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.
37 use Fcntl qw( :DEFAULT :flock :seek );
41 use DBM::Deep::Engine;
44 use vars qw( $VERSION );
45 $VERSION = q(0.99_01);
48 # Setup constants for users to pass to new()
50 sub TYPE_HASH () { DBM::Deep::Engine->SIG_HASH }
51 sub TYPE_ARRAY () { DBM::Deep::Engine->SIG_ARRAY }
59 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
64 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
65 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
70 $args = { file => shift };
78 # Class constructor method for Perl OO interface.
79 # Calls tie() and returns blessed reference to tied hash or array,
80 # providing a hybrid OO/tie interface.
83 my $args = $class->_get_args( @_ );
86 # Check if we want a tied hash or array.
89 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
90 $class = 'DBM::Deep::Array';
91 require DBM::Deep::Array;
92 tie @$self, $class, %$args;
95 $class = 'DBM::Deep::Hash';
96 require DBM::Deep::Hash;
97 tie %$self, $class, %$args;
100 return bless $self, $class;
103 # This initializer is called from the various TIE* methods. new() calls tie(),
104 # which allows for a single point of entry.
109 $args->{fileobj} = DBM::Deep::File->new( $args )
110 unless exists $args->{fileobj};
112 # locking implicitly enables autoflush
113 if ($args->{locking}) { $args->{autoflush} = 1; }
115 # These are the defaults to be optionally overridden below
118 base_offset => undef,
125 $self->{engine} = DBM::Deep::Engine->new( { %{$args}, obj => $self } );
127 # Grab the parameters we want to use
128 foreach my $param ( keys %$self ) {
129 next unless exists $args->{$param};
130 $self->{$param} = $args->{$param};
133 $self->{engine}->setup_fh( $self );
135 $self->{fileobj}->set_db( $self );
142 require DBM::Deep::Hash;
143 return DBM::Deep::Hash->TIEHASH( @_ );
148 require DBM::Deep::Array;
149 return DBM::Deep::Array->TIEARRAY( @_ );
153 my $self = shift->_get_self;
154 return $self->_fileobj->lock( $self, @_ );
158 my $self = shift->_get_self;
159 return $self->_fileobj->unlock( $self, @_ );
163 my $self = shift->_get_self;
164 my ($spot, $value) = @_;
169 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
170 ${$spot} = $value->_repr;
171 $value->_copy_node( ${$spot} );
174 my $r = Scalar::Util::reftype( $value );
175 my $c = Scalar::Util::blessed( $value );
176 if ( $r eq 'ARRAY' ) {
177 ${$spot} = [ @{$value} ];
180 ${$spot} = { %{$value} };
182 ${$spot} = bless ${$spot}, $c
190 die "Must be implemented in a child class\n";
194 die "Must be implemented in a child class\n";
199 # Recursively export into standard Perl hashes and arrays.
201 my $self = shift->_get_self;
203 my $temp = $self->_repr;
206 $self->_copy_node( $temp );
214 # Recursively import Perl hash/array structure
216 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
218 my $self = shift->_get_self;
221 # struct is not a reference, so just import based on our type
223 $struct = $self->_repr( @_ );
226 return $self->_import( $struct );
231 # Rebuild entire database into new file, then move
232 # it back on top of original.
234 my $self = shift->_get_self;
236 #XXX Need to create a new test for this
237 # if ($self->_fileobj->{links} > 1) {
238 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
241 my $db_temp = DBM::Deep->new(
242 file => $self->_fileobj->{file} . '.tmp',
247 $self->_copy_node( $db_temp );
251 # Attempt to copy user, group and permissions over to new file
253 my @stats = stat($self->_fh);
254 my $perms = $stats[2] & 07777;
257 chown( $uid, $gid, $self->_fileobj->{file} . '.tmp' );
258 chmod( $perms, $self->_fileobj->{file} . '.tmp' );
260 # q.v. perlport for more information on this variable
261 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
263 # Potential race condition when optmizing on Win32 with locking.
264 # The Windows filesystem requires that the filehandle be closed
265 # before it is overwritten with rename(). This could be redone
269 $self->_fileobj->close;
272 if (!rename $self->_fileobj->{file} . '.tmp', $self->_fileobj->{file}) {
273 unlink $self->_fileobj->{file} . '.tmp';
275 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
279 $self->_fileobj->close;
280 $self->_fileobj->open;
281 $self->{engine}->setup_fh( $self );
288 # Make copy of object and return
290 my $self = shift->_get_self;
292 return DBM::Deep->new(
293 type => $self->_type,
294 base_offset => $self->_base_offset,
295 fileobj => $self->_fileobj,
300 my %is_legal_filter = map {
303 store_key store_value
304 fetch_key fetch_value
309 # Setup filter function for storing or fetching the key or value
311 my $self = shift->_get_self;
315 if ( $is_legal_filter{$type} ) {
316 $self->_fileobj->{"filter_$type"} = $func;
325 my $self = shift->_get_self;
326 $self->_fileobj->begin_transaction;
331 my $self = shift->_get_self;
332 $self->_fileobj->end_transaction;
337 my $self = shift->_get_self;
338 # At this point, we need to replay the actions taken
339 $self->_fileobj->end_transaction;
348 my $self = $_[0]->_get_self;
349 return $self->{fileobj};
353 my $self = $_[0]->_get_self;
354 return $self->{type};
358 my $self = $_[0]->_get_self;
359 return $self->{base_offset};
363 my $self = $_[0]->_get_self;
364 return $self->_fileobj->{fh};
372 die "DBM::Deep: $_[1]\n";
377 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
382 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
389 if ( my $parent = $self->{parent} ) {
393 $parent->_type eq TYPE_HASH
394 ? "\{$child->{parent_key}\}"
395 : "\[$child->{parent_key}\]"
399 $parent = $parent->{parent};
403 return '$db->' . $base;
408 # Store single hash key/value or array element in database.
