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 our $VERSION = q(0.99_01);
39 use Fcntl qw( :DEFAULT :flock :seek );
43 use DBM::Deep::Engine;
47 # Setup constants for users to pass to new()
49 sub TYPE_HASH () { DBM::Deep::Engine->SIG_HASH }
50 sub TYPE_ARRAY () { DBM::Deep::Engine->SIG_ARRAY }
58 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
63 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
64 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
69 $args = { file => shift };
77 # Class constructor method for Perl OO interface.
78 # Calls tie() and returns blessed reference to tied hash or array,
79 # providing a hybrid OO/tie interface.
82 my $args = $class->_get_args( @_ );
85 # Check if we want a tied hash or array.
88 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
89 $class = 'DBM::Deep::Array';
90 require DBM::Deep::Array;
91 tie @$self, $class, %$args;
94 $class = 'DBM::Deep::Hash';
95 require DBM::Deep::Hash;
96 tie %$self, $class, %$args;
99 return bless $self, $class;
102 # This initializer is called from the various TIE* methods. new() calls tie(),
103 # which allows for a single point of entry.
108 $args->{fileobj} = DBM::Deep::File->new( $args )
109 unless exists $args->{fileobj};
111 # locking implicitly enables autoflush
112 if ($args->{locking}) { $args->{autoflush} = 1; }
114 # These are the defaults to be optionally overridden below
117 base_offset => undef,
124 $self->{engine} = DBM::Deep::Engine->new( { %{$args}, obj => $self } );
126 # Grab the parameters we want to use
127 foreach my $param ( keys %$self ) {
128 next unless exists $args->{$param};
129 $self->{$param} = $args->{$param};
132 $self->_engine->setup_fh( $self );
134 $self->{fileobj}->set_db( $self );
141 require DBM::Deep::Hash;
142 return DBM::Deep::Hash->TIEHASH( @_ );
147 require DBM::Deep::Array;
148 return DBM::Deep::Array->TIEARRAY( @_ );
152 my $self = shift->_get_self;
153 return $self->_fileobj->lock( $self, @_ );
157 my $self = shift->_get_self;
158 return $self->_fileobj->unlock( $self, @_ );
162 my $self = shift->_get_self;
163 my ($spot, $value) = @_;
168 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
169 ${$spot} = $value->_repr;
170 $value->_copy_node( ${$spot} );
173 my $r = Scalar::Util::reftype( $value );
174 my $c = Scalar::Util::blessed( $value );
175 if ( $r eq 'ARRAY' ) {
176 ${$spot} = [ @{$value} ];
179 ${$spot} = { %{$value} };
181 ${$spot} = bless ${$spot}, $c
189 die "Must be implemented in a child class\n";
193 die "Must be implemented in a child class\n";
198 # Recursively export into standard Perl hashes and arrays.
200 my $self = shift->_get_self;
202 my $temp = $self->_repr;
205 $self->_copy_node( $temp );
213 # Recursively import Perl hash/array structure
215 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
217 my $self = shift->_get_self;
220 # struct is not a reference, so just import based on our type
222 $struct = $self->_repr( @_ );
225 return $self->_import( $struct );
230 # Rebuild entire database into new file, then move
231 # it back on top of original.
233 my $self = shift->_get_self;
235 #XXX Need to create a new test for this
236 # if ($self->_fileobj->{links} > 1) {
237 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
240 my $db_temp = DBM::Deep->new(
241 file => $self->_fileobj->{file} . '.tmp',
246 $self->_copy_node( $db_temp );
250 # Attempt to copy user, group and permissions over to new file
252 my @stats = stat($self->_fh);
253 my $perms = $stats[2] & 07777;
256 chown( $uid, $gid, $self->_fileobj->{file} . '.tmp' );
257 chmod( $perms, $self->_fileobj->{file} . '.tmp' );
259 # q.v. perlport for more information on this variable
260 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
262 # Potential race condition when optmizing on Win32 with locking.
263 # The Windows filesystem requires that the filehandle be closed
264 # before it is overwritten with rename(). This could be redone
268 $self->_fileobj->close;
271 if (!rename $self->_fileobj->{file} . '.tmp', $self->_fileobj->{file}) {
272 unlink $self->_fileobj->{file} . '.tmp';
274 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
278 $self->_fileobj->close;
279 $self->_fileobj->open;
280 $self->_engine->setup_fh( $self );
287 # Make copy of object and return
289 my $self = shift->_get_self;
291 return DBM::Deep->new(
292 type => $self->_type,
293 base_offset => $self->_base_offset,
294 fileobj => $self->_fileobj,
299 my %is_legal_filter = map {
302 store_key store_value
303 fetch_key fetch_value
308 # Setup filter function for storing or fetching the key or value
310 my $self = shift->_get_self;
314 if ( $is_legal_filter{$type} ) {
315 $self->_fileobj->{"filter_$type"} = $func;
324 my $self = shift->_get_self;
325 $self->_fileobj->begin_transaction;
330 my $self = shift->_get_self;
331 $self->_fileobj->end_transaction;
336 my $self = shift->_get_self;
337 $self->_fileobj->commit_transaction;
346 my $self = $_[0]->_get_self;
347 return $self->{engine};
351 my $self = $_[0]->_get_self;
352 return $self->{fileobj};
356 my $self = $_[0]->_get_self;
357 return $self->{type};
361 my $self = $_[0]->_get_self;
362 return $self->{base_offset};
366 my $self = $_[0]->_get_self;
367 return $self->_fileobj->{fh};
375 die "DBM::Deep: $_[1]\n";
380 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
385 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
392 #XXX This if() is redundant
393 if ( my $parent = $self->{parent} ) {
395 while ( $parent->{parent} ) {
397 $parent->_type eq TYPE_HASH
398 ? "\{$child->{parent_key}\}"
399 : "\[$child->{parent_key}\]"
403 $parent = $parent->{parent};
406 $base = "\$db->get( '$child->{parent_key}' )->" . $base;
409 $base = "\$db->get( '$child->{parent_key}' )";
417 # Store single hash key/value or array element in database.
