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 if ( $self->_type eq TYPE_HASH ) {
645 my $key = $self->first_key;
647 my $next_key = $self->next_key( $key );
648 my $md5 = $self->_engine->{digest}->($key);
649 my $tag = $self->_engine->find_blist( $self->_base_offset, $md5 );
650 $self->_engine->delete_bucket( $tag, $md5, $key );
655 my $size = $self->FETCHSIZE;
656 for my $key ( map { pack ( $self->_engine->{long_pack}, $_ ) } 0 .. $size - 1 ) {
657 my $md5 = $self->_engine->{digest}->($key);
658 my $tag = $self->_engine->find_blist( $self->_base_offset, $md5 );
659 $self->_engine->delete_bucket( $tag, $md5, $key );
661 $self->STORESIZE( 0 );
663 #XXX This needs updating to use _release_space
664 # $self->_engine->write_tag(
665 # $self->_base_offset, $self->_type,
666 # chr(0)x$self->_engine->{index_size},
675 # Public method aliases
677 sub put { (shift)->STORE( @_ ) }
678 sub store { (shift)->STORE( @_ ) }
679 sub get { (shift)->FETCH( @_ ) }
680 sub fetch { (shift)->FETCH( @_ ) }
681 sub delete { (shift)->DELETE( @_ ) }
682 sub exists { (shift)->EXISTS( @_ ) }
683 sub clear { (shift)->CLEAR( @_ ) }
690 DBM::Deep - A pure perl multi-level hash/array DBM
695 my $db = DBM::Deep->new( "foo.db" );
697 $db->{key} = 'value';
700 $db->put('key' => 'value');
701 print $db->get('key');
703 # true multi-level support
704 $db->{my_complex} = [
705 'hello', { perl => 'rules' },
709 tie my %db, 'DBM::Deep', 'foo.db';
713 tied(%db)->put('key' => 'value');
714 print tied(%db)->get('key');
718 A unique flat-file database module, written in pure perl. True multi-level
719 hash/array support (unlike MLDBM, which is faked), hybrid OO / tie()
720 interface, cross-platform FTPable files, ACID transactions, and is quite fast.
721 Can handle millions of keys and unlimited levels without significant
722 slow-down. Written from the ground-up in pure perl -- this is NOT a wrapper
723 around a C-based DBM. Out-of-the-box compatibility with Unix, Mac OS X and
726 =head1 VERSION DIFFERENCES
728 B<NOTE>: 0.99_01 and above have significant file format differences from 0.983 and
729 before. There will be a backwards-compatibility layer in 1.00, but that is
730 slated for a later 0.99_x release. This version is B<NOT> backwards compatible
731 with 0.983 and before.
735 Construction can be done OO-style (which is the recommended way), or using
736 Perl's tie() function. Both are examined here.
738 =head2 OO CONSTRUCTION
740 The recommended way to construct a DBM::Deep object is to use the new()
741 method, which gets you a blessed I<and> tied hash (or array) reference.
743 my $db = DBM::Deep->new( "foo.db" );
745 This opens a new database handle, mapped to the file "foo.db". If this
746 file does not exist, it will automatically be created. DB files are
747 opened in "r+" (read/write) mode, and the type of object returned is a
748 hash, unless otherwise specified (see L<OPTIONS> below).
750 You can pass a number of options to the constructor to specify things like
751 locking, autoflush, etc. This is done by passing an inline hash (or hashref):
753 my $db = DBM::Deep->new(
759 Notice that the filename is now specified I<inside> the hash with
760 the "file" parameter, as opposed to being the sole argument to the
761 constructor. This is required if any options are specified.
762 See L<OPTIONS> below for the complete list.
764 You can also start with an array instead of a hash. For this, you must
765 specify the C<type> parameter:
767 my $db = DBM::Deep->new(
769 type => DBM::Deep->TYPE_ARRAY
772 B<Note:> Specifing the C<type> parameter only takes effect when beginning
773 a new DB file. If you create a DBM::Deep object with an existing file, the
774 C<type> will be loaded from the file header, and an error will be thrown if
775 the wrong type is passed in.
777 =head2 TIE CONSTRUCTION
779 Alternately, you can create a DBM::Deep handle by using Perl's built-in
780 tie() function. The object returned from tie() can be used to call methods,
781 such as lock() and unlock(). (That object can be retrieved from the tied
782 variable at any time using tied() - please see L<perltie/> for more info.