410 my $self = shift->_get_self;
411 my ($key, $value, $orig_key) = @_;
414 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
415 $self->_throw_error( 'Cannot write to a readonly filehandle' );
418 # if ( my $afh = $self->_fileobj->{audit_fh} ) {
419 if ( defined $orig_key ) {
420 my $lhs = $self->_find_parent;
421 if ( $self->_type eq TYPE_HASH ) {
422 $lhs .= "\{$orig_key\}";
425 $lhs .= "\[$orig_key\]";
430 my $r = Scalar::Util::reftype( $value ) || '';
431 if ( $r eq 'HASH' ) {
434 elsif ( $r eq 'ARRAY' ) {
438 if ( defined $value ) {
446 if ( my $c = Scalar::Util::blessed( $value ) ) {
447 $rhs = "bless $rhs, '$c'";
450 $self->_fileobj->audit( "$lhs = $rhs;" );
451 # flock( $afh, LOCK_EX );
452 # print( $afh "$lhs = $rhs; # " . localtime(time) . "\n" );
453 # flock( $afh, LOCK_UN );
458 # Request exclusive lock for writing
460 $self->lock( LOCK_EX );
462 my $md5 = $self->{engine}{digest}->($key);
464 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5, { create => 1 } );
466 # User may be storing a hash, in which case we do not want it run
467 # through the filtering system
468 if ( !ref($value) && $self->_fileobj->{filter_store_value} ) {
469 $value = $self->_fileobj->{filter_store_value}->( $value );
473 # Add key/value to bucket list
475 my $result = $self->{engine}->add_bucket( $tag, $md5, $key, $value, undef, $orig_key );
484 # Fetch single value or element given plain key or array index
486 my $self = shift->_get_self;
487 my ($key, $orig_key) = @_;
489 my $md5 = $self->{engine}{digest}->($key);
492 # Request shared lock for reading
494 $self->lock( LOCK_SH );
496 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
503 # Get value from bucket list
505 my $result = $self->{engine}->get_bucket_value( $tag, $md5, $orig_key );
509 # Filters only apply to scalar values, so the ref check is making
510 # sure the fetched bucket is a scalar, not a child hash or array.
511 return ($result && !ref($result) && $self->_fileobj->{filter_fetch_value})
512 ? $self->_fileobj->{filter_fetch_value}->($result)
518 # Delete single key/value pair or element given plain key or array index
520 my $self = shift->_get_self;
521 my ($key, $orig_key) = @_;
523 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
524 $self->_throw_error( 'Cannot write to a readonly filehandle' );
527 if ( my $afh = $self->_fileobj->{audit_fh} ) {
528 if ( defined $orig_key ) {
529 my $lhs = $self->_find_parent;
530 if ( $self->_type eq TYPE_HASH ) {
531 $lhs .= "\{$orig_key\}";
534 $lhs .= "\[$orig_key]\]";
537 flock( $afh, LOCK_EX );
538 print( $afh "delete $lhs; # " . localtime(time) . "\n" );
539 flock( $afh, LOCK_UN );
544 # Request exclusive lock for writing
546 $self->lock( LOCK_EX );
548 my $md5 = $self->{engine}{digest}->($key);
550 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
559 my $value = $self->{engine}->get_bucket_value( $tag, $md5 );
561 if (defined $value && !ref($value) && $self->_fileobj->{filter_fetch_value}) {
562 $value = $self->_fileobj->{filter_fetch_value}->($value);
565 my $result = $self->{engine}->delete_bucket( $tag, $md5, $orig_key );
568 # If this object is an array and the key deleted was on the end of the stack,
569 # decrement the length variable.
579 # Check if a single key or element exists given plain key or array index
581 my $self = shift->_get_self;
584 my $md5 = $self->{engine}{digest}->($key);
587 # Request shared lock for reading
589 $self->lock( LOCK_SH );
591 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
596 # For some reason, the built-in exists() function returns '' for false
602 # Check if bucket exists and return 1 or ''
604 my $result = $self->{engine}->bucket_exists( $tag, $md5 ) || '';
613 # Clear all keys from hash, or all elements from array.
615 my $self = shift->_get_self;
617 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
618 $self->_throw_error( 'Cannot write to a readonly filehandle' );
621 if ( my $afh = $self->_fileobj->{audit_fh} ) {
622 my $lhs = $self->_find_parent;
625 if ( $self->_type eq TYPE_HASH ) {
626 $lhs = '%{' . $lhs . '}';
629 $lhs = '@{' . $lhs . '}';
632 flock( $afh, LOCK_EX );
633 print( $afh "$lhs = $rhs; # " . localtime(time) . "\n" );
634 flock( $afh, LOCK_UN );
638 # Request exclusive lock for writing
640 $self->lock( LOCK_EX );
644 seek($fh, $self->_base_offset + $self->_fileobj->{file_offset}, SEEK_SET);
650 #XXX This needs updating to use _release_space
651 $self->{engine}->write_tag(
652 $self->_base_offset, $self->_type,
653 chr(0)x$self->{engine}{index_size},
662 # Public method aliases
664 sub put { (shift)->STORE( @_ ) }
665 sub store { (shift)->STORE( @_ ) }
666 sub get { (shift)->FETCH( @_ ) }
667 sub fetch { (shift)->FETCH( @_ ) }
668 sub delete { (shift)->DELETE( @_ ) }
669 sub exists { (shift)->EXISTS( @_ ) }
670 sub clear { (shift)->CLEAR( @_ ) }
677 DBM::Deep - A pure perl multi-level hash/array DBM
682 my $db = DBM::Deep->new( "foo.db" );
684 $db->{key} = 'value'; # tie() style
687 $db->put('key' => 'value'); # OO style
688 print $db->get('key');