419 my $self = shift->_get_self;
420 my ($key, $value, $orig_key) = @_;
423 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
424 $self->_throw_error( 'Cannot write to a readonly filehandle' );
427 #XXX The second condition needs to disappear
428 if ( defined $orig_key && !( $self->_type eq TYPE_ARRAY && $orig_key eq 'length') ) {
431 my $r = Scalar::Util::reftype( $value ) || '';
432 if ( $r eq 'HASH' ) {
435 elsif ( $r eq 'ARRAY' ) {
438 elsif ( defined $value ) {
445 if ( my $c = Scalar::Util::blessed( $value ) ) {
446 $rhs = "bless $rhs, '$c'";
449 my $lhs = $self->_find_parent;
451 if ( $self->_type eq TYPE_HASH ) {
452 $lhs .= "->\{$orig_key\}";
455 $lhs .= "->\[$orig_key\]";
461 $lhs = "\$db->put('$orig_key',$rhs);";
464 $self->_fileobj->audit($lhs);
468 # Request exclusive lock for writing
470 $self->lock( LOCK_EX );
472 my $md5 = $self->_engine->{digest}->($key);
474 my $tag = $self->_engine->find_blist( $self->_base_offset, $md5, { create => 1 } );
476 # User may be storing a hash, in which case we do not want it run
477 # through the filtering system
478 if ( !ref($value) && $self->_fileobj->{filter_store_value} ) {
479 $value = $self->_fileobj->{filter_store_value}->( $value );
483 # Add key/value to bucket list
485 $self->_engine->add_bucket( $tag, $md5, $key, $value, undef, $orig_key );
494 # Fetch single value or element given plain key or array index
496 my $self = shift->_get_self;
497 my ($key, $orig_key) = @_;
499 my $md5 = $self->_engine->{digest}->($key);
502 # Request shared lock for reading
504 $self->lock( LOCK_SH );
506 my $tag = $self->_engine->find_blist( $self->_base_offset, $md5 );#, { create => 1 } );
507 #XXX This needs to autovivify
514 # Get value from bucket list
516 my $result = $self->_engine->get_bucket_value( $tag, $md5, $orig_key );
520 # Filters only apply to scalar values, so the ref check is making
521 # sure the fetched bucket is a scalar, not a child hash or array.
522 return ($result && !ref($result) && $self->_fileobj->{filter_fetch_value})
523 ? $self->_fileobj->{filter_fetch_value}->($result)
529 # Delete single key/value pair or element given plain key or array index
531 my $self = shift->_get_self;
532 my ($key, $orig_key) = @_;
534 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
535 $self->_throw_error( 'Cannot write to a readonly filehandle' );
538 if ( defined $orig_key ) {
539 my $lhs = $self->_find_parent;
541 $self->_fileobj->audit( "delete $lhs;" );
544 $self->_fileobj->audit( "\$db->delete('$orig_key');" );
549 # Request exclusive lock for writing
551 $self->lock( LOCK_EX );
553 my $md5 = $self->_engine->{digest}->($key);
555 my $tag = $self->_engine->find_blist( $self->_base_offset, $md5 );
564 my $value = $self->_engine->get_bucket_value( $tag, $md5 );
566 if (defined $value && !ref($value) && $self->_fileobj->{filter_fetch_value}) {
567 $value = $self->_fileobj->{filter_fetch_value}->($value);
570 my $result = $self->_engine->delete_bucket( $tag, $md5, $orig_key );
573 # If this object is an array and the key deleted was on the end of the stack,
574 # decrement the length variable.
584 # Check if a single key or element exists given plain key or array index
586 my $self = shift->_get_self;
589 my $md5 = $self->_engine->{digest}->($key);
592 # Request shared lock for reading
594 $self->lock( LOCK_SH );
596 my $tag = $self->_engine->find_blist( $self->_base_offset, $md5 );
601 # For some reason, the built-in exists() function returns '' for false
607 # Check if bucket exists and return 1 or ''
609 my $result = $self->_engine->bucket_exists( $tag, $md5 ) || '';
618 # Clear all keys from hash, or all elements from array.
620 my $self = shift->_get_self;
622 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
623 $self->_throw_error( 'Cannot write to a readonly filehandle' );
627 my $lhs = $self->_find_parent;
629 if ( $self->_type eq TYPE_HASH ) {
630 $lhs = '%{' . $lhs . '}';
633 $lhs = '@{' . $lhs . '}';
636 $self->_fileobj->audit( "$lhs = ();" );
640 # Request exclusive lock for writing
642 $self->lock( LOCK_EX );
644 #XXX This needs updating to use _release_space
645 $self->_engine->write_tag(
646 $self->_base_offset, $self->_type,
647 chr(0)x$self->_engine->{index_size},
656 # Public method aliases
658 sub put { (shift)->STORE( @_ ) }
659 sub store { (shift)->STORE( @_ ) }
660 sub get { (shift)->FETCH( @_ ) }
661 sub fetch { (shift)->FETCH( @_ ) }
662 sub delete { (shift)->DELETE( @_ ) }
663 sub exists { (shift)->EXISTS( @_ ) }
664 sub clear { (shift)->CLEAR( @_ ) }
671 DBM::Deep - A pure perl multi-level hash/array DBM
676 my $db = DBM::Deep->new( "foo.db" );
678 $db->{key} = 'value'; # tie() style
681 $db->put('key' => 'value'); # OO style
682 print $db->get('key');
684 # true multi-level support
685 $db->{my_complex} = [
686 'hello', { perl => 'rules' },
692 A unique flat-file database module, written in pure perl. True
693 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
694 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
695 handle millions of keys and unlimited hash levels without significant
696 slow-down. Written from the ground-up in pure perl -- this is NOT a
697 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
698 Mac OS X and Windows.
700 =head1 VERSION DIFFERENCES
702 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
703 before. While attempts have been made to be backwards compatible, no guarantees.
