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( $self ); }
192 require DBM::Deep::Hash;
193 return DBM::Deep::Hash->TIEHASH( @_ );
198 require DBM::Deep::Array;
199 return DBM::Deep::Array->TIEARRAY( @_ );
202 #XXX Unneeded now ...
208 # Delete single key/value pair given tag and MD5 digested key.
211 my ($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 $key ) {
236 # Matched key -- delete bucket and return
238 seek($fh, $tag->{offset} + ($i * $BUCKET_SIZE) + $self->_root->{file_offset}, SEEK_SET);
239 print( $fh substr($keys, ($i+1) * $BUCKET_SIZE ) );
240 print( $fh chr(0) x $BUCKET_SIZE );
250 # Check existence of single key given tag and MD5 digested key.
253 my ($tag, $md5) = @_;
254 my $keys = $tag->{content};
257 # Iterate through buckets, looking for a key match
260 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
261 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
262 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
266 # Hit end of list, no match
271 if ( $md5 ne $key ) {
276 # Matched key -- return true
284 sub _find_bucket_list {
286 # Locate offset for bucket list, given digested key
292 # Locate offset for bucket list using digest index system
295 my $tag = $self->{engine}->load_tag($self, $self->_base_offset);
296 if (!$tag) { return; }
298 while ($tag->{signature} ne SIG_BLIST) {
299 $tag = $self->{engine}->index_lookup($self, $tag, ord(substr($md5, $ch, 1)));
300 if (!$tag) { return; }
307 sub _traverse_index {
309 # Scan index and recursively step into deeper levels, looking for next key.
311 my ($self, $offset, $ch, $force_return_next) = @_;
312 $force_return_next = undef unless $force_return_next;
314 my $tag = $self->{engine}->load_tag($self, $offset );
318 if ($tag->{signature} ne SIG_BLIST) {
319 my $content = $tag->{content};
321 if ($self->{return_next}) { $start = 0; }
322 else { $start = ord(substr($self->{prev_md5}, $ch, 1)); }
324 for (my $index = $start; $index < 256; $index++) {
325 my $subloc = unpack($LONG_PACK, substr($content, $index * $LONG_SIZE, $LONG_SIZE) );
327 my $result = $self->_traverse_index( $subloc, $ch + 1, $force_return_next );
328 if (defined($result)) { return $result; }
332 $self->{return_next} = 1;
335 elsif ($tag->{signature} eq SIG_BLIST) {
336 my $keys = $tag->{content};
337 if ($force_return_next) { $self->{return_next} = 1; }
340 # Iterate through buckets, looking for a key match
342 for (my $i=0; $i<$MAX_BUCKETS; $i++) {
343 my $key = substr($keys, $i * $BUCKET_SIZE, $HASH_SIZE);
344 my $subloc = unpack($LONG_PACK, substr($keys, ($i * $BUCKET_SIZE) + $HASH_SIZE, $LONG_SIZE));
348 # End of bucket list -- return to outer loop
350 $self->{return_next} = 1;
353 elsif ($key eq $self->{prev_md5}) {
355 # Located previous key -- return next one found
357 $self->{return_next} = 1;
360 elsif ($self->{return_next}) {
362 # Seek to bucket location and skip over signature
364 seek($fh, $subloc + SIG_SIZE + $self->_root->{file_offset}, SEEK_SET);
367 # Skip over value to get to plain key
370 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
371 if ($size) { seek($fh, $size, SEEK_CUR); }
374 # Read in plain key and return as scalar
377 read( $fh, $size, $DATA_LENGTH_SIZE); $size = unpack($DATA_LENGTH_PACK, $size);
378 if ($size) { read( $fh, $plain_key, $size); }
384 $self->{return_next} = 1;
385 } # tag is a bucket list
392 # Locate next key, given digested previous one
394 my $self = $_[0]->_get_self;
396 $self->{prev_md5} = $_[1] ? $_[1] : undef;
397 $self->{return_next} = 0;
400 # If the previous key was not specifed, start at the top and
401 # return the first one found.
403 if (!$self->{prev_md5}) {
404 $self->{prev_md5} = chr(0) x $HASH_SIZE;
405 $self->{return_next} = 1;
408 return $self->_traverse_index( $self->_base_offset, 0 );
413 # If db locking is set, flock() the db file. If called multiple
414 # times before unlock(), then the same number of unlocks() must
415 # be called before the lock is released.
417 my $self = $_[0]->_get_self;
419 $type = LOCK_EX unless defined $type;
421 if (!defined($self->_fh)) { return; }
423 if ($self->_root->{locking}) {
424 if (!$self->_root->{locked}) {
425 flock($self->_fh, $type);
427 # refresh end counter in case file has changed size
428 my @stats = stat($self->_root->{file});
429 $self->_root->{end} = $stats[7];
431 # double-check file inode, in case another process
432 # has optimize()d our file while we were waiting.
433 if ($stats[1] != $self->_root->{inode}) {
434 $self->{engine}->open( $self ); # re-open
435 flock($self->_fh, $type); # re-lock
436 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
439 $self->_root->{locked}++;
449 # If db locking is set, unlock the db file. See note in lock()
450 # regarding calling lock() multiple times.
452 my $self = $_[0]->_get_self;
454 if (!defined($self->_fh)) { return; }
456 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
457 $self->_root->{locked}--;
458 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
467 my $self = shift->_get_self;
468 my ($spot, $value) = @_;
473 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
474 my $type = $value->_type;
475 ${$spot} = $type eq TYPE_HASH ? {} : [];
476 $value->_copy_node( ${$spot} );
479 my $r = Scalar::Util::reftype( $value );
480 my $c = Scalar::Util::blessed( $value );
481 if ( $r eq 'ARRAY' ) {
482 ${$spot} = [ @{$value} ];
485 ${$spot} = { %{$value} };
487 ${$spot} = bless ${$spot}, $c
496 # Copy single level of keys or elements to new DB handle.