690 # true multi-level support
691 $db->{my_complex} = [
692 'hello', { perl => 'rules' },
698 A unique flat-file database module, written in pure perl. True
699 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
700 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
701 handle millions of keys and unlimited hash levels without significant
702 slow-down. Written from the ground-up in pure perl -- this is NOT a
703 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
704 Mac OS X and Windows.
706 =head1 VERSION DIFFERENCES
708 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
709 before. While attempts have been made to be backwards compatible, no guarantees.
713 Hopefully you are using Perl's excellent CPAN module, which will download
714 and install the module for you. If not, get the tarball, and run these
726 Construction can be done OO-style (which is the recommended way), or using
727 Perl's tie() function. Both are examined here.
729 =head2 OO CONSTRUCTION
731 The recommended way to construct a DBM::Deep object is to use the new()
732 method, which gets you a blessed, tied hash or array reference.
734 my $db = DBM::Deep->new( "foo.db" );
736 This opens a new database handle, mapped to the file "foo.db". If this
737 file does not exist, it will automatically be created. DB files are
738 opened in "r+" (read/write) mode, and the type of object returned is a
739 hash, unless otherwise specified (see L<OPTIONS> below).
741 You can pass a number of options to the constructor to specify things like
742 locking, autoflush, etc. This is done by passing an inline hash:
744 my $db = DBM::Deep->new(
750 Notice that the filename is now specified I<inside> the hash with
751 the "file" parameter, as opposed to being the sole argument to the
752 constructor. This is required if any options are specified.
753 See L<OPTIONS> below for the complete list.
757 You can also start with an array instead of a hash. For this, you must
758 specify the C<type> parameter:
760 my $db = DBM::Deep->new(
762 type => DBM::Deep->TYPE_ARRAY
765 B<Note:> Specifing the C<type> parameter only takes effect when beginning
766 a new DB file. If you create a DBM::Deep object with an existing file, the
767 C<type> will be loaded from the file header, and an error will be thrown if
768 the wrong type is passed in.
770 =head2 TIE CONSTRUCTION
772 Alternately, you can create a DBM::Deep handle by using Perl's built-in
773 tie() function. The object returned from tie() can be used to call methods,
774 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
775 file (as expected with most tie'd objects).
778 my $db = tie %hash, "DBM::Deep", "foo.db";
781 my $db = tie @array, "DBM::Deep", "bar.db";
783 As with the OO constructor, you can replace the DB filename parameter with
784 a hash containing one or more options (see L<OPTIONS> just below for the
787 tie %hash, "DBM::Deep", {
795 There are a number of options that can be passed in when constructing your
796 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
802 Filename of the DB file to link the handle to. You can pass a full absolute
803 filesystem path, partial path, or a plain filename if the file is in the
804 current working directory. This is a required parameter (though q.v. fh).
808 If you want, you can pass in the fh instead of the file. This is most useful for doing
811 my $db = DBM::Deep->new( { fh => \*DATA } );
813 You are responsible for making sure that the fh has been opened appropriately for your
814 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
815 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
816 needs to read from the fh.
820 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
821 not need to set this. However, it's there if you want it.
823 If you pass in fh and do not set this, it will be set appropriately.
827 This parameter specifies what type of object to create, a hash or array. Use
828 one of these two constants:
832 =item * C<DBM::Deep-E<gt>TYPE_HASH>
834 =item * C<DBM::Deep-E<gt>TYPE_ARRAY>.
838 This only takes effect when beginning a new file. This is an optional
839 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
843 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
844 function to lock the database in exclusive mode for writes, and shared mode for
845 reads. Pass any true value to enable. This affects the base DB handle I<and
846 any child hashes or arrays> that use the same DB file. This is an optional
847 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
851 Specifies whether autoflush is to be enabled on the underlying filehandle.
852 This obviously slows down write operations, but is required if you may have
853 multiple processes accessing the same DB file (also consider enable I<locking>).
854 Pass any true value to enable. This is an optional parameter, and defaults to 0
859 If I<autobless> mode is enabled, DBM::Deep will preserve the class something
860 is blessed into, and restores it when fetched. This is an optional parameter, and defaults to 1 (enabled).
862 B<Note:> If you use the OO-interface, you will not be able to call any methods
863 of DBM::Deep on the blessed item. This is considered to be a feature.
867 See L</FILTERS> below.
873 With DBM::Deep you can access your databases using Perl's standard hash/array
874 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
875 treat them as such. DBM::Deep will intercept all reads/writes and direct them
876 to the right place -- the DB file. This has nothing to do with the
877 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
878 using regular hashes and arrays, rather than calling functions like C<get()>
879 and C<put()> (although those work too). It is entirely up to you how to want
880 to access your databases.
