7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
37 use Fcntl qw( :DEFAULT :flock :seek );
41 use DBM::Deep::Engine;
44 use vars qw( $VERSION );
45 $VERSION = q(0.99_01);
48 # Setup constants for users to pass to new()
50 sub TYPE_HASH () { DBM::Deep::Engine->SIG_HASH }
51 sub TYPE_ARRAY () { DBM::Deep::Engine->SIG_ARRAY }
59 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
64 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
65 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
70 $args = { file => shift };
78 # Class constructor method for Perl OO interface.
79 # Calls tie() and returns blessed reference to tied hash or array,
80 # providing a hybrid OO/tie interface.
83 my $args = $class->_get_args( @_ );
86 # Check if we want a tied hash or array.
89 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
90 $class = 'DBM::Deep::Array';
91 require DBM::Deep::Array;
92 tie @$self, $class, %$args;
95 $class = 'DBM::Deep::Hash';
96 require DBM::Deep::Hash;
97 tie %$self, $class, %$args;
100 return bless $self, $class;
103 # This initializer is called from the various TIE* methods. new() calls tie(),
104 # which allows for a single point of entry.
109 $args->{fileobj} = DBM::Deep::File->new( $args )
110 unless exists $args->{fileobj};
112 # locking implicitly enables autoflush
113 if ($args->{locking}) { $args->{autoflush} = 1; }
115 # These are the defaults to be optionally overridden below
118 base_offset => undef,
125 $self->{engine} = DBM::Deep::Engine->new( { %{$args}, obj => $self } );
127 # Grab the parameters we want to use
128 foreach my $param ( keys %$self ) {
129 next unless exists $args->{$param};
130 $self->{$param} = $args->{$param};
133 $self->{engine}->setup_fh( $self );
135 $self->{fileobj}->set_db( $self );
142 require DBM::Deep::Hash;
143 return DBM::Deep::Hash->TIEHASH( @_ );
148 require DBM::Deep::Array;
149 return DBM::Deep::Array->TIEARRAY( @_ );
153 my $self = shift->_get_self;
154 return $self->_fileobj->lock( $self, @_ );
158 my $self = shift->_get_self;
159 return $self->_fileobj->unlock( $self, @_ );
163 my $self = shift->_get_self;
164 my ($spot, $value) = @_;
169 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
170 ${$spot} = $value->_repr;
171 $value->_copy_node( ${$spot} );
174 my $r = Scalar::Util::reftype( $value );
175 my $c = Scalar::Util::blessed( $value );
176 if ( $r eq 'ARRAY' ) {
177 ${$spot} = [ @{$value} ];
180 ${$spot} = { %{$value} };
182 ${$spot} = bless ${$spot}, $c
190 die "Must be implemented in a child class\n";
194 die "Must be implemented in a child class\n";
199 # Recursively export into standard Perl hashes and arrays.
201 my $self = shift->_get_self;
203 my $temp = $self->_repr;
206 $self->_copy_node( $temp );
214 # Recursively import Perl hash/array structure
216 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
218 my $self = shift->_get_self;
221 # struct is not a reference, so just import based on our type
223 $struct = $self->_repr( @_ );
226 return $self->_import( $struct );
231 # Rebuild entire database into new file, then move
232 # it back on top of original.
234 my $self = shift->_get_self;
236 #XXX Need to create a new test for this
237 # if ($self->_fileobj->{links} > 1) {
238 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
241 my $db_temp = DBM::Deep->new(
242 file => $self->_fileobj->{file} . '.tmp',
247 $self->_copy_node( $db_temp );
251 # Attempt to copy user, group and permissions over to new file
253 my @stats = stat($self->_fh);
254 my $perms = $stats[2] & 07777;
257 chown( $uid, $gid, $self->_fileobj->{file} . '.tmp' );
258 chmod( $perms, $self->_fileobj->{file} . '.tmp' );
260 # q.v. perlport for more information on this variable
261 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
263 # Potential race condition when optmizing on Win32 with locking.
264 # The Windows filesystem requires that the filehandle be closed
265 # before it is overwritten with rename(). This could be redone
269 $self->_fileobj->close;
272 if (!rename $self->_fileobj->{file} . '.tmp', $self->_fileobj->{file}) {
273 unlink $self->_fileobj->{file} . '.tmp';
275 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
279 $self->_fileobj->close;
280 $self->_fileobj->open;
281 $self->{engine}->setup_fh( $self );
288 # Make copy of object and return
290 my $self = shift->_get_self;
292 return DBM::Deep->new(
293 type => $self->_type,
294 base_offset => $self->_base_offset,
295 fileobj => $self->_fileobj,
300 my %is_legal_filter = map {
303 store_key store_value
304 fetch_key fetch_value
309 # Setup filter function for storing or fetching the key or value
311 my $self = shift->_get_self;
315 if ( $is_legal_filter{$type} ) {
316 $self->_fileobj->{"filter_$type"} = $func;
325 my $self = shift->_get_self;
326 $self->_fileobj->begin_transaction;
331 my $self = shift->_get_self;
332 $self->_fileobj->end_transaction;
337 my $self = shift->_get_self;
338 # At this point, we need to replay the actions taken
339 $self->_fileobj->end_transaction;
348 my $self = $_[0]->_get_self;
349 return $self->{fileobj};
353 my $self = $_[0]->_get_self;
354 return $self->{type};
358 my $self = $_[0]->_get_self;
359 return $self->{base_offset};
363 my $self = $_[0]->_get_self;
364 return $self->_fileobj->{fh};
372 die "DBM::Deep: $_[1]\n";
377 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
382 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
387 if ( $self->{parent} ) {
388 my $base = $self->{parent}->_find_parent();
389 if ( $self->{parent}->_type eq TYPE_HASH ) {
390 return $base . "\{$self->{parent_key}\}";
392 return $base . "\[$self->{parent_key}\]";
399 # Store single hash key/value or array element in database.