785 my $db = tie %hash, "DBM::Deep", "foo.db";
788 my $db = tie @array, "DBM::Deep", "bar.db";
790 As with the OO constructor, you can replace the DB filename parameter with
791 a hash containing one or more options (see L<OPTIONS> just below for the
794 tie %hash, "DBM::Deep", {
802 There are a number of options that can be passed in when constructing your
803 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
809 Filename of the DB file to link the handle to. You can pass a full absolute
810 filesystem path, partial path, or a plain filename if the file is in the
811 current working directory. This is a required parameter (though q.v. fh).
815 If you want, you can pass in the fh instead of the file. This is most useful for doing
818 my $db = DBM::Deep->new( { fh => \*DATA } );
820 You are responsible for making sure that the fh has been opened appropriately for your
821 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
822 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
823 needs to read from the fh.
825 =item * audit_file / audit_fh
827 These are just like file/fh, except for auditing. Please see L</AUDITING> for
832 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
833 not need to set this. However, it's there if you want it.
835 If you pass in fh and do not set this, it will be set appropriately.
839 This parameter specifies what type of object to create, a hash or array. Use
840 one of these two constants:
844 =item * C<DBM::Deep-E<gt>TYPE_HASH>
846 =item * C<DBM::Deep-E<gt>TYPE_ARRAY>.
850 This only takes effect when beginning a new file. This is an optional
851 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
855 Specifies whether locking is to be enabled. DBM::Deep uses Perl's flock()
856 function to lock the database in exclusive mode for writes, and shared mode
857 for reads. Pass any true value to enable. This affects the base DB handle
858 I<and any child hashes or arrays> that use the same DB file. This is an
859 optional parameter, and defaults to 0 (disabled). See L<LOCKING> below for
864 Specifies whether autoflush is to be enabled on the underlying filehandle.
865 This obviously slows down write operations, but is required if you may have
866 multiple processes accessing the same DB file (also consider enable I<locking>).
867 Pass any true value to enable. This is an optional parameter, and defaults to 0
872 If I<autobless> mode is enabled, DBM::Deep will preserve the class something
873 is blessed into, and restores it when fetched. This is an optional parameter, and defaults to 1 (enabled).
875 B<Note:> If you use the OO-interface, you will not be able to call any methods
876 of DBM::Deep on the blessed item. This is considered to be a feature.
880 See L</FILTERS> below.
886 With DBM::Deep you can access your databases using Perl's standard hash/array
887 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
888 treat them as such. DBM::Deep will intercept all reads/writes and direct them
889 to the right place -- the DB file. This has nothing to do with the
890 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
891 using regular hashes and arrays, rather than calling functions like C<get()>
892 and C<put()> (although those work too). It is entirely up to you how to want
893 to access your databases.
897 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
898 or even nested hashes (or arrays) using standard Perl syntax:
900 my $db = DBM::Deep->new( "foo.db" );
902 $db->{mykey} = "myvalue";
904 $db->{myhash}->{subkey} = "subvalue";
906 print $db->{myhash}->{subkey} . "\n";
908 You can even step through hash keys using the normal Perl C<keys()> function:
910 foreach my $key (keys %$db) {
911 print "$key: " . $db->{$key} . "\n";
914 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
915 pushes them onto an array, all before the loop even begins. If you have an
916 extremely large hash, this may exhaust Perl's memory. Instead, consider using
917 Perl's C<each()> function, which pulls keys/values one at a time, using very
920 while (my ($key, $value) = each %$db) {
921 print "$key: $value\n";
924 Please note that when using C<each()>, you should always pass a direct
925 hash reference, not a lookup. Meaning, you should B<never> do this:
928 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
930 This causes an infinite loop, because for each iteration, Perl is calling
931 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
932 it effectively keeps returning the first key over and over again. Instead,
933 assign a temporary variable to C<$db->{foo}>, then pass that to each().
937 As with hashes, you can treat any DBM::Deep object like a normal Perl array
938 reference. This includes inserting, removing and manipulating elements,
939 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
940 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
941 or simply be a nested array reference inside a hash. Example:
943 my $db = DBM::Deep->new(
944 file => "foo-array.db",
945 type => DBM::Deep->TYPE_ARRAY
949 push @$db, "bar", "baz";
952 my $last_elem = pop @$db; # baz
953 my $first_elem = shift @$db; # bah
954 my $second_elem = $db->[1]; # bar
956 my $num_elements = scalar @$db;
960 In addition to the I<tie()> interface, you can also use a standard OO interface
961 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
962 array) has its own methods, but both types share the following common methods:
963 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>. C<fetch()> and
964 C<store(> are aliases to C<put()> and C<get()>, respectively.
968 =item * new() / clone()
970 These are the constructor and copy-functions.
972 =item * put() / store()
974 Stores a new hash key/value pair, or sets an array element value. Takes two
975 arguments, the hash key or array index, and the new value. The value can be
976 a scalar, hash ref or array ref. Returns true on success, false on failure.
