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 engine => DBM::Deep::Engine->new( $args ),
119 base_offset => undef,
123 # Grab the parameters we want to use
124 foreach my $param ( keys %$self ) {
125 next unless exists $args->{$param};
126 $self->{$param} = $args->{$param};
129 $self->{engine}->setup_fh( $self );
136 require DBM::Deep::Hash;
137 return DBM::Deep::Hash->TIEHASH( @_ );
142 require DBM::Deep::Array;
143 return DBM::Deep::Array->TIEARRAY( @_ );
146 #XXX Unneeded now ...
152 # If db locking is set, flock() the db file. If called multiple
153 # times before unlock(), then the same number of unlocks() must
154 # be called before the lock is released.
156 my $self = shift->_get_self;
158 $type = LOCK_EX unless defined $type;
160 if (!defined($self->_fh)) { return; }
162 if ($self->_fileobj->{locking}) {
163 if (!$self->_fileobj->{locked}) {
164 flock($self->_fh, $type);
166 # refresh end counter in case file has changed size
167 my @stats = stat($self->_fh);
168 $self->_fileobj->{end} = $stats[7];
170 # double-check file inode, in case another process
171 # has optimize()d our file while we were waiting.
172 if ($stats[1] != $self->_fileobj->{inode}) {
173 $self->_fileobj->close;
174 $self->_fileobj->open;
175 $self->{engine}->setup_fh( $self );
176 flock($self->_fh, $type); # re-lock
178 # This may not be necessary after re-opening
179 $self->_fileobj->{end} = (stat($self->_fh))[7]; # re-end
182 $self->_fileobj->{locked}++;
192 # If db locking is set, unlock the db file. See note in lock()
193 # regarding calling lock() multiple times.
195 my $self = shift->_get_self;
197 if (!defined($self->_fh)) { return; }
199 if ($self->_fileobj->{locking} && $self->_fileobj->{locked} > 0) {
200 $self->_fileobj->{locked}--;
201 if (!$self->_fileobj->{locked}) { flock($self->_fh, LOCK_UN); }
210 my $self = shift->_get_self;
211 my ($spot, $value) = @_;
216 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
217 ${$spot} = $value->_repr;
218 $value->_copy_node( ${$spot} );
221 my $r = Scalar::Util::reftype( $value );
222 my $c = Scalar::Util::blessed( $value );
223 if ( $r eq 'ARRAY' ) {
224 ${$spot} = [ @{$value} ];
227 ${$spot} = { %{$value} };
229 ${$spot} = bless ${$spot}, $c
237 die "Must be implemented in a child class\n";
241 die "Must be implemented in a child class\n";
246 # Recursively export into standard Perl hashes and arrays.
248 my $self = shift->_get_self;
250 my $temp = $self->_repr;
253 $self->_copy_node( $temp );
261 # Recursively import Perl hash/array structure
263 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
265 my $self = shift->_get_self;
268 # struct is not a reference, so just import based on our type
270 $struct = $self->_repr( @_ );
273 return $self->_import( $struct );
278 # Rebuild entire database into new file, then move
279 # it back on top of original.
281 my $self = shift->_get_self;
283 #XXX Need to create a new test for this
284 # if ($self->_fileobj->{links} > 1) {
285 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
288 my $db_temp = DBM::Deep->new(
289 file => $self->_fileobj->{file} . '.tmp',
294 $self->_copy_node( $db_temp );
298 # Attempt to copy user, group and permissions over to new file
300 my @stats = stat($self->_fh);
301 my $perms = $stats[2] & 07777;
304 chown( $uid, $gid, $self->_fileobj->{file} . '.tmp' );
305 chmod( $perms, $self->_fileobj->{file} . '.tmp' );
307 # q.v. perlport for more information on this variable
308 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
310 # Potential race condition when optmizing on Win32 with locking.