401 my $self = shift->_get_self;
402 my ($key, $value, $orig_key) = @_;
404 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
405 $self->_throw_error( 'Cannot write to a readonly filehandle' );
408 if ( my $afh = $self->_fileobj->{audit_fh} ) {
409 unless ( $self->_type eq TYPE_ARRAY && $orig_key eq 'length' ) {
410 my $lhs = $self->_find_parent;
411 if ( $self->_type eq TYPE_HASH ) {
412 $lhs .= "\{$orig_key\}";
415 $lhs .= "\[$orig_key\]";
420 my $r = Scalar::Util::reftype( $value ) || '';
421 if ( $r eq 'HASH' ) {
424 elsif ( $r eq 'ARRAY' ) {
431 if ( my $c = Scalar::Util::blessed( $value ) ) {
432 $rhs = "bless $rhs, '$c'";
435 flock( $afh, LOCK_EX );
436 print( $afh "$lhs = $rhs; # " . localtime(time) . "\n" );
437 flock( $afh, LOCK_UN );
442 # Request exclusive lock for writing
444 $self->lock( LOCK_EX );
446 my $md5 = $self->{engine}{digest}->($key);
448 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5, { create => 1 } );
450 # User may be storing a hash, in which case we do not want it run
451 # through the filtering system
452 if ( !ref($value) && $self->_fileobj->{filter_store_value} ) {
453 $value = $self->_fileobj->{filter_store_value}->( $value );
457 # Add key/value to bucket list
459 my $result = $self->{engine}->add_bucket( $tag, $md5, $key, $value, undef, $orig_key );
468 # Fetch single value or element given plain key or array index
470 my $self = shift->_get_self;
473 my $md5 = $self->{engine}{digest}->($key);
476 # Request shared lock for reading
478 $self->lock( LOCK_SH );
480 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
487 # Get value from bucket list
489 my $result = $self->{engine}->get_bucket_value( $tag, $md5 );
493 # Filters only apply to scalar values, so the ref check is making
494 # sure the fetched bucket is a scalar, not a child hash or array.
495 return ($result && !ref($result) && $self->_fileobj->{filter_fetch_value})
496 ? $self->_fileobj->{filter_fetch_value}->($result)
502 # Delete single key/value pair or element given plain key or array index
504 my $self = $_[0]->_get_self;
507 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
508 $self->_throw_error( 'Cannot write to a readonly filehandle' );
512 # Request exclusive lock for writing
514 $self->lock( LOCK_EX );
516 my $md5 = $self->{engine}{digest}->($key);
518 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
527 my $value = $self->{engine}->get_bucket_value( $tag, $md5 );
529 if (defined $value && !ref($value) && $self->_fileobj->{filter_fetch_value}) {
530 $value = $self->_fileobj->{filter_fetch_value}->($value);
533 my $result = $self->{engine}->delete_bucket( $tag, $md5 );
536 # If this object is an array and the key deleted was on the end of the stack,
537 # decrement the length variable.
547 # Check if a single key or element exists given plain key or array index
549 my $self = $_[0]->_get_self;
552 my $md5 = $self->{engine}{digest}->($key);
555 # Request shared lock for reading
557 $self->lock( LOCK_SH );
559 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
564 # For some reason, the built-in exists() function returns '' for false
570 # Check if bucket exists and return 1 or ''
572 my $result = $self->{engine}->bucket_exists( $tag, $md5 ) || '';
581 # Clear all keys from hash, or all elements from array.
583 my $self = $_[0]->_get_self;
585 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
586 $self->_throw_error( 'Cannot write to a readonly filehandle' );
590 # Request exclusive lock for writing
592 $self->lock( LOCK_EX );
596 seek($fh, $self->_base_offset + $self->_fileobj->{file_offset}, SEEK_SET);
602 #XXX This needs updating to use _release_space
603 $self->{engine}->write_tag(
604 $self->_base_offset, $self->_type,
605 chr(0)x$self->{engine}{index_size},
614 # Public method aliases
616 sub put { (shift)->STORE( @_ ) }
617 sub store { (shift)->STORE( @_ ) }
618 sub get { (shift)->FETCH( @_ ) }
619 sub fetch { (shift)->FETCH( @_ ) }
620 sub delete { (shift)->DELETE( @_ ) }
621 sub exists { (shift)->EXISTS( @_ ) }
622 sub clear { (shift)->CLEAR( @_ ) }
629 DBM::Deep - A pure perl multi-level hash/array DBM
634 my $db = DBM::Deep->new( "foo.db" );
636 $db->{key} = 'value'; # tie() style
639 $db->put('key' => 'value'); # OO style
640 print $db->get('key');
642 # true multi-level support
643 $db->{my_complex} = [
644 'hello', { perl => 'rules' },
650 A unique flat-file database module, written in pure perl. True
651 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
652 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
653 handle millions of keys and unlimited hash levels without significant
654 slow-down. Written from the ground-up in pure perl -- this is NOT a
655 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
656 Mac OS X and Windows.
658 =head1 VERSION DIFFERENCES
660 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
661 before. While attempts have been made to be backwards compatible, no guarantees.
