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 $self->_fileobj->commit_transaction;
347 my $self = $_[0]->_get_self;
348 return $self->{fileobj};
352 my $self = $_[0]->_get_self;
353 return $self->{type};
357 my $self = $_[0]->_get_self;
358 return $self->{base_offset};
362 my $self = $_[0]->_get_self;
363 return $self->_fileobj->{fh};
371 die "DBM::Deep: $_[1]\n";
376 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
381 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
388 #XXX This if() is redundant
389 if ( my $parent = $self->{parent} ) {
391 while ( $parent->{parent} ) {
393 $parent->_type eq TYPE_HASH
394 ? "\{$child->{parent_key}\}"
395 : "\[$child->{parent_key}\]"
399 $parent = $parent->{parent};
402 $base = "\$db->get( '$child->{parent_key}' )->" . $base;
405 $base = "\$db->get( '$child->{parent_key}' )";
413 # Store single hash key/value or array element in database.
415 my $self = shift->_get_self;
416 my ($key, $value, $orig_key) = @_;
419 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
420 $self->_throw_error( 'Cannot write to a readonly filehandle' );
423 #XXX The second condition needs to disappear
424 if ( defined $orig_key && !( $self->_type eq TYPE_ARRAY && $orig_key eq 'length') ) {
427 my $r = Scalar::Util::reftype( $value ) || '';
428 if ( $r eq 'HASH' ) {
431 elsif ( $r eq 'ARRAY' ) {
434 elsif ( defined $value ) {
441 if ( my $c = Scalar::Util::blessed( $value ) ) {
442 $rhs = "bless $rhs, '$c'";
445 my $lhs = $self->_find_parent;
447 if ( $self->_type eq TYPE_HASH ) {
448 $lhs .= "->\{$orig_key\}";
451 $lhs .= "->\[$orig_key\]";
457 $lhs = "\$db->put('$orig_key',$rhs);";
460 $self->_fileobj->audit($lhs);
464 # Request exclusive lock for writing
466 $self->lock( LOCK_EX );
468 my $md5 = $self->{engine}{digest}->($key);
470 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5, { create => 1 } );
472 # User may be storing a hash, in which case we do not want it run
473 # through the filtering system
474 if ( !ref($value) && $self->_fileobj->{filter_store_value} ) {
475 $value = $self->_fileobj->{filter_store_value}->( $value );
479 # Add key/value to bucket list
481 my $result = $self->{engine}->add_bucket( $tag, $md5, $key, $value, undef, $orig_key );
490 # Fetch single value or element given plain key or array index
492 my $self = shift->_get_self;
493 my ($key, $orig_key) = @_;
495 my $md5 = $self->{engine}{digest}->($key);
498 # Request shared lock for reading
500 $self->lock( LOCK_SH );
502 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );#, { create => 1 } );
503 #XXX This needs to autovivify
510 # Get value from bucket list
512 my $result = $self->{engine}->get_bucket_value( $tag, $md5, $orig_key );
516 # Filters only apply to scalar values, so the ref check is making
517 # sure the fetched bucket is a scalar, not a child hash or array.
518 return ($result && !ref($result) && $self->_fileobj->{filter_fetch_value})
519 ? $self->_fileobj->{filter_fetch_value}->($result)
525 # Delete single key/value pair or element given plain key or array index
527 my $self = shift->_get_self;
528 my ($key, $orig_key) = @_;
530 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
531 $self->_throw_error( 'Cannot write to a readonly filehandle' );
534 if ( defined $orig_key ) {
535 my $lhs = $self->_find_parent;
537 $self->_fileobj->audit( "delete $lhs;" );
540 $self->_fileobj->audit( "\$db->delete('$orig_key');" );
545 # Request exclusive lock for writing
547 $self->lock( LOCK_EX );
549 my $md5 = $self->{engine}{digest}->($key);
551 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
560 my $value = $self->{engine}->get_bucket_value( $tag, $md5 );
562 if (defined $value && !ref($value) && $self->_fileobj->{filter_fetch_value}) {
563 $value = $self->_fileobj->{filter_fetch_value}->($value);
566 my $result = $self->{engine}->delete_bucket( $tag, $md5, $orig_key );
569 # If this object is an array and the key deleted was on the end of the stack,
570 # decrement the length variable.
580 # Check if a single key or element exists given plain key or array index
582 my $self = shift->_get_self;
585 my $md5 = $self->{engine}{digest}->($key);
588 # Request shared lock for reading
590 $self->lock( LOCK_SH );
592 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
597 # For some reason, the built-in exists() function returns '' for false
603 # Check if bucket exists and return 1 or ''
605 my $result = $self->{engine}->bucket_exists( $tag, $md5 ) || '';
614 # Clear all keys from hash, or all elements from array.