497 # Recurse for nested structures
499 my $self = shift->_get_self;
502 if ($self->_type eq TYPE_HASH) {
503 my $key = $self->first_key();
505 my $value = $self->get($key);
506 $self->_copy_value( \$db_temp->{$key}, $value );
507 $key = $self->next_key($key);
511 my $length = $self->length();
512 for (my $index = 0; $index < $length; $index++) {
513 my $value = $self->get($index);
514 $self->_copy_value( \$db_temp->[$index], $value );
523 # Recursively export into standard Perl hashes and arrays.
525 my $self = $_[0]->_get_self;
528 if ($self->_type eq TYPE_HASH) { $temp = {}; }
529 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
532 $self->_copy_node( $temp );
540 # Recursively import Perl hash/array structure
542 #XXX This use of ref() seems to be ok
543 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
545 my $self = $_[0]->_get_self;
548 #XXX This use of ref() seems to be ok
551 # struct is not a reference, so just import based on our type
555 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
556 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
559 my $r = Scalar::Util::reftype($struct) || '';
560 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
561 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
563 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
564 $self->push( @$struct );
567 return $self->_throw_error("Cannot import: type mismatch");
575 # Rebuild entire database into new file, then move
576 # it back on top of original.
578 my $self = $_[0]->_get_self;
580 #XXX Need to create a new test for this
581 # if ($self->_root->{links} > 1) {
582 # return $self->_throw_error("Cannot optimize: reference count is greater than 1");
585 my $db_temp = DBM::Deep->new(
586 file => $self->_root->{file} . '.tmp',
590 return $self->_throw_error("Cannot optimize: failed to open temp file: $!");
594 $self->_copy_node( $db_temp );
598 # Attempt to copy user, group and permissions over to new file
600 my @stats = stat($self->_fh);
601 my $perms = $stats[2] & 07777;
604 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
605 chmod( $perms, $self->_root->{file} . '.tmp' );
607 # q.v. perlport for more information on this variable
608 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
610 # Potential race condition when optmizing on Win32 with locking.
611 # The Windows filesystem requires that the filehandle be closed
612 # before it is overwritten with rename(). This could be redone
616 $self->{engine}->close( $self );
619 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
620 unlink $self->_root->{file} . '.tmp';
622 return $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
626 $self->{engine}->close( $self );
627 $self->{engine}->open( $self );
634 # Make copy of object and return
636 my $self = $_[0]->_get_self;
638 return DBM::Deep->new(
639 type => $self->_type,
640 base_offset => $self->_base_offset,
646 my %is_legal_filter = map {
649 store_key store_value
650 fetch_key fetch_value
655 # Setup filter function for storing or fetching the key or value
657 my $self = $_[0]->_get_self;
659 my $func = $_[2] ? $_[2] : undef;
661 if ( $is_legal_filter{$type} ) {
662 $self->_root->{"filter_$type"} = $func;
676 # Get access to the root structure
678 my $self = $_[0]->_get_self;
679 return $self->{root};
684 # Get access to the raw fh
686 #XXX It will be useful, though, when we split out HASH and ARRAY
687 my $self = $_[0]->_get_self;
688 return $self->_root->{fh};
693 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
695 my $self = $_[0]->_get_self;
696 return $self->{type};
701 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
703 my $self = $_[0]->_get_self;
704 return $self->{base_offset};
712 die "DBM::Deep: $_[1]\n";
717 # Precalculate index, bucket and bucket list sizes
720 #XXX I don't like this ...
721 set_pack() unless defined $LONG_SIZE;
723 $INDEX_SIZE = 256 * $LONG_SIZE;
724 $BUCKET_SIZE = $HASH_SIZE + $LONG_SIZE;
725 $BUCKET_LIST_SIZE = $MAX_BUCKETS * $BUCKET_SIZE;
730 # Set pack/unpack modes (see file header for more)
732 my ($long_s, $long_p, $data_s, $data_p) = @_;
734 $LONG_SIZE = $long_s ? $long_s : 4;
735 $LONG_PACK = $long_p ? $long_p : 'N';
737 $DATA_LENGTH_SIZE = $data_s ? $data_s : 4;
738 $DATA_LENGTH_PACK = $data_p ? $data_p : 'N';
745 # Set key digest function (default is MD5)
747 my ($digest_func, $hash_size) = @_;
749 $DIGEST_FUNC = $digest_func ? $digest_func : \&Digest::MD5::md5;
750 $HASH_SIZE = $hash_size ? $hash_size : 16;
757 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
762 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
766 # tie() methods (hashes and arrays)
771 # Store single hash key/value or array element in database.
773 my $self = $_[0]->_get_self;
776 # User may be storing a hash, in which case we do not want it run
777 # through the filtering system
778 my $value = ($self->_root->{filter_store_value} && !ref($_[2]))
779 ? $self->_root->{filter_store_value}->($_[2])
782 my $md5 = $DIGEST_FUNC->($key);
784 unless ( _is_writable( $self->_fh ) ) {
785 $self->_throw_error( 'Cannot write to a readonly filehandle' );
789 # Request exclusive lock for writing
791 $self->lock( LOCK_EX );
796 # Locate offset for bucket list using digest index system
798 my $tag = $self->{engine}->load_tag($self, $self->_base_offset);
800 $tag = $self->{engine}->create_tag($self, $self->_base_offset, SIG_INDEX, chr(0) x $INDEX_SIZE);
804 while ($tag->{signature} ne SIG_BLIST) {
805 my $num = ord(substr($md5, $ch, 1));
807 my $ref_loc = $tag->{offset} + ($num * $LONG_SIZE);
808 my $new_tag = $self->{engine}->index_lookup($self, $tag, $num);
811 seek($fh, $ref_loc + $self->_root->{file_offset}, SEEK_SET);
812 print( $fh pack($LONG_PACK, $self->_root->{end}) );
814 $tag = $self->{engine}->create_tag($self, $self->_root->{end}, SIG_BLIST, chr(0) x $BUCKET_LIST_SIZE);
816 $tag->{ref_loc} = $ref_loc;
824 $tag->{ref_loc} = $ref_loc;
831 # Add key/value to bucket list
833 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
842 # Fetch single value or element given plain key or array index
844 my $self = shift->_get_self;
847 my $md5 = $DIGEST_FUNC->($key);
850 # Request shared lock for reading
852 $self->lock( LOCK_SH );
854 my $tag = $self->_find_bucket_list( $md5 );
861 # Get value from bucket list
863 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
867 #XXX What is ref() checking here?
