7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use DBM::Deep::Engine;
40 use vars qw( $VERSION );
41 $VERSION = q(0.99_01);
44 # Setup constants for users to pass to new()
46 sub TYPE_HASH () { DBM::Deep::Engine->SIG_HASH }
47 sub TYPE_ARRAY () { DBM::Deep::Engine->SIG_ARRAY }
55 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
60 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
61 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
66 $args = { file => shift };
74 # Class constructor method for Perl OO interface.
75 # Calls tie() and returns blessed reference to tied hash or array,
76 # providing a hybrid OO/tie interface.
79 my $args = $class->_get_args( @_ );
82 # Check if we want a tied hash or array.
85 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
86 $class = 'DBM::Deep::Array';
87 require DBM::Deep::Array;
88 tie @$self, $class, %$args;
91 $class = 'DBM::Deep::Hash';
92 require DBM::Deep::Hash;
93 tie %$self, $class, %$args;
96 return bless $self, $class;
101 # Setup $self and bless into this class.
106 # These are the defaults to be optionally overridden below
109 engine => DBM::Deep::Engine->new,
110 base_offset => undef,
113 foreach my $param ( keys %$self ) {
114 next unless exists $args->{$param};
115 $self->{$param} = delete $args->{$param}
118 # locking implicitly enables autoflush
119 if ($args->{locking}) { $args->{autoflush} = 1; }
121 $self->{root} = exists $args->{root}
123 : DBM::Deep::_::Root->new( $args );
125 $self->{engine}->setup_fh( $self );
132 require DBM::Deep::Hash;
133 return DBM::Deep::Hash->TIEHASH( @_ );
138 require DBM::Deep::Array;
139 return DBM::Deep::Array->TIEARRAY( @_ );
142 #XXX Unneeded now ...
148 # If db locking is set, flock() the db file. If called multiple
149 # times before unlock(), then the same number of unlocks() must
150 # be called before the lock is released.
152 my $self = shift->_get_self;
154 $type = LOCK_EX unless defined $type;
156 if (!defined($self->_fh)) { return; }
158 if ($self->_root->{locking}) {
159 if (!$self->_root->{locked}) {
160 flock($self->_fh, $type);
162 # refresh end counter in case file has changed size
163 my @stats = stat($self->_fh);
164 $self->_root->{end} = $stats[7];
166 # double-check file inode, in case another process
167 # has optimize()d our file while we were waiting.
168 if ($stats[1] != $self->_root->{inode}) {
169 $self->{engine}->close_fh( $self );
170 $self->{engine}->setup_fh( $self );
171 flock($self->_fh, $type); # re-lock
173 # This may not be necessary after re-opening
174 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
177 $self->_root->{locked}++;
187 # If db locking is set, unlock the db file. See note in lock()
188 # regarding calling lock() multiple times.
190 my $self = shift->_get_self;
192 if (!defined($self->_fh)) { return; }
194 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
195 $self->_root->{locked}--;
196 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
205 my $self = shift->_get_self;
206 my ($spot, $value) = @_;
211 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
212 my $type = $value->_type;
213 ${$spot} = $type eq TYPE_HASH ? {} : [];
214 $value->_copy_node( ${$spot} );
217 my $r = Scalar::Util::reftype( $value );
218 my $c = Scalar::Util::blessed( $value );
219 if ( $r eq 'ARRAY' ) {
220 ${$spot} = [ @{$value} ];
223 ${$spot} = { %{$value} };
225 ${$spot} = bless ${$spot}, $c
234 # Copy single level of keys or elements to new DB handle.
235 # Recurse for nested structures
237 my $self = shift->_get_self;
240 if ($self->_type eq TYPE_HASH) {
241 my $key = $self->first_key();
243 my $value = $self->get($key);
244 $self->_copy_value( \$db_temp->{$key}, $value );
245 $key = $self->next_key($key);
249 my $length = $self->length();
250 for (my $index = 0; $index < $length; $index++) {
251 my $value = $self->get($index);
252 $self->_copy_value( \$db_temp->[$index], $value );
261 # Recursively export into standard Perl hashes and arrays.
263 my $self = shift->_get_self;
266 if ($self->_type eq TYPE_HASH) { $temp = {}; }
267 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
270 $self->_copy_node( $temp );
278 # Recursively import Perl hash/array structure
280 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
282 my $self = shift->_get_self;
285 # struct is not a reference, so just import based on our type
287 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
288 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
291 my $r = Scalar::Util::reftype($struct) || '';
292 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
293 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
295 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
296 $self->push( @$struct );
299 $self->_throw_error("Cannot import: type mismatch");
307 # Rebuild entire database into new file, then move
308 # it back on top of original.
310 my $self = shift->_get_self;
312 #XXX Need to create a new test for this
313 # if ($self->_root->{links} > 1) {
314 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
317 my $db_temp = DBM::Deep->new(
318 file => $self->_root->{file} . '.tmp',
323 $self->_copy_node( $db_temp );
327 # Attempt to copy user, group and permissions over to new file
329 my @stats = stat($self->_fh);
330 my $perms = $stats[2] & 07777;
333 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
334 chmod( $perms, $self->_root->{file} . '.tmp' );
336 # q.v. perlport for more information on this variable
337 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
339 # Potential race condition when optmizing on Win32 with locking.