616 my $self = shift->_get_self;
618 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
619 $self->_throw_error( 'Cannot write to a readonly filehandle' );
623 my $lhs = $self->_find_parent;
625 if ( $self->_type eq TYPE_HASH ) {
626 $lhs = '%{' . $lhs . '}';
629 $lhs = '@{' . $lhs . '}';
632 $self->_fileobj->audit( "$lhs = ();" );
636 # Request exclusive lock for writing
638 $self->lock( LOCK_EX );
640 #XXX This needs updating to use _release_space
641 $self->{engine}->write_tag(
642 $self->_base_offset, $self->_type,
643 chr(0)x$self->{engine}{index_size},
652 # Public method aliases
654 sub put { (shift)->STORE( @_ ) }
655 sub store { (shift)->STORE( @_ ) }
656 sub get { (shift)->FETCH( @_ ) }
657 sub fetch { (shift)->FETCH( @_ ) }
658 sub delete { (shift)->DELETE( @_ ) }
659 sub exists { (shift)->EXISTS( @_ ) }
660 sub clear { (shift)->CLEAR( @_ ) }
667 DBM::Deep - A pure perl multi-level hash/array DBM
672 my $db = DBM::Deep->new( "foo.db" );
674 $db->{key} = 'value'; # tie() style
677 $db->put('key' => 'value'); # OO style
678 print $db->get('key');
680 # true multi-level support
681 $db->{my_complex} = [
682 'hello', { perl => 'rules' },
688 A unique flat-file database module, written in pure perl. True
689 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
690 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
691 handle millions of keys and unlimited hash levels without significant
692 slow-down. Written from the ground-up in pure perl -- this is NOT a
693 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
694 Mac OS X and Windows.
696 =head1 VERSION DIFFERENCES
698 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
699 before. While attempts have been made to be backwards compatible, no guarantees.
703 Hopefully you are using Perl's excellent CPAN module, which will download
704 and install the module for you. If not, get the tarball, and run these
716 Construction can be done OO-style (which is the recommended way), or using
717 Perl's tie() function. Both are examined here.
719 =head2 OO CONSTRUCTION
721 The recommended way to construct a DBM::Deep object is to use the new()
722 method, which gets you a blessed, tied hash or array reference.
724 my $db = DBM::Deep->new( "foo.db" );
726 This opens a new database handle, mapped to the file "foo.db". If this
727 file does not exist, it will automatically be created. DB files are
728 opened in "r+" (read/write) mode, and the type of object returned is a
729 hash, unless otherwise specified (see L<OPTIONS> below).
731 You can pass a number of options to the constructor to specify things like
732 locking, autoflush, etc. This is done by passing an inline hash:
734 my $db = DBM::Deep->new(
740 Notice that the filename is now specified I<inside> the hash with
741 the "file" parameter, as opposed to being the sole argument to the
742 constructor. This is required if any options are specified.
743 See L<OPTIONS> below for the complete list.
747 You can also start with an array instead of a hash. For this, you must
748 specify the C<type> parameter:
750 my $db = DBM::Deep->new(
752 type => DBM::Deep->TYPE_ARRAY
755 B<Note:> Specifing the C<type> parameter only takes effect when beginning
756 a new DB file. If you create a DBM::Deep object with an existing file, the
757 C<type> will be loaded from the file header, and an error will be thrown if
758 the wrong type is passed in.
760 =head2 TIE CONSTRUCTION
762 Alternately, you can create a DBM::Deep handle by using Perl's built-in
763 tie() function. The object returned from tie() can be used to call methods,
764 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
765 file (as expected with most tie'd objects).
768 my $db = tie %hash, "DBM::Deep", "foo.db";
771 my $db = tie @array, "DBM::Deep", "bar.db";
773 As with the OO constructor, you can replace the DB filename parameter with
774 a hash containing one or more options (see L<OPTIONS> just below for the
777 tie %hash, "DBM::Deep", {
785 There are a number of options that can be passed in when constructing your
786 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
792 Filename of the DB file to link the handle to. You can pass a full absolute
793 filesystem path, partial path, or a plain filename if the file is in the
794 current working directory. This is a required parameter (though q.v. fh).
798 If you want, you can pass in the fh instead of the file. This is most useful for doing
801 my $db = DBM::Deep->new( { fh => \*DATA } );
803 You are responsible for making sure that the fh has been opened appropriately for your
804 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
805 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
806 needs to read from the fh.
810 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
811 not need to set this. However, it's there if you want it.
813 If you pass in fh and do not set this, it will be set appropriately.
817 This parameter specifies what type of object to create, a hash or array. Use
818 one of these two constants:
822 =item * C<DBM::Deep-E<gt>TYPE_HASH>
824 =item * C<DBM::Deep-E<gt>TYPE_ARRAY>.
828 This only takes effect when beginning a new file. This is an optional
829 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
833 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
834 function to lock the database in exclusive mode for writes, and shared mode for
835 reads. Pass any true value to enable. This affects the base DB handle I<and
836 any child hashes or arrays> that use the same DB file. This is an optional
837 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
841 Specifies whether autoflush is to be enabled on the underlying filehandle.
842 This obviously slows down write operations, but is required if you may have
843 multiple processes accessing the same DB file (also consider enable I<locking>).