665 Hopefully you are using Perl's excellent CPAN module, which will download
666 and install the module for you. If not, get the tarball, and run these
678 Construction can be done OO-style (which is the recommended way), or using
679 Perl's tie() function. Both are examined here.
681 =head2 OO CONSTRUCTION
683 The recommended way to construct a DBM::Deep object is to use the new()
684 method, which gets you a blessed, tied hash or array reference.
686 my $db = DBM::Deep->new( "foo.db" );
688 This opens a new database handle, mapped to the file "foo.db". If this
689 file does not exist, it will automatically be created. DB files are
690 opened in "r+" (read/write) mode, and the type of object returned is a
691 hash, unless otherwise specified (see L<OPTIONS> below).
693 You can pass a number of options to the constructor to specify things like
694 locking, autoflush, etc. This is done by passing an inline hash:
696 my $db = DBM::Deep->new(
702 Notice that the filename is now specified I<inside> the hash with
703 the "file" parameter, as opposed to being the sole argument to the
704 constructor. This is required if any options are specified.
705 See L<OPTIONS> below for the complete list.
709 You can also start with an array instead of a hash. For this, you must
710 specify the C<type> parameter:
712 my $db = DBM::Deep->new(
714 type => DBM::Deep->TYPE_ARRAY
717 B<Note:> Specifing the C<type> parameter only takes effect when beginning
718 a new DB file. If you create a DBM::Deep object with an existing file, the
719 C<type> will be loaded from the file header, and an error will be thrown if
720 the wrong type is passed in.
722 =head2 TIE CONSTRUCTION
724 Alternately, you can create a DBM::Deep handle by using Perl's built-in
725 tie() function. The object returned from tie() can be used to call methods,
726 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
727 file (as expected with most tie'd objects).
730 my $db = tie %hash, "DBM::Deep", "foo.db";
733 my $db = tie @array, "DBM::Deep", "bar.db";
735 As with the OO constructor, you can replace the DB filename parameter with
736 a hash containing one or more options (see L<OPTIONS> just below for the
739 tie %hash, "DBM::Deep", {
747 There are a number of options that can be passed in when constructing your
748 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
754 Filename of the DB file to link the handle to. You can pass a full absolute
755 filesystem path, partial path, or a plain filename if the file is in the
756 current working directory. This is a required parameter (though q.v. fh).
760 If you want, you can pass in the fh instead of the file. This is most useful for doing
763 my $db = DBM::Deep->new( { fh => \*DATA } );
765 You are responsible for making sure that the fh has been opened appropriately for your
766 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
767 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
768 needs to read from the fh.
772 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
773 not need to set this. However, it's there if you want it.
775 If you pass in fh and do not set this, it will be set appropriately.
779 This parameter specifies what type of object to create, a hash or array. Use
780 one of these two constants:
784 =item * C<DBM::Deep-E<gt>TYPE_HASH>
786 =item * C<DBM::Deep-E<gt>TYPE_ARRAY>.
790 This only takes effect when beginning a new file. This is an optional
791 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
795 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
796 function to lock the database in exclusive mode for writes, and shared mode for
797 reads. Pass any true value to enable. This affects the base DB handle I<and
798 any child hashes or arrays> that use the same DB file. This is an optional
799 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
803 Specifies whether autoflush is to be enabled on the underlying filehandle.
804 This obviously slows down write operations, but is required if you may have
805 multiple processes accessing the same DB file (also consider enable I<locking>).
806 Pass any true value to enable. This is an optional parameter, and defaults to 0
811 If I<autobless> mode is enabled, DBM::Deep will preserve the class something
812 is blessed into, and restores it when fetched. This is an optional parameter, and defaults to 1 (enabled).
814 B<Note:> If you use the OO-interface, you will not be able to call any methods
815 of DBM::Deep on the blessed item. This is considered to be a feature.
819 See L</FILTERS> below.
825 With DBM::Deep you can access your databases using Perl's standard hash/array
826 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
827 treat them as such. DBM::Deep will intercept all reads/writes and direct them
828 to the right place -- the DB file. This has nothing to do with the
829 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
830 using regular hashes and arrays, rather than calling functions like C<get()>
831 and C<put()> (although those work too). It is entirely up to you how to want
832 to access your databases.
836 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
837 or even nested hashes (or arrays) using standard Perl syntax:
839 my $db = DBM::Deep->new( "foo.db" );
841 $db->{mykey} = "myvalue";
843 $db->{myhash}->{subkey} = "subvalue";
845 print $db->{myhash}->{subkey} . "\n";
847 You can even step through hash keys using the normal Perl C<keys()> function:
849 foreach my $key (keys %$db) {
850 print "$key: " . $db->{$key} . "\n";
853 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
854 pushes them onto an array, all before the loop even begins. If you have an
855 extra large hash, this may exhaust Perl's memory. Instead, consider using
856 Perl's C<each()> function, which pulls keys/values one at a time, using very
859 while (my ($key, $value) = each %$db) {
860 print "$key: $value\n";
863 Please note that when using C<each()>, you should always pass a direct
864 hash reference, not a lookup. Meaning, you should B<never> do this:
867 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
869 This causes an infinite loop, because for each iteration, Perl is calling
870 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
871 it effectively keeps returning the first key over and over again. Instead,
872 assign a temporary variable to C<$db->{foo}>, then pass that to each().