340 # The Windows filesystem requires that the filehandle be closed
341 # before it is overwritten with rename(). This could be redone
345 $self->{engine}->close_fh( $self );
348 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
349 unlink $self->_root->{file} . '.tmp';
351 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
355 $self->{engine}->close_fh( $self );
356 $self->{engine}->setup_fh( $self );
363 # Make copy of object and return
365 my $self = shift->_get_self;
367 return DBM::Deep->new(
368 type => $self->_type,
369 base_offset => $self->_base_offset,
375 my %is_legal_filter = map {
378 store_key store_value
379 fetch_key fetch_value
384 # Setup filter function for storing or fetching the key or value
386 my $self = shift->_get_self;
390 if ( $is_legal_filter{$type} ) {
391 $self->_root->{"filter_$type"} = $func;
405 # Get access to the root structure
407 my $self = $_[0]->_get_self;
408 return $self->{root};
413 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
415 my $self = $_[0]->_get_self;
416 return $self->{type};
421 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
423 my $self = $_[0]->_get_self;
424 return $self->{base_offset};
429 # Get access to the raw fh
431 my $self = $_[0]->_get_self;
432 return $self->_root->{fh};
440 die "DBM::Deep: $_[1]\n";
445 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
450 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
455 # Store single hash key/value or array element in database.
457 my $self = shift->_get_self;
458 my ($key, $value) = @_;
460 unless ( _is_writable( $self->_fh ) ) {
461 $self->_throw_error( 'Cannot write to a readonly filehandle' );
465 # Request exclusive lock for writing
467 $self->lock( LOCK_EX );
469 my $md5 = $self->{engine}{digest}->($key);
471 my $tag = $self->{engine}->find_bucket_list( $self, $md5, { create => 1 } );
473 # User may be storing a hash, in which case we do not want it run
474 # through the filtering system
475 if ( !ref($value) && $self->_root->{filter_store_value} ) {
476 $value = $self->_root->{filter_store_value}->( $value );
480 # Add key/value to bucket list
482 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
491 # Fetch single value or element given plain key or array index
493 my $self = shift->_get_self;
496 my $md5 = $self->{engine}{digest}->($key);
499 # Request shared lock for reading
501 $self->lock( LOCK_SH );
503 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
510 # Get value from bucket list
512 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
516 # Filters only apply to scalar values, so the ref check is making
517 # sure the fetched bucket is a scalar, not a child hash or array.
518 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
519 ? $self->_root->{filter_fetch_value}->($result)
525 # Delete single key/value pair or element given plain key or array index
527 my $self = $_[0]->_get_self;
530 unless ( _is_writable( $self->_fh ) ) {
531 $self->_throw_error( 'Cannot write to a readonly filehandle' );
535 # Request exclusive lock for writing
537 $self->lock( LOCK_EX );
539 my $md5 = $self->{engine}{digest}->($key);
541 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
550 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
552 if (defined $value && !ref($value) && $self->_root->{filter_fetch_value}) {
553 $value = $self->_root->{filter_fetch_value}->($value);
556 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
559 # If this object is an array and the key deleted was on the end of the stack,
560 # decrement the length variable.
570 # Check if a single key or element exists given plain key or array index
572 my $self = $_[0]->_get_self;
575 my $md5 = $self->{engine}{digest}->($key);
578 # Request shared lock for reading
580 $self->lock( LOCK_SH );
582 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
587 # For some reason, the built-in exists() function returns '' for false
593 # Check if bucket exists and return 1 or ''
595 my $result = $self->{engine}->bucket_exists( $self, $tag, $md5 ) || '';
604 # Clear all keys from hash, or all elements from array.
606 my $self = $_[0]->_get_self;
608 unless ( _is_writable( $self->_fh ) ) {
609 $self->_throw_error( 'Cannot write to a readonly filehandle' );
613 # Request exclusive lock for writing
615 $self->lock( LOCK_EX );
619 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
625 $self->{engine}->create_tag(
626 $self, $self->_base_offset, $self->_type,
627 chr(0)x$self->{engine}{index_size},
636 # Public method aliases
638 sub put { (shift)->STORE( @_ ) }
639 sub store { (shift)->STORE( @_ ) }
640 sub get { (shift)->FETCH( @_ ) }
641 sub fetch { (shift)->FETCH( @_ ) }
642 sub delete { (shift)->DELETE( @_ ) }
643 sub exists { (shift)->EXISTS( @_ ) }
644 sub clear { (shift)->CLEAR( @_ ) }
646 package DBM::Deep::_::Root;
661 filter_store_key => undef,
662 filter_store_value => undef,
663 filter_fetch_key => undef,
664 filter_fetch_value => undef,
668 if ( $self->{fh} && !$self->{file_offset} ) {
669 $self->{file_offset} = tell( $self->{fh} );
679 close $self->{fh} if $self->{fh};
689 DBM::Deep - A pure perl multi-level hash/array DBM
694 my $db = DBM::Deep->new( "foo.db" );
696 $db->{key} = 'value'; # tie() style
699 $db->put('key' => 'value'); # OO style
700 print $db->get('key');
702 # true multi-level support
703 $db->{my_complex} = [
704 'hello', { perl => 'rules' },
710 A unique flat-file database module, written in pure perl. True
711 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
712 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
713 handle millions of keys and unlimited hash levels without significant
714 slow-down. Written from the ground-up in pure perl -- this is NOT a
715 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
716 Mac OS X and Windows.
718 =head1 VERSION DIFFERENCES
720 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
721 before. While attempts have been made to be backwards compatible, no guarantees.