707 Hopefully you are using Perl's excellent CPAN module, which will download
708 and install the module for you. If not, get the tarball, and run these
720 Construction can be done OO-style (which is the recommended way), or using
721 Perl's tie() function. Both are examined here.
723 =head2 OO CONSTRUCTION
725 The recommended way to construct a DBM::Deep object is to use the new()
726 method, which gets you a blessed, tied hash or array reference.
728 my $db = DBM::Deep->new( "foo.db" );
730 This opens a new database handle, mapped to the file "foo.db". If this
731 file does not exist, it will automatically be created. DB files are
732 opened in "r+" (read/write) mode, and the type of object returned is a
733 hash, unless otherwise specified (see L<OPTIONS> below).
735 You can pass a number of options to the constructor to specify things like
736 locking, autoflush, etc. This is done by passing an inline hash:
738 my $db = DBM::Deep->new(
744 Notice that the filename is now specified I<inside> the hash with
745 the "file" parameter, as opposed to being the sole argument to the
746 constructor. This is required if any options are specified.
747 See L<OPTIONS> below for the complete list.
751 You can also start with an array instead of a hash. For this, you must
752 specify the C<type> parameter:
754 my $db = DBM::Deep->new(
756 type => DBM::Deep->TYPE_ARRAY
759 B<Note:> Specifing the C<type> parameter only takes effect when beginning
760 a new DB file. If you create a DBM::Deep object with an existing file, the
761 C<type> will be loaded from the file header, and an error will be thrown if
762 the wrong type is passed in.
764 =head2 TIE CONSTRUCTION
766 Alternately, you can create a DBM::Deep handle by using Perl's built-in
767 tie() function. The object returned from tie() can be used to call methods,
768 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
769 file (as expected with most tie'd objects).
772 my $db = tie %hash, "DBM::Deep", "foo.db";
775 my $db = tie @array, "DBM::Deep", "bar.db";
777 As with the OO constructor, you can replace the DB filename parameter with
778 a hash containing one or more options (see L<OPTIONS> just below for the
781 tie %hash, "DBM::Deep", {
789 There are a number of options that can be passed in when constructing your
790 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
796 Filename of the DB file to link the handle to. You can pass a full absolute
797 filesystem path, partial path, or a plain filename if the file is in the
798 current working directory. This is a required parameter (though q.v. fh).
802 If you want, you can pass in the fh instead of the file. This is most useful for doing
805 my $db = DBM::Deep->new( { fh => \*DATA } );
807 You are responsible for making sure that the fh has been opened appropriately for your
808 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
809 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
810 needs to read from the fh.
814 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
815 not need to set this. However, it's there if you want it.
817 If you pass in fh and do not set this, it will be set appropriately.
821 This parameter specifies what type of object to create, a hash or array. Use
822 one of these two constants:
826 =item * C<DBM::Deep-E<gt>TYPE_HASH>
828 =item * C<DBM::Deep-E<gt>TYPE_ARRAY>.
832 This only takes effect when beginning a new file. This is an optional
833 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
837 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
838 function to lock the database in exclusive mode for writes, and shared mode for
839 reads. Pass any true value to enable. This affects the base DB handle I<and
840 any child hashes or arrays> that use the same DB file. This is an optional
841 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
845 Specifies whether autoflush is to be enabled on the underlying filehandle.
846 This obviously slows down write operations, but is required if you may have
847 multiple processes accessing the same DB file (also consider enable I<locking>).
848 Pass any true value to enable. This is an optional parameter, and defaults to 0
853 If I<autobless> mode is enabled, DBM::Deep will preserve the class something
854 is blessed into, and restores it when fetched. This is an optional parameter, and defaults to 1 (enabled).
856 B<Note:> If you use the OO-interface, you will not be able to call any methods
857 of DBM::Deep on the blessed item. This is considered to be a feature.
861 See L</FILTERS> below.
867 With DBM::Deep you can access your databases using Perl's standard hash/array
868 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
869 treat them as such. DBM::Deep will intercept all reads/writes and direct them
870 to the right place -- the DB file. This has nothing to do with the
871 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
872 using regular hashes and arrays, rather than calling functions like C<get()>
873 and C<put()> (although those work too). It is entirely up to you how to want
874 to access your databases.