876 As with hashes, you can treat any DBM::Deep object like a normal Perl array
877 reference. This includes inserting, removing and manipulating elements,
878 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
879 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
880 or simply be a nested array reference inside a hash. Example:
882 my $db = DBM::Deep->new(
883 file => "foo-array.db",
884 type => DBM::Deep->TYPE_ARRAY
888 push @$db, "bar", "baz";
891 my $last_elem = pop @$db; # baz
892 my $first_elem = shift @$db; # bah
893 my $second_elem = $db->[1]; # bar
895 my $num_elements = scalar @$db;
899 In addition to the I<tie()> interface, you can also use a standard OO interface
900 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
901 array) has its own methods, but both types share the following common methods:
902 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
906 =item * new() / clone()
908 These are the constructor and copy-functions.
910 =item * put() / store()
912 Stores a new hash key/value pair, or sets an array element value. Takes two
913 arguments, the hash key or array index, and the new value. The value can be
914 a scalar, hash ref or array ref. Returns true on success, false on failure.
916 $db->put("foo", "bar"); # for hashes
917 $db->put(1, "bar"); # for arrays
919 =item * get() / fetch()
921 Fetches the value of a hash key or array element. Takes one argument: the hash
922 key or array index. Returns a scalar, hash ref or array ref, depending on the
925 my $value = $db->get("foo"); # for hashes
926 my $value = $db->get(1); # for arrays
930 Checks if a hash key or array index exists. Takes one argument: the hash key
931 or array index. Returns true if it exists, false if not.
933 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
934 if ($db->exists(1)) { print "yay!\n"; } # for arrays
938 Deletes one hash key/value pair or array element. Takes one argument: the hash
939 key or array index. Returns true on success, false if not found. For arrays,
940 the remaining elements located after the deleted element are NOT moved over.
941 The deleted element is essentially just undefined, which is exactly how Perl's
942 internal arrays work. Please note that the space occupied by the deleted
943 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
944 below for details and workarounds.
946 $db->delete("foo"); # for hashes
947 $db->delete(1); # for arrays
951 Deletes B<all> hash keys or array elements. Takes no arguments. No return
952 value. Please note that the space occupied by the deleted keys/values or
953 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
954 details and workarounds.
956 $db->clear(); # hashes or arrays
958 =item * lock() / unlock()
964 Recover lost disk space.
966 =item * import() / export()
968 Data going in and out.
974 For hashes, DBM::Deep supports all the common methods described above, and the
975 following additional methods: C<first_key()> and C<next_key()>.
981 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
982 fetched in an undefined order (which appears random). Takes no arguments,
983 returns the key as a scalar value.
985 my $key = $db->first_key();
989 Returns the "next" key in the hash, given the previous one as the sole argument.
990 Returns undef if there are no more keys to be fetched.
992 $key = $db->next_key($key);
996 Here are some examples of using hashes:
998 my $db = DBM::Deep->new( "foo.db" );
1000 $db->put("foo", "bar");
1001 print "foo: " . $db->get("foo") . "\n";
1003 $db->put("baz", {}); # new child hash ref
1004 $db->get("baz")->put("buz", "biz");
1005 print "buz: " . $db->get("baz")->get("buz") . "\n";
1007 my $key = $db->first_key();
1009 print "$key: " . $db->get($key) . "\n";
1010 $key = $db->next_key($key);
1013 if ($db->exists("foo")) { $db->delete("foo"); }
1017 For arrays, DBM::Deep supports all the common methods described above, and the
1018 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1019 C<unshift()> and C<splice()>.
1025 Returns the number of elements in the array. Takes no arguments.
1027 my $len = $db->length();
1031 Adds one or more elements onto the end of the array. Accepts scalars, hash
1032 refs or array refs. No return value.
1034 $db->push("foo", "bar", {});
1038 Fetches the last element in the array, and deletes it. Takes no arguments.
1039 Returns undef if array is empty. Returns the element value.
1041 my $elem = $db->pop();
1045 Fetches the first element in the array, deletes it, then shifts all the
1046 remaining elements over to take up the space. Returns the element value. This
1047 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1050 my $elem = $db->shift();
1054 Inserts one or more elements onto the beginning of the array, shifting all
1055 existing elements over to make room. Accepts scalars, hash refs or array refs.
1056 No return value. This method is not recommended with large arrays -- see
1057 <LARGE ARRAYS> below for details.
1059 $db->unshift("foo", "bar", {});
1063 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1064 -f splice> for usage -- it is too complicated to document here. This method is
1065 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1069 Here are some examples of using arrays:
1071 my $db = DBM::Deep->new(
1073 type => DBM::Deep->TYPE_ARRAY
1076 $db->push("bar", "baz");
1077 $db->unshift("foo");
1080 my $len = $db->length();
1081 print "length: $len\n"; # 4
1083 for (my $k=0; $k<$len; $k++) {
1084 print "$k: " . $db->get($k) . "\n";
1087 $db->splice(1, 2, "biz", "baf");
1089 while (my $elem = shift @$db) {
1090 print "shifted: $elem\n";
1095 Enable automatic file locking by passing a true value to the C<locking>
1096 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1098 my $db = DBM::Deep->new(
1103 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1104 mode for writes, and shared mode for reads. This is required if you have
1105 multiple processes accessing the same database file, to avoid file corruption.
1106 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1107 NFS> below for more.