868 #YYY Filters only apply on scalar values, so the ref check is making
869 #YYY sure the fetched bucket is a scalar, not a child hash or array.
870 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
871 ? $self->_root->{filter_fetch_value}->($result)
877 # Delete single key/value pair or element given plain key or array index
879 my $self = $_[0]->_get_self;
882 my $md5 = $DIGEST_FUNC->($key);
885 # Request exclusive lock for writing
887 $self->lock( LOCK_EX );
889 my $tag = $self->_find_bucket_list( $md5 );
898 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
899 if ($value && !ref($value) && $self->_root->{filter_fetch_value}) {
900 $value = $self->_root->{filter_fetch_value}->($value);
903 my $result = $self->_delete_bucket( $tag, $md5 );
906 # If this object is an array and the key deleted was on the end of the stack,
907 # decrement the length variable.
917 # Check if a single key or element exists given plain key or array index
919 my $self = $_[0]->_get_self;
922 my $md5 = $DIGEST_FUNC->($key);
925 # Request shared lock for reading
927 $self->lock( LOCK_SH );
929 my $tag = $self->_find_bucket_list( $md5 );
932 # For some reason, the built-in exists() function returns '' for false
940 # Check if bucket exists and return 1 or ''
942 my $result = $self->_bucket_exists( $tag, $md5 ) || '';
951 # Clear all keys from hash, or all elements from array.
953 my $self = $_[0]->_get_self;
956 # Request exclusive lock for writing
958 $self->lock( LOCK_EX );
962 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
968 $self->{engine}->create_tag($self, $self->_base_offset, $self->_type, chr(0) x $INDEX_SIZE);
976 # Public method aliases
978 sub put { (shift)->STORE( @_ ) }
979 sub store { (shift)->STORE( @_ ) }
980 sub get { (shift)->FETCH( @_ ) }
981 sub fetch { (shift)->FETCH( @_ ) }
982 sub delete { (shift)->DELETE( @_ ) }
983 sub exists { (shift)->EXISTS( @_ ) }
984 sub clear { (shift)->CLEAR( @_ ) }
986 package DBM::Deep::_::Root;
1000 filter_store_key => undef,
1001 filter_store_value => undef,
1002 filter_fetch_key => undef,
1003 filter_fetch_value => undef,
1009 if ( $self->{fh} && !$self->{file_offset} ) {
1010 $self->{file_offset} = tell( $self->{fh} );
1018 return unless $self;
1020 close $self->{fh} if $self->{fh};
1031 DBM::Deep - A pure perl multi-level hash/array DBM
1036 my $db = DBM::Deep->new( "foo.db" );
1038 $db->{key} = 'value'; # tie() style
1041 $db->put('key' => 'value'); # OO style
1042 print $db->get('key');
1044 # true multi-level support
1045 $db->{my_complex} = [
1046 'hello', { perl => 'rules' },
1052 A unique flat-file database module, written in pure perl. True
1053 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
1054 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
1055 handle millions of keys and unlimited hash levels without significant
1056 slow-down. Written from the ground-up in pure perl -- this is NOT a
1057 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
1058 Mac OS X and Windows.
1062 Hopefully you are using Perl's excellent CPAN module, which will download
1063 and install the module for you. If not, get the tarball, and run these
1075 Construction can be done OO-style (which is the recommended way), or using
1076 Perl's tie() function. Both are examined here.
1078 =head2 OO CONSTRUCTION
1080 The recommended way to construct a DBM::Deep object is to use the new()
1081 method, which gets you a blessed, tied hash or array reference.
1083 my $db = DBM::Deep->new( "foo.db" );
1085 This opens a new database handle, mapped to the file "foo.db". If this
1086 file does not exist, it will automatically be created. DB files are
1087 opened in "r+" (read/write) mode, and the type of object returned is a
1088 hash, unless otherwise specified (see L<OPTIONS> below).
1090 You can pass a number of options to the constructor to specify things like
1091 locking, autoflush, etc. This is done by passing an inline hash:
1093 my $db = DBM::Deep->new(
1099 Notice that the filename is now specified I<inside> the hash with
1100 the "file" parameter, as opposed to being the sole argument to the
1101 constructor. This is required if any options are specified.
1102 See L<OPTIONS> below for the complete list.
1106 You can also start with an array instead of a hash. For this, you must
1107 specify the C<type> parameter:
1109 my $db = DBM::Deep->new(
1111 type => DBM::Deep->TYPE_ARRAY
1114 B<Note:> Specifing the C<type> parameter only takes effect when beginning
1115 a new DB file. If you create a DBM::Deep object with an existing file, the
1116 C<type> will be loaded from the file header, and an error will be thrown if
1117 the wrong type is passed in.
1119 =head2 TIE CONSTRUCTION
1121 Alternately, you can create a DBM::Deep handle by using Perl's built-in
1122 tie() function. The object returned from tie() can be used to call methods,
1123 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
1124 file (as expected with most tie'd objects).
1127 my $db = tie %hash, "DBM::Deep", "foo.db";
1130 my $db = tie @array, "DBM::Deep", "bar.db";
1132 As with the OO constructor, you can replace the DB filename parameter with
1133 a hash containing one or more options (see L<OPTIONS> just below for the
1136 tie %hash, "DBM::Deep", {
1144 There are a number of options that can be passed in when constructing your
1145 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
1151 Filename of the DB file to link the handle to. You can pass a full absolute
1152 filesystem path, partial path, or a plain filename if the file is in the
1153 current working directory. This is a required parameter (though q.v. fh).
1157 If you want, you can pass in the fh instead of the file. This is most useful for doing
1160 my $db = DBM::Deep->new( { fh => \*DATA } );
1162 You are responsible for making sure that the fh has been opened appropriately for your
1163 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
1164 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
1165 needs to read from the fh.