725 Hopefully you are using Perl's excellent CPAN module, which will download
726 and install the module for you. If not, get the tarball, and run these
738 Construction can be done OO-style (which is the recommended way), or using
739 Perl's tie() function. Both are examined here.
741 =head2 OO CONSTRUCTION
743 The recommended way to construct a DBM::Deep object is to use the new()
744 method, which gets you a blessed, tied hash or array reference.
746 my $db = DBM::Deep->new( "foo.db" );
748 This opens a new database handle, mapped to the file "foo.db". If this
749 file does not exist, it will automatically be created. DB files are
750 opened in "r+" (read/write) mode, and the type of object returned is a
751 hash, unless otherwise specified (see L<OPTIONS> below).
753 You can pass a number of options to the constructor to specify things like
754 locking, autoflush, etc. This is done by passing an inline hash:
756 my $db = DBM::Deep->new(
762 Notice that the filename is now specified I<inside> the hash with
763 the "file" parameter, as opposed to being the sole argument to the
764 constructor. This is required if any options are specified.
765 See L<OPTIONS> below for the complete list.
769 You can also start with an array instead of a hash. For this, you must
770 specify the C<type> parameter:
772 my $db = DBM::Deep->new(
774 type => DBM::Deep->TYPE_ARRAY
777 B<Note:> Specifing the C<type> parameter only takes effect when beginning
778 a new DB file. If you create a DBM::Deep object with an existing file, the
779 C<type> will be loaded from the file header, and an error will be thrown if
780 the wrong type is passed in.
782 =head2 TIE CONSTRUCTION
784 Alternately, you can create a DBM::Deep handle by using Perl's built-in
785 tie() function. The object returned from tie() can be used to call methods,
786 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
787 file (as expected with most tie'd objects).
790 my $db = tie %hash, "DBM::Deep", "foo.db";
793 my $db = tie @array, "DBM::Deep", "bar.db";
795 As with the OO constructor, you can replace the DB filename parameter with
796 a hash containing one or more options (see L<OPTIONS> just below for the
799 tie %hash, "DBM::Deep", {
807 There are a number of options that can be passed in when constructing your
808 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
814 Filename of the DB file to link the handle to. You can pass a full absolute
815 filesystem path, partial path, or a plain filename if the file is in the
816 current working directory. This is a required parameter (though q.v. fh).
820 If you want, you can pass in the fh instead of the file. This is most useful for doing
823 my $db = DBM::Deep->new( { fh => \*DATA } );
825 You are responsible for making sure that the fh has been opened appropriately for your
826 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
827 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
828 needs to read from the fh.
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: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
841 This only takes effect when beginning a new file. This is an optional
842 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
846 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
847 function to lock the database in exclusive mode for writes, and shared mode for
848 reads. Pass any true value to enable. This affects the base DB handle I<and
849 any child hashes or arrays> that use the same DB file. This is an optional
850 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
854 Specifies whether autoflush is to be enabled on the underlying filehandle.
855 This obviously slows down write operations, but is required if you may have
856 multiple processes accessing the same DB file (also consider enable I<locking>).
857 Pass any true value to enable. This is an optional parameter, and defaults to 0
862 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
863 restore them when fetched. This is an B<experimental> feature, and does have
864 side-effects. Basically, when hashes are re-blessed into their original
865 classes, they are no longer blessed into the DBM::Deep class! So you won't be
866 able to call any DBM::Deep methods on them. You have been warned.
867 This is an optional parameter, and defaults to 0 (disabled).
871 See L<FILTERS> below.
877 With DBM::Deep you can access your databases using Perl's standard hash/array
878 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
879 treat them as such. DBM::Deep will intercept all reads/writes and direct them
880 to the right place -- the DB file. This has nothing to do with the
881 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
882 using regular hashes and arrays, rather than calling functions like C<get()>
883 and C<put()> (although those work too). It is entirely up to you how to want
884 to access your databases.
888 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
889 or even nested hashes (or arrays) using standard Perl syntax:
891 my $db = DBM::Deep->new( "foo.db" );
893 $db->{mykey} = "myvalue";
895 $db->{myhash}->{subkey} = "subvalue";
897 print $db->{myhash}->{subkey} . "\n";
899 You can even step through hash keys using the normal Perl C<keys()> function:
901 foreach my $key (keys %$db) {
902 print "$key: " . $db->{$key} . "\n";
905 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
906 pushes them onto an array, all before the loop even begins. If you have an
907 extra large hash, this may exhaust Perl's memory. Instead, consider using
908 Perl's C<each()> function, which pulls keys/values one at a time, using very
911 while (my ($key, $value) = each %$db) {
912 print "$key: $value\n";
915 Please note that when using C<each()>, you should always pass a direct
916 hash reference, not a lookup. Meaning, you should B<never> do this:
919 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
921 This causes an infinite loop, because for each iteration, Perl is calling
922 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
923 it effectively keeps returning the first key over and over again. Instead,
924 assign a temporary variable to C<$db->{foo}>, then pass that to each().