884 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
885 or even nested hashes (or arrays) using standard Perl syntax:
887 my $db = DBM::Deep->new( "foo.db" );
889 $db->{mykey} = "myvalue";
891 $db->{myhash}->{subkey} = "subvalue";
893 print $db->{myhash}->{subkey} . "\n";
895 You can even step through hash keys using the normal Perl C<keys()> function:
897 foreach my $key (keys %$db) {
898 print "$key: " . $db->{$key} . "\n";
901 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
902 pushes them onto an array, all before the loop even begins. If you have an
903 extra large hash, this may exhaust Perl's memory. Instead, consider using
904 Perl's C<each()> function, which pulls keys/values one at a time, using very
907 while (my ($key, $value) = each %$db) {
908 print "$key: $value\n";
911 Please note that when using C<each()>, you should always pass a direct
912 hash reference, not a lookup. Meaning, you should B<never> do this:
915 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
917 This causes an infinite loop, because for each iteration, Perl is calling
918 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
919 it effectively keeps returning the first key over and over again. Instead,
920 assign a temporary variable to C<$db->{foo}>, then pass that to each().
924 As with hashes, you can treat any DBM::Deep object like a normal Perl array
925 reference. This includes inserting, removing and manipulating elements,
926 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
927 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
928 or simply be a nested array reference inside a hash. Example:
930 my $db = DBM::Deep->new(
931 file => "foo-array.db",
932 type => DBM::Deep->TYPE_ARRAY
936 push @$db, "bar", "baz";
939 my $last_elem = pop @$db; # baz
940 my $first_elem = shift @$db; # bah
941 my $second_elem = $db->[1]; # bar
943 my $num_elements = scalar @$db;
947 In addition to the I<tie()> interface, you can also use a standard OO interface
948 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
949 array) has its own methods, but both types share the following common methods:
950 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
954 =item * new() / clone()
956 These are the constructor and copy-functions.
958 =item * put() / store()
960 Stores a new hash key/value pair, or sets an array element value. Takes two
961 arguments, the hash key or array index, and the new value. The value can be
962 a scalar, hash ref or array ref. Returns true on success, false on failure.
964 $db->put("foo", "bar"); # for hashes
965 $db->put(1, "bar"); # for arrays
967 =item * get() / fetch()
969 Fetches the value of a hash key or array element. Takes one argument: the hash
970 key or array index. Returns a scalar, hash ref or array ref, depending on the
973 my $value = $db->get("foo"); # for hashes
974 my $value = $db->get(1); # for arrays
978 Checks if a hash key or array index exists. Takes one argument: the hash key
979 or array index. Returns true if it exists, false if not.
981 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
982 if ($db->exists(1)) { print "yay!\n"; } # for arrays
986 Deletes one hash key/value pair or array element. Takes one argument: the hash
987 key or array index. Returns true on success, false if not found. For arrays,
988 the remaining elements located after the deleted element are NOT moved over.
989 The deleted element is essentially just undefined, which is exactly how Perl's
990 internal arrays work. Please note that the space occupied by the deleted
991 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
992 below for details and workarounds.
994 $db->delete("foo"); # for hashes
995 $db->delete(1); # for arrays
999 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1000 value. Please note that the space occupied by the deleted keys/values or
1001 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1002 details and workarounds.
1004 $db->clear(); # hashes or arrays
1006 =item * lock() / unlock()
1012 Recover lost disk space.
1014 =item * import() / export()
1016 Data going in and out.
1022 For hashes, DBM::Deep supports all the common methods described above, and the
1023 following additional methods: C<first_key()> and C<next_key()>.
1029 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1030 fetched in an undefined order (which appears random). Takes no arguments,
1031 returns the key as a scalar value.
1033 my $key = $db->first_key();
1037 Returns the "next" key in the hash, given the previous one as the sole argument.
1038 Returns undef if there are no more keys to be fetched.
1040 $key = $db->next_key($key);
1044 Here are some examples of using hashes:
1046 my $db = DBM::Deep->new( "foo.db" );
1048 $db->put("foo", "bar");
1049 print "foo: " . $db->get("foo") . "\n";
1051 $db->put("baz", {}); # new child hash ref
1052 $db->get("baz")->put("buz", "biz");
1053 print "buz: " . $db->get("baz")->get("buz") . "\n";
1055 my $key = $db->first_key();
1057 print "$key: " . $db->get($key) . "\n";
1058 $key = $db->next_key($key);
1061 if ($db->exists("foo")) { $db->delete("foo"); }
1065 For arrays, DBM::Deep supports all the common methods described above, and the
1066 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1067 C<unshift()> and C<splice()>.
1073 Returns the number of elements in the array. Takes no arguments.
1075 my $len = $db->length();
1079 Adds one or more elements onto the end of the array. Accepts scalars, hash
1080 refs or array refs. No return value.
1082 $db->push("foo", "bar", {});
1086 Fetches the last element in the array, and deletes it. Takes no arguments.
1087 Returns undef if array is empty. Returns the element value.
1089 my $elem = $db->pop();
1093 Fetches the first element in the array, deletes it, then shifts all the
1094 remaining elements over to take up the space. Returns the element value. This
1095 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1098 my $elem = $db->shift();
1102 Inserts one or more elements onto the beginning of the array, shifting all
1103 existing elements over to make room. Accepts scalars, hash refs or array refs.
1104 No return value. This method is not recommended with large arrays -- see
1105 <LARGE ARRAYS> below for details.
1107 $db->unshift("foo", "bar", {});
1111 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1112 -f splice> for usage -- it is too complicated to document here. This method is
1113 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1117 Here are some examples of using arrays:
1119 my $db = DBM::Deep->new(
1121 type => DBM::Deep->TYPE_ARRAY
1124 $db->push("bar", "baz");
1125 $db->unshift("foo");
1128 my $len = $db->length();
1129 print "length: $len\n"; # 4
1131 for (my $k=0; $k<$len; $k++) {
1132 print "$k: " . $db->get($k) . "\n";
1135 $db->splice(1, 2, "biz", "baf");
1137 while (my $elem = shift @$db) {
1138 print "shifted: $elem\n";
1143 Enable automatic file locking by passing a true value to the C<locking>
1144 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1146 my $db = DBM::Deep->new(
1151 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1152 mode for writes, and shared mode for reads. This is required if you have
1153 multiple processes accessing the same database file, to avoid file corruption.
