strange things can happen.
2022 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
2023 returns a new, blessed, tied hash or array to the same level in the DB.
2025 my $copy = $db->clone();
2027 B<Note>: Since clone() here is cloning the object, not the database location, any
2028 modifications to either $db or $copy will be visible in both.
2032 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
2033 These functions cause every element in the array to move, which can be murder
2034 on DBM::Deep, as every element has to be fetched from disk, then stored again in
2035 a different location. This will be addressed in the forthcoming version 1.00.
2037 =head2 WRITEONLY FILES
2039 If you pass in a filehandle to new(), you may have opened it in either a readonly or
2040 writeonly mode. STORE will verify that the filehandle is writable. However, there
2041 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
2042 filehandle isn't readable, it's not clear what will happen. So, don't do that.
2046 This section discusses DBM::Deep's speed and memory usage.
2050 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
2051 the almighty I<BerkeleyDB>. But it makes up for it in features like true
2052 multi-level hash/array support, and cross-platform FTPable files. Even so,
2053 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
2054 with huge databases. Here is some test data:
2056 Adding 1,000,000 keys to new DB file...
2058 At 100 keys, avg. speed is 2,703 keys/sec
2059 At 200 keys, avg. speed is 2,642 keys/sec
2060 At 300 keys, avg. speed is 2,598 keys/sec
2061 At 400 keys, avg. speed is 2,578 keys/sec
2062 At 500 keys, avg. speed is 2,722 keys/sec
2063 At 600 keys, avg. speed is 2,628 keys/sec
2064 At 700 keys, avg. speed is 2,700 keys/sec
2065 At 800 keys, avg. speed is 2,607 keys/sec
2066 At 900 keys, avg. speed is 2,190 keys/sec
2067 At 1,000 keys, avg. speed is 2,570 keys/sec
2068 At 2,000 keys, avg. speed is 2,417 keys/sec
2069 At 3,000 keys, avg. speed is 1,982 keys/sec
2070 At 4,000 keys, avg. speed is 1,568 keys/sec
2071 At 5,000 keys, avg. speed is 1,533 keys/sec
2072 At 6,000 keys, avg. speed is 1,787 keys/sec
2073 At 7,000 keys, avg. speed is 1,977 keys/sec
2074 At 8,000 keys, avg. speed is 2,028 keys/sec
2075 At 9,000 keys, avg. speed is 2,077 keys/sec
2076 At 10,000 keys, avg. speed is 2,031 keys/sec
2077 At 20,000 keys, avg. speed is 1,970 keys/sec
2078 At 30,000 keys, avg. speed is 2,050 keys/sec
2079 At 40,000 keys, avg. speed is 2,073 keys/sec
2080 At 50,000 keys, avg. speed is 1,973 keys/sec
2081 At 60,000 keys, avg. speed is 1,914 keys/sec
2082 At 70,000 keys, avg. speed is 2,091 keys/sec
2083 At 80,000 keys, avg. speed is 2,103 keys/sec
2084 At 90,000 keys, avg. speed is 1,886 keys/sec
2085 At 100,000 keys, avg. speed is 1,970 keys/sec
2086 At 200,000 keys, avg. speed is 2,053 keys/sec
2087 At 300,000 keys, avg. speed is 1,697 keys/sec
2088 At 400,000 keys, avg. speed is 1,838 keys/sec
2089 At 500,000 keys, avg. speed is 1,941 keys/sec
2090 At 600,000 keys, avg. speed is 1,930 keys/sec
2091 At 700,000 keys, avg. speed is 1,735 keys/sec
2092 At 800,000 keys, avg. speed is 1,795 keys/sec
2093 At 900,000 keys, avg. speed is 1,221 keys/sec
2094 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2096 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2097 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2098 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2099 Run time was 12 min 3 sec.
2103 One of the great things about DBM::Deep is that it uses very little memory.
2104 Even with huge databases (1,000,000+ keys) you will not see much increased
2105 memory on your process. DBM::Deep relies solely on the filesystem for storing
2106 and fetching data. Here is output from I</usr/bin/top> before even opening a
2109 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2110 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2112 Basically the process is taking 2,716K of memory. And here is the same
2113 process after storing and fetching 1,000,000 keys:
2115 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2116 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2118 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2119 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2121 =head1 DB FILE FORMAT
2123 In case you were interested in the underlying DB file format, it is documented
2124 here in this section. You don't need to know this to use the module, it's just
2125 included for reference.