928 As with hashes, you can treat any DBM::Deep object like a normal Perl array
929 reference. This includes inserting, removing and manipulating elements,
930 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
931 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
932 or simply be a nested array reference inside a hash. Example:
934 my $db = DBM::Deep->new(
935 file => "foo-array.db",
936 type => DBM::Deep->TYPE_ARRAY
940 push @$db, "bar", "baz";
943 my $last_elem = pop @$db; # baz
944 my $first_elem = shift @$db; # bah
945 my $second_elem = $db->[1]; # bar
947 my $num_elements = scalar @$db;
951 In addition to the I<tie()> interface, you can also use a standard OO interface
952 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
953 array) has its own methods, but both types share the following common methods:
954 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
958 =item * new() / clone()
960 These are the constructor and copy-functions.
962 =item * put() / store()
964 Stores a new hash key/value pair, or sets an array element value. Takes two
965 arguments, the hash key or array index, and the new value. The value can be
966 a scalar, hash ref or array ref. Returns true on success, false on failure.
968 $db->put("foo", "bar"); # for hashes
969 $db->put(1, "bar"); # for arrays
971 =item * get() / fetch()
973 Fetches the value of a hash key or array element. Takes one argument: the hash
974 key or array index. Returns a scalar, hash ref or array ref, depending on the
977 my $value = $db->get("foo"); # for hashes
978 my $value = $db->get(1); # for arrays
982 Checks if a hash key or array index exists. Takes one argument: the hash key
983 or array index. Returns true if it exists, false if not.
985 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
986 if ($db->exists(1)) { print "yay!\n"; } # for arrays
990 Deletes one hash key/value pair or array element. Takes one argument: the hash
991 key or array index. Returns true on success, false if not found. For arrays,
992 the remaining elements located after the deleted element are NOT moved over.
993 The deleted element is essentially just undefined, which is exactly how Perl's
994 internal arrays work. Please note that the space occupied by the deleted
995 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
996 below for details and workarounds.
998 $db->delete("foo"); # for hashes
999 $db->delete(1); # for arrays
1003 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1004 value. Please note that the space occupied by the deleted keys/values or
1005 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1006 details and workarounds.
1008 $db->clear(); # hashes or arrays
1010 =item * lock() / unlock()
1016 Recover lost disk space.
1018 =item * import() / export()
1020 Data going in and out.
1022 =item * set_digest() / set_pack() / set_filter()
1024 q.v. adjusting the interal parameters.
1030 For hashes, DBM::Deep supports all the common methods described above, and the
1031 following additional methods: C<first_key()> and C<next_key()>.
1037 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1038 fetched in an undefined order (which appears random). Takes no arguments,
1039 returns the key as a scalar value.
1041 my $key = $db->first_key();
1045 Returns the "next" key in the hash, given the previous one as the sole argument.
1046 Returns undef if there are no more keys to be fetched.
1048 $key = $db->next_key($key);
1052 Here are some examples of using hashes:
1054 my $db = DBM::Deep->new( "foo.db" );
1056 $db->put("foo", "bar");
1057 print "foo: " . $db->get("foo") . "\n";
1059 $db->put("baz", {}); # new child hash ref
1060 $db->get("baz")->put("buz", "biz");
1061 print "buz: " . $db->get("baz")->get("buz") . "\n";
1063 my $key = $db->first_key();
1065 print "$key: " . $db->get($key) . "\n";
1066 $key = $db->next_key($key);
1069 if ($db->exists("foo")) { $db->delete("foo"); }
1073 For arrays, DBM::Deep supports all the common methods described above, and the
1074 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1075 C<unshift()> and C<splice()>.
1081 Returns the number of elements in the array. Takes no arguments.
1083 my $len = $db->length();
1087 Adds one or more elements onto the end of the array. Accepts scalars, hash
1088 refs or array refs. No return value.
1090 $db->push("foo", "bar", {});
1094 Fetches the last element in the array, and deletes it. Takes no arguments.
1095 Returns undef if array is empty. Returns the element value.
1097 my $elem = $db->pop();
1101 Fetches the first element in the array, deletes it, then shifts all the
1102 remaining elements over to take up the space. Returns the element value. This
1103 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1106 my $elem = $db->shift();
1110 Inserts one or more elements onto the beginning of the array, shifting all
1111 existing elements over to make room. Accepts scalars, hash refs or array refs.
1112 No return value. This method is not recommended with large arrays -- see
1113 <LARGE ARRAYS> below for details.
1115 $db->unshift("foo", "bar", {});
1119 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1120 -f splice> for usage -- it is too complicated to document here. This method is
1121 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1125 Here are some examples of using arrays:
1127 my $db = DBM::Deep->new(
1129 type => DBM::Deep->TYPE_ARRAY
1132 $db->push("bar", "baz");
1133 $db->unshift("foo");
1136 my $len = $db->length();
1137 print "length: $len\n"; # 4
1139 for (my $k=0; $k<$len; $k++) {
1140 print "$k: " . $db->get($k) . "\n";
1143 $db->splice(1, 2, "biz", "baf");
1145 while (my $elem = shift @$db) {
1146 print "shifted: $elem\n";
1151 Enable automatic file locking by passing a true value to the C<locking>
1152 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1154 my $db = DBM::Deep->new(
1159 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1160 mode for writes, and shared mode for reads. This is required if you have
1161 multiple processes accessing the same database file, to avoid file corruption.
1162 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1163 NFS> below for more.
1165 =head2 EXPLICIT LOCKING
1167 You can explicitly lock a database, so it remains locked for multiple
1168 transactions. This is done by calling the C<lock()> method, and passing an
1169 optional lock mode argument (defaults to exclusive mode). This is particularly
1170 useful for things like counters, where the current value needs to be fetched,
1171 then incremented, then stored again.