1154 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1155 NFS> below for more.
1157 =head2 EXPLICIT LOCKING
1159 You can explicitly lock a database, so it remains locked for multiple
1160 transactions. This is done by calling the C<lock()> method, and passing an
1161 optional lock mode argument (defaults to exclusive mode). This is particularly
1162 useful for things like counters, where the current value needs to be fetched,
1163 then incremented, then stored again.
1166 my $counter = $db->get("counter");
1168 $db->put("counter", $counter);
1177 You can pass C<lock()> an optional argument, which specifies which mode to use
1178 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1179 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1180 same as the constants defined in Perl's C<Fcntl> module.
1182 $db->lock( DBM::Deep->LOCK_SH );
1186 =head1 IMPORTING/EXPORTING
1188 You can import existing complex structures by calling the C<import()> method,
1189 and export an entire database into an in-memory structure using the C<export()>
1190 method. Both are examined here.
1194 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1195 walking the structure and adding keys/elements to the database as you go,
1196 simply pass a reference to the C<import()> method. This recursively adds
1197 everything to an existing DBM::Deep object for you. Here is an example:
1202 array1 => [ "elem0", "elem1", "elem2" ],
1204 subkey1 => "subvalue1",
1205 subkey2 => "subvalue2"
1209 my $db = DBM::Deep->new( "foo.db" );
1210 $db->import( $struct );
1212 print $db->{key1} . "\n"; # prints "value1"
1214 This recursively imports the entire C<$struct> object into C<$db>, including
1215 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1216 keys are merged with the existing ones, replacing if they already exist.
1217 The C<import()> method can be called on any database level (not just the base
1218 level), and works with both hash and array DB types.
1220 B<Note:> Make sure your existing structure has no circular references in it.
1221 These will cause an infinite loop when importing.
1225 Calling the C<export()> method on an existing DBM::Deep object will return
1226 a reference to a new in-memory copy of the database. The export is done
1227 recursively, so all nested hashes/arrays are all exported to standard Perl
1228 objects. Here is an example:
1230 my $db = DBM::Deep->new( "foo.db" );
1232 $db->{key1} = "value1";
1233 $db->{key2} = "value2";
1235 $db->{hash1}->{subkey1} = "subvalue1";
1236 $db->{hash1}->{subkey2} = "subvalue2";
1238 my $struct = $db->export();
1240 print $struct->{key1} . "\n"; # prints "value1"
1242 This makes a complete copy of the database in memory, and returns a reference
1243 to it. The C<export()> method can be called on any database level (not just
1244 the base level), and works with both hash and array DB types. Be careful of
1245 large databases -- you can store a lot more data in a DBM::Deep object than an
1246 in-memory Perl structure.
1248 B<Note:> Make sure your database has no circular references in it.
1249 These will cause an infinite loop when exporting.
1253 DBM::Deep has a number of hooks where you can specify your own Perl function
1254 to perform filtering on incoming or outgoing data. This is a perfect
1255 way to extend the engine, and implement things like real-time compression or
1256 encryption. Filtering applies to the base DB level, and all child hashes /
1257 arrays. Filter hooks can be specified when your DBM::Deep object is first
1258 constructed, or by calling the C<set_filter()> method at any time. There are
1259 four available filter hooks, described below:
1263 =item * filter_store_key
1265 This filter is called whenever a hash key is stored. It
1266 is passed the incoming key, and expected to return a transformed key.
1268 =item * filter_store_value
1270 This filter is called whenever a hash key or array element is stored. It
1271 is passed the incoming value, and expected to return a transformed value.
1273 =item * filter_fetch_key
1275 This filter is called whenever a hash key is fetched (i.e. via
1276 C<first_key()> or C<next_key()>). It is passed the transformed key,
1277 and expected to return the plain key.
1279 =item * filter_fetch_value
1281 This filter is called whenever a hash key or array element is fetched.
1282 It is passed the transformed value, and expected to return the plain value.
1286 Here are the two ways to setup a filter hook:
1288 my $db = DBM::Deep->new(
1290 filter_store_value => \&my_filter_store,
1291 filter_fetch_value => \&my_filter_fetch
1296 $db->set_filter( "filter_store_value", \&my_filter_store );
1297 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1299 Your filter function will be called only when dealing with SCALAR keys or
1300 values. When nested hashes and arrays are being stored/fetched, filtering
1301 is bypassed. Filters are called as static functions, passed a single SCALAR
1302 argument, and expected to return a single SCALAR value. If you want to
1303 remove a filter, set the function reference to C<undef>:
1305 $db->set_filter( "filter_store_value", undef );
1307 =head2 REAL-TIME ENCRYPTION EXAMPLE
1309 Here is a working example that uses the I<Crypt::Blowfish> module to
1310 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1311 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1312 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1315 use Crypt::Blowfish;
1318 my $cipher = Crypt::CBC->new({
1319 'key' => 'my secret key',
1320 'cipher' => 'Blowfish',
1322 'regenerate_key' => 0,
1323 'padding' => 'space',
1327 my $db = DBM::Deep->new(
1328 file => "foo-encrypt.db",
1329 filter_store_key => \&my_encrypt,
1330 filter_store_value => \&my_encrypt,
1331 filter_fetch_key => \&my_decrypt,
1332 filter_fetch_value => \&my_decrypt,
1335 $db->{key1} = "value1";
1336 $db->{key2} = "value2";
1337 print "key1: " . $db->{key1} . "\n";
1338 print "key2: " . $db->{key2} . "\n";
1344 return $cipher->encrypt( $_[0] );
1347 return $cipher->decrypt( $_[0] );
1350 =head2 REAL-TIME COMPRESSION EXAMPLE
1352 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1353 compression / decompression of keys & values with DBM::Deep Filters.