978 $db->put("foo", "bar"); # for hashes
979 $db->put(1, "bar"); # for arrays
981 =item * get() / fetch()
983 Fetches the value of a hash key or array element. Takes one argument: the hash
984 key or array index. Returns a scalar, hash ref or array ref, depending on the
987 my $value = $db->get("foo"); # for hashes
988 my $value = $db->get(1); # for arrays
992 Checks if a hash key or array index exists. Takes one argument: the hash key
993 or array index. Returns true if it exists, false if not.
995 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
996 if ($db->exists(1)) { print "yay!\n"; } # for arrays
1000 Deletes one hash key/value pair or array element. Takes one argument: the hash
1001 key or array index. Returns true on success, false if not found. For arrays,
1002 the remaining elements located after the deleted element are NOT moved over.
1003 The deleted element is essentially just undefined, which is exactly how Perl's
1004 internal arrays work. Please note that the space occupied by the deleted
1005 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1006 below for details and workarounds.
1008 $db->delete("foo"); # for hashes
1009 $db->delete(1); # for arrays
1013 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1014 value. Please note that the space occupied by the deleted keys/values or
1015 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1016 details and workarounds.
1018 $db->clear(); # hashes or arrays
1020 =item * lock() / unlock()
1026 Recover lost disk space. This is important to do, especially if you use
1029 =item * import() / export()
1031 Data going in and out.
1037 For hashes, DBM::Deep supports all the common methods described above, and the
1038 following additional methods: C<first_key()> and C<next_key()>.
1044 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1045 fetched in an undefined order (which appears random). Takes no arguments,
1046 returns the key as a scalar value.
1048 my $key = $db->first_key();
1052 Returns the "next" key in the hash, given the previous one as the sole argument.
1053 Returns undef if there are no more keys to be fetched.
1055 $key = $db->next_key($key);
1059 Here are some examples of using hashes:
1061 my $db = DBM::Deep->new( "foo.db" );
1063 $db->put("foo", "bar");
1064 print "foo: " . $db->get("foo") . "\n";
1066 $db->put("baz", {}); # new child hash ref
1067 $db->get("baz")->put("buz", "biz");
1068 print "buz: " . $db->get("baz")->get("buz") . "\n";
1070 my $key = $db->first_key();
1072 print "$key: " . $db->get($key) . "\n";
1073 $key = $db->next_key($key);
1076 if ($db->exists("foo")) { $db->delete("foo"); }
1080 For arrays, DBM::Deep supports all the common methods described above, and the
1081 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1082 C<unshift()> and C<splice()>.
1088 Returns the number of elements in the array. Takes no arguments.
1090 my $len = $db->length();
1094 Adds one or more elements onto the end of the array. Accepts scalars, hash
1095 refs or array refs. No return value.
1097 $db->push("foo", "bar", {});
1101 Fetches the last element in the array, and deletes it. Takes no arguments.
1102 Returns undef if array is empty. Returns the element value.
1104 my $elem = $db->pop();
1108 Fetches the first element in the array, deletes it, then shifts all the
1109 remaining elements over to take up the space. Returns the element value. This
1110 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1113 my $elem = $db->shift();
1117 Inserts one or more elements onto the beginning of the array, shifting all
1118 existing elements over to make room. Accepts scalars, hash refs or array refs.
1119 No return value. This method is not recommended with large arrays -- see
1120 <LARGE ARRAYS> below for details.
1122 $db->unshift("foo", "bar", {});
1126 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1127 -f splice> for usage -- it is too complicated to document here. This method is
1128 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1132 Here are some examples of using arrays:
1134 my $db = DBM::Deep->new(
1136 type => DBM::Deep->TYPE_ARRAY
1139 $db->push("bar", "baz");
1140 $db->unshift("foo");
1143 my $len = $db->length();
1144 print "length: $len\n"; # 4
1146 for (my $k=0; $k<$len; $k++) {
1147 print "$k: " . $db->get($k) . "\n";
1150 $db->splice(1, 2, "biz", "baf");
1152 while (my $elem = shift @$db) {
1153 print "shifted: $elem\n";
1158 Enable automatic file locking by passing a true value to the C<locking>
1159 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1161 my $db = DBM::Deep->new(
1166 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1167 mode for writes, and shared mode for reads. This is required if you have
1168 multiple processes accessing the same database file, to avoid file corruption.
1169 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1170 NFS> below for more.
1172 =head2 EXPLICIT LOCKING
1174 You can explicitly lock a database, so it remains locked for multiple
1175 transactions. This is done by calling the C<lock()> method, and passing an
1176 optional lock mode argument (defaults to exclusive mode). This is particularly
1177 useful for things like counters, where the current value needs to be fetched,
1178 then incremented, then stored again.