1174 my $counter = $db->get("counter");
1176 $db->put("counter", $counter);
1185 You can pass C<lock()> an optional argument, which specifies which mode to use
1186 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1187 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1188 same as the constants defined in Perl's C<Fcntl> module.
1190 $db->lock( DBM::Deep->LOCK_SH );
1194 =head1 IMPORTING/EXPORTING
1196 You can import existing complex structures by calling the C<import()> method,
1197 and export an entire database into an in-memory structure using the C<export()>
1198 method. Both are examined here.
1202 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1203 walking the structure and adding keys/elements to the database as you go,
1204 simply pass a reference to the C<import()> method. This recursively adds
1205 everything to an existing DBM::Deep object for you. Here is an example:
1210 array1 => [ "elem0", "elem1", "elem2" ],
1212 subkey1 => "subvalue1",
1213 subkey2 => "subvalue2"
1217 my $db = DBM::Deep->new( "foo.db" );
1218 $db->import( $struct );
1220 print $db->{key1} . "\n"; # prints "value1"
1222 This recursively imports the entire C<$struct> object into C<$db>, including
1223 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1224 keys are merged with the existing ones, replacing if they already exist.
1225 The C<import()> method can be called on any database level (not just the base
1226 level), and works with both hash and array DB types.
1228 B<Note:> Make sure your existing structure has no circular references in it.
1229 These will cause an infinite loop when importing.
1233 Calling the C<export()> method on an existing DBM::Deep object will return
1234 a reference to a new in-memory copy of the database. The export is done
1235 recursively, so all nested hashes/arrays are all exported to standard Perl
1236 objects. Here is an example:
1238 my $db = DBM::Deep->new( "foo.db" );
1240 $db->{key1} = "value1";
1241 $db->{key2} = "value2";
1243 $db->{hash1}->{subkey1} = "subvalue1";
1244 $db->{hash1}->{subkey2} = "subvalue2";
1246 my $struct = $db->export();
1248 print $struct->{key1} . "\n"; # prints "value1"
1250 This makes a complete copy of the database in memory, and returns a reference
1251 to it. The C<export()> method can be called on any database level (not just
1252 the base level), and works with both hash and array DB types. Be careful of
1253 large databases -- you can store a lot more data in a DBM::Deep object than an
1254 in-memory Perl structure.
1256 B<Note:> Make sure your database has no circular references in it.
1257 These will cause an infinite loop when exporting.
1261 DBM::Deep has a number of hooks where you can specify your own Perl function
1262 to perform filtering on incoming or outgoing data. This is a perfect
1263 way to extend the engine, and implement things like real-time compression or
1264 encryption. Filtering applies to the base DB level, and all child hashes /
1265 arrays. Filter hooks can be specified when your DBM::Deep object is first
1266 constructed, or by calling the C<set_filter()> method at any time. There are
1267 four available filter hooks, described below:
1271 =item * filter_store_key
1273 This filter is called whenever a hash key is stored. It
1274 is passed the incoming key, and expected to return a transformed key.
1276 =item * filter_store_value
1278 This filter is called whenever a hash key or array element is stored. It
1279 is passed the incoming value, and expected to return a transformed value.
1281 =item * filter_fetch_key
1283 This filter is called whenever a hash key is fetched (i.e. via
1284 C<first_key()> or C<next_key()>). It is passed the transformed key,
1285 and expected to return the plain key.
1287 =item * filter_fetch_value
1289 This filter is called whenever a hash key or array element is fetched.
1290 It is passed the transformed value, and expected to return the plain value.
1294 Here are the two ways to setup a filter hook:
1296 my $db = DBM::Deep->new(
1298 filter_store_value => \&my_filter_store,
1299 filter_fetch_value => \&my_filter_fetch
1304 $db->set_filter( "filter_store_value", \&my_filter_store );
1305 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1307 Your filter function will be called only when dealing with SCALAR keys or
1308 values. When nested hashes and arrays are being stored/fetched, filtering
1309 is bypassed. Filters are called as static functions, passed a single SCALAR
1310 argument, and expected to return a single SCALAR value. If you want to
1311 remove a filter, set the function reference to C<undef>:
1313 $db->set_filter( "filter_store_value", undef );
1315 =head2 REAL-TIME ENCRYPTION EXAMPLE
1317 Here is a working example that uses the I<Crypt::Blowfish> module to
1318 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1319 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1320 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1323 use Crypt::Blowfish;
1326 my $cipher = Crypt::CBC->new({
1327 'key' => 'my secret key',
1328 'cipher' => 'Blowfish',
1330 'regenerate_key' => 0,
1331 'padding' => 'space',
1335 my $db = DBM::Deep->new(
1336 file => "foo-encrypt.db",
1337 filter_store_key => \&my_encrypt,
1338 filter_store_value => \&my_encrypt,
1339 filter_fetch_key => \&my_decrypt,
1340 filter_fetch_value => \&my_decrypt,
1343 $db->{key1} = "value1";
1344 $db->{key2} = "value2";
1345 print "key1: " . $db->{key1} . "\n";
1346 print "key2: " . $db->{key2} . "\n";
1352 return $cipher->encrypt( $_[0] );
1355 return $cipher->decrypt( $_[0] );
1358 =head2 REAL-TIME COMPRESSION EXAMPLE
1360 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1361 compression / decompression of keys & values with DBM::Deep Filters.