2129 DBM::Deep files always start with a 32-bit signature to identify the file type.
2130 This is at offset 0. The signature is "DPDB" in network byte order. This is
2131 checked for when the file is opened and an error will be thrown if it's not found.
2135 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2136 has a standard header containing the type of data, the length of data, and then
2137 the data itself. The type is a single character (1 byte), the length is a
2138 32-bit unsigned long in network byte order, and the data is, well, the data.
2139 Here is how it unfolds:
2143 Immediately after the 32-bit file signature is the I<Master Index> record.
2144 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2145 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2146 depending on how the DBM::Deep object was constructed.
2148 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2149 number). The first 8-bit char of the MD5 signature is the offset into the
2150 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2151 index element is a file offset of the next tag for the key/element in question,
2152 which is usually a I<Bucket List> tag (see below).
2154 The next tag I<could> be another index, depending on how many keys/elements
2155 exist. See L<RE-INDEXING> below for details.
2159 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2160 file offsets to where the actual data is stored. It starts with a standard
2161 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2162 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2163 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2164 When the list fills up, a I<Re-Index> operation is performed (See
2165 L<RE-INDEXING> below).
2169 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2170 index/value pair (in array mode). It starts with a standard tag header with
2171 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2172 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2173 header. The size reported in the tag header is only for the value, but then,
2174 just after the value is another size (32-bit unsigned long) and then the plain
2175 key itself. Since the value is likely to be fetched more often than the plain
2176 key, I figured it would be I<slightly> faster to store the value first.
2178 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2179 record for the nested structure, where the process begins all over again.
2183 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2184 exhausted. Then, when another key/element comes in, the list is converted to a
2185 new index record. However, this index will look at the next char in the MD5
2186 hash, and arrange new Bucket List pointers accordingly. This process is called
2187 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2188 17 (16 + new one) keys/elements are removed from the old Bucket List and
2189 inserted into the new index. Several new Bucket Lists are created in the
2190 process, as a new MD5 char from the key is being examined (it is unlikely that
2191 the keys will all share the same next char of their MD5s).
2193 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2194 when the Bucket Lists will turn into indexes, but the first round tends to
2195 happen right around 4,000 keys. You will see a I<slight> decrease in
2196 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2197 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2198 right around 900,000 keys. This process can continue nearly indefinitely --
2199 right up until the point the I<MD5> signatures start colliding with each other,
2200 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2201 getting struck by lightning while you are walking to cash in your tickets.
2202 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2203 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2204 this is 340 unodecillion, but don't quote me).
2208 When a new key/element is stored, the key (or index number) is first run through
2209 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2210 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2211 for the first char of the signature (in this case I<b0>). If it does not exist,
2212 a new I<Bucket List> is created for our key (and the next 15 future keys that
2213 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2214 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2215 this point, unless we are replacing an existing I<Bucket>), where the actual
2216 data will be stored.
2220 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2221 (or index number), then walking along the indexes. If there are enough
2222 keys/elements in this DB level, there might be nested indexes, each linked to
2223 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2224 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2225 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2226 plain key are stored.
2228 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2229 methods. In this process the indexes are walked systematically, and each key
2230 fetched in increasing MD5 order (which is why it appears random). Once the
2231 I<Bucket> is found, the value is skipped and the plain key returned instead.
2232 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2233 alphabetically sorted. This only happens on an index-level -- as soon as the
2234 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2235 so it's pretty much undefined how the keys will come out -- just like Perl's
2238 =head1 CODE COVERAGE
2240 We use B<Devel::Cover> to test the code coverage of our tests, below is the
2241 B<Devel::Cover> report on this module's test suite.
2243 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2244 File stmt bran cond sub pod time total
2245 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2246 blib/lib/DBM/Deep.pm 95.2 83.8 70.0 98.2 100.0 58.0 91.0
2247 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 n/a 26.7 98.0
2248 blib/lib/DBM/Deep/Hash.pm 95.3 80.0 100.0 100.0 n/a 15.3 92.4
2249 Total 96.2 84.8 74.4 98.8 100.0 100.0 92.4
2250 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2252 =head1 MORE INFORMATION
2254 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
2255 or send email to L<DBM-Deep@googlegroups.com>.
2259 Joseph Huckaby, L<jhuckaby@cpan.org>
2261 Rob Kinyon, L<rkinyon@cpan.org>
2263 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2267 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2268 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2272 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
2273 This is free software, you may use it and distribute it under the
2274 same terms as Perl itself.