311 # The Windows filesystem requires that the filehandle be closed
312 # before it is overwritten with rename(). This could be redone
316 $self->_fileobj->close;
319 if (!rename $self->_fileobj->{file} . '.tmp', $self->_fileobj->{file}) {
320 unlink $self->_fileobj->{file} . '.tmp';
322 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
326 $self->_fileobj->close;
327 $self->_fileobj->open;
328 $self->{engine}->setup_fh( $self );
335 # Make copy of object and return
337 my $self = shift->_get_self;
339 return DBM::Deep->new(
340 type => $self->_type,
341 base_offset => $self->_base_offset,
342 fileobj => $self->_fileobj,
347 my %is_legal_filter = map {
350 store_key store_value
351 fetch_key fetch_value
356 # Setup filter function for storing or fetching the key or value
358 my $self = shift->_get_self;
362 if ( $is_legal_filter{$type} ) {
363 $self->_fileobj->{"filter_$type"} = $func;
372 my $self = shift->_get_self;
376 my $self = shift->_get_self;
380 # my $self = shift->_get_self;
389 # Get access to the root structure
391 my $self = $_[0]->_get_self;
392 return $self->{fileobj};
397 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
399 my $self = $_[0]->_get_self;
400 return $self->{type};
405 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
407 my $self = $_[0]->_get_self;
408 return $self->{base_offset};
413 # Get access to the raw fh
415 my $self = $_[0]->_get_self;
416 return $self->_fileobj->{fh};
424 die "DBM::Deep: $_[1]\n";
429 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
434 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
439 # Store single hash key/value or array element in database.
441 my $self = shift->_get_self;
442 my ($key, $value) = @_;
444 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
445 $self->_throw_error( 'Cannot write to a readonly filehandle' );
449 # Request exclusive lock for writing
451 $self->lock( LOCK_EX );
453 my $md5 = $self->{engine}{digest}->($key);
455 my $tag = $self->{engine}->find_bucket_list( $self, $md5, { create => 1 } );
457 # User may be storing a hash, in which case we do not want it run
458 # through the filtering system
459 if ( !ref($value) && $self->_fileobj->{filter_store_value} ) {
460 $value = $self->_fileobj->{filter_store_value}->( $value );
464 # Add key/value to bucket list
466 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
475 # Fetch single value or element given plain key or array index
477 my $self = shift->_get_self;
480 my $md5 = $self->{engine}{digest}->($key);
483 # Request shared lock for reading
485 $self->lock( LOCK_SH );
487 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
494 # Get value from bucket list
496 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
500 # Filters only apply to scalar values, so the ref check is making
501 # sure the fetched bucket is a scalar, not a child hash or array.
502 return ($result && !ref($result) && $self->_fileobj->{filter_fetch_value})
503 ? $self->_fileobj->{filter_fetch_value}->($result)
509 # Delete single key/value pair or element given plain key or array index
511 my $self = $_[0]->_get_self;
514 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
515 $self->_throw_error( 'Cannot write to a readonly filehandle' );
519 # Request exclusive lock for writing
521 $self->lock( LOCK_EX );
523 my $md5 = $self->{engine}{digest}->($key);
525 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
534 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
536 if (defined $value && !ref($value) && $self->_fileobj->{filter_fetch_value}) {
537 $value = $self->_fileobj->{filter_fetch_value}->($value);
540 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
543 # If this object is an array and the key deleted was on the end of the stack,
544 # decrement the length variable.
554 # Check if a single key or element exists given plain key or array index
556 my $self = $_[0]->_get_self;
559 my $md5 = $self->{engine}{digest}->($key);
562 # Request shared lock for reading
564 $self->lock( LOCK_SH );
566 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
571 # For some reason, the built-in exists() function returns '' for false
577 # Check if bucket exists and return 1 or ''
579 my $result = $self->{engine}->bucket_exists( $self, $tag, $md5 ) || '';
588 # Clear all keys from hash, or all elements from array.
590 my $self = $_[0]->_get_self;
592 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
593 $self->_throw_error( 'Cannot write to a readonly filehandle' );
597 # Request exclusive lock for writing
599 $self->lock( LOCK_EX );
603 seek($fh, $self->_base_offset + $self->_fileobj->{file_offset}, SEEK_SET);
609 #XXX This needs updating to use _release_space
610 $self->{engine}->write_tag(
611 $self, $self->_base_offset, $self->_type,
612 chr(0)x$self->{engine}{index_size},
621 # Public method aliases
623 sub put { (shift)->STORE( @_ ) }
624 sub store { (shift)->STORE( @_ ) }
625 sub get { (shift)->FETCH( @_ ) }
626 sub fetch { (shift)->FETCH( @_ ) }
627 sub delete { (shift)->DELETE( @_ ) }
628 sub exists { (shift)->EXISTS( @_ ) }
629 sub clear { (shift)->CLEAR( @_ ) }
636 DBM::Deep - A pure perl multi-level hash/array DBM
641 my $db = DBM::Deep->new( "foo.db" );
643 $db->{key} = 'value'; # tie() style
646 $db->put('key' => 'value'); # OO style
647 print $db->get('key');
649 # true multi-level support
650 $db->{my_complex} = [
651 'hello', { perl => 'rules' },
657 A unique flat-file database module, written in pure perl. True
658 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
659 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
660 handle millions of keys and unlimited hash levels without significant
661 slow-down. Written from the ground-up in pure perl -- this is NOT a
662 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
663 Mac OS X and Windows.
665 =head1 VERSION DIFFERENCES
667 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
668 before. While attempts have been made to be backwards compatible, no guarantees.