844 Pass any true value to enable. This is an optional parameter, and defaults to 0
849 If I<autobless> mode is enabled, DBM::Deep will preserve the class something
850 is blessed into, and restores it when fetched. This is an optional parameter, and defaults to 1 (enabled).
852 B<Note:> If you use the OO-interface, you will not be able to call any methods
853 of DBM::Deep on the blessed item. This is considered to be a feature.
857 See L</FILTERS> below.
863 With DBM::Deep you can access your databases using Perl's standard hash/array
864 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
865 treat them as such. DBM::Deep will intercept all reads/writes and direct them
866 to the right place -- the DB file. This has nothing to do with the
867 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
868 using regular hashes and arrays, rather than calling functions like C<get()>
869 and C<put()> (although those work too). It is entirely up to you how to want
870 to access your databases.
874 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
875 or even nested hashes (or arrays) using standard Perl syntax:
877 my $db = DBM::Deep->new( "foo.db" );
879 $db->{mykey} = "myvalue";
881 $db->{myhash}->{subkey} = "subvalue";
883 print $db->{myhash}->{subkey} . "\n";
885 You can even step through hash keys using the normal Perl C<keys()> function:
887 foreach my $key (keys %$db) {
888 print "$key: " . $db->{$key} . "\n";
891 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
892 pushes them onto an array, all before the loop even begins. If you have an
893 extra large hash, this may exhaust Perl's memory. Instead, consider using
894 Perl's C<each()> function, which pulls keys/values one at a time, using very
897 while (my ($key, $value) = each %$db) {
898 print "$key: $value\n";
901 Please note that when using C<each()>, you should always pass a direct
902 hash reference, not a lookup. Meaning, you should B<never> do this:
905 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
907 This causes an infinite loop, because for each iteration, Perl is calling
908 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
909 it effectively keeps returning the first key over and over again. Instead,
910 assign a temporary variable to C<$db->{foo}>, then pass that to each().
914 As with hashes, you can treat any DBM::Deep object like a normal Perl array
915 reference. This includes inserting, removing and manipulating elements,
916 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
917 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
918 or simply be a nested array reference inside a hash. Example:
920 my $db = DBM::Deep->new(
921 file => "foo-array.db",
922 type => DBM::Deep->TYPE_ARRAY
926 push @$db, "bar", "baz";
929 my $last_elem = pop @$db; # baz
930 my $first_elem = shift @$db; # bah
931 my $second_elem = $db->[1]; # bar
933 my $num_elements = scalar @$db;
937 In addition to the I<tie()> interface, you can also use a standard OO interface
938 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
939 array) has its own methods, but both types share the following common methods:
940 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
944 =item * new() / clone()
946 These are the constructor and copy-functions.
948 =item * put() / store()
950 Stores a new hash key/value pair, or sets an array element value. Takes two
951 arguments, the hash key or array index, and the new value. The value can be
952 a scalar, hash ref or array ref. Returns true on success, false on failure.
954 $db->put("foo", "bar"); # for hashes
955 $db->put(1, "bar"); # for arrays
957 =item * get() / fetch()
959 Fetches the value of a hash key or array element. Takes one argument: the hash
960 key or array index. Returns a scalar, hash ref or array ref, depending on the
963 my $value = $db->get("foo"); # for hashes
964 my $value = $db->get(1); # for arrays
968 Checks if a hash key or array index exists. Takes one argument: the hash key
969 or array index. Returns true if it exists, false if not.
971 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
972 if ($db->exists(1)) { print "yay!\n"; } # for arrays
976 Deletes one hash key/value pair or array element. Takes one argument: the hash
977 key or array index. Returns true on success, false if not found. For arrays,
978 the remaining elements located after the deleted element are NOT moved over.
979 The deleted element is essentially just undefined, which is exactly how Perl's
980 internal arrays work. Please note that the space occupied by the deleted
981 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
982 below for details and workarounds.
984 $db->delete("foo"); # for hashes
985 $db->delete(1); # for arrays
989 Deletes B<all> hash keys or array elements. Takes no arguments. No return
990 value. Please note that the space occupied by the deleted keys/values or
991 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
992 details and workarounds.
994 $db->clear(); # hashes or arrays
996 =item * lock() / unlock()
1002 Recover lost disk space.
1004 =item * import() / export()
1006 Data going in and out.
1012 For hashes, DBM::Deep supports all the common methods described above, and the
1013 following additional methods: C<first_key()> and C<next_key()>.
1019 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1020 fetched in an undefined order (which appears random). Takes no arguments,
1021 returns the key as a scalar value.