1109 =head2 EXPLICIT LOCKING
1111 You can explicitly lock a database, so it remains locked for multiple
1112 transactions. This is done by calling the C<lock()> method, and passing an
1113 optional lock mode argument (defaults to exclusive mode). This is particularly
1114 useful for things like counters, where the current value needs to be fetched,
1115 then incremented, then stored again.
1118 my $counter = $db->get("counter");
1120 $db->put("counter", $counter);
1129 You can pass C<lock()> an optional argument, which specifies which mode to use
1130 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1131 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1132 same as the constants defined in Perl's C<Fcntl> module.
1134 $db->lock( DBM::Deep->LOCK_SH );
1138 =head1 IMPORTING/EXPORTING
1140 You can import existing complex structures by calling the C<import()> method,
1141 and export an entire database into an in-memory structure using the C<export()>
1142 method. Both are examined here.
1146 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1147 walking the structure and adding keys/elements to the database as you go,
1148 simply pass a reference to the C<import()> method. This recursively adds
1149 everything to an existing DBM::Deep object for you. Here is an example:
1154 array1 => [ "elem0", "elem1", "elem2" ],
1156 subkey1 => "subvalue1",
1157 subkey2 => "subvalue2"
1161 my $db = DBM::Deep->new( "foo.db" );
1162 $db->import( $struct );
1164 print $db->{key1} . "\n"; # prints "value1"
1166 This recursively imports the entire C<$struct> object into C<$db>, including
1167 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1168 keys are merged with the existing ones, replacing if they already exist.
1169 The C<import()> method can be called on any database level (not just the base
1170 level), and works with both hash and array DB types.
1172 B<Note:> Make sure your existing structure has no circular references in it.
1173 These will cause an infinite loop when importing.
1177 Calling the C<export()> method on an existing DBM::Deep object will return
1178 a reference to a new in-memory copy of the database. The export is done
1179 recursively, so all nested hashes/arrays are all exported to standard Perl
1180 objects. Here is an example:
1182 my $db = DBM::Deep->new( "foo.db" );
1184 $db->{key1} = "value1";
1185 $db->{key2} = "value2";
1187 $db->{hash1}->{subkey1} = "subvalue1";
1188 $db->{hash1}->{subkey2} = "subvalue2";
1190 my $struct = $db->export();
1192 print $struct->{key1} . "\n"; # prints "value1"
1194 This makes a complete copy of the database in memory, and returns a reference
1195 to it. The C<export()> method can be called on any database level (not just
1196 the base level), and works with both hash and array DB types. Be careful of
1197 large databases -- you can store a lot more data in a DBM::Deep object than an
1198 in-memory Perl structure.
1200 B<Note:> Make sure your database has no circular references in it.
1201 These will cause an infinite loop when exporting.
1205 DBM::Deep has a number of hooks where you can specify your own Perl function
1206 to perform filtering on incoming or outgoing data. This is a perfect
1207 way to extend the engine, and implement things like real-time compression or
1208 encryption. Filtering applies to the base DB level, and all child hashes /
1209 arrays. Filter hooks can be specified when your DBM::Deep object is first
1210 constructed, or by calling the C<set_filter()> method at any time. There are
1211 four available filter hooks, described below:
1215 =item * filter_store_key
1217 This filter is called whenever a hash key is stored. It
1218 is passed the incoming key, and expected to return a transformed key.
1220 =item * filter_store_value
1222 This filter is called whenever a hash key or array element is stored. It
1223 is passed the incoming value, and expected to return a transformed value.
1225 =item * filter_fetch_key
1227 This filter is called whenever a hash key is fetched (i.e. via
1228 C<first_key()> or C<next_key()>). It is passed the transformed key,
1229 and expected to return the plain key.
1231 =item * filter_fetch_value
1233 This filter is called whenever a hash key or array element is fetched.
1234 It is passed the transformed value, and expected to return the plain value.
1238 Here are the two ways to setup a filter hook:
1240 my $db = DBM::Deep->new(
1242 filter_store_value => \&my_filter_store,
1243 filter_fetch_value => \&my_filter_fetch
1248 $db->set_filter( "filter_store_value", \&my_filter_store );
1249 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1251 Your filter function will be called only when dealing with SCALAR keys or
1252 values. When nested hashes and arrays are being stored/fetched, filtering
1253 is bypassed. Filters are called as static functions, passed a single SCALAR
1254 argument, and expected to return a single SCALAR value. If you want to
1255 remove a filter, set the function reference to C<undef>:
1257 $db->set_filter( "filter_store_value", undef );
1259 =head2 REAL-TIME ENCRYPTION EXAMPLE
1261 Here is a working example that uses the I<Crypt::Blowfish> module to
1262 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1263 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1264 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1267 use Crypt::Blowfish;
1270 my $cipher = Crypt::CBC->new({
1271 'key' => 'my secret key',
1272 'cipher' => 'Blowfish',
1274 'regenerate_key' => 0,
1275 'padding' => 'space',
1279 my $db = DBM::Deep->new(
1280 file => "foo-encrypt.db",
1281 filter_store_key => \&my_encrypt,
1282 filter_store_value => \&my_encrypt,
1283 filter_fetch_key => \&my_decrypt,
1284 filter_fetch_value => \&my_decrypt,
1287 $db->{key1} = "value1";
1288 $db->{key2} = "value2";
1289 print "key1: " . $db->{key1} . "\n";
1290 print "key2: " . $db->{key2} . "\n";
1296 return $cipher->encrypt( $_[0] );
1299 return $cipher->decrypt( $_[0] );
1302 =head2 REAL-TIME COMPRESSION EXAMPLE
1304 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1305 compression / decompression of keys & values with DBM::Deep Filters.