1169 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
1170 not need to set this. However, it's there if you want it.
1172 If you pass in fh and do not set this, it will be set appropriately.
1176 This parameter specifies what type of object to create, a hash or array. Use
1177 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
1178 This only takes effect when beginning a new file. This is an optional
1179 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
1183 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
1184 function to lock the database in exclusive mode for writes, and shared mode for
1185 reads. Pass any true value to enable. This affects the base DB handle I<and
1186 any child hashes or arrays> that use the same DB file. This is an optional
1187 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
1191 Specifies whether autoflush is to be enabled on the underlying filehandle.
1192 This obviously slows down write operations, but is required if you may have
1193 multiple processes accessing the same DB file (also consider enable I<locking>).
1194 Pass any true value to enable. This is an optional parameter, and defaults to 0
1199 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
1200 restore them when fetched. This is an B<experimental> feature, and does have
1201 side-effects. Basically, when hashes are re-blessed into their original
1202 classes, they are no longer blessed into the DBM::Deep class! So you won't be
1203 able to call any DBM::Deep methods on them. You have been warned.
1204 This is an optional parameter, and defaults to 0 (disabled).
1208 See L<FILTERS> below.
1212 Setting I<debug> mode will make all errors non-fatal, dump them out to
1213 STDERR, and continue on. This is for debugging purposes only, and probably
1214 not what you want. This is an optional parameter, and defaults to 0 (disabled).
1216 B<NOTE>: This parameter is considered deprecated and should not be used anymore.
1220 =head1 TIE INTERFACE
1222 With DBM::Deep you can access your databases using Perl's standard hash/array
1223 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
1224 treat them as such. DBM::Deep will intercept all reads/writes and direct them
1225 to the right place -- the DB file. This has nothing to do with the
1226 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
1227 using regular hashes and arrays, rather than calling functions like C<get()>
1228 and C<put()> (although those work too). It is entirely up to you how to want
1229 to access your databases.
1233 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
1234 or even nested hashes (or arrays) using standard Perl syntax:
1236 my $db = DBM::Deep->new( "foo.db" );
1238 $db->{mykey} = "myvalue";
1240 $db->{myhash}->{subkey} = "subvalue";
1242 print $db->{myhash}->{subkey} . "\n";
1244 You can even step through hash keys using the normal Perl C<keys()> function:
1246 foreach my $key (keys %$db) {
1247 print "$key: " . $db->{$key} . "\n";
1250 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
1251 pushes them onto an array, all before the loop even begins. If you have an
1252 extra large hash, this may exhaust Perl's memory. Instead, consider using
1253 Perl's C<each()> function, which pulls keys/values one at a time, using very
1256 while (my ($key, $value) = each %$db) {
1257 print "$key: $value\n";
1260 Please note that when using C<each()>, you should always pass a direct
1261 hash reference, not a lookup. Meaning, you should B<never> do this:
1264 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
1266 This causes an infinite loop, because for each iteration, Perl is calling
1267 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
1268 it effectively keeps returning the first key over and over again. Instead,
1269 assign a temporary variable to C<$db->{foo}>, then pass that to each().
1273 As with hashes, you can treat any DBM::Deep object like a normal Perl array
1274 reference. This includes inserting, removing and manipulating elements,
1275 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
1276 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
1277 or simply be a nested array reference inside a hash. Example:
1279 my $db = DBM::Deep->new(
1280 file => "foo-array.db",
1281 type => DBM::Deep->TYPE_ARRAY
1285 push @$db, "bar", "baz";
1286 unshift @$db, "bah";
1288 my $last_elem = pop @$db; # baz
1289 my $first_elem = shift @$db; # bah
1290 my $second_elem = $db->[1]; # bar
1292 my $num_elements = scalar @$db;
1296 In addition to the I<tie()> interface, you can also use a standard OO interface
1297 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
1298 array) has its own methods, but both types share the following common methods:
1299 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
1303 =item * new() / clone()
1305 These are the constructor and copy-functions.
1307 =item * put() / store()
1309 Stores a new hash key/value pair, or sets an array element value. Takes two
1310 arguments, the hash key or array index, and the new value. The value can be
1311 a scalar, hash ref or array ref. Returns true on success, false on failure.
1313 $db->put("foo", "bar"); # for hashes
1314 $db->put(1, "bar"); # for arrays
1316 =item * get() / fetch()
1318 Fetches the value of a hash key or array element. Takes one argument: the hash
1319 key or array index. Returns a scalar, hash ref or array ref, depending on the
1322 my $value = $db->get("foo"); # for hashes
1323 my $value = $db->get(1); # for arrays
1327 Checks if a hash key or array index exists. Takes one argument: the hash key
1328 or array index. Returns true if it exists, false if not.
1330 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
1331 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1335 Deletes one hash key/value pair or array element. Takes one argument: the hash
1336 key or array index. Returns true on success, false if not found. For arrays,
1337 the remaining elements located after the deleted element are NOT moved over.
1338 The deleted element is essentially just undefined, which is exactly how Perl's
1339 internal arrays work. Please note that the space occupied by the deleted
1340 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1341 below for details and workarounds.
1343 $db->delete("foo"); # for hashes
1344 $db->delete(1); # for arrays
1348 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1349 value. Please note that the space occupied by the deleted keys/values or
1350 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1351 details and workarounds.
1353 $db->clear(); # hashes or arrays
1355 =item * lock() / unlock()
1361 Recover lost disk space.
1363 =item * import() / export()
1365 Data going in and out.
1367 =item * set_digest() / set_pack() / set_filter()
1369 q.v. adjusting the interal parameters.
1375 For hashes, DBM::Deep supports all the common methods described above, and the
1376 following additional methods: C<first_key()> and C<next_key()>.
1382 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1383 fetched in an undefined order (which appears random). Takes no arguments,
1384 returns the key as a scalar value.