878 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
879 or even nested hashes (or arrays) using standard Perl syntax:
881 my $db = DBM::Deep->new( "foo.db" );
883 $db->{mykey} = "myvalue";
885 $db->{myhash}->{subkey} = "subvalue";
887 print $db->{myhash}->{subkey} . "\n";
889 You can even step through hash keys using the normal Perl C<keys()> function:
891 foreach my $key (keys %$db) {
892 print "$key: " . $db->{$key} . "\n";
895 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
896 pushes them onto an array, all before the loop even begins. If you have an
897 extra large hash, this may exhaust Perl's memory. Instead, consider using
898 Perl's C<each()> function, which pulls keys/values one at a time, using very
901 while (my ($key, $value) = each %$db) {
902 print "$key: $value\n";
905 Please note that when using C<each()>, you should always pass a direct
906 hash reference, not a lookup. Meaning, you should B<never> do this:
909 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
911 This causes an infinite loop, because for each iteration, Perl is calling
912 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
913 it effectively keeps returning the first key over and over again. Instead,
914 assign a temporary variable to C<$db->{foo}>, then pass that to each().
918 As with hashes, you can treat any DBM::Deep object like a normal Perl array
919 reference. This includes inserting, removing and manipulating elements,
920 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
921 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
922 or simply be a nested array reference inside a hash. Example:
924 my $db = DBM::Deep->new(
925 file => "foo-array.db",
926 type => DBM::Deep->TYPE_ARRAY
930 push @$db, "bar", "baz";
933 my $last_elem = pop @$db; # baz
934 my $first_elem = shift @$db; # bah
935 my $second_elem = $db->[1]; # bar
937 my $num_elements = scalar @$db;
941 In addition to the I<tie()> interface, you can also use a standard OO interface
942 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
943 array) has its own methods, but both types share the following common methods:
944 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
948 =item * new() / clone()
950 These are the constructor and copy-functions.
952 =item * put() / store()
954 Stores a new hash key/value pair, or sets an array element value. Takes two
955 arguments, the hash key or array index, and the new value. The value can be
956 a scalar, hash ref or array ref. Returns true on success, false on failure.
958 $db->put("foo", "bar"); # for hashes
959 $db->put(1, "bar"); # for arrays
961 =item * get() / fetch()
963 Fetches the value of a hash key or array element. Takes one argument: the hash
964 key or array index. Returns a scalar, hash ref or array ref, depending on the
967 my $value = $db->get("foo"); # for hashes
968 my $value = $db->get(1); # for arrays
972 Checks if a hash key or array index exists. Takes one argument: the hash key
973 or array index. Returns true if it exists, false if not.
975 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
976 if ($db->exists(1)) { print "yay!\n"; } # for arrays
980 Deletes one hash key/value pair or array element. Takes one argument: the hash
981 key or array index. Returns true on success, false if not found. For arrays,
982 the remaining elements located after the deleted element are NOT moved over.
983 The deleted element is essentially just undefined, which is exactly how Perl's
984 internal arrays work. Please note that the space occupied by the deleted
985 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
986 below for details and workarounds.
988 $db->delete("foo"); # for hashes
989 $db->delete(1); # for arrays
993 Deletes B<all> hash keys or array elements. Takes no arguments. No return
994 value. Please note that the space occupied by the deleted keys/values or
995 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
996 details and workarounds.
998 $db->clear(); # hashes or arrays
1000 =item * lock() / unlock()
1006 Recover lost disk space.
1008 =item * import() / export()
1010 Data going in and out.
1016 For hashes, DBM::Deep supports all the common methods described above, and the
1017 following additional methods: C<first_key()> and C<next_key()>.
1023 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1024 fetched in an undefined order (which appears random). Takes no arguments,
1025 returns the key as a scalar value.
1027 my $key = $db->first_key();
1031 Returns the "next" key in the hash, given the previous one as the sole argument.
1032 Returns undef if there are no more keys to be fetched.
1034 $key = $db->next_key($key);
1038 Here are some examples of using hashes:
1040 my $db = DBM::Deep->new( "foo.db" );
1042 $db->put("foo", "bar");
1043 print "foo: " . $db->get("foo") . "\n";
1045 $db->put("baz", {}); # new child hash ref
1046 $db->get("baz")->put("buz", "biz");
1047 print "buz: " . $db->get("baz")->get("buz") . "\n";
1049 my $key = $db->first_key();
1051 print "$key: " . $db->get($key) . "\n";
1052 $key = $db->next_key($key);
1055 if ($db->exists("foo")) { $db->delete("foo"); }
1059 For arrays, DBM::Deep supports all the common methods described above, and the
1060 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1061 C<unshift()> and C<splice()>.
1067 Returns the number of elements in the array. Takes no arguments.
1069 my $len = $db->length();
1073 Adds one or more elements onto the end of the array. Accepts scalars, hash
1074 refs or array refs. No return value.
1076 $db->push("foo", "bar", {});
1080 Fetches the last element in the array, and deletes it. Takes no arguments.
1081 Returns undef if array is empty. Returns the element value.
1083 my $elem = $db->pop();
1087 Fetches the first element in the array, deletes it, then shifts all the
1088 remaining elements over to take up the space. Returns the element value. This
1089 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1092 my $elem = $db->shift();
1096 Inserts one or more elements onto the beginning of the array, shifting all
1097 existing elements over to make room. Accepts scalars, hash refs or array refs.
1098 No return value. This method is not recommended with large arrays -- see
1099 <LARGE ARRAYS> below for details.
1101 $db->unshift("foo", "bar", {});
1105 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1106 -f splice> for usage -- it is too complicated to document here. This method is
1107 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1111 Here are some examples of using arrays:
1113 my $db = DBM::Deep->new(
1115 type => DBM::Deep->TYPE_ARRAY
1118 $db->push("bar", "baz");
1119 $db->unshift("foo");
1122 my $len = $db->length();
1123 print "length: $len\n"; # 4
1125 for (my $k=0; $k<$len; $k++) {
1126 print "$k: " . $db->get($k) . "\n";
1129 $db->splice(1, 2, "biz", "baf");
1131 while (my $elem = shift @$db) {
1132 print "shifted: $elem\n";
1137 Enable automatic file locking by passing a true value to the C<locking>
1138 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1140 my $db = DBM::Deep->new(
1145 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1146 mode for writes, and shared mode for reads. This is required if you have
1147 multiple processes accessing the same database file, to avoid file corruption.