1023 my $key = $db->first_key();
1027 Returns the "next" key in the hash, given the previous one as the sole argument.
1028 Returns undef if there are no more keys to be fetched.
1030 $key = $db->next_key($key);
1034 Here are some examples of using hashes:
1036 my $db = DBM::Deep->new( "foo.db" );
1038 $db->put("foo", "bar");
1039 print "foo: " . $db->get("foo") . "\n";
1041 $db->put("baz", {}); # new child hash ref
1042 $db->get("baz")->put("buz", "biz");
1043 print "buz: " . $db->get("baz")->get("buz") . "\n";
1045 my $key = $db->first_key();
1047 print "$key: " . $db->get($key) . "\n";
1048 $key = $db->next_key($key);
1051 if ($db->exists("foo")) { $db->delete("foo"); }
1055 For arrays, DBM::Deep supports all the common methods described above, and the
1056 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1057 C<unshift()> and C<splice()>.
1063 Returns the number of elements in the array. Takes no arguments.
1065 my $len = $db->length();
1069 Adds one or more elements onto the end of the array. Accepts scalars, hash
1070 refs or array refs. No return value.
1072 $db->push("foo", "bar", {});
1076 Fetches the last element in the array, and deletes it. Takes no arguments.
1077 Returns undef if array is empty. Returns the element value.
1079 my $elem = $db->pop();
1083 Fetches the first element in the array, deletes it, then shifts all the
1084 remaining elements over to take up the space. Returns the element value. This
1085 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1088 my $elem = $db->shift();
1092 Inserts one or more elements onto the beginning of the array, shifting all
1093 existing elements over to make room. Accepts scalars, hash refs or array refs.
1094 No return value. This method is not recommended with large arrays -- see
1095 <LARGE ARRAYS> below for details.
1097 $db->unshift("foo", "bar", {});
1101 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1102 -f splice> for usage -- it is too complicated to document here. This method is
1103 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1107 Here are some examples of using arrays:
1109 my $db = DBM::Deep->new(
1111 type => DBM::Deep->TYPE_ARRAY
1114 $db->push("bar", "baz");
1115 $db->unshift("foo");
1118 my $len = $db->length();
1119 print "length: $len\n"; # 4
1121 for (my $k=0; $k<$len; $k++) {
1122 print "$k: " . $db->get($k) . "\n";
1125 $db->splice(1, 2, "biz", "baf");
1127 while (my $elem = shift @$db) {
1128 print "shifted: $elem\n";
1133 Enable automatic file locking by passing a true value to the C<locking>
1134 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1136 my $db = DBM::Deep->new(
1141 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1142 mode for writes, and shared mode for reads. This is required if you have
1143 multiple processes accessing the same database file, to avoid file corruption.
1144 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1145 NFS> below for more.
1147 =head2 EXPLICIT LOCKING
1149 You can explicitly lock a database, so it remains locked for multiple
1150 transactions. This is done by calling the C<lock()> method, and passing an
1151 optional lock mode argument (defaults to exclusive mode). This is particularly
1152 useful for things like counters, where the current value needs to be fetched,
1153 then incremented, then stored again.
1156 my $counter = $db->get("counter");
1158 $db->put("counter", $counter);
1167 You can pass C<lock()> an optional argument, which specifies which mode to use
1168 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1169 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1170 same as the constants defined in Perl's C<Fcntl> module.
1172 $db->lock( DBM::Deep->LOCK_SH );
1176 =head1 IMPORTING/EXPORTING
1178 You can import existing complex structures by calling the C<import()> method,
1179 and export an entire database into an in-memory structure using the C<export()>
1180 method. Both are examined here.
1184 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1185 walking the structure and adding keys/elements to the database as you go,
1186 simply pass a reference to the C<import()> method. This recursively adds
1187 everything to an existing DBM::Deep object for you. Here is an example:
1192 array1 => [ "elem0", "elem1", "elem2" ],
1194 subkey1 => "subvalue1",
1195 subkey2 => "subvalue2"
1199 my $db = DBM::Deep->new( "foo.db" );
1200 $db->import( $struct );
1202 print $db->{key1} . "\n"; # prints "value1"
1204 This recursively imports the entire C<$struct> object into C<$db>, including
1205 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1206 keys are merged with the existing ones, replacing if they already exist.
1207 The C<import()> method can be called on any database level (not just the base
1208 level), and works with both hash and array DB types.
1210 B<Note:> Make sure your existing structure has no circular references in it.
1211 These will cause an infinite loop when importing.
1215 Calling the C<export()> method on an existing DBM::Deep object will return
1216 a reference to a new in-memory copy of the database. The export is done
1217 recursively, so all nested hashes/arrays are all exported to standard Perl
1218 objects. Here is an example:
1220 my $db = DBM::Deep->new( "foo.db" );
1222 $db->{key1} = "value1";
1223 $db->{key2} = "value2";
1225 $db->{hash1}->{subkey1} = "subvalue1";
1226 $db->{hash1}->{subkey2} = "subvalue2";
1228 my $struct = $db->export();
1230 print $struct->{key1} . "\n"; # prints "value1"
1232 This makes a complete copy of the database in memory, and returns a reference
1233 to it. The C<export()> method can be called on any database level (not just
1234 the base level), and works with both hash and array DB types. Be careful of
1235 large databases -- you can store a lot more data in a DBM::Deep object than an
1236 in-memory Perl structure.