672 Hopefully you are using Perl's excellent CPAN module, which will download
673 and install the module for you. If not, get the tarball, and run these
685 Construction can be done OO-style (which is the recommended way), or using
686 Perl's tie() function. Both are examined here.
688 =head2 OO CONSTRUCTION
690 The recommended way to construct a DBM::Deep object is to use the new()
691 method, which gets you a blessed, tied hash or array reference.
693 my $db = DBM::Deep->new( "foo.db" );
695 This opens a new database handle, mapped to the file "foo.db". If this
696 file does not exist, it will automatically be created. DB files are
697 opened in "r+" (read/write) mode, and the type of object returned is a
698 hash, unless otherwise specified (see L<OPTIONS> below).
700 You can pass a number of options to the constructor to specify things like
701 locking, autoflush, etc. This is done by passing an inline hash:
703 my $db = DBM::Deep->new(
709 Notice that the filename is now specified I<inside> the hash with
710 the "file" parameter, as opposed to being the sole argument to the
711 constructor. This is required if any options are specified.
712 See L<OPTIONS> below for the complete list.
716 You can also start with an array instead of a hash. For this, you must
717 specify the C<type> parameter:
719 my $db = DBM::Deep->new(
721 type => DBM::Deep->TYPE_ARRAY
724 B<Note:> Specifing the C<type> parameter only takes effect when beginning
725 a new DB file. If you create a DBM::Deep object with an existing file, the
726 C<type> will be loaded from the file header, and an error will be thrown if
727 the wrong type is passed in.
729 =head2 TIE CONSTRUCTION
731 Alternately, you can create a DBM::Deep handle by using Perl's built-in
732 tie() function. The object returned from tie() can be used to call methods,
733 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
734 file (as expected with most tie'd objects).
737 my $db = tie %hash, "DBM::Deep", "foo.db";
740 my $db = tie @array, "DBM::Deep", "bar.db";
742 As with the OO constructor, you can replace the DB filename parameter with
743 a hash containing one or more options (see L<OPTIONS> just below for the
746 tie %hash, "DBM::Deep", {
754 There are a number of options that can be passed in when constructing your
755 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
761 Filename of the DB file to link the handle to. You can pass a full absolute
762 filesystem path, partial path, or a plain filename if the file is in the
763 current working directory. This is a required parameter (though q.v. fh).
767 If you want, you can pass in the fh instead of the file. This is most useful for doing
770 my $db = DBM::Deep->new( { fh => \*DATA } );
772 You are responsible for making sure that the fh has been opened appropriately for your
773 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
774 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
775 needs to read from the fh.
779 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
780 not need to set this. However, it's there if you want it.
782 If you pass in fh and do not set this, it will be set appropriately.
786 This parameter specifies what type of object to create, a hash or array. Use
787 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
788 This only takes effect when beginning a new file. This is an optional
789 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
793 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
794 function to lock the database in exclusive mode for writes, and shared mode for
795 reads. Pass any true value to enable. This affects the base DB handle I<and
796 any child hashes or arrays> that use the same DB file. This is an optional
797 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
801 Specifies whether autoflush is to be enabled on the underlying filehandle.
802 This obviously slows down write operations, but is required if you may have
803 multiple processes accessing the same DB file (also consider enable I<locking>).
804 Pass any true value to enable. This is an optional parameter, and defaults to 0
809 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
810 restore them when fetched. This is an B<experimental> feature, and does have
811 side-effects. Basically, when hashes are re-blessed into their original
812 classes, they are no longer blessed into the DBM::Deep class! So you won't be
813 able to call any DBM::Deep methods on them. You have been warned.
814 This is an optional parameter, and defaults to 0 (disabled).
818 See L<FILTERS> below.
824 With DBM::Deep you can access your databases using Perl's standard hash/array
825 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
826 treat them as such. DBM::Deep will intercept all reads/writes and direct them
827 to the right place -- the DB file. This has nothing to do with the
828 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
829 using regular hashes and arrays, rather than calling functions like C<get()>
830 and C<put()> (although those work too). It is entirely up to you how to want
831 to access your databases.