1181 my $counter = $db->get("counter");
1183 $db->put("counter", $counter);
1192 You can pass C<lock()> an optional argument, which specifies which mode to use
1193 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1194 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1195 same as the constants defined in Perl's C<Fcntl> module.
1197 $db->lock( DBM::Deep->LOCK_SH );
1201 =head1 IMPORTING/EXPORTING
1203 You can import existing complex structures by calling the C<import()> method,
1204 and export an entire database into an in-memory structure using the C<export()>
1205 method. Both are examined here.
1209 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1210 walking the structure and adding keys/elements to the database as you go,
1211 simply pass a reference to the C<import()> method. This recursively adds
1212 everything to an existing DBM::Deep object for you. Here is an example:
1217 array1 => [ "elem0", "elem1", "elem2" ],
1219 subkey1 => "subvalue1",
1220 subkey2 => "subvalue2"
1224 my $db = DBM::Deep->new( "foo.db" );
1225 $db->import( $struct );
1227 print $db->{key1} . "\n"; # prints "value1"
1229 This recursively imports the entire C<$struct> object into C<$db>, including
1230 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1231 keys are merged with the existing ones, replacing if they already exist.
1232 The C<import()> method can be called on any database level (not just the base
1233 level), and works with both hash and array DB types.
1235 B<Note:> Make sure your existing structure has no circular references in it.
1236 These will cause an infinite loop when importing. There are plans to fix this
1241 Calling the C<export()> method on an existing DBM::Deep object will return
1242 a reference to a new in-memory copy of the database. The export is done
1243 recursively, so all nested hashes/arrays are all exported to standard Perl
1244 objects. Here is an example:
1246 my $db = DBM::Deep->new( "foo.db" );
1248 $db->{key1} = "value1";
1249 $db->{key2} = "value2";
1251 $db->{hash1}->{subkey1} = "subvalue1";
1252 $db->{hash1}->{subkey2} = "subvalue2";
1254 my $struct = $db->export();
1256 print $struct->{key1} . "\n"; # prints "value1"
1258 This makes a complete copy of the database in memory, and returns a reference
1259 to it. The C<export()> method can be called on any database level (not just
1260 the base level), and works with both hash and array DB types. Be careful of
1261 large databases -- you can store a lot more data in a DBM::Deep object than an
1262 in-memory Perl structure.
1264 B<Note:> Make sure your database has no circular references in it.
1265 These will cause an infinite loop when exporting. There are plans to fix this
1270 DBM::Deep has a number of hooks where you can specify your own Perl function
1271 to perform filtering on incoming or outgoing data. This is a perfect
1272 way to extend the engine, and implement things like real-time compression or
1273 encryption. Filtering applies to the base DB level, and all child hashes /
1274 arrays. Filter hooks can be specified when your DBM::Deep object is first
1275 constructed, or by calling the C<set_filter()> method at any time. There are
1276 four available filter hooks, described below:
1280 =item * filter_store_key
1282 This filter is called whenever a hash key is stored. It
1283 is passed the incoming key, and expected to return a transformed key.
1285 =item * filter_store_value
1287 This filter is called whenever a hash key or array element is stored. It
1288 is passed the incoming value, and expected to return a transformed value.
1290 =item * filter_fetch_key
1292 This filter is called whenever a hash key is fetched (i.e. via
1293 C<first_key()> or C<next_key()>). It is passed the transformed key,
1294 and expected to return the plain key.
1296 =item * filter_fetch_value
1298 This filter is called whenever a hash key or array element is fetched.
1299 It is passed the transformed value, and expected to return the plain value.
1303 Here are the two ways to setup a filter hook:
1305 my $db = DBM::Deep->new(
1307 filter_store_value => \&my_filter_store,
1308 filter_fetch_value => \&my_filter_fetch
1313 $db->set_filter( "filter_store_value", \&my_filter_store );
1314 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1316 Your filter function will be called only when dealing with SCALAR keys or
1317 values. When nested hashes and arrays are being stored/fetched, filtering
1318 is bypassed. Filters are called as static functions, passed a single SCALAR
1319 argument, and expected to return a single SCALAR value. If you want to
1320 remove a filter, set the function reference to C<undef>:
1322 $db->set_filter( "filter_store_value", undef );
1324 =head2 REAL-TIME ENCRYPTION EXAMPLE
1326 Here is a working example that uses the I<Crypt::Blowfish> module to
1327 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1328 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1329 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1332 use Crypt::Blowfish;
1335 my $cipher = Crypt::CBC->new({
1336 'key' => 'my secret key',
1337 'cipher' => 'Blowfish',
1339 'regenerate_key' => 0,
1340 'padding' => 'space',
1344 my $db = DBM::Deep->new(
1345 file => "foo-encrypt.db",
1346 filter_store_key => \&my_encrypt,
1347 filter_store_value => \&my_encrypt,
1348 filter_fetch_key => \&my_decrypt,
1349 filter_fetch_value => \&my_decrypt,
1352 $db->{key1} = "value1";
1353 $db->{key2} = "value2";
1354 print "key1: " . $db->{key1} . "\n";
1355 print "key2: " . $db->{key2} . "\n";
1361 return $cipher->encrypt( $_[0] );
1364 return $cipher->decrypt( $_[0] );
1367 =head2 REAL-TIME COMPRESSION EXAMPLE
1369 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1370 compression / decompression of keys & values with DBM::Deep Filters.