1238 B<Note:> Make sure your database has no circular references in it.
1239 These will cause an infinite loop when exporting.
1243 DBM::Deep has a number of hooks where you can specify your own Perl function
1244 to perform filtering on incoming or outgoing data. This is a perfect
1245 way to extend the engine, and implement things like real-time compression or
1246 encryption. Filtering applies to the base DB level, and all child hashes /
1247 arrays. Filter hooks can be specified when your DBM::Deep object is first
1248 constructed, or by calling the C<set_filter()> method at any time. There are
1249 four available filter hooks, described below:
1253 =item * filter_store_key
1255 This filter is called whenever a hash key is stored. It
1256 is passed the incoming key, and expected to return a transformed key.
1258 =item * filter_store_value
1260 This filter is called whenever a hash key or array element is stored. It
1261 is passed the incoming value, and expected to return a transformed value.
1263 =item * filter_fetch_key
1265 This filter is called whenever a hash key is fetched (i.e. via
1266 C<first_key()> or C<next_key()>). It is passed the transformed key,
1267 and expected to return the plain key.
1269 =item * filter_fetch_value
1271 This filter is called whenever a hash key or array element is fetched.
1272 It is passed the transformed value, and expected to return the plain value.
1276 Here are the two ways to setup a filter hook:
1278 my $db = DBM::Deep->new(
1280 filter_store_value => \&my_filter_store,
1281 filter_fetch_value => \&my_filter_fetch
1286 $db->set_filter( "filter_store_value", \&my_filter_store );
1287 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1289 Your filter function will be called only when dealing with SCALAR keys or
1290 values. When nested hashes and arrays are being stored/fetched, filtering
1291 is bypassed. Filters are called as static functions, passed a single SCALAR
1292 argument, and expected to return a single SCALAR value. If you want to
1293 remove a filter, set the function reference to C<undef>:
1295 $db->set_filter( "filter_store_value", undef );
1297 =head2 REAL-TIME ENCRYPTION EXAMPLE
1299 Here is a working example that uses the I<Crypt::Blowfish> module to
1300 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1301 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1302 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1305 use Crypt::Blowfish;
1308 my $cipher = Crypt::CBC->new({
1309 'key' => 'my secret key',
1310 'cipher' => 'Blowfish',
1312 'regenerate_key' => 0,
1313 'padding' => 'space',
1317 my $db = DBM::Deep->new(
1318 file => "foo-encrypt.db",
1319 filter_store_key => \&my_encrypt,
1320 filter_store_value => \&my_encrypt,
1321 filter_fetch_key => \&my_decrypt,
1322 filter_fetch_value => \&my_decrypt,
1325 $db->{key1} = "value1";
1326 $db->{key2} = "value2";
1327 print "key1: " . $db->{key1} . "\n";
1328 print "key2: " . $db->{key2} . "\n";
1334 return $cipher->encrypt( $_[0] );
1337 return $cipher->decrypt( $_[0] );
1340 =head2 REAL-TIME COMPRESSION EXAMPLE
1342 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1343 compression / decompression of keys & values with DBM::Deep Filters.
1344 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1345 more on I<Compress::Zlib>.
1350 my $db = DBM::Deep->new(
1351 file => "foo-compress.db",
1352 filter_store_key => \&my_compress,
1353 filter_store_value => \&my_compress,
1354 filter_fetch_key => \&my_decompress,
1355 filter_fetch_value => \&my_decompress,
1358 $db->{key1} = "value1";
1359 $db->{key2} = "value2";
1360 print "key1: " . $db->{key1} . "\n";
1361 print "key2: " . $db->{key2} . "\n";
1367 return Compress::Zlib::memGzip( $_[0] ) ;
1370 return Compress::Zlib::memGunzip( $_[0] ) ;
1373 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1374 actually numerical index numbers, and are not filtered.
1376 =head1 ERROR HANDLING
1378 Most DBM::Deep methods return a true value for success, and call die() on
1379 failure. You can wrap calls in an eval block to catch the die.
1381 my $db = DBM::Deep->new( "foo.db" ); # create hash
1382 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1384 print $@; # prints error message
1386 =head1 LARGEFILE SUPPORT
1388 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1389 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1390 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1391 by specifying the 'pack_size' parameter when constructing the file.
1394 filename => $filename,
1395 pack_size => 'large',
1398 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1399 instead of 32-bit longs. After setting these values your DB files have a
1400 theoretical maximum size of 16 XB (exabytes).