1354 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1355 more on I<Compress::Zlib>.
1360 my $db = DBM::Deep->new(
1361 file => "foo-compress.db",
1362 filter_store_key => \&my_compress,
1363 filter_store_value => \&my_compress,
1364 filter_fetch_key => \&my_decompress,
1365 filter_fetch_value => \&my_decompress,
1368 $db->{key1} = "value1";
1369 $db->{key2} = "value2";
1370 print "key1: " . $db->{key1} . "\n";
1371 print "key2: " . $db->{key2} . "\n";
1377 return Compress::Zlib::memGzip( $_[0] ) ;
1380 return Compress::Zlib::memGunzip( $_[0] ) ;
1383 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1384 actually numerical index numbers, and are not filtered.
1386 =head1 ERROR HANDLING
1388 Most DBM::Deep methods return a true value for success, and call die() on
1389 failure. You can wrap calls in an eval block to catch the die.
1391 my $db = DBM::Deep->new( "foo.db" ); # create hash
1392 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1394 print $@; # prints error message
1396 =head1 LARGEFILE SUPPORT
1398 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1399 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1400 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1401 by specifying the 'pack_size' parameter when constructing the file.
1404 filename => $filename,
1405 pack_size => 'large',
1408 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1409 instead of 32-bit longs. After setting these values your DB files have a
1410 theoretical maximum size of 16 XB (exabytes).
1412 You can also use C<pack_size =E<gt> 'small'> in order to use 16-bit file
1415 B<Note:> Changing these values will B<NOT> work for existing database files.
1416 Only change this for new files. Once the value has been set, it is stored in
1417 the file's header and cannot be changed for the life of the file. These
1418 parameters are per-file, meaning you can access 32-bit and 64-bit files, as
1421 B<Note:> We have not personally tested files larger than 2 GB -- all my
1422 systems have only a 32-bit Perl. However, I have received user reports that
1423 this does indeed work!
1425 =head1 LOW-LEVEL ACCESS
1427 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1428 you can call the C<_fh()> method, which returns the handle:
1430 my $fh = $db->_fh();
1432 This method can be called on the root level of the datbase, or any child
1433 hashes or arrays. All levels share a I<root> structure, which contains things
1434 like the filehandle, a reference counter, and all the options specified
1435 when you created the object. You can get access to this file object by
1436 calling the C<_fileobj()> method.
1438 my $file_obj = $db->_fileobj();
1440 This is useful for changing options after the object has already been created,
1441 such as enabling/disabling locking. You can also store your own temporary user
1442 data in this structure (be wary of name collision), which is then accessible from
1443 any child hash or array.
1445 =head1 CUSTOM DIGEST ALGORITHM
1447 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1448 keys. However you can override this, and use another algorithm (such as SHA-256)
1449 or even write your own. But please note that DBM::Deep currently expects zero
1450 collisions, so your algorithm has to be I<perfect>, so to speak. Collision
1451 detection may be introduced in a later version.
1453 You can specify a custom digest algorithm by passing it into the parameter
1454 list for new(), passing a reference to a subroutine as the 'digest' parameter,
1455 and the length of the algorithm's hashes (in bytes) as the 'hash_size'
1456 parameter. Here is a working example that uses a 256-bit hash from the
1457 I<Digest::SHA256> module. Please see
1458 L<http://search.cpan.org/search?module=Digest::SHA256> for more information.
1463 my $context = Digest::SHA256::new(256);
1465 my $db = DBM::Deep->new(
1466 filename => "foo-sha.db",
1467 digest => \&my_digest,
1471 $db->{key1} = "value1";
1472 $db->{key2} = "value2";
1473 print "key1: " . $db->{key1} . "\n";
1474 print "key2: " . $db->{key2} . "\n";
1480 return substr( $context->hash($_[0]), 0, 32 );
1483 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1484 of bytes you specify in the hash_size parameter (in this case 32).
1486 B<Note:> If you do choose to use a custom digest algorithm, you must set it
1487 every time you access this file. Otherwise, the default (MD5) will be used.
1489 =head1 CIRCULAR REFERENCES
1491 DBM::Deep has B<experimental> support for circular references. Meaning you
1492 can have a nested hash key or array element that points to a parent object.
1493 This relationship is stored in the DB file, and is preserved between sessions.
1496 my $db = DBM::Deep->new( "foo.db" );
1499 $db->{circle} = $db; # ref to self
1501 print $db->{foo} . "\n"; # prints "bar"
1502 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1504 B<Note>: Passing the object to a function that recursively walks the
1505 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1506 C<export()> methods) will result in an infinite loop. This will be fixed in
1509 =head1 CAVEATS / ISSUES / BUGS
1511 This section describes all the known issues with DBM::Deep. It you have found
1512 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1514 =head2 UNUSED SPACE RECOVERY
1516 One major caveat with DBM::Deep is that space occupied by existing keys and
1517 values is not recovered when they are deleted. Meaning if you keep deleting
1518 and adding new keys, your file will continuously grow. I am working on this,
1519 but in the meantime you can call the built-in C<optimize()> method from time to
1520 time (perhaps in a crontab or something) to recover all your unused space.