1362 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1363 more on I<Compress::Zlib>.
1368 my $db = DBM::Deep->new(
1369 file => "foo-compress.db",
1370 filter_store_key => \&my_compress,
1371 filter_store_value => \&my_compress,
1372 filter_fetch_key => \&my_decompress,
1373 filter_fetch_value => \&my_decompress,
1376 $db->{key1} = "value1";
1377 $db->{key2} = "value2";
1378 print "key1: " . $db->{key1} . "\n";
1379 print "key2: " . $db->{key2} . "\n";
1385 return Compress::Zlib::memGzip( $_[0] ) ;
1388 return Compress::Zlib::memGunzip( $_[0] ) ;
1391 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1392 actually numerical index numbers, and are not filtered.
1394 =head1 ERROR HANDLING
1396 Most DBM::Deep methods return a true value for success, and call die() on
1397 failure. You can wrap calls in an eval block to catch the die.
1399 my $db = DBM::Deep->new( "foo.db" ); # create hash
1400 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1402 print $@; # prints error message
1404 =head1 LARGEFILE SUPPORT
1406 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1407 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1408 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1409 by calling the static C<set_pack()> method before you do anything else.
1411 DBM::Deep::set_pack(8, 'Q');
1413 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1414 instead of 32-bit longs. After setting these values your DB files have a
1415 theoretical maximum size of 16 XB (exabytes).
1417 B<Note:> Changing these values will B<NOT> work for existing database files.
1418 Only change this for new files, and make sure it stays set consistently
1419 throughout the file's life. If you do set these values, you can no longer
1420 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1421 back to 32-bit mode.
1423 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1424 only a 32-bit Perl. However, I have received user reports that this does
1427 =head1 LOW-LEVEL ACCESS
1429 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1430 you can call the C<_fh()> method, which returns the handle:
1432 my $fh = $db->_fh();
1434 This method can be called on the root level of the datbase, or any child
1435 hashes or arrays. All levels share a I<root> structure, which contains things
1436 like the filehandle, a reference counter, and all the options specified
1437 when you created the object. You can get access to this root structure by
1438 calling the C<root()> method.
1440 my $root = $db->_root();
1442 This is useful for changing options after the object has already been created,
1443 such as enabling/disabling locking. You can also store your own temporary user
1444 data in this structure (be wary of name collision), which is then accessible from
1445 any child hash or array.
1447 =head1 CUSTOM DIGEST ALGORITHM
1449 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1450 keys. However you can override this, and use another algorithm (such as SHA-256)
1451 or even write your own. But please note that DBM::Deep currently expects zero
1452 collisions, so your algorithm has to be I<perfect>, so to speak.
1453 Collision detection may be introduced in a later version.
1457 You can specify a custom digest algorithm by calling the static C<set_digest()>
1458 function, passing a reference to a subroutine, and the length of the algorithm's
1459 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1460 objects. Here is a working example that uses a 256-bit hash from the
1461 I<Digest::SHA256> module. Please see
1462 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1467 my $context = Digest::SHA256::new(256);
1469 DBM::Deep::set_digest( \&my_digest, 32 );
1471 my $db = DBM::Deep->new( "foo-sha.db" );
1473 $db->{key1} = "value1";
1474 $db->{key2} = "value2";
1475 print "key1: " . $db->{key1} . "\n";
1476 print "key2: " . $db->{key2} . "\n";
1482 return substr( $context->hash($_[0]), 0, 32 );
1485 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1486 of bytes you specify in the C<set_digest()> function (in this case 32).
1488 =head1 CIRCULAR REFERENCES
1490 DBM::Deep has B<experimental> support for circular references. Meaning you
1491 can have a nested hash key or array element that points to a parent object.
1492 This relationship is stored in the DB file, and is preserved between sessions.
1495 my $db = DBM::Deep->new( "foo.db" );
1498 $db->{circle} = $db; # ref to self
1500 print $db->{foo} . "\n"; # prints "foo"
1501 print $db->{circle}->{foo} . "\n"; # prints "foo" again
1503 B<Note>: Passing the object to a function that recursively walks the
1504 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1505 C<export()> methods) will result in an infinite loop. This will be fixed in
1508 =head1 CAVEATS / ISSUES / BUGS
1510 This section describes all the known issues with DBM::Deep. It you have found
1511 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1513 =head2 UNUSED SPACE RECOVERY
1515 One major caveat with DBM::Deep is that space occupied by existing keys and
1516 values is not recovered when they are deleted. Meaning if you keep deleting
1517 and adding new keys, your file will continuously grow. I am working on this,
1518 but in the meantime you can call the built-in C<optimize()> method from time to
1519 time (perhaps in a crontab or something) to recover all your unused space.
1521 $db->optimize(); # returns true on success
1523 This rebuilds the ENTIRE database into a new file, then moves it on top of
1524 the original. The new file will have no unused space, thus it will take up as
1525 little disk space as possible. Please note that this operation can take
1526 a long time for large files, and you need enough disk space to temporarily hold
1527 2 copies of your DB file. The temporary file is created in the same directory
1528 as the original, named with a ".tmp" extension, and is deleted when the
1529 operation completes. Oh, and if locking is enabled, the DB is automatically
1530 locked for the entire duration of the copy.