1386 my $key = $db->first_key();
1390 Returns the "next" key in the hash, given the previous one as the sole argument.
1391 Returns undef if there are no more keys to be fetched.
1393 $key = $db->next_key($key);
1397 Here are some examples of using hashes:
1399 my $db = DBM::Deep->new( "foo.db" );
1401 $db->put("foo", "bar");
1402 print "foo: " . $db->get("foo") . "\n";
1404 $db->put("baz", {}); # new child hash ref
1405 $db->get("baz")->put("buz", "biz");
1406 print "buz: " . $db->get("baz")->get("buz") . "\n";
1408 my $key = $db->first_key();
1410 print "$key: " . $db->get($key) . "\n";
1411 $key = $db->next_key($key);
1414 if ($db->exists("foo")) { $db->delete("foo"); }
1418 For arrays, DBM::Deep supports all the common methods described above, and the
1419 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1420 C<unshift()> and C<splice()>.
1426 Returns the number of elements in the array. Takes no arguments.
1428 my $len = $db->length();
1432 Adds one or more elements onto the end of the array. Accepts scalars, hash
1433 refs or array refs. No return value.
1435 $db->push("foo", "bar", {});
1439 Fetches the last element in the array, and deletes it. Takes no arguments.
1440 Returns undef if array is empty. Returns the element value.
1442 my $elem = $db->pop();
1446 Fetches the first element in the array, deletes it, then shifts all the
1447 remaining elements over to take up the space. Returns the element value. This
1448 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1451 my $elem = $db->shift();
1455 Inserts one or more elements onto the beginning of the array, shifting all
1456 existing elements over to make room. Accepts scalars, hash refs or array refs.
1457 No return value. This method is not recommended with large arrays -- see
1458 <LARGE ARRAYS> below for details.
1460 $db->unshift("foo", "bar", {});
1464 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1465 -f splice> for usage -- it is too complicated to document here. This method is
1466 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1470 Here are some examples of using arrays:
1472 my $db = DBM::Deep->new(
1474 type => DBM::Deep->TYPE_ARRAY
1477 $db->push("bar", "baz");
1478 $db->unshift("foo");
1481 my $len = $db->length();
1482 print "length: $len\n"; # 4
1484 for (my $k=0; $k<$len; $k++) {
1485 print "$k: " . $db->get($k) . "\n";
1488 $db->splice(1, 2, "biz", "baf");
1490 while (my $elem = shift @$db) {
1491 print "shifted: $elem\n";
1496 Enable automatic file locking by passing a true value to the C<locking>
1497 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1499 my $db = DBM::Deep->new(
1504 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1505 mode for writes, and shared mode for reads. This is required if you have
1506 multiple processes accessing the same database file, to avoid file corruption.
1507 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1508 NFS> below for more.
1510 =head2 EXPLICIT LOCKING
1512 You can explicitly lock a database, so it remains locked for multiple
1513 transactions. This is done by calling the C<lock()> method, and passing an
1514 optional lock mode argument (defaults to exclusive mode). This is particularly
1515 useful for things like counters, where the current value needs to be fetched,
1516 then incremented, then stored again.
1519 my $counter = $db->get("counter");
1521 $db->put("counter", $counter);
1530 You can pass C<lock()> an optional argument, which specifies which mode to use
1531 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1532 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1533 same as the constants defined in Perl's C<Fcntl> module.
1535 $db->lock( DBM::Deep->LOCK_SH );
1539 =head1 IMPORTING/EXPORTING
1541 You can import existing complex structures by calling the C<import()> method,
1542 and export an entire database into an in-memory structure using the C<export()>
1543 method. Both are examined here.
1547 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1548 walking the structure and adding keys/elements to the database as you go,
1549 simply pass a reference to the C<import()> method. This recursively adds
1550 everything to an existing DBM::Deep object for you. Here is an example:
1555 array1 => [ "elem0", "elem1", "elem2" ],
1557 subkey1 => "subvalue1",
1558 subkey2 => "subvalue2"
1562 my $db = DBM::Deep->new( "foo.db" );
1563 $db->import( $struct );
1565 print $db->{key1} . "\n"; # prints "value1"
1567 This recursively imports the entire C<$struct> object into C<$db>, including
1568 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1569 keys are merged with the existing ones, replacing if they already exist.
1570 The C<import()> method can be called on any database level (not just the base
1571 level), and works with both hash and array DB types.
1573 B<Note:> Make sure your existing structure has no circular references in it.
1574 These will cause an infinite loop when importing.
1578 Calling the C<export()> method on an existing DBM::Deep object will return
1579 a reference to a new in-memory copy of the database. The export is done
1580 recursively, so all nested hashes/arrays are all exported to standard Perl
1581 objects. Here is an example:
1583 my $db = DBM::Deep->new( "foo.db" );
1585 $db->{key1} = "value1";
1586 $db->{key2} = "value2";
1588 $db->{hash1}->{subkey1} = "subvalue1";
1589 $db->{hash1}->{subkey2} = "subvalue2";
1591 my $struct = $db->export();
1593 print $struct->{key1} . "\n"; # prints "value1"
1595 This makes a complete copy of the database in memory, and returns a reference
1596 to it. The C<export()> method can be called on any database level (not just
1597 the base level), and works with both hash and array DB types. Be careful of
1598 large databases -- you can store a lot more data in a DBM::Deep object than an
1599 in-memory Perl structure.
1601 B<Note:> Make sure your database has no circular references in it.
1602 These will cause an infinite loop when exporting.
1606 DBM::Deep has a number of hooks where you can specify your own Perl function
1607 to perform filtering on incoming or outgoing data. This is a perfect
1608 way to extend the engine, and implement things like real-time compression or
1609 encryption. Filtering applies to the base DB level, and all child hashes /
1610 arrays. Filter hooks can be specified when your DBM::Deep object is first
1611 constructed, or by calling the C<set_filter()> method at any time. There are
1612 four available filter hooks, described below:
1616 =item * filter_store_key
1618 This filter is called whenever a hash key is stored. It
1619 is passed the incoming key, and expected to return a transformed key.
1621 =item * filter_store_value
1623 This filter is called whenever a hash key or array element is stored. It
1624 is passed the incoming value, and expected to return a transformed value.