1371 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1372 more on I<Compress::Zlib>.
1377 my $db = DBM::Deep->new(
1378 file => "foo-compress.db",
1379 filter_store_key => \&my_compress,
1380 filter_store_value => \&my_compress,
1381 filter_fetch_key => \&my_decompress,
1382 filter_fetch_value => \&my_decompress,
1385 $db->{key1} = "value1";
1386 $db->{key2} = "value2";
1387 print "key1: " . $db->{key1} . "\n";
1388 print "key2: " . $db->{key2} . "\n";
1394 return Compress::Zlib::memGzip( $_[0] ) ;
1397 return Compress::Zlib::memGunzip( $_[0] ) ;
1400 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1401 actually numerical index numbers, and are not filtered.
1403 =head1 ERROR HANDLING
1405 Most DBM::Deep methods return a true value for success, and call die() on
1406 failure. You can wrap calls in an eval block to catch the die.
1408 my $db = DBM::Deep->new( "foo.db" ); # create hash
1409 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1411 print $@; # prints error message
1413 =head1 LARGEFILE SUPPORT
1415 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1416 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1417 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1418 by specifying the 'pack_size' parameter when constructing the file.
1421 filename => $filename,
1422 pack_size => 'large',
1425 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1426 instead of 32-bit longs. After setting these values your DB files have a
1427 theoretical maximum size of 16 XB (exabytes).
1429 You can also use C<pack_size =E<gt> 'small'> in order to use 16-bit file
1432 B<Note:> Changing these values will B<NOT> work for existing database files.
1433 Only change this for new files. Once the value has been set, it is stored in
1434 the file's header and cannot be changed for the life of the file. These
1435 parameters are per-file, meaning you can access 32-bit and 64-bit files, as
1438 B<Note:> We have not personally tested files larger than 2 GB -- all my
1439 systems have only a 32-bit Perl. However, I have received user reports that
1440 this does indeed work!
1442 =head1 LOW-LEVEL ACCESS
1444 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1445 you can call the C<_fh()> method, which returns the handle:
1447 my $fh = $db->_fh();
1449 This method can be called on the root level of the datbase, or any child
1450 hashes or arrays. All levels share a I<root> structure, which contains things
1451 like the filehandle, a reference counter, and all the options specified
1452 when you created the object. You can get access to this file object by
1453 calling the C<_fileobj()> method.
1455 my $file_obj = $db->_fileobj();
1457 This is useful for changing options after the object has already been created,
1458 such as enabling/disabling locking. You can also store your own temporary user
1459 data in this structure (be wary of name collision), which is then accessible from
1460 any child hash or array.
1462 =head1 CUSTOM DIGEST ALGORITHM
1464 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1465 keys. However you can override this, and use another algorithm (such as SHA-256)
1466 or even write your own. But please note that DBM::Deep currently expects zero
1467 collisions, so your algorithm has to be I<perfect>, so to speak. Collision
1468 detection may be introduced in a later version.
1470 You can specify a custom digest algorithm by passing it into the parameter
1471 list for new(), passing a reference to a subroutine as the 'digest' parameter,
1472 and the length of the algorithm's hashes (in bytes) as the 'hash_size'
1473 parameter. Here is a working example that uses a 256-bit hash from the
1474 I<Digest::SHA256> module. Please see
1475 L<http://search.cpan.org/search?module=Digest::SHA256> for more information.
1480 my $context = Digest::SHA256::new(256);
1482 my $db = DBM::Deep->new(
1483 filename => "foo-sha.db",
1484 digest => \&my_digest,
1488 $db->{key1} = "value1";
1489 $db->{key2} = "value2";
1490 print "key1: " . $db->{key1} . "\n";
1491 print "key2: " . $db->{key2} . "\n";
1497 return substr( $context->hash($_[0]), 0, 32 );
1500 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1501 of bytes you specify in the hash_size parameter (in this case 32).