1532 B<WARNING:> Only call optimize() on the top-level node of the database, and
1533 make sure there are no child references lying around. DBM::Deep keeps a reference
1534 counter, and if it is greater than 1, optimize() will abort and return undef.
1536 =head2 AUTOVIVIFICATION
1538 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1539 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1540 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1541 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1542 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1545 $db->{foo}->{bar} = "hello";
1547 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1548 being an empty hash. Try this instead, which works fine:
1550 $db->{foo} = { bar => "hello" };
1552 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1553 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1554 Probably a bug in Perl.
1558 (The reasons given assume a high level of Perl understanding, specifically of
1559 references. You can safely skip this section.)
1561 Currently, the only references supported are HASH and ARRAY. The other reference
1562 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1568 These are things like filehandles and other sockets. They can't be supported
1569 because it's completely unclear how DBM::Deep should serialize them.
1571 =item * SCALAR / REF
1573 The discussion here refers to the following type of example:
1580 # In some other process ...
1582 my $val = ${ $db->{key1} };
1584 is( $val, 50, "What actually gets stored in the DB file?" );
1586 The problem is one of synchronization. When the variable being referred to
1587 changes value, the reference isn't notified. This means that the new value won't
1588 be stored in the datafile for other processes to read. There is no TIEREF.
1590 It is theoretically possible to store references to values already within a
1591 DBM::Deep object because everything already is synchronized, but the change to
1592 the internals would be quite large. Specifically, DBM::Deep would have to tie
1593 every single value that is stored. This would bloat the RAM footprint of
1594 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1595 all to support a feature that has never been requested.
1599 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1600 mechanism for serializing coderefs, including saving off all closure state.
1601 However, just as for SCALAR and REF, that closure state may change without
1602 notifying the DBM::Deep object storing the reference.
1606 =head2 FILE CORRUPTION
1608 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1609 for a 32-bit signature when opened, but other corruption in files can cause
1610 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1611 stuck in an infinite loop depending on the level of corruption. File write
1612 operations are not checked for failure (for speed), so if you happen to run
1613 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1614 be addressed in a later version of DBM::Deep.
1618 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1619 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1620 about setting up your NFS server with a locking daemon, then using lockf() to
1621 lock your files, but your mileage may vary there as well. From what I
1622 understand, there is no real way to do it. However, if you need access to the
1623 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1624 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1626 =head2 COPYING OBJECTS
1628 Beware of copying tied objects in Perl. Very strange things can happen.
1629 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1630 returns a new, blessed, tied hash or array to the same level in the DB.
1632 my $copy = $db->clone();
1634 B<Note>: Since clone() here is cloning the object, not the database location, any
1635 modifications to either $db or $copy will be visible in both.
1639 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1640 These functions cause every element in the array to move, which can be murder
1641 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1642 a different location. This will be addressed in the forthcoming version 1.00.
1644 =head2 WRITEONLY FILES
1646 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1647 writeonly mode. STORE will verify that the filehandle is writable. However, there
1648 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1649 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1653 This section discusses DBM::Deep's speed and memory usage.
1657 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1658 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1659 multi-level hash/array support, and cross-platform FTPable files. Even so,
1660 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1661 with huge databases. Here is some test data:
1663 Adding 1,000,000 keys to new DB file...
1665 At 100 keys, avg. speed is 2,703 keys/sec
1666 At 200 keys, avg. speed is 2,642 keys/sec
1667 At 300 keys, avg. speed is 2,598 keys/sec
1668 At 400 keys, avg. speed is 2,578 keys/sec
1669 At 500 keys, avg. speed is 2,722 keys/sec
1670 At 600 keys, avg. speed is 2,628 keys/sec
1671 At 700 keys, avg. speed is 2,700 keys/sec
1672 At 800 keys, avg. speed is 2,607 keys/sec
1673 At 900 keys, avg. speed is 2,190 keys/sec
1674 At 1,000 keys, avg. speed is 2,570 keys/sec
1675 At 2,000 keys, avg. speed is 2,417 keys/sec
1676 At 3,000 keys, avg. speed is 1,982 keys/sec
1677 At 4,000 keys, avg. speed is 1,568 keys/sec
1678 At 5,000 keys, avg. speed is 1,533 keys/sec
1679 At 6,000 keys, avg. speed is 1,787 keys/sec
1680 At 7,000 keys, avg. speed is 1,977 keys/sec
1681 At 8,000 keys, avg. speed is 2,028 keys/sec
1682 At 9,000 keys, avg. speed is 2,077 keys/sec
1683 At 10,000 keys, avg. speed is 2,031 keys/sec
1684 At 20,000 keys, avg. speed is 1,970 keys/sec
1685 At 30,000 keys, avg. speed is 2,050 keys/sec
1686 At 40,000 keys, avg. speed is 2,073 keys/sec
1687 At 50,000 keys, avg. speed is 1,973 keys/sec
1688 At 60,000 keys, avg. speed is 1,914 keys/sec
1689 At 70,000 keys, avg. speed is 2,091 keys/sec
1690 At 80,000 keys, avg. speed is 2,103 keys/sec
1691 At 90,000 keys, avg. speed is 1,886 keys/sec
1692 At 100,000 keys, avg. speed is 1,970 keys/sec
1693 At 200,000 keys, avg. speed is 2,053 keys/sec
1694 At 300,000 keys, avg. speed is 1,697 keys/sec
1695 At 400,000 keys, avg. speed is 1,838 keys/sec
1696 At 500,000 keys, avg. speed is 1,941 keys/sec
1697 At 600,000 keys, avg. speed is 1,930 keys/sec
1698 At 700,000 keys, avg. speed is 1,735 keys/sec
1699 At 800,000 keys, avg. speed is 1,795 keys/sec
1700 At 900,000 keys, avg. speed is 1,221 keys/sec
1701 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1703 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1704 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1705 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1706 Run time was 12 min 3 sec.