1626 =item * filter_fetch_key
1628 This filter is called whenever a hash key is fetched (i.e. via
1629 C<first_key()> or C<next_key()>). It is passed the transformed key,
1630 and expected to return the plain key.
1632 =item * filter_fetch_value
1634 This filter is called whenever a hash key or array element is fetched.
1635 It is passed the transformed value, and expected to return the plain value.
1639 Here are the two ways to setup a filter hook:
1641 my $db = DBM::Deep->new(
1643 filter_store_value => \&my_filter_store,
1644 filter_fetch_value => \&my_filter_fetch
1649 $db->set_filter( "filter_store_value", \&my_filter_store );
1650 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1652 Your filter function will be called only when dealing with SCALAR keys or
1653 values. When nested hashes and arrays are being stored/fetched, filtering
1654 is bypassed. Filters are called as static functions, passed a single SCALAR
1655 argument, and expected to return a single SCALAR value. If you want to
1656 remove a filter, set the function reference to C<undef>:
1658 $db->set_filter( "filter_store_value", undef );
1660 =head2 REAL-TIME ENCRYPTION EXAMPLE
1662 Here is a working example that uses the I<Crypt::Blowfish> module to
1663 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1664 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1665 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1668 use Crypt::Blowfish;
1671 my $cipher = Crypt::CBC->new({
1672 'key' => 'my secret key',
1673 'cipher' => 'Blowfish',
1675 'regenerate_key' => 0,
1676 'padding' => 'space',
1680 my $db = DBM::Deep->new(
1681 file => "foo-encrypt.db",
1682 filter_store_key => \&my_encrypt,
1683 filter_store_value => \&my_encrypt,
1684 filter_fetch_key => \&my_decrypt,
1685 filter_fetch_value => \&my_decrypt,
1688 $db->{key1} = "value1";
1689 $db->{key2} = "value2";
1690 print "key1: " . $db->{key1} . "\n";
1691 print "key2: " . $db->{key2} . "\n";
1697 return $cipher->encrypt( $_[0] );
1700 return $cipher->decrypt( $_[0] );
1703 =head2 REAL-TIME COMPRESSION EXAMPLE
1705 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1706 compression / decompression of keys & values with DBM::Deep Filters.
1707 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1708 more on I<Compress::Zlib>.
1713 my $db = DBM::Deep->new(
1714 file => "foo-compress.db",
1715 filter_store_key => \&my_compress,
1716 filter_store_value => \&my_compress,
1717 filter_fetch_key => \&my_decompress,
1718 filter_fetch_value => \&my_decompress,
1721 $db->{key1} = "value1";
1722 $db->{key2} = "value2";
1723 print "key1: " . $db->{key1} . "\n";
1724 print "key2: " . $db->{key2} . "\n";
1730 return Compress::Zlib::memGzip( $_[0] ) ;
1733 return Compress::Zlib::memGunzip( $_[0] ) ;
1736 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1737 actually numerical index numbers, and are not filtered.
1739 =head1 ERROR HANDLING
1741 Most DBM::Deep methods return a true value for success, and call die() on
1742 failure. You can wrap calls in an eval block to catch the die.
1744 my $db = DBM::Deep->new( "foo.db" ); # create hash
1745 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1747 print $@; # prints error message
1749 =head1 LARGEFILE SUPPORT
1751 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1752 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1753 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1754 by calling the static C<set_pack()> method before you do anything else.
1756 DBM::Deep::set_pack(8, 'Q');
1758 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1759 instead of 32-bit longs. After setting these values your DB files have a
1760 theoretical maximum size of 16 XB (exabytes).
1762 B<Note:> Changing these values will B<NOT> work for existing database files.
1763 Only change this for new files, and make sure it stays set consistently
1764 throughout the file's life. If you do set these values, you can no longer
1765 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1766 back to 32-bit mode.
1768 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1769 only a 32-bit Perl. However, I have received user reports that this does
1772 =head1 LOW-LEVEL ACCESS
1774 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1775 you can call the C<_fh()> method, which returns the handle:
1777 my $fh = $db->_fh();
1779 This method can be called on the root level of the datbase, or any child
1780 hashes or arrays. All levels share a I<root> structure, which contains things
1781 like the filehandle, a reference counter, and all the options specified
1782 when you created the object. You can get access to this root structure by
1783 calling the C<root()> method.
1785 my $root = $db->_root();
1787 This is useful for changing options after the object has already been created,
1788 such as enabling/disabling locking, or debug modes. You can also
1789 store your own temporary user data in this structure (be wary of name
1790 collision), which is then accessible from any child hash or array.
1792 =head1 CUSTOM DIGEST ALGORITHM
1794 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1795 keys. However you can override this, and use another algorithm (such as SHA-256)
1796 or even write your own. But please note that DBM::Deep currently expects zero
1797 collisions, so your algorithm has to be I<perfect>, so to speak.
1798 Collision detection may be introduced in a later version.
1802 You can specify a custom digest algorithm by calling the static C<set_digest()>
1803 function, passing a reference to a subroutine, and the length of the algorithm's
1804 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1805 objects. Here is a working example that uses a 256-bit hash from the
1806 I<Digest::SHA256> module. Please see
1807 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1812 my $context = Digest::SHA256::new(256);
1814 DBM::Deep::set_digest( \&my_digest, 32 );
1816 my $db = DBM::Deep->new( "foo-sha.db" );
1818 $db->{key1} = "value1";
1819 $db->{key2} = "value2";
1820 print "key1: " . $db->{key1} . "\n";
1821 print "key2: " . $db->{key2} . "\n";
1827 return substr( $context->hash($_[0]), 0, 32 );
1830 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1831 of bytes you specify in the C<set_digest()> function (in this case 32).
1833 =head1 CIRCULAR REFERENCES
1835 DBM::Deep has B<experimental> support for circular references. Meaning you
1836 can have a nested hash key or array element that points to a parent object.