1503 B<Note:> If you do choose to use a custom digest algorithm, you must set it
1504 every time you access this file. Otherwise, the default (MD5) will be used.
1506 =head1 CIRCULAR REFERENCES
1508 DBM::Deep has B<experimental> support for circular references. Meaning you
1509 can have a nested hash key or array element that points to a parent object.
1510 This relationship is stored in the DB file, and is preserved between sessions.
1513 my $db = DBM::Deep->new( "foo.db" );
1516 $db->{circle} = $db; # ref to self
1518 print $db->{foo} . "\n"; # prints "bar"
1519 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1521 B<Note>: Passing the object to a function that recursively walks the
1522 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1523 C<export()> methods) will result in an infinite loop. This will be fixed in
1528 New in 0.99_01 is the ability to audit your databases actions. By passing in
1529 audit_file (or audit_fh) to the constructor, all actions will be logged to
1530 that file. The format is one that is suitable for eval'ing against the
1531 database to replay the actions. Please see t/33_audit_trail.t for an example
1536 New in 0.99_01 is ACID transactions. Every DBM::Deep object is completely
1537 transaction-ready - it is not an option you have to turn on. Three new methods
1538 have been added to support them. They are:
1542 =item * begin_work()
1544 This starts a transaction.
1548 This applies the changes done within the transaction to the mainline and ends
1553 This discards the changes done within the transaction to the mainline and ends
1558 Transactions in DBM::Deep are done using the MVCC method, the same method used
1559 by the InnoDB MySQL table type.
1561 =head1 CAVEATS / ISSUES / BUGS
1563 This section describes all the known issues with DBM::Deep. It you have found
1564 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1566 =head2 UNUSED SPACE RECOVERY
1568 One major caveat with DBM::Deep is that space occupied by existing keys and
1569 values is not recovered when they are deleted. Meaning if you keep deleting
1570 and adding new keys, your file will continuously grow. I am working on this,
1571 but in the meantime you can call the built-in C<optimize()> method from time to
1572 time (perhaps in a crontab or something) to recover all your unused space.
1574 $db->optimize(); # returns true on success
1576 This rebuilds the ENTIRE database into a new file, then moves it on top of
1577 the original. The new file will have no unused space, thus it will take up as
1578 little disk space as possible. Please note that this operation can take
1579 a long time for large files, and you need enough disk space to temporarily hold
1580 2 copies of your DB file. The temporary file is created in the same directory
1581 as the original, named with a ".tmp" extension, and is deleted when the
1582 operation completes. Oh, and if locking is enabled, the DB is automatically
1583 locked for the entire duration of the copy.
1585 B<WARNING:> Only call optimize() on the top-level node of the database, and
1586 make sure there are no child references lying around. DBM::Deep keeps a reference
1587 counter, and if it is greater than 1, optimize() will abort and return undef.
1591 (The reasons given assume a high level of Perl understanding, specifically of
1592 references. You can safely skip this section.)
1594 Currently, the only references supported are HASH and ARRAY. The other reference
1595 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1601 These are things like filehandles and other sockets. They can't be supported
1602 because it's completely unclear how DBM::Deep should serialize them.
1604 =item * SCALAR / REF
1606 The discussion here refers to the following type of example:
1613 # In some other process ...
1615 my $val = ${ $db->{key1} };
1617 is( $val, 50, "What actually gets stored in the DB file?" );
1619 The problem is one of synchronization. When the variable being referred to
1620 changes value, the reference isn't notified. This means that the new value won't
1621 be stored in the datafile for other processes to read. There is no TIEREF.
1623 It is theoretically possible to store references to values already within a
1624 DBM::Deep object because everything already is synchronized, but the change to
1625 the internals would be quite large. Specifically, DBM::Deep would have to tie
1626 every single value that is stored. This would bloat the RAM footprint of
1627 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1628 all to support a feature that has never been requested.
1632 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1633 mechanism for serializing coderefs, including saving off all closure state.
1634 However, just as for SCALAR and REF, that closure state may change without
1635 notifying the DBM::Deep object storing the reference.
1639 =head2 FILE CORRUPTION
1641 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1642 for a 32-bit signature when opened, but other corruption in files can cause
1643 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1644 stuck in an infinite loop depending on the level of corruption. File write
1645 operations are not checked for failure (for speed), so if you happen to run
1646 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1647 be addressed in a later version of DBM::Deep.