1402 You can also use C<pack_size =E<gt> 'small'> in order to use 16-bit file
1405 B<Note:> Changing these values will B<NOT> work for existing database files.
1406 Only change this for new files. Once the value has been set, it is stored in
1407 the file's header and cannot be changed for the life of the file. These
1408 parameters are per-file, meaning you can access 32-bit and 64-bit files, as
1411 B<Note:> We have not personally tested files larger than 2 GB -- all my
1412 systems have only a 32-bit Perl. However, I have received user reports that
1413 this does indeed work!
1415 =head1 LOW-LEVEL ACCESS
1417 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1418 you can call the C<_fh()> method, which returns the handle:
1420 my $fh = $db->_fh();
1422 This method can be called on the root level of the datbase, or any child
1423 hashes or arrays. All levels share a I<root> structure, which contains things
1424 like the filehandle, a reference counter, and all the options specified
1425 when you created the object. You can get access to this file object by
1426 calling the C<_fileobj()> method.
1428 my $file_obj = $db->_fileobj();
1430 This is useful for changing options after the object has already been created,
1431 such as enabling/disabling locking. You can also store your own temporary user
1432 data in this structure (be wary of name collision), which is then accessible from
1433 any child hash or array.
1435 =head1 CUSTOM DIGEST ALGORITHM
1437 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1438 keys. However you can override this, and use another algorithm (such as SHA-256)
1439 or even write your own. But please note that DBM::Deep currently expects zero
1440 collisions, so your algorithm has to be I<perfect>, so to speak. Collision
1441 detection may be introduced in a later version.
1443 You can specify a custom digest algorithm by passing it into the parameter
1444 list for new(), passing a reference to a subroutine as the 'digest' parameter,
1445 and the length of the algorithm's hashes (in bytes) as the 'hash_size'
1446 parameter. Here is a working example that uses a 256-bit hash from the
1447 I<Digest::SHA256> module. Please see
1448 L<http://search.cpan.org/search?module=Digest::SHA256> for more information.
1453 my $context = Digest::SHA256::new(256);
1455 my $db = DBM::Deep->new(
1456 filename => "foo-sha.db",
1457 digest => \&my_digest,
1461 $db->{key1} = "value1";
1462 $db->{key2} = "value2";
1463 print "key1: " . $db->{key1} . "\n";
1464 print "key2: " . $db->{key2} . "\n";
1470 return substr( $context->hash($_[0]), 0, 32 );
1473 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1474 of bytes you specify in the hash_size parameter (in this case 32).
1476 B<Note:> If you do choose to use a custom digest algorithm, you must set it
1477 every time you access this file. Otherwise, the default (MD5) will be used.
1479 =head1 CIRCULAR REFERENCES
1481 DBM::Deep has B<experimental> support for circular references. Meaning you
1482 can have a nested hash key or array element that points to a parent object.
1483 This relationship is stored in the DB file, and is preserved between sessions.
1486 my $db = DBM::Deep->new( "foo.db" );
1489 $db->{circle} = $db; # ref to self
1491 print $db->{foo} . "\n"; # prints "bar"
1492 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1494 B<Note>: Passing the object to a function that recursively walks the
1495 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1496 C<export()> methods) will result in an infinite loop. This will be fixed in
1499 =head1 CAVEATS / ISSUES / BUGS
1501 This section describes all the known issues with DBM::Deep. It you have found
1502 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1504 =head2 UNUSED SPACE RECOVERY
1506 One major caveat with DBM::Deep is that space occupied by existing keys and
1507 values is not recovered when they are deleted. Meaning if you keep deleting
1508 and adding new keys, your file will continuously grow. I am working on this,
1509 but in the meantime you can call the built-in C<optimize()> method from time to
1510 time (perhaps in a crontab or something) to recover all your unused space.
1512 $db->optimize(); # returns true on success
1514 This rebuilds the ENTIRE database into a new file, then moves it on top of
1515 the original. The new file will have no unused space, thus it will take up as
1516 little disk space as possible. Please note that this operation can take
1517 a long time for large files, and you need enough disk space to temporarily hold
1518 2 copies of your DB file. The temporary file is created in the same directory
1519 as the original, named with a ".tmp" extension, and is deleted when the
1520 operation completes. Oh, and if locking is enabled, the DB is automatically
1521 locked for the entire duration of the copy.
1523 B<WARNING:> Only call optimize() on the top-level node of the database, and
1524 make sure there are no child references lying around. DBM::Deep keeps a reference
1525 counter, and if it is greater than 1, optimize() will abort and return undef.
1529 (The reasons given assume a high level of Perl understanding, specifically of
1530 references. You can safely skip this section.)
1532 Currently, the only references supported are HASH and ARRAY. The other reference
1533 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1539 These are things like filehandles and other sockets. They can't be supported
1540 because it's completely unclear how DBM::Deep should serialize them.
1542 =item * SCALAR / REF
1544 The discussion here refers to the following type of example:
1551 # In some other process ...