1837 This relationship is stored in the DB file, and is preserved between sessions.
1840 my $db = DBM::Deep->new( "foo.db" );
1843 $db->{circle} = $db; # ref to self
1845 print $db->{foo} . "\n"; # prints "foo"
1846 print $db->{circle}->{foo} . "\n"; # prints "foo" again
1848 One catch is, passing the object to a function that recursively walks the
1849 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1850 C<export()> methods) will result in an infinite loop. The other catch is,
1851 if you fetch the I<key> of a circular reference (i.e. using the C<first_key()>
1852 or C<next_key()> methods), you will get the I<target object's key>, not the
1853 ref's key. This gets even more interesting with the above example, where
1854 the I<circle> key points to the base DB object, which technically doesn't
1855 have a key. So I made DBM::Deep return "[base]" as the key name in that
1858 =head1 CAVEATS / ISSUES / BUGS
1860 This section describes all the known issues with DBM::Deep. It you have found
1861 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1863 =head2 UNUSED SPACE RECOVERY
1865 One major caveat with DBM::Deep is that space occupied by existing keys and
1866 values is not recovered when they are deleted. Meaning if you keep deleting
1867 and adding new keys, your file will continuously grow. I am working on this,
1868 but in the meantime you can call the built-in C<optimize()> method from time to
1869 time (perhaps in a crontab or something) to recover all your unused space.
1871 $db->optimize(); # returns true on success
1873 This rebuilds the ENTIRE database into a new file, then moves it on top of
1874 the original. The new file will have no unused space, thus it will take up as
1875 little disk space as possible. Please note that this operation can take
1876 a long time for large files, and you need enough disk space to temporarily hold
1877 2 copies of your DB file. The temporary file is created in the same directory
1878 as the original, named with a ".tmp" extension, and is deleted when the
1879 operation completes. Oh, and if locking is enabled, the DB is automatically
1880 locked for the entire duration of the copy.
1882 B<WARNING:> Only call optimize() on the top-level node of the database, and
1883 make sure there are no child references lying around. DBM::Deep keeps a reference
1884 counter, and if it is greater than 1, optimize() will abort and return undef.
1886 =head2 AUTOVIVIFICATION
1888 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1889 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1890 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1891 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1892 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1895 $db->{foo}->{bar} = "hello";
1897 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1898 being an empty hash. Try this instead, which works fine:
1900 $db->{foo} = { bar => "hello" };
1902 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1903 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1904 Probably a bug in Perl.
1906 =head2 FILE CORRUPTION
1908 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1909 for a 32-bit signature when opened, but other corruption in files can cause
1910 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1911 stuck in an infinite loop depending on the level of corruption. File write
1912 operations are not checked for failure (for speed), so if you happen to run
1913 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1914 be addressed in a later version of DBM::Deep.
1918 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1919 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1920 about setting up your NFS server with a locking daemon, then using lockf() to
1921 lock your files, but your mileage may vary there as well. From what I
1922 understand, there is no real way to do it. However, if you need access to the
1923 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1924 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1926 =head2 COPYING OBJECTS
1928 Beware of copying tied objects in Perl. Very strange things can happen.
1929 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1930 returns a new, blessed, tied hash or array to the same level in the DB.
1932 my $copy = $db->clone();
1934 B<Note>: Since clone() here is cloning the object, not the database location, any
1935 modifications to either $db or $copy will be visible in both.
1939 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1940 These functions cause every element in the array to move, which can be murder
1941 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1942 a different location. This will be addressed in the forthcoming version 1.00.
1944 =head2 WRITEONLY FILES
1946 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1947 writeonly mode. STORE will verify that the filehandle is writable. However, there
1948 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1949 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1953 This section discusses DBM::Deep's speed and memory usage.
1957 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1958 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1959 multi-level hash/array support, and cross-platform FTPable files. Even so,
1960 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1961 with huge databases. Here is some test data:
1963 Adding 1,000,000 keys to new DB file...
1965 At 100 keys, avg. speed is 2,703 keys/sec
1966 At 200 keys, avg. speed is 2,642 keys/sec
1967 At 300 keys, avg. speed is 2,598 keys/sec
1968 At 400 keys, avg. speed is 2,578 keys/sec
1969 At 500 keys, avg. speed is 2,722 keys/sec
1970 At 600 keys, avg. speed is 2,628 keys/sec
1971 At 700 keys, avg. speed is 2,700 keys/sec
1972 At 800 keys, avg. speed is 2,607 keys/sec
1973 At 900 keys, avg. speed is 2,190 keys/sec
1974 At 1,000 keys, avg. speed is 2,570 keys/sec
1975 At 2,000 keys, avg. speed is 2,417 keys/sec
1976 At 3,000 keys, avg. speed is 1,982 keys/sec
1977 At 4,000 keys, avg. speed is 1,568 keys/sec
1978 At 5,000 keys, avg. speed is 1,533 keys/sec
1979 At 6,000 keys, avg. speed is 1,787 keys/sec
1980 At 7,000 keys, avg. speed is 1,977 keys/sec
1981 At 8,000 keys, avg. speed is 2,028 keys/sec
1982 At 9,000 keys, avg. speed is 2,077 keys/sec
1983 At 10,000 keys, avg. speed is 2,031 keys/sec
1984 At 20,000 keys, avg. speed is 1,970 keys/sec
1985 At 30,000 keys, avg. speed is 2,050 keys/sec
1986 At 40,000 keys, avg. speed is 2,073 keys/sec
1987 At 50,000 keys, avg. speed is 1,973 keys/sec
1988 At 60,000 keys, avg. speed is 1,914 keys/sec
1989 At 70,000 keys, avg. speed is 2,091 keys/sec
1990 At 80,000 keys, avg. speed is 2,103 keys/sec
1991 At 90,000 keys, avg. speed is 1,886 keys/sec
1992 At 100,000 keys, avg. speed is 1,970 keys/sec
1993 At 200,000 keys, avg. speed is 2,053 keys/sec
1994 At 300,000 keys, avg. speed is 1,697 keys/sec
1995 At 400,000 keys, avg. speed is 1,838 keys/sec
1996 At 500,000 keys, avg. speed is 1,941 keys/sec
1997 At 600,000 keys, avg. speed is 1,930 keys/sec
1998 At 700,000 keys, avg. speed is 1,735 keys/sec
1999 At 800,000 keys, avg. speed is 1,795 keys/sec
2000 At 900,000 keys, avg. speed is 1,221 keys/sec
2001 At 1,000,000 keys, avg. speed is 1,077 keys/sec
2003 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
2004 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
2005 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
2006 Run time was 12 min 3 sec.
2010 One of the great things about DBM::Deep is that it uses very little memory.
2011 Even with huge databases (1,000,000+ keys) you will not see much increased
2012 memory on your process. DBM::Deep relies solely on the filesystem for storing
2013 and fetching data. Here is output from I</usr/bin/top> before even opening a
2016 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2017 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
2019 Basically the process is taking 2,716K of memory. And here is the same
2020 process after storing and fetching 1,000,000 keys:
2022 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2023 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
2025 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
2026 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
2028 =head1 DB FILE FORMAT
2030 In case you were interested in the underlying DB file format, it is documented
2031 here in this section. You don't need to know this to use the module, it's just
2032 included for reference.
2036 DBM::Deep files always start with a 32-bit signature to identify the file type.
2037 This is at offset 0. The signature is "DPDB" in network byte order. This is
2038 checked for when the file is opened and an error will be thrown if it's not found.
2042 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
2043 has a standard header containing the type of data, the length of data, and then
2044 the data itself. The type is a single character (1 byte), the length is a
2045 32-bit unsigned long in network byte order, and the data is, well, the data.
2046 Here is how it unfolds:
2050 Immediately after the 32-bit file signature is the I<Master Index> record.
2051 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
2052 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
2053 depending on how the DBM::Deep object was constructed.
2055 The index works by looking at a I<MD5 Hash> of the hash key (or array index
2056 number). The first 8-bit char of the MD5 signature is the offset into the
2057 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
2058 index element is a file offset of the next tag for the key/element in question,
2059 which is usually a I<Bucket List> tag (see below).
2061 The next tag I<could> be another index, depending on how many keys/elements
2062 exist. See L<RE-INDEXING> below for details.
2066 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
2067 file offsets to where the actual data is stored. It starts with a standard
2068 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
2069 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
2070 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
2071 When the list fills up, a I<Re-Index> operation is performed (See
2072 L<RE-INDEXING> below).
2076 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
2077 index/value pair (in array mode). It starts with a standard tag header with
2078 type I<D> for scalar data (string, binary, etc.), or it could be a nested
2079 hash (type I<H>) or array (type I<A>). The value comes just after the tag
2080 header. The size reported in the tag header is only for the value, but then,
2081 just after the value is another size (32-bit unsigned long) and then the plain
2082 key itself. Since the value is likely to be fetched more often than the plain
2083 key, I figured it would be I<slightly> faster to store the value first.
2085 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
2086 record for the nested structure, where the process begins all over again.
2090 After a I<Bucket List> grows to 16 records, its allocated space in the file is
2091 exhausted. Then, when another key/element comes in, the list is converted to a
2092 new index record. However, this index will look at the next char in the MD5
2093 hash, and arrange new Bucket List pointers accordingly. This process is called
2094 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
2095 17 (16 + new one) keys/elements are removed from the old Bucket List and
2096 inserted into the new index. Several new Bucket Lists are created in the
2097 process, as a new MD5 char from the key is being examined (it is unlikely that
2098 the keys will all share the same next char of their MD5s).
2100 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
2101 when the Bucket Lists will turn into indexes, but the first round tends to
2102 happen right around 4,000 keys. You will see a I<slight> decrease in
2103 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
2104 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
2105 right around 900,000 keys. This process can continue nearly indefinitely --
2106 right up until the point the I<MD5> signatures start colliding with each other,
2107 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
2108 getting struck by lightning while you are walking to cash in your tickets.
2109 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
2110 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
2111 this is 340 unodecillion, but don't quote me).
2115 When a new key/element is stored, the key (or index number) is first run through
2116 I<Digest::MD5> to get a 128-bit signature (example, in hex:
2117 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
2118 for the first char of the signature (in this case I<b0>). If it does not exist,
2119 a new I<Bucket List> is created for our key (and the next 15 future keys that
2120 happen to also have I<b> as their first MD5 char). The entire MD5 is written
2121 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
2122 this point, unless we are replacing an existing I<Bucket>), where the actual
2123 data will be stored.
2127 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
2128 (or index number), then walking along the indexes. If there are enough
2129 keys/elements in this DB level, there might be nested indexes, each linked to
2130 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
2131 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
2132 question. If we found a match, the I<Bucket> tag is loaded, where the value and
2133 plain key are stored.
2135 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
2136 methods. In this process the indexes are walked systematically, and each key
2137 fetched in increasing MD5 order (which is why it appears random). Once the
2138 I<Bucket> is found, the value is skipped and the plain key returned instead.
2139 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
2140 alphabetically sorted. This only happens on an index-level -- as soon as the
2141 I<Bucket Lists> are hit, the keys will come out in the order they went in --
2142 so it's pretty much undefined how the keys will come out -- just like Perl's
2145 =head1 CODE COVERAGE
2147 We use B<Devel::Cover> to test the code coverage of our tests, below is the
2148 B<Devel::Cover> report on this module's test suite.
2150 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2151 File stmt bran cond sub pod time total
2152 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2153 blib/lib/DBM/Deep.pm 95.2 83.8 70.0 98.2 100.0 58.0 91.0
2154 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 n/a 26.7 98.0
2155 blib/lib/DBM/Deep/Hash.pm 95.3 80.0 100.0 100.0 n/a 15.3 92.4
2156 Total 96.2 84.8 74.4 98.8 100.0 100.0 92.4
2157 ---------------------------- ------ ------ ------ ------ ------ ------ ------
2159 =head1 MORE INFORMATION
2161 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
2162 or send email to L<DBM-Deep@googlegroups.com>.
2166 Joseph Huckaby, L<jhuckaby@cpan.org>
2168 Rob Kinyon, L<rkinyon@cpan.org>
2170 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
2174 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
2175 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
2179 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
2180 This is free software, you may use it and distribute it under the
2181 same terms as Perl itself.