1651 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1652 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1653 about setting up your NFS server with a locking daemon, then using lockf() to
1654 lock your files, but your mileage may vary there as well. From what I
1655 understand, there is no real way to do it. However, if you need access to the
1656 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1657 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1659 =head2 COPYING OBJECTS
1661 Beware of copying tied objects in Perl. Very strange things can happen.
1662 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1663 returns a new, blessed, tied hash or array to the same level in the DB.
1665 my $copy = $db->clone();
1667 B<Note>: Since clone() here is cloning the object, not the database location, any
1668 modifications to either $db or $copy will be visible in both.
1672 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1673 These functions cause every element in the array to move, which can be murder
1674 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1675 a different location. This will be addressed in the forthcoming version 1.00.
1677 =head2 WRITEONLY FILES
1679 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1680 writeonly mode. STORE will verify that the filehandle is writable. However, there
1681 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1682 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1686 This section discusses DBM::Deep's speed and memory usage.
1690 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1691 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1692 multi-level hash/array support, and cross-platform FTPable files. Even so,
1693 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1694 with huge databases. Here is some test data:
1696 Adding 1,000,000 keys to new DB file...
1698 At 100 keys, avg. speed is 2,703 keys/sec
1699 At 200 keys, avg. speed is 2,642 keys/sec
1700 At 300 keys, avg. speed is 2,598 keys/sec
1701 At 400 keys, avg. speed is 2,578 keys/sec
1702 At 500 keys, avg. speed is 2,722 keys/sec
1703 At 600 keys, avg. speed is 2,628 keys/sec
1704 At 700 keys, avg. speed is 2,700 keys/sec
1705 At 800 keys, avg. speed is 2,607 keys/sec
1706 At 900 keys, avg. speed is 2,190 keys/sec
1707 At 1,000 keys, avg. speed is 2,570 keys/sec
1708 At 2,000 keys, avg. speed is 2,417 keys/sec
1709 At 3,000 keys, avg. speed is 1,982 keys/sec
1710 At 4,000 keys, avg. speed is 1,568 keys/sec
1711 At 5,000 keys, avg. speed is 1,533 keys/sec
1712 At 6,000 keys, avg. speed is 1,787 keys/sec
1713 At 7,000 keys, avg. speed is 1,977 keys/sec
1714 At 8,000 keys, avg. speed is 2,028 keys/sec
1715 At 9,000 keys, avg. speed is 2,077 keys/sec
1716 At 10,000 keys, avg. speed is 2,031 keys/sec
1717 At 20,000 keys, avg. speed is 1,970 keys/sec
1718 At 30,000 keys, avg. speed is 2,050 keys/sec
1719 At 40,000 keys, avg. speed is 2,073 keys/sec
1720 At 50,000 keys, avg. speed is 1,973 keys/sec
1721 At 60,000 keys, avg. speed is 1,914 keys/sec
1722 At 70,000 keys, avg. speed is 2,091 keys/sec
1723 At 80,000 keys, avg. speed is 2,103 keys/sec
1724 At 90,000 keys, avg. speed is 1,886 keys/sec
1725 At 100,000 keys, avg. speed is 1,970 keys/sec
1726 At 200,000 keys, avg. speed is 2,053 keys/sec
1727 At 300,000 keys, avg. speed is 1,697 keys/sec
1728 At 400,000 keys, avg. speed is 1,838 keys/sec
1729 At 500,000 keys, avg. speed is 1,941 keys/sec
1730 At 600,000 keys, avg. speed is 1,930 keys/sec
1731 At 700,000 keys, avg. speed is 1,735 keys/sec
1732 At 800,000 keys, avg. speed is 1,795 keys/sec
1733 At 900,000 keys, avg. speed is 1,221 keys/sec
1734 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1736 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1737 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1738 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1739 Run time was 12 min 3 sec.
1743 One of the great things about DBM::Deep is that it uses very little memory.
1744 Even with huge databases (1,000,000+ keys) you will not see much increased
1745 memory on your process. DBM::Deep relies solely on the filesystem for storing
1746 and fetching data. Here is output from I</usr/bin/top> before even opening a
1749 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1750 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1752 Basically the process is taking 2,716K of memory. And here is the same
1753 process after storing and fetching 1,000,000 keys:
1755 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1756 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1758 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1759 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1761 =head1 DB FILE FORMAT
1763 In case you were interested in the underlying DB file format, it is documented
1764 here in this section. You don't need to know this to use the module, it's just
1765 included for reference.
1769 DBM::Deep files always start with a 32-bit signature to identify the file type.
1770 This is at offset 0. The signature is "DPDB" in network byte order. This is
1771 checked for when the file is opened and an error will be thrown if it's not found.
1775 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1776 has a standard header containing the type of data, the length of data, and then
1777 the data itself. The type is a single character (1 byte), the length is a
1778 32-bit unsigned long in network byte order, and the data is, well, the data.