1522 $db->optimize(); # returns true on success
1524 This rebuilds the ENTIRE database into a new file, then moves it on top of
1525 the original. The new file will have no unused space, thus it will take up as
1526 little disk space as possible. Please note that this operation can take
1527 a long time for large files, and you need enough disk space to temporarily hold
1528 2 copies of your DB file. The temporary file is created in the same directory
1529 as the original, named with a ".tmp" extension, and is deleted when the
1530 operation completes. Oh, and if locking is enabled, the DB is automatically
1531 locked for the entire duration of the copy.
1533 B<WARNING:> Only call optimize() on the top-level node of the database, and
1534 make sure there are no child references lying around. DBM::Deep keeps a reference
1535 counter, and if it is greater than 1, optimize() will abort and return undef.
1539 (The reasons given assume a high level of Perl understanding, specifically of
1540 references. You can safely skip this section.)
1542 Currently, the only references supported are HASH and ARRAY. The other reference
1543 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1549 These are things like filehandles and other sockets. They can't be supported
1550 because it's completely unclear how DBM::Deep should serialize them.
1552 =item * SCALAR / REF
1554 The discussion here refers to the following type of example:
1561 # In some other process ...
1563 my $val = ${ $db->{key1} };
1565 is( $val, 50, "What actually gets stored in the DB file?" );
1567 The problem is one of synchronization. When the variable being referred to
1568 changes value, the reference isn't notified. This means that the new value won't
1569 be stored in the datafile for other processes to read. There is no TIEREF.
1571 It is theoretically possible to store references to values already within a
1572 DBM::Deep object because everything already is synchronized, but the change to
1573 the internals would be quite large. Specifically, DBM::Deep would have to tie
1574 every single value that is stored. This would bloat the RAM footprint of
1575 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1576 all to support a feature that has never been requested.
1580 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1581 mechanism for serializing coderefs, including saving off all closure state.
1582 However, just as for SCALAR and REF, that closure state may change without
1583 notifying the DBM::Deep object storing the reference.
1587 =head2 FILE CORRUPTION
1589 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1590 for a 32-bit signature when opened, but other corruption in files can cause
1591 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1592 stuck in an infinite loop depending on the level of corruption. File write
1593 operations are not checked for failure (for speed), so if you happen to run
1594 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1595 be addressed in a later version of DBM::Deep.
1599 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1600 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1601 about setting up your NFS server with a locking daemon, then using lockf() to
1602 lock your files, but your mileage may vary there as well. From what I
1603 understand, there is no real way to do it. However, if you need access to the
1604 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1605 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1607 =head2 COPYING OBJECTS
1609 Beware of copying tied objects in Perl. Very strange things can happen.
1610 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1611 returns a new, blessed, tied hash or array to the same level in the DB.
1613 my $copy = $db->clone();
1615 B<Note>: Since clone() here is cloning the object, not the database location, any
1616 modifications to either $db or $copy will be visible in both.
1620 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1621 These functions cause every element in the array to move, which can be murder
1622 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1623 a different location. This will be addressed in the forthcoming version 1.00.
1625 =head2 WRITEONLY FILES
1627 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1628 writeonly mode. STORE will verify that the filehandle is writable. However, there
1629 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1630 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1634 This section discusses DBM::Deep's speed and memory usage.
1638 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1639 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1640 multi-level hash/array support, and cross-platform FTPable files. Even so,
1641 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1642 with huge databases. Here is some test data:
1644 Adding 1,000,000 keys to new DB file...
1646 At 100 keys, avg. speed is 2,703 keys/sec
1647 At 200 keys, avg. speed is 2,642 keys/sec
1648 At 300 keys, avg. speed is 2,598 keys/sec
1649 At 400 keys, avg. speed is 2,578 keys/sec
1650 At 500 keys, avg. speed is 2,722 keys/sec
1651 At 600 keys, avg. speed is 2,628 keys/sec
1652 At 700 keys, avg. speed is 2,700 keys/sec
1653 At 800 keys, avg. speed is 2,607 keys/sec
1654 At 900 keys, avg. speed is 2,190 keys/sec
1655 At 1,000 keys, avg. speed is 2,570 keys/sec
1656 At 2,000 keys, avg. speed is 2,417 keys/sec
1657 At 3,000 keys, avg. speed is 1,982 keys/sec
1658 At 4,000 keys, avg. speed is 1,568 keys/sec
1659 At 5,000 keys, avg. speed is 1,533 keys/sec
1660 At 6,000 keys, avg. speed is 1,787 keys/sec
1661 At 7,000 keys, avg. speed is 1,977 keys/sec
1662 At 8,000 keys, avg. speed is 2,028 keys/sec
1663 At 9,000 keys, avg. speed is 2,077 keys/sec
1664 At 10,000 keys, avg. speed is 2,031 keys/sec
1665 At 20,000 keys, avg. speed is 1,970 keys/sec
1666 At 30,000 keys, avg. speed is 2,050 keys/sec
1667 At 40,000 keys, avg. speed is 2,073 keys/sec
1668 At 50,000 keys, avg. speed is 1,973 keys/sec
1669 At 60,000 keys, avg. speed is 1,914 keys/sec
1670 At 70,000 keys, avg. speed is 2,091 keys/sec
1671 At 80,000 keys, avg. speed is 2,103 keys/sec
1672 At 90,000 keys, avg. speed is 1,886 keys/sec
1673 At 100,000 keys, avg. speed is 1,970 keys/sec
1674 At 200,000 keys, avg. speed is 2,053 keys/sec
1675 At 300,000 keys, avg. speed is 1,697 keys/sec
1676 At 400,000 keys, avg. speed is 1,838 keys/sec
1677 At 500,000 keys, avg. speed is 1,941 keys/sec
1678 At 600,000 keys, avg. speed is 1,930 keys/sec
1679 At 700,000 keys, avg. speed is 1,735 keys/sec
1680 At 800,000 keys, avg. speed is 1,795 keys/sec
1681 At 900,000 keys, avg. speed is 1,221 keys/sec
1682 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1684 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1685 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1686 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1687 Run time was 12 min 3 sec.