1306 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1307 more on I<Compress::Zlib>.
1312 my $db = DBM::Deep->new(
1313 file => "foo-compress.db",
1314 filter_store_key => \&my_compress,
1315 filter_store_value => \&my_compress,
1316 filter_fetch_key => \&my_decompress,
1317 filter_fetch_value => \&my_decompress,
1320 $db->{key1} = "value1";
1321 $db->{key2} = "value2";
1322 print "key1: " . $db->{key1} . "\n";
1323 print "key2: " . $db->{key2} . "\n";
1329 return Compress::Zlib::memGzip( $_[0] ) ;
1332 return Compress::Zlib::memGunzip( $_[0] ) ;
1335 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1336 actually numerical index numbers, and are not filtered.
1338 =head1 ERROR HANDLING
1340 Most DBM::Deep methods return a true value for success, and call die() on
1341 failure. You can wrap calls in an eval block to catch the die.
1343 my $db = DBM::Deep->new( "foo.db" ); # create hash
1344 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1346 print $@; # prints error message
1348 =head1 LARGEFILE SUPPORT
1350 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1351 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1352 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1353 by specifying the 'pack_size' parameter when constructing the file.
1356 filename => $filename,
1357 pack_size => 'large',
1360 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1361 instead of 32-bit longs. After setting these values your DB files have a
1362 theoretical maximum size of 16 XB (exabytes).
1364 You can also use C<pack_size =E<gt> 'small'> in order to use 16-bit file
1367 B<Note:> Changing these values will B<NOT> work for existing database files.
1368 Only change this for new files. Once the value has been set, it is stored in
1369 the file's header and cannot be changed for the life of the file. These
1370 parameters are per-file, meaning you can access 32-bit and 64-bit files, as
1373 B<Note:> We have not personally tested files larger than 2 GB -- all my
1374 systems have only a 32-bit Perl. However, I have received user reports that
1375 this does indeed work!
1377 =head1 LOW-LEVEL ACCESS
1379 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1380 you can call the C<_fh()> method, which returns the handle:
1382 my $fh = $db->_fh();
1384 This method can be called on the root level of the datbase, or any child
1385 hashes or arrays. All levels share a I<root> structure, which contains things
1386 like the filehandle, a reference counter, and all the options specified
1387 when you created the object. You can get access to this file object by
1388 calling the C<_fileobj()> method.
1390 my $file_obj = $db->_fileobj();
1392 This is useful for changing options after the object has already been created,
1393 such as enabling/disabling locking. You can also store your own temporary user
1394 data in this structure (be wary of name collision), which is then accessible from
1395 any child hash or array.
1397 =head1 CUSTOM DIGEST ALGORITHM
1399 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1400 keys. However you can override this, and use another algorithm (such as SHA-256)
1401 or even write your own. But please note that DBM::Deep currently expects zero
1402 collisions, so your algorithm has to be I<perfect>, so to speak. Collision
1403 detection may be introduced in a later version.
1405 You can specify a custom digest algorithm by passing it into the parameter
1406 list for new(), passing a reference to a subroutine as the 'digest' parameter,
1407 and the length of the algorithm's hashes (in bytes) as the 'hash_size'
1408 parameter. Here is a working example that uses a 256-bit hash from the
1409 I<Digest::SHA256> module. Please see
1410 L<http://search.cpan.org/search?module=Digest::SHA256> for more information.
1415 my $context = Digest::SHA256::new(256);
1417 my $db = DBM::Deep->new(
1418 filename => "foo-sha.db",
1419 digest => \&my_digest,
1423 $db->{key1} = "value1";
1424 $db->{key2} = "value2";
1425 print "key1: " . $db->{key1} . "\n";
1426 print "key2: " . $db->{key2} . "\n";
1432 return substr( $context->hash($_[0]), 0, 32 );
1435 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1436 of bytes you specify in the hash_size parameter (in this case 32).
1438 B<Note:> If you do choose to use a custom digest algorithm, you must set it
1439 every time you access this file. Otherwise, the default (MD5) will be used.
1441 =head1 CIRCULAR REFERENCES
1443 DBM::Deep has B<experimental> support for circular references. Meaning you
1444 can have a nested hash key or array element that points to a parent object.
1445 This relationship is stored in the DB file, and is preserved between sessions.
1448 my $db = DBM::Deep->new( "foo.db" );
1451 $db->{circle} = $db; # ref to self
1453 print $db->{foo} . "\n"; # prints "bar"
1454 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1456 B<Note>: Passing the object to a function that recursively walks the
1457 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1458 C<export()> methods) will result in an infinite loop. This will be fixed in
1461 =head1 CAVEATS / ISSUES / BUGS
1463 This section describes all the known issues with DBM::Deep. It you have found
1464 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1466 =head2 UNUSED SPACE RECOVERY
1468 One major caveat with DBM::Deep is that space occupied by existing keys and
1469 values is not recovered when they are deleted. Meaning if you keep deleting
1470 and adding new keys, your file will continuously grow. I am working on this,
1471 but in the meantime you can call the built-in C<optimize()> method from time to
1472 time (perhaps in a crontab or something) to recover all your unused space.