1710 One of the great things about DBM::Deep is that it uses very little memory.
1711 Even with huge databases (1,000,000+ keys) you will not see much increased
1712 memory on your process. DBM::Deep relies solely on the filesystem for storing
1713 and fetching data. Here is output from I</usr/bin/top> before even opening a
1716 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1717 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1719 Basically the process is taking 2,716K of memory. And here is the same
1720 process after storing and fetching 1,000,000 keys:
1722 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1723 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1725 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1726 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1728 =head1 DB FILE FORMAT
1730 In case you were interested in the underlying DB file format, it is documented
1731 here in this section. You don't need to know this to use the module, it's just
1732 included for reference.
1736 DBM::Deep files always start with a 32-bit signature to identify the file type.
1737 This is at offset 0. The signature is "DPDB" in network byte order. This is
1738 checked for when the file is opened and an error will be thrown if it's not found.
1742 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1743 has a standard header containing the type of data, the length of data, and then
1744 the data itself. The type is a single character (1 byte), the length is a
1745 32-bit unsigned long in network byte order, and the data is, well, the data.
1746 Here is how it unfolds:
1750 Immediately after the 32-bit file signature is the I<Master Index> record.
1751 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1752 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1753 depending on how the DBM::Deep object was constructed.
1755 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1756 number). The first 8-bit char of the MD5 signature is the offset into the
1757 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1758 index element is a file offset of the next tag for the key/element in question,
1759 which is usually a I<Bucket List> tag (see below).
1761 The next tag I<could> be another index, depending on how many keys/elements
1762 exist. See L<RE-INDEXING> below for details.
1766 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1767 file offsets to where the actual data is stored. It starts with a standard
1768 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1769 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1770 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1771 When the list fills up, a I<Re-Index> operation is performed (See
1772 L<RE-INDEXING> below).
1776 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1777 index/value pair (in array mode). It starts with a standard tag header with
1778 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1779 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1780 header. The size reported in the tag header is only for the value, but then,
1781 just after the value is another size (32-bit unsigned long) and then the plain
1782 key itself. Since the value is likely to be fetched more often than the plain
1783 key, I figured it would be I<slightly> faster to store the value first.
1785 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1786 record for the nested structure, where the process begins all over again.
1790 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1791 exhausted. Then, when another key/element comes in, the list is converted to a
1792 new index record. However, this index will look at the next char in the MD5
1793 hash, and arrange new Bucket List pointers accordingly. This process is called
1794 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1795 17 (16 + new one) keys/elements are removed from the old Bucket List and
1796 inserted into the new index. Several new Bucket Lists are created in the
1797 process, as a new MD5 char from the key is being examined (it is unlikely that
1798 the keys will all share the same next char of their MD5s).
1800 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1801 when the Bucket Lists will turn into indexes, but the first round tends to
1802 happen right around 4,000 keys. You will see a I<slight> decrease in
1803 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1804 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1805 right around 900,000 keys. This process can continue nearly indefinitely --
1806 right up until the point the I<MD5> signatures start colliding with each other,
1807 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1808 getting struck by lightning while you are walking to cash in your tickets.
1809 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1810 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1811 this is 340 unodecillion, but don't quote me).
1815 When a new key/element is stored, the key (or index number) is first run through
1816 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1817 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1818 for the first char of the signature (in this case I<b0>). If it does not exist,
1819 a new I<Bucket List> is created for our key (and the next 15 future keys that
1820 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1821 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1822 this point, unless we are replacing an existing I<Bucket>), where the actual
1823 data will be stored.
1827 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1828 (or index number), then walking along the indexes. If there are enough
1829 keys/elements in this DB level, there might be nested indexes, each linked to
1830 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1831 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1832 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1833 plain key are stored.
1835 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1836 methods. In this process the indexes are walked systematically, and each key
1837 fetched in increasing MD5 order (which is why it appears random). Once the
1838 I<Bucket> is found, the value is skipped and the plain key returned instead.
1839 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1840 alphabetically sorted. This only happens on an index-level -- as soon as the
1841 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1842 so it's pretty much undefined how the keys will come out -- just like Perl's
1845 =head1 CODE COVERAGE
1847 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1848 B<Devel::Cover> report on this module's test suite.
1850 ----------------------------------- ------ ------ ------ ------ ------ ------
1851 File stmt bran cond sub time total
1852 ----------------------------------- ------ ------ ------ ------ ------ ------
1853 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1854 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1855 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1856 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1857 Total 97.9 85.9 79.7 100.0 100.0 94.3
1858 ----------------------------------- ------ ------ ------ ------ ------ ------
1860 =head1 MORE INFORMATION
1862 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1863 or send email to L<DBM-Deep@googlegroups.com>.
1867 Joseph Huckaby, L<jhuckaby@cpan.org>
1869 Rob Kinyon, L<rkinyon@cpan.org>
1871 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1875 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1876 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1880 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1881 This is free software, you may use it and distribute it under the
1882 same terms as Perl itself.