1553 my $val = ${ $db->{key1} };
1555 is( $val, 50, "What actually gets stored in the DB file?" );
1557 The problem is one of synchronization. When the variable being referred to
1558 changes value, the reference isn't notified. This means that the new value won't
1559 be stored in the datafile for other processes to read. There is no TIEREF.
1561 It is theoretically possible to store references to values already within a
1562 DBM::Deep object because everything already is synchronized, but the change to
1563 the internals would be quite large. Specifically, DBM::Deep would have to tie
1564 every single value that is stored. This would bloat the RAM footprint of
1565 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1566 all to support a feature that has never been requested.
1570 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1571 mechanism for serializing coderefs, including saving off all closure state.
1572 However, just as for SCALAR and REF, that closure state may change without
1573 notifying the DBM::Deep object storing the reference.
1577 =head2 FILE CORRUPTION
1579 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1580 for a 32-bit signature when opened, but other corruption in files can cause
1581 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1582 stuck in an infinite loop depending on the level of corruption. File write
1583 operations are not checked for failure (for speed), so if you happen to run
1584 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1585 be addressed in a later version of DBM::Deep.
1589 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1590 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1591 about setting up your NFS server with a locking daemon, then using lockf() to
1592 lock your files, but your mileage may vary there as well. From what I
1593 understand, there is no real way to do it. However, if you need access to the
1594 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1595 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1597 =head2 COPYING OBJECTS
1599 Beware of copying tied objects in Perl. Very strange things can happen.
1600 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1601 returns a new, blessed, tied hash or array to the same level in the DB.
1603 my $copy = $db->clone();
1605 B<Note>: Since clone() here is cloning the object, not the database location, any
1606 modifications to either $db or $copy will be visible in both.
1610 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1611 These functions cause every element in the array to move, which can be murder
1612 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1613 a different location. This will be addressed in the forthcoming version 1.00.
1615 =head2 WRITEONLY FILES
1617 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1618 writeonly mode. STORE will verify that the filehandle is writable. However, there
1619 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1620 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1624 This section discusses DBM::Deep's speed and memory usage.
1628 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1629 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1630 multi-level hash/array support, and cross-platform FTPable files. Even so,
1631 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1632 with huge databases. Here is some test data:
1634 Adding 1,000,000 keys to new DB file...
1636 At 100 keys, avg. speed is 2,703 keys/sec
1637 At 200 keys, avg. speed is 2,642 keys/sec
1638 At 300 keys, avg. speed is 2,598 keys/sec
1639 At 400 keys, avg. speed is 2,578 keys/sec
1640 At 500 keys, avg. speed is 2,722 keys/sec
1641 At 600 keys, avg. speed is 2,628 keys/sec
1642 At 700 keys, avg. speed is 2,700 keys/sec
1643 At 800 keys, avg. speed is 2,607 keys/sec
1644 At 900 keys, avg. speed is 2,190 keys/sec
1645 At 1,000 keys, avg. speed is 2,570 keys/sec
1646 At 2,000 keys, avg. speed is 2,417 keys/sec
1647 At 3,000 keys, avg. speed is 1,982 keys/sec
1648 At 4,000 keys, avg. speed is 1,568 keys/sec
1649 At 5,000 keys, avg. speed is 1,533 keys/sec
1650 At 6,000 keys, avg. speed is 1,787 keys/sec
1651 At 7,000 keys, avg. speed is 1,977 keys/sec
1652 At 8,000 keys, avg. speed is 2,028 keys/sec
1653 At 9,000 keys, avg. speed is 2,077 keys/sec
1654 At 10,000 keys, avg. speed is 2,031 keys/sec
1655 At 20,000 keys, avg. speed is 1,970 keys/sec
1656 At 30,000 keys, avg. speed is 2,050 keys/sec
1657 At 40,000 keys, avg. speed is 2,073 keys/sec
1658 At 50,000 keys, avg. speed is 1,973 keys/sec
1659 At 60,000 keys, avg. speed is 1,914 keys/sec
1660 At 70,000 keys, avg. speed is 2,091 keys/sec
1661 At 80,000 keys, avg. speed is 2,103 keys/sec
1662 At 90,000 keys, avg. speed is 1,886 keys/sec
1663 At 100,000 keys, avg. speed is 1,970 keys/sec
1664 At 200,000 keys, avg. speed is 2,053 keys/sec
1665 At 300,000 keys, avg. speed is 1,697 keys/sec
1666 At 400,000 keys, avg. speed is 1,838 keys/sec
1667 At 500,000 keys, avg. speed is 1,941 keys/sec
1668 At 600,000 keys, avg. speed is 1,930 keys/sec
1669 At 700,000 keys, avg. speed is 1,735 keys/sec
1670 At 800,000 keys, avg. speed is 1,795 keys/sec
1671 At 900,000 keys, avg. speed is 1,221 keys/sec
1672 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1674 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1675 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1676 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1677 Run time was 12 min 3 sec.