1474 $db->optimize(); # returns true on success
1476 This rebuilds the ENTIRE database into a new file, then moves it on top of
1477 the original. The new file will have no unused space, thus it will take up as
1478 little disk space as possible. Please note that this operation can take
1479 a long time for large files, and you need enough disk space to temporarily hold
1480 2 copies of your DB file. The temporary file is created in the same directory
1481 as the original, named with a ".tmp" extension, and is deleted when the
1482 operation completes. Oh, and if locking is enabled, the DB is automatically
1483 locked for the entire duration of the copy.
1485 B<WARNING:> Only call optimize() on the top-level node of the database, and
1486 make sure there are no child references lying around. DBM::Deep keeps a reference
1487 counter, and if it is greater than 1, optimize() will abort and return undef.
1491 (The reasons given assume a high level of Perl understanding, specifically of
1492 references. You can safely skip this section.)
1494 Currently, the only references supported are HASH and ARRAY. The other reference
1495 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1501 These are things like filehandles and other sockets. They can't be supported
1502 because it's completely unclear how DBM::Deep should serialize them.
1504 =item * SCALAR / REF
1506 The discussion here refers to the following type of example:
1513 # In some other process ...
1515 my $val = ${ $db->{key1} };
1517 is( $val, 50, "What actually gets stored in the DB file?" );
1519 The problem is one of synchronization. When the variable being referred to
1520 changes value, the reference isn't notified. This means that the new value won't
1521 be stored in the datafile for other processes to read. There is no TIEREF.
1523 It is theoretically possible to store references to values already within a
1524 DBM::Deep object because everything already is synchronized, but the change to
1525 the internals would be quite large. Specifically, DBM::Deep would have to tie
1526 every single value that is stored. This would bloat the RAM footprint of
1527 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1528 all to support a feature that has never been requested.
1532 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1533 mechanism for serializing coderefs, including saving off all closure state.
1534 However, just as for SCALAR and REF, that closure state may change without
1535 notifying the DBM::Deep object storing the reference.
1539 =head2 FILE CORRUPTION
1541 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1542 for a 32-bit signature when opened, but other corruption in files can cause
1543 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1544 stuck in an infinite loop depending on the level of corruption. File write
1545 operations are not checked for failure (for speed), so if you happen to run
1546 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1547 be addressed in a later version of DBM::Deep.
1551 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1552 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1553 about setting up your NFS server with a locking daemon, then using lockf() to
1554 lock your files, but your mileage may vary there as well. From what I
1555 understand, there is no real way to do it. However, if you need access to the
1556 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1557 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1559 =head2 COPYING OBJECTS
1561 Beware of copying tied objects in Perl. Very strange things can happen.
1562 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1563 returns a new, blessed, tied hash or array to the same level in the DB.
1565 my $copy = $db->clone();
1567 B<Note>: Since clone() here is cloning the object, not the database location, any
1568 modifications to either $db or $copy will be visible in both.
1572 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1573 These functions cause every element in the array to move, which can be murder
1574 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1575 a different location. This will be addressed in the forthcoming version 1.00.
1577 =head2 WRITEONLY FILES
1579 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1580 writeonly mode. STORE will verify that the filehandle is writable. However, there
1581 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1582 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1586 This section discusses DBM::Deep's speed and memory usage.
1590 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1591 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1592 multi-level hash/array support, and cross-platform FTPable files. Even so,
1593 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1594 with huge databases. Here is some test data:
1596 Adding 1,000,000 keys to new DB file...
1598 At 100 keys, avg. speed is 2,703 keys/sec
1599 At 200 keys, avg. speed is 2,642 keys/sec
1600 At 300 keys, avg. speed is 2,598 keys/sec
1601 At 400 keys, avg. speed is 2,578 keys/sec
1602 At 500 keys, avg. speed is 2,722 keys/sec
1603 At 600 keys, avg. speed is 2,628 keys/sec
1604 At 700 keys, avg. speed is 2,700 keys/sec
1605 At 800 keys, avg. speed is 2,607 keys/sec
1606 At 900 keys, avg. speed is 2,190 keys/sec
1607 At 1,000 keys, avg. speed is 2,570 keys/sec
1608 At 2,000 keys, avg. speed is 2,417 keys/sec
1609 At 3,000 keys, avg. speed is 1,982 keys/sec
1610 At 4,000 keys, avg. speed is 1,568 keys/sec
1611 At 5,000 keys, avg. speed is 1,533 keys/sec
1612 At 6,000 keys, avg. speed is 1,787 keys/sec
1613 At 7,000 keys, avg. speed is 1,977 keys/sec
1614 At 8,000 keys, avg. speed is 2,028 keys/sec
1615 At 9,000 keys, avg. speed is 2,077 keys/sec
1616 At 10,000 keys, avg. speed is 2,031 keys/sec
1617 At 20,000 keys, avg. speed is 1,970 keys/sec
1618 At 30,000 keys, avg. speed is 2,050 keys/sec
1619 At 40,000 keys, avg. speed is 2,073 keys/sec
1620 At 50,000 keys, avg. speed is 1,973 keys/sec
1621 At 60,000 keys, avg. speed is 1,914 keys/sec
1622 At 70,000 keys, avg. speed is 2,091 keys/sec
1623 At 80,000 keys, avg. speed is 2,103 keys/sec
1624 At 90,000 keys, avg. speed is 1,886 keys/sec
1625 At 100,000 keys, avg. speed is 1,970 keys/sec
1626 At 200,000 keys, avg. speed is 2,053 keys/sec
1627 At 300,000 keys, avg. speed is 1,697 keys/sec
1628 At 400,000 keys, avg. speed is 1,838 keys/sec
1629 At 500,000 keys, avg. speed is 1,941 keys/sec
1630 At 600,000 keys, avg. speed is 1,930 keys/sec
1631 At 700,000 keys, avg. speed is 1,735 keys/sec
1632 At 800,000 keys, avg. speed is 1,795 keys/sec
1633 At 900,000 keys, avg. speed is 1,221 keys/sec
1634 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1636 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1637 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1638 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1639 Run time was 12 min 3 sec.