1148 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1149 NFS> below for more.
1151 =head2 EXPLICIT LOCKING
1153 You can explicitly lock a database, so it remains locked for multiple
1154 transactions. This is done by calling the C<lock()> method, and passing an
1155 optional lock mode argument (defaults to exclusive mode). This is particularly
1156 useful for things like counters, where the current value needs to be fetched,
1157 then incremented, then stored again.
1160 my $counter = $db->get("counter");
1162 $db->put("counter", $counter);
1171 You can pass C<lock()> an optional argument, which specifies which mode to use
1172 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1173 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1174 same as the constants defined in Perl's C<Fcntl> module.
1176 $db->lock( DBM::Deep->LOCK_SH );
1180 =head1 IMPORTING/EXPORTING
1182 You can import existing complex structures by calling the C<import()> method,
1183 and export an entire database into an in-memory structure using the C<export()>
1184 method. Both are examined here.
1188 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1189 walking the structure and adding keys/elements to the database as you go,
1190 simply pass a reference to the C<import()> method. This recursively adds
1191 everything to an existing DBM::Deep object for you. Here is an example:
1196 array1 => [ "elem0", "elem1", "elem2" ],
1198 subkey1 => "subvalue1",
1199 subkey2 => "subvalue2"
1203 my $db = DBM::Deep->new( "foo.db" );
1204 $db->import( $struct );
1206 print $db->{key1} . "\n"; # prints "value1"
1208 This recursively imports the entire C<$struct> object into C<$db>, including
1209 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1210 keys are merged with the existing ones, replacing if they already exist.
1211 The C<import()> method can be called on any database level (not just the base
1212 level), and works with both hash and array DB types.
1214 B<Note:> Make sure your existing structure has no circular references in it.
1215 These will cause an infinite loop when importing.
1219 Calling the C<export()> method on an existing DBM::Deep object will return
1220 a reference to a new in-memory copy of the database. The export is done
1221 recursively, so all nested hashes/arrays are all exported to standard Perl
1222 objects. Here is an example:
1224 my $db = DBM::Deep->new( "foo.db" );
1226 $db->{key1} = "value1";
1227 $db->{key2} = "value2";
1229 $db->{hash1}->{subkey1} = "subvalue1";
1230 $db->{hash1}->{subkey2} = "subvalue2";
1232 my $struct = $db->export();
1234 print $struct->{key1} . "\n"; # prints "value1"
1236 This makes a complete copy of the database in memory, and returns a reference
1237 to it. The C<export()> method can be called on any database level (not just
1238 the base level), and works with both hash and array DB types. Be careful of
1239 large databases -- you can store a lot more data in a DBM::Deep object than an
1240 in-memory Perl structure.
1242 B<Note:> Make sure your database has no circular references in it.
1243 These will cause an infinite loop when exporting.
1247 DBM::Deep has a number of hooks where you can specify your own Perl function
1248 to perform filtering on incoming or outgoing data. This is a perfect
1249 way to extend the engine, and implement things like real-time compression or
1250 encryption. Filtering applies to the base DB level, and all child hashes /
1251 arrays. Filter hooks can be specified when your DBM::Deep object is first
1252 constructed, or by calling the C<set_filter()> method at any time. There are
1253 four available filter hooks, described below:
1257 =item * filter_store_key
1259 This filter is called whenever a hash key is stored. It
1260 is passed the incoming key, and expected to return a transformed key.
1262 =item * filter_store_value
1264 This filter is called whenever a hash key or array element is stored. It
1265 is passed the incoming value, and expected to return a transformed value.
1267 =item * filter_fetch_key
1269 This filter is called whenever a hash key is fetched (i.e. via
1270 C<first_key()> or C<next_key()>). It is passed the transformed key,
1271 and expected to return the plain key.
1273 =item * filter_fetch_value
1275 This filter is called whenever a hash key or array element is fetched.
1276 It is passed the transformed value, and expected to return the plain value.
1280 Here are the two ways to setup a filter hook:
1282 my $db = DBM::Deep->new(
1284 filter_store_value => \&my_filter_store,
1285 filter_fetch_value => \&my_filter_fetch
1290 $db->set_filter( "filter_store_value", \&my_filter_store );
1291 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1293 Your filter function will be called only when dealing with SCALAR keys or
1294 values. When nested hashes and arrays are being stored/fetched, filtering
1295 is bypassed. Filters are called as static functions, passed a single SCALAR
1296 argument, and expected to return a single SCALAR value. If you want to
1297 remove a filter, set the function reference to C<undef>:
1299 $db->set_filter( "filter_store_value", undef );
1301 =head2 REAL-TIME ENCRYPTION EXAMPLE
1303 Here is a working example that uses the I<Crypt::Blowfish> module to
1304 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1305 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1306 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1309 use Crypt::Blowfish;
1312 my $cipher = Crypt::CBC->new({
1313 'key' => 'my secret key',
1314 'cipher' => 'Blowfish',
1316 'regenerate_key' => 0,
1317 'padding' => 'space',
1321 my $db = DBM::Deep->new(
1322 file => "foo-encrypt.db",
1323 filter_store_key => \&my_encrypt,
1324 filter_store_value => \&my_encrypt,
1325 filter_fetch_key => \&my_decrypt,
1326 filter_fetch_value => \&my_decrypt,
1329 $db->{key1} = "value1";
1330 $db->{key2} = "value2";
1331 print "key1: " . $db->{key1} . "\n";
1332 print "key2: " . $db->{key2} . "\n";
1338 return $cipher->encrypt( $_[0] );
1341 return $cipher->decrypt( $_[0] );
1344 =head2 REAL-TIME COMPRESSION EXAMPLE
1346 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1347 compression / decompression of keys & values with DBM::Deep Filters.