1681 One of the great things about DBM::Deep is that it uses very little memory.
1682 Even with huge databases (1,000,000+ keys) you will not see much increased
1683 memory on your process. DBM::Deep relies solely on the filesystem for storing
1684 and fetching data. Here is output from I</usr/bin/top> before even opening a
1687 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1688 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1690 Basically the process is taking 2,716K of memory. And here is the same
1691 process after storing and fetching 1,000,000 keys:
1693 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1694 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1696 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1697 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1699 =head1 DB FILE FORMAT
1701 In case you were interested in the underlying DB file format, it is documented
1702 here in this section. You don't need to know this to use the module, it's just
1703 included for reference.
1707 DBM::Deep files always start with a 32-bit signature to identify the file type.
1708 This is at offset 0. The signature is "DPDB" in network byte order. This is
1709 checked for when the file is opened and an error will be thrown if it's not found.
1713 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1714 has a standard header containing the type of data, the length of data, and then
1715 the data itself. The type is a single character (1 byte), the length is a
1716 32-bit unsigned long in network byte order, and the data is, well, the data.
1717 Here is how it unfolds:
1721 Immediately after the 32-bit file signature is the I<Master Index> record.
1722 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1723 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1724 depending on how the DBM::Deep object was constructed.
1726 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1727 number). The first 8-bit char of the MD5 signature is the offset into the
1728 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1729 index element is a file offset of the next tag for the key/element in question,
1730 which is usually a I<Bucket List> tag (see below).
1732 The next tag I<could> be another index, depending on how many keys/elements
1733 exist. See L<RE-INDEXING> below for details.
1737 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1738 file offsets to where the actual data is stored. It starts with a standard
1739 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1740 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1741 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1742 When the list fills up, a I<Re-Index> operation is performed (See
1743 L<RE-INDEXING> below).
1747 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1748 index/value pair (in array mode). It starts with a standard tag header with
1749 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1750 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1751 header. The size reported in the tag header is only for the value, but then,
1752 just after the value is another size (32-bit unsigned long) and then the plain
1753 key itself. Since the value is likely to be fetched more often than the plain
1754 key, I figured it would be I<slightly> faster to store the value first.
1756 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1757 record for the nested structure, where the process begins all over again.
1761 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1762 exhausted. Then, when another key/element comes in, the list is converted to a
1763 new index record. However, this index will look at the next char in the MD5
1764 hash, and arrange new Bucket List pointers accordingly. This process is called
1765 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1766 17 (16 + new one) keys/elements are removed from the old Bucket List and
1767 inserted into the new index. Several new Bucket Lists are created in the
1768 process, as a new MD5 char from the key is being examined (it is unlikely that
1769 the keys will all share the same next char of their MD5s).
1771 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1772 when the Bucket Lists will turn into indexes, but the first round tends to
1773 happen right around 4,000 keys. You will see a I<slight> decrease in
1774 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1775 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1776 right around 900,000 keys. This process can continue nearly indefinitely --
1777 right up until the point the I<MD5> signatures start colliding with each other,
1778 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1779 getting struck by lightning while you are walking to cash in your tickets.
1780 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1781 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1782 this is 340 unodecillion, but don't quote me).
1786 When a new key/element is stored, the key (or index number) is first run through
1787 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1788 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1789 for the first char of the signature (in this case I<b0>). If it does not exist,
1790 a new I<Bucket List> is created for our key (and the next 15 future keys that
1791 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1792 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1793 this point, unless we are replacing an existing I<Bucket>), where the actual
1794 data will be stored.
1798 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1799 (or index number), then walking along the indexes. If there are enough
1800 keys/elements in this DB level, there might be nested indexes, each linked to
1801 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1802 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1803 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1804 plain key are stored.
1806 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1807 methods. In this process the indexes are walked systematically, and each key
1808 fetched in increasing MD5 order (which is why it appears random). Once the
1809 I<Bucket> is found, the value is skipped and the plain key returned instead.
1810 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1811 alphabetically sorted. This only happens on an index-level -- as soon as the
1812 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1813 so it's pretty much undefined how the keys will come out -- just like Perl's
1816 =head1 CODE COVERAGE
1818 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1819 B<Devel::Cover> report on this module's test suite.
1821 ----------------------------------- ------ ------ ------ ------ ------ ------
1822 File stmt bran cond sub time total
1823 ----------------------------------- ------ ------ ------ ------ ------ ------
1824 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1825 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1826 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1827 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1828 Total 97.9 85.9 79.7 100.0 100.0 94.3
1829 ----------------------------------- ------ ------ ------ ------ ------ ------
1831 =head1 MORE INFORMATION
1833 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1834 or send email to L<DBM-Deep@googlegroups.com>.
1838 Joseph Huckaby, L<jhuckaby@cpan.org>
1840 Rob Kinyon, L<rkinyon@cpan.org>
1842 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1846 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1847 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1851 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1852 This is free software, you may use it and distribute it under the
1853 same terms as Perl itself.