1643 One of the great things about DBM::Deep is that it uses very little memory.
1644 Even with huge databases (1,000,000+ keys) you will not see much increased
1645 memory on your process. DBM::Deep relies solely on the filesystem for storing
1646 and fetching data. Here is output from I</usr/bin/top> before even opening a
1649 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1650 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1652 Basically the process is taking 2,716K of memory. And here is the same
1653 process after storing and fetching 1,000,000 keys:
1655 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1656 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1658 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1659 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1661 =head1 DB FILE FORMAT
1663 In case you were interested in the underlying DB file format, it is documented
1664 here in this section. You don't need to know this to use the module, it's just
1665 included for reference.
1669 DBM::Deep files always start with a 32-bit signature to identify the file type.
1670 This is at offset 0. The signature is "DPDB" in network byte order. This is
1671 checked for when the file is opened and an error will be thrown if it's not found.
1675 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1676 has a standard header containing the type of data, the length of data, and then
1677 the data itself. The type is a single character (1 byte), the length is a
1678 32-bit unsigned long in network byte order, and the data is, well, the data.
1679 Here is how it unfolds:
1683 Immediately after the 32-bit file signature is the I<Master Index> record.
1684 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1685 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1686 depending on how the DBM::Deep object was constructed.
1688 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1689 number). The first 8-bit char of the MD5 signature is the offset into the
1690 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1691 index element is a file offset of the next tag for the key/element in question,
1692 which is usually a I<Bucket List> tag (see below).
1694 The next tag I<could> be another index, depending on how many keys/elements
1695 exist. See L<RE-INDEXING> below for details.
1699 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1700 file offsets to where the actual data is stored. It starts with a standard
1701 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1702 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1703 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1704 When the list fills up, a I<Re-Index> operation is performed (See
1705 L<RE-INDEXING> below).
1709 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1710 index/value pair (in array mode). It starts with a standard tag header with
1711 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1712 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1713 header. The size reported in the tag header is only for the value, but then,
1714 just after the value is another size (32-bit unsigned long) and then the plain
1715 key itself. Since the value is likely to be fetched more often than the plain
1716 key, I figured it would be I<slightly> faster to store the value first.
1718 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1719 record for the nested structure, where the process begins all over again.
1723 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1724 exhausted. Then, when another key/element comes in, the list is converted to a
1725 new index record. However, this index will look at the next char in the MD5
1726 hash, and arrange new Bucket List pointers accordingly. This process is called
1727 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1728 17 (16 + new one) keys/elements are removed from the old Bucket List and
1729 inserted into the new index. Several new Bucket Lists are created in the
1730 process, as a new MD5 char from the key is being examined (it is unlikely that
1731 the keys will all share the same next char of their MD5s).
1733 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1734 when the Bucket Lists will turn into indexes, but the first round tends to
1735 happen right around 4,000 keys. You will see a I<slight> decrease in
1736 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1737 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1738 right around 900,000 keys. This process can continue nearly indefinitely --
1739 right up until the point the I<MD5> signatures start colliding with each other,
1740 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1741 getting struck by lightning while you are walking to cash in your tickets.
1742 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1743 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1744 this is 340 unodecillion, but don't quote me).
1748 When a new key/element is stored, the key (or index number) is first run through
1749 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1750 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1751 for the first char of the signature (in this case I<b0>). If it does not exist,
1752 a new I<Bucket List> is created for our key (and the next 15 future keys that
1753 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1754 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1755 this point, unless we are replacing an existing I<Bucket>), where the actual
1756 data will be stored.
1760 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1761 (or index number), then walking along the indexes. If there are enough
1762 keys/elements in this DB level, there might be nested indexes, each linked to
1763 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1764 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1765 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1766 plain key are stored.
1768 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1769 methods. In this process the indexes are walked systematically, and each key
1770 fetched in increasing MD5 order (which is why it appears random). Once the
1771 I<Bucket> is found, the value is skipped and the plain key returned instead.
1772 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1773 alphabetically sorted. This only happens on an index-level -- as soon as the
1774 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1775 so it's pretty much undefined how the keys will come out -- just like Perl's
1778 =head1 CODE COVERAGE
1780 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1781 B<Devel::Cover> report on this module's test suite.
1783 ----------------------------------- ------ ------ ------ ------ ------ ------
1784 File stmt bran cond sub time total
1785 ----------------------------------- ------ ------ ------ ------ ------ ------
1786 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1787 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1788 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1789 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1790 Total 97.9 85.9 79.7 100.0 100.0 94.3
1791 ----------------------------------- ------ ------ ------ ------ ------ ------
1793 =head1 MORE INFORMATION
1795 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1796 or send email to L<DBM-Deep@googlegroups.com>.
1800 Joseph Huckaby, L<jhuckaby@cpan.org>
1802 Rob Kinyon, L<rkinyon@cpan.org>
1804 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1808 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1809 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1813 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1814 This is free software, you may use it and distribute it under the
1815 same terms as Perl itself.