1348 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1349 more on I<Compress::Zlib>.
1354 my $db = DBM::Deep->new(
1355 file => "foo-compress.db",
1356 filter_store_key => \&my_compress,
1357 filter_store_value => \&my_compress,
1358 filter_fetch_key => \&my_decompress,
1359 filter_fetch_value => \&my_decompress,
1362 $db->{key1} = "value1";
1363 $db->{key2} = "value2";
1364 print "key1: " . $db->{key1} . "\n";
1365 print "key2: " . $db->{key2} . "\n";
1371 return Compress::Zlib::memGzip( $_[0] ) ;
1374 return Compress::Zlib::memGunzip( $_[0] ) ;
1377 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1378 actually numerical index numbers, and are not filtered.
1380 =head1 ERROR HANDLING
1382 Most DBM::Deep methods return a true value for success, and call die() on
1383 failure. You can wrap calls in an eval block to catch the die.
1385 my $db = DBM::Deep->new( "foo.db" ); # create hash
1386 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1388 print $@; # prints error message
1390 =head1 LARGEFILE SUPPORT
1392 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1393 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1394 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1395 by specifying the 'pack_size' parameter when constructing the file.
1398 filename => $filename,
1399 pack_size => 'large',
1402 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1403 instead of 32-bit longs. After setting these values your DB files have a
1404 theoretical maximum size of 16 XB (exabytes).
1406 You can also use C<pack_size =E<gt> 'small'> in order to use 16-bit file
1409 B<Note:> Changing these values will B<NOT> work for existing database files.
1410 Only change this for new files. Once the value has been set, it is stored in
1411 the file's header and cannot be changed for the life of the file. These
1412 parameters are per-file, meaning you can access 32-bit and 64-bit files, as
1415 B<Note:> We have not personally tested files larger than 2 GB -- all my
1416 systems have only a 32-bit Perl. However, I have received user reports that
1417 this does indeed work!
1419 =head1 LOW-LEVEL ACCESS
1421 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1422 you can call the C<_fh()> method, which returns the handle:
1424 my $fh = $db->_fh();
1426 This method can be called on the root level of the datbase, or any child
1427 hashes or arrays. All levels share a I<root> structure, which contains things
1428 like the filehandle, a reference counter, and all the options specified
1429 when you created the object. You can get access to this file object by
1430 calling the C<_fileobj()> method.
1432 my $file_obj = $db->_fileobj();
1434 This is useful for changing options after the object has already been created,
1435 such as enabling/disabling locking. You can also store your own temporary user
1436 data in this structure (be wary of name collision), which is then accessible from
1437 any child hash or array.
1439 =head1 CUSTOM DIGEST ALGORITHM
1441 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1442 keys. However you can override this, and use another algorithm (such as SHA-256)
1443 or even write your own. But please note that DBM::Deep currently expects zero
1444 collisions, so your algorithm has to be I<perfect>, so to speak. Collision
1445 detection may be introduced in a later version.
1447 You can specify a custom digest algorithm by passing it into the parameter
1448 list for new(), passing a reference to a subroutine as the 'digest' parameter,
1449 and the length of the algorithm's hashes (in bytes) as the 'hash_size'
1450 parameter. Here is a working example that uses a 256-bit hash from the
1451 I<Digest::SHA256> module. Please see
1452 L<http://search.cpan.org/search?module=Digest::SHA256> for more information.
1457 my $context = Digest::SHA256::new(256);
1459 my $db = DBM::Deep->new(
1460 filename => "foo-sha.db",
1461 digest => \&my_digest,
1465 $db->{key1} = "value1";
1466 $db->{key2} = "value2";
1467 print "key1: " . $db->{key1} . "\n";
1468 print "key2: " . $db->{key2} . "\n";
1474 return substr( $context->hash($_[0]), 0, 32 );
1477 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1478 of bytes you specify in the hash_size parameter (in this case 32).
1480 B<Note:> If you do choose to use a custom digest algorithm, you must set it
1481 every time you access this file. Otherwise, the default (MD5) will be used.
1483 =head1 CIRCULAR REFERENCES
1485 DBM::Deep has B<experimental> support for circular references. Meaning you
1486 can have a nested hash key or array element that points to a parent object.
1487 This relationship is stored in the DB file, and is preserved between sessions.
1490 my $db = DBM::Deep->new( "foo.db" );
1493 $db->{circle} = $db; # ref to self
1495 print $db->{foo} . "\n"; # prints "bar"
1496 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1498 B<Note>: Passing the object to a function that recursively walks the
1499 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1500 C<export()> methods) will result in an infinite loop. This will be fixed in
1503 =head1 CAVEATS / ISSUES / BUGS
1505 This section describes all the known issues with DBM::Deep. It you have found
1506 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1508 =head2 UNUSED SPACE RECOVERY
1510 One major caveat with DBM::Deep is that space occupied by existing keys and
1511 values is not recovered when they are deleted. Meaning if you keep deleting
1512 and adding new keys, your file will continuously grow. I am working on this,
1513 but in the meantime you can call the built-in C<optimize()> method from time to
1514 time (perhaps in a crontab or something) to recover all your unused space.