1691 One of the great things about DBM::Deep is that it uses very little memory.
1692 Even with huge databases (1,000,000+ keys) you will not see much increased
1693 memory on your process. DBM::Deep relies solely on the filesystem for storing
1694 and fetching data. Here is output from I</usr/bin/top> before even opening a
1697 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1698 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1700 Basically the process is taking 2,716K of memory. And here is the same
1701 process after storing and fetching 1,000,000 keys:
1703 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1704 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1706 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1707 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1709 =head1 DB FILE FORMAT
1711 In case you were interested in the underlying DB file format, it is documented
1712 here in this section. You don't need to know this to use the module, it's just
1713 included for reference.
1717 DBM::Deep files always start with a 32-bit signature to identify the file type.
1718 This is at offset 0. The signature is "DPDB" in network byte order. This is
1719 checked for when the file is opened and an error will be thrown if it's not found.
1723 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1724 has a standard header containing the type of data, the length of data, and then
1725 the data itself. The type is a single character (1 byte), the length is a
1726 32-bit unsigned long in network byte order, and the data is, well, the data.
1727 Here is how it unfolds:
1731 Immediately after the 32-bit file signature is the I<Master Index> record.
1732 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1733 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1734 depending on how the DBM::Deep object was constructed.
1736 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1737 number). The first 8-bit char of the MD5 signature is the offset into the
1738 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1739 index element is a file offset of the next tag for the key/element in question,
1740 which is usually a I<Bucket List> tag (see below).
1742 The next tag I<could> be another index, depending on how many keys/elements
1743 exist. See L<RE-INDEXING> below for details.
1747 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1748 file offsets to where the actual data is stored. It starts with a standard
1749 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1750 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1751 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1752 When the list fills up, a I<Re-Index> operation is performed (See
1753 L<RE-INDEXING> below).
1757 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1758 index/value pair (in array mode). It starts with a standard tag header with
1759 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1760 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1761 header. The size reported in the tag header is only for the value, but then,
1762 just after the value is another size (32-bit unsigned long) and then the plain
1763 key itself. Since the value is likely to be fetched more often than the plain
1764 key, I figured it would be I<slightly> faster to store the value first.
1766 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1767 record for the nested structure, where the process begins all over again.
1771 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1772 exhausted. Then, when another key/element comes in, the list is converted to a
1773 new index record. However, this index will look at the next char in the MD5
1774 hash, and arrange new Bucket List pointers accordingly. This process is called
1775 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1776 17 (16 + new one) keys/elements are removed from the old Bucket List and
1777 inserted into the new index. Several new Bucket Lists are created in the
1778 process, as a new MD5 char from the key is being examined (it is unlikely that
1779 the keys will all share the same next char of their MD5s).
1781 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1782 when the Bucket Lists will turn into indexes, but the first round tends to
1783 happen right around 4,000 keys. You will see a I<slight> decrease in
1784 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1785 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1786 right around 900,000 keys. This process can continue nearly indefinitely --
1787 right up until the point the I<MD5> signatures start colliding with each other,
1788 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1789 getting struck by lightning while you are walking to cash in your tickets.
1790 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1791 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1792 this is 340 unodecillion, but don't quote me).
1796 When a new key/element is stored, the key (or index number) is first run through
1797 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1798 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1799 for the first char of the signature (in this case I<b0>). If it does not exist,
1800 a new I<Bucket List> is created for our key (and the next 15 future keys that
1801 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1802 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1803 this point, unless we are replacing an existing I<Bucket>), where the actual
1804 data will be stored.
1808 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1809 (or index number), then walking along the indexes. If there are enough
1810 keys/elements in this DB level, there might be nested indexes, each linked to
1811 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1812 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1813 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1814 plain key are stored.
1816 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1817 methods. In this process the indexes are walked systematically, and each key
1818 fetched in increasing MD5 order (which is why it appears random). Once the
1819 I<Bucket> is found, the value is skipped and the plain key returned instead.
1820 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1821 alphabetically sorted. This only happens on an index-level -- as soon as the
1822 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1823 so it's pretty much undefined how the keys will come out -- just like Perl's
1826 =head1 CODE COVERAGE
1828 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1829 B<Devel::Cover> report on this module's test suite.
1831 ----------------------------------- ------ ------ ------ ------ ------ ------
1832 File stmt bran cond sub time total
1833 ----------------------------------- ------ ------ ------ ------ ------ ------
1834 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1835 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1836 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1837 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1838 Total 97.9 85.9 79.7 100.0 100.0 94.3
1839 ----------------------------------- ------ ------ ------ ------ ------ ------
1841 =head1 MORE INFORMATION
1843 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1844 or send email to L<DBM-Deep@googlegroups.com>.
1848 Joseph Huckaby, L<jhuckaby@cpan.org>
1850 Rob Kinyon, L<rkinyon@cpan.org>
1852 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1856 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1857 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1861 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1862 This is free software, you may use it and distribute it under the
1863 same terms as Perl itself.