1779 Here is how it unfolds:
1783 Immediately after the 32-bit file signature is the I<Master Index> record.
1784 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1785 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1786 depending on how the DBM::Deep object was constructed.
1788 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1789 number). The first 8-bit char of the MD5 signature is the offset into the
1790 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1791 index element is a file offset of the next tag for the key/element in question,
1792 which is usually a I<Bucket List> tag (see below).
1794 The next tag I<could> be another index, depending on how many keys/elements
1795 exist. See L<RE-INDEXING> below for details.
1799 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1800 file offsets to where the actual data is stored. It starts with a standard
1801 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1802 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1803 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1804 When the list fills up, a I<Re-Index> operation is performed (See
1805 L<RE-INDEXING> below).
1809 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1810 index/value pair (in array mode). It starts with a standard tag header with
1811 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1812 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1813 header. The size reported in the tag header is only for the value, but then,
1814 just after the value is another size (32-bit unsigned long) and then the plain
1815 key itself. Since the value is likely to be fetched more often than the plain
1816 key, I figured it would be I<slightly> faster to store the value first.
1818 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1819 record for the nested structure, where the process begins all over again.
1823 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1824 exhausted. Then, when another key/element comes in, the list is converted to a
1825 new index record. However, this index will look at the next char in the MD5
1826 hash, and arrange new Bucket List pointers accordingly. This process is called
1827 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1828 17 (16 + new one) keys/elements are removed from the old Bucket List and
1829 inserted into the new index. Several new Bucket Lists are created in the
1830 process, as a new MD5 char from the key is being examined (it is unlikely that
1831 the keys will all share the same next char of their MD5s).
1833 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1834 when the Bucket Lists will turn into indexes, but the first round tends to
1835 happen right around 4,000 keys. You will see a I<slight> decrease in
1836 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1837 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1838 right around 900,000 keys. This process can continue nearly indefinitely --
1839 right up until the point the I<MD5> signatures start colliding with each other,
1840 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1841 getting struck by lightning while you are walking to cash in your tickets.
1842 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1843 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1844 this is 340 unodecillion, but don't quote me).
1848 When a new key/element is stored, the key (or index number) is first run through
1849 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1850 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1851 for the first char of the signature (in this case I<b0>). If it does not exist,
1852 a new I<Bucket List> is created for our key (and the next 15 future keys that
1853 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1854 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1855 this point, unless we are replacing an existing I<Bucket>), where the actual
1856 data will be stored.
1860 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1861 (or index number), then walking along the indexes. If there are enough
1862 keys/elements in this DB level, there might be nested indexes, each linked to
1863 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1864 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1865 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1866 plain key are stored.
1868 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1869 methods. In this process the indexes are walked systematically, and each key
1870 fetched in increasing MD5 order (which is why it appears random). Once the
1871 I<Bucket> is found, the value is skipped and the plain key returned instead.
1872 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1873 alphabetically sorted. This only happens on an index-level -- as soon as the
1874 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1875 so it's pretty much undefined how the keys will come out -- just like Perl's
1878 =head1 CODE COVERAGE
1880 B<Devel::Cover> is used to test the code coverage of the tests. Below is the
1881 B<Devel::Cover> report on this distribution's test suite.
1883 ---------------------------- ------ ------ ------ ------ ------ ------ ------
1884 File stmt bran cond sub pod time total
1885 ---------------------------- ------ ------ ------ ------ ------ ------ ------
1886 blib/lib/DBM/Deep.pm 96.2 89.0 75.0 95.8 89.5 36.0 92.9
1887 blib/lib/DBM/Deep/Array.pm 96.1 88.3 100.0 96.4 100.0 15.9 94.7
1888 blib/lib/DBM/Deep/Engine.pm 96.6 86.6 89.5 100.0 0.0 20.0 91.0
1889 blib/lib/DBM/Deep/File.pm 99.4 88.3 55.6 100.0 0.0 19.6 89.5
1890 blib/lib/DBM/Deep/Hash.pm 98.5 83.3 100.0 100.0 100.0 8.5 96.3
1891 Total 96.9 87.4 81.2 98.0 38.5 100.0 92.1
1892 ---------------------------- ------ ------ ------ ------ ------ ------ ------
1894 =head1 MORE INFORMATION
1896 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1897 or send email to L<DBM-Deep@googlegroups.com>. You can also visit #dbm-deep on
1902 Rob Kinyon, L<rkinyon@cpan.org>
1904 Originally written by Joseph Huckaby, L<jhuckaby@cpan.org>
1906 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1910 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1911 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1915 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1916 This is free software, you may use it and distribute it under the
1917 same terms as Perl itself.