1516 $db->optimize(); # returns true on success
1518 This rebuilds the ENTIRE database into a new file, then moves it on top of
1519 the original. The new file will have no unused space, thus it will take up as
1520 little disk space as possible. Please note that this operation can take
1521 a long time for large files, and you need enough disk space to temporarily hold
1522 2 copies of your DB file. The temporary file is created in the same directory
1523 as the original, named with a ".tmp" extension, and is deleted when the
1524 operation completes. Oh, and if locking is enabled, the DB is automatically
1525 locked for the entire duration of the copy.
1527 B<WARNING:> Only call optimize() on the top-level node of the database, and
1528 make sure there are no child references lying around. DBM::Deep keeps a reference
1529 counter, and if it is greater than 1, optimize() will abort and return undef.
1533 (The reasons given assume a high level of Perl understanding, specifically of
1534 references. You can safely skip this section.)
1536 Currently, the only references supported are HASH and ARRAY. The other reference
1537 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1543 These are things like filehandles and other sockets. They can't be supported
1544 because it's completely unclear how DBM::Deep should serialize them.
1546 =item * SCALAR / REF
1548 The discussion here refers to the following type of example:
1555 # In some other process ...
1557 my $val = ${ $db->{key1} };
1559 is( $val, 50, "What actually gets stored in the DB file?" );
1561 The problem is one of synchronization. When the variable being referred to
1562 changes value, the reference isn't notified. This means that the new value won't
1563 be stored in the datafile for other processes to read. There is no TIEREF.
1565 It is theoretically possible to store references to values already within a
1566 DBM::Deep object because everything already is synchronized, but the change to
1567 the internals would be quite large. Specifically, DBM::Deep would have to tie
1568 every single value that is stored. This would bloat the RAM footprint of
1569 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1570 all to support a feature that has never been requested.
1574 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1575 mechanism for serializing coderefs, including saving off all closure state.
1576 However, just as for SCALAR and REF, that closure state may change without
1577 notifying the DBM::Deep object storing the reference.
1581 =head2 FILE CORRUPTION
1583 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1584 for a 32-bit signature when opened, but other corruption in files can cause
1585 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1586 stuck in an infinite loop depending on the level of corruption. File write
1587 operations are not checked for failure (for speed), so if you happen to run
1588 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1589 be addressed in a later version of DBM::Deep.
1593 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1594 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1595 about setting up your NFS server with a locking daemon, then using lockf() to
1596 lock your files, but your mileage may vary there as well. From what I
1597 understand, there is no real way to do it. However, if you need access to the
1598 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1599 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1601 =head2 COPYING OBJECTS
1603 Beware of copying tied objects in Perl. Very strange things can happen.
1604 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1605 returns a new, blessed, tied hash or array to the same level in the DB.
1607 my $copy = $db->clone();
1609 B<Note>: Since clone() here is cloning the object, not the database location, any
1610 modifications to either $db or $copy will be visible in both.
1614 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1615 These functions cause every element in the array to move, which can be murder
1616 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1617 a different location. This will be addressed in the forthcoming version 1.00.
1619 =head2 WRITEONLY FILES
1621 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1622 writeonly mode. STORE will verify that the filehandle is writable. However, there
1623 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1624 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1628 This section discusses DBM::Deep's speed and memory usage.
1632 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1633 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1634 multi-level hash/array support, and cross-platform FTPable files. Even so,
1635 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1636 with huge databases. Here is some test data:
1638 Adding 1,000,000 keys to new DB file...
1640 At 100 keys, avg. speed is 2,703 keys/sec
1641 At 200 keys, avg. speed is 2,642 keys/sec
1642 At 300 keys, avg. speed is 2,598 keys/sec
1643 At 400 keys, avg. speed is 2,578 keys/sec
1644 At 500 keys, avg. speed is 2,722 keys/sec
1645 At 600 keys, avg. speed is 2,628 keys/sec
1646 At 700 keys, avg. speed is 2,700 keys/sec
1647 At 800 keys, avg. speed is 2,607 keys/sec
1648 At 900 keys, avg. speed is 2,190 keys/sec
1649 At 1,000 keys, avg. speed is 2,570 keys/sec
1650 At 2,000 keys, avg. speed is 2,417 keys/sec
1651 At 3,000 keys, avg. speed is 1,982 keys/sec
1652 At 4,000 keys, avg. speed is 1,568 keys/sec
1653 At 5,000 keys, avg. speed is 1,533 keys/sec
1654 At 6,000 keys, avg. speed is 1,787 keys/sec
1655 At 7,000 keys, avg. speed is 1,977 keys/sec
1656 At 8,000 keys, avg. speed is 2,028 keys/sec
1657 At 9,000 keys, avg. speed is 2,077 keys/sec
1658 At 10,000 keys, avg. speed is 2,031 keys/sec
1659 At 20,000 keys, avg. speed is 1,970 keys/sec
1660 At 30,000 keys, avg. speed is 2,050 keys/sec
1661 At 40,000 keys, avg. speed is 2,073 keys/sec
1662 At 50,000 keys, avg. speed is 1,973 keys/sec
1663 At 60,000 keys, avg. speed is 1,914 keys/sec
1664 At 70,000 keys, avg. speed is 2,091 keys/sec
1665 At 80,000 keys, avg. speed is 2,103 keys/sec
1666 At 90,000 keys, avg. speed is 1,886 keys/sec
1667 At 100,000 keys, avg. speed is 1,970 keys/sec
1668 At 200,000 keys, avg. speed is 2,053 keys/sec
1669 At 300,000 keys, avg. speed is 1,697 keys/sec
1670 At 400,000 keys, avg. speed is 1,838 keys/sec
1671 At 500,000 keys, avg. speed is 1,941 keys/sec
1672 At 600,000 keys, avg. speed is 1,930 keys/sec
1673 At 700,000 keys, avg. speed is 1,735 keys/sec
1674 At 800,000 keys, avg. speed is 1,795 keys/sec
1675 At 900,000 keys, avg. speed is 1,221 keys/sec
1676 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1678 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1679 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1680 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1681 Run time was 12 min 3 sec.
1685 One of the great things about DBM::Deep is that it uses very little memory.
1686 Even with huge databases (1,000,000+ keys) you will not see much increased
1687 memory on your process. DBM::Deep relies solely on the filesystem for storing
1688 and fetching data. Here is output from I</usr/bin/top> before even opening a
1691 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1692 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1694 Basically the process is taking 2,716K of memory. And here is the same
1695 process after storing and fetching 1,000,000 keys:
1697 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1698 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1700 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1701 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1703 =head1 DB FILE FORMAT
1705 In case you were interested in the underlying DB file format, it is documented
1706 here in this section. You don't need to know this to use the module, it's just
1707 included for reference.
1711 DBM::Deep files always start with a 32-bit signature to identify the file type.
1712 This is at offset 0. The signature is "DPDB" in network byte order. This is
1713 checked for when the file is opened and an error will be thrown if it's not found.
1717 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1718 has a standard header containing the type of data, the length of data, and then
1719 the data itself. The type is a single character (1 byte), the length is a
1720 32-bit unsigned long in network byte order, and the data is, well, the data.
1721 Here is how it unfolds:
1725 Immediately after the 32-bit file signature is the I<Master Index> record.
1726 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1727 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1728 depending on how the DBM::Deep object was constructed.
1730 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1731 number). The first 8-bit char of the MD5 signature is the offset into the
1732 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1733 index element is a file offset of the next tag for the key/element in question,
1734 which is usually a I<Bucket List> tag (see below).
1736 The next tag I<could> be another index, depending on how many keys/elements
1737 exist. See L<RE-INDEXING> below for details.
1741 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1742 file offsets to where the actual data is stored. It starts with a standard
1743 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1744 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1745 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1746 When the list fills up, a I<Re-Index> operation is performed (See
1747 L<RE-INDEXING> below).
1751 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1752 index/value pair (in array mode). It starts with a standard tag header with
1753 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1754 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1755 header. The size reported in the tag header is only for the value, but then,
1756 just after the value is another size (32-bit unsigned long) and then the plain
1757 key itself. Since the value is likely to be fetched more often than the plain
1758 key, I figured it would be I<slightly> faster to store the value first.
1760 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1761 record for the nested structure, where the process begins all over again.
1765 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1766 exhausted. Then, when another key/element comes in, the list is converted to a
1767 new index record. However, this index will look at the next char in the MD5
1768 hash, and arrange new Bucket List pointers accordingly. This process is called
1769 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1770 17 (16 + new one) keys/elements are removed from the old Bucket List and
1771 inserted into the new index. Several new Bucket Lists are created in the
1772 process, as a new MD5 char from the key is being examined (it is unlikely that
1773 the keys will all share the same next char of their MD5s).
1775 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1776 when the Bucket Lists will turn into indexes, but the first round tends to
1777 happen right around 4,000 keys. You will see a I<slight> decrease in
1778 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1779 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1780 right around 900,000 keys. This process can continue nearly indefinitely --
1781 right up until the point the I<MD5> signatures start colliding with each other,
1782 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1783 getting struck by lightning while you are walking to cash in your tickets.
1784 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1785 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1786 this is 340 unodecillion, but don't quote me).
1790 When a new key/element is stored, the key (or index number) is first run through
1791 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1792 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1793 for the first char of the signature (in this case I<b0>). If it does not exist,
1794 a new I<Bucket List> is created for our key (and the next 15 future keys that
1795 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1796 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1797 this point, unless we are replacing an existing I<Bucket>), where the actual
1798 data will be stored.
1802 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1803 (or index number), then walking along the indexes. If there are enough
1804 keys/elements in this DB level, there might be nested indexes, each linked to
1805 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1806 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1807 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1808 plain key are stored.
1810 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1811 methods. In this process the indexes are walked systematically, and each key
1812 fetched in increasing MD5 order (which is why it appears random). Once the
1813 I<Bucket> is found, the value is skipped and the plain key returned instead.
1814 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1815 alphabetically sorted. This only happens on an index-level -- as soon as the
1816 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1817 so it's pretty much undefined how the keys will come out -- just like Perl's
1820 =head1 CODE COVERAGE
1822 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1823 B<Devel::Cover> report on this module's test suite.
1825 ----------------------------------- ------ ------ ------ ------ ------ ------
1826 File stmt bran cond sub time total
1827 ----------------------------------- ------ ------ ------ ------ ------ ------
1828 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1829 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1830 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1831 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1832 Total 97.9 85.9 79.7 100.0 100.0 94.3
1833 ----------------------------------- ------ ------ ------ ------ ------ ------
1835 =head1 MORE INFORMATION
1837 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1838 or send email to L<DBM-Deep@googlegroups.com>.
1842 Joseph Huckaby, L<jhuckaby@cpan.org>
1844 Rob Kinyon, L<rkinyon@cpan.org>
1846 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1850 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1851 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1855 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1856 This is free software, you may use it and distribute it under the
1857 same terms as Perl itself.