835 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
836 or even nested hashes (or arrays) using standard Perl syntax:
838 my $db = DBM::Deep->new( "foo.db" );
840 $db->{mykey} = "myvalue";
842 $db->{myhash}->{subkey} = "subvalue";
844 print $db->{myhash}->{subkey} . "\n";
846 You can even step through hash keys using the normal Perl C<keys()> function:
848 foreach my $key (keys %$db) {
849 print "$key: " . $db->{$key} . "\n";
852 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
853 pushes them onto an array, all before the loop even begins. If you have an
854 extra large hash, this may exhaust Perl's memory. Instead, consider using
855 Perl's C<each()> function, which pulls keys/values one at a time, using very
858 while (my ($key, $value) = each %$db) {
859 print "$key: $value\n";
862 Please note that when using C<each()>, you should always pass a direct
863 hash reference, not a lookup. Meaning, you should B<never> do this:
866 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
868 This causes an infinite loop, because for each iteration, Perl is calling
869 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
870 it effectively keeps returning the first key over and over again. Instead,
871 assign a temporary variable to C<$db->{foo}>, then pass that to each().
875 As with hashes, you can treat any DBM::Deep object like a normal Perl array
876 reference. This includes inserting, removing and manipulating elements,
877 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
878 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
879 or simply be a nested array reference inside a hash. Example:
881 my $db = DBM::Deep->new(
882 file => "foo-array.db",
883 type => DBM::Deep->TYPE_ARRAY
887 push @$db, "bar", "baz";
890 my $last_elem = pop @$db; # baz
891 my $first_elem = shift @$db; # bah
892 my $second_elem = $db->[1]; # bar
894 my $num_elements = scalar @$db;
898 In addition to the I<tie()> interface, you can also use a standard OO interface
899 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
900 array) has its own methods, but both types share the following common methods:
901 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
905 =item * new() / clone()
907 These are the constructor and copy-functions.
909 =item * put() / store()
911 Stores a new hash key/value pair, or sets an array element value. Takes two
912 arguments, the hash key or array index, and the new value. The value can be
913 a scalar, hash ref or array ref. Returns true on success, false on failure.
915 $db->put("foo", "bar"); # for hashes
916 $db->put(1, "bar"); # for arrays
918 =item * get() / fetch()
920 Fetches the value of a hash key or array element. Takes one argument: the hash
921 key or array index. Returns a scalar, hash ref or array ref, depending on the
924 my $value = $db->get("foo"); # for hashes
925 my $value = $db->get(1); # for arrays
929 Checks if a hash key or array index exists. Takes one argument: the hash key
930 or array index. Returns true if it exists, false if not.
932 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
933 if ($db->exists(1)) { print "yay!\n"; } # for arrays
937 Deletes one hash key/value pair or array element. Takes one argument: the hash
938 key or array index. Returns true on success, false if not found. For arrays,
939 the remaining elements located after the deleted element are NOT moved over.
940 The deleted element is essentially just undefined, which is exactly how Perl's
941 internal arrays work. Please note that the space occupied by the deleted
942 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
943 below for details and workarounds.
945 $db->delete("foo"); # for hashes
946 $db->delete(1); # for arrays
950 Deletes B<all> hash keys or array elements. Takes no arguments. No return
951 value. Please note that the space occupied by the deleted keys/values or
952 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
953 details and workarounds.
955 $db->clear(); # hashes or arrays
957 =item * lock() / unlock()
963 Recover lost disk space.
965 =item * import() / export()
967 Data going in and out.
973 For hashes, DBM::Deep supports all the common methods described above, and the
974 following additional methods: C<first_key()> and C<next_key()>.
980 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
981 fetched in an undefined order (which appears random). Takes no arguments,
982 returns the key as a scalar value.
984 my $key = $db->first_key();
988 Returns the "next" key in the hash, given the previous one as the sole argument.
989 Returns undef if there are no more keys to be fetched.
991 $key = $db->next_key($key);
995 Here are some examples of using hashes:
997 my $db = DBM::Deep->new( "foo.db" );
999 $db->put("foo", "bar");
1000 print "foo: " . $db->get("foo") . "\n";
1002 $db->put("baz", {}); # new child hash ref
1003 $db->get("baz")->put("buz", "biz");
1004 print "buz: " . $db->get("baz")->get("buz") . "\n";
1006 my $key = $db->first_key();
1008 print "$key: " . $db->get($key) . "\n";
1009 $key = $db->next_key($key);
1012 if ($db->exists("foo")) { $db->delete("foo"); }
1016 For arrays, DBM::Deep supports all the common methods described above, and the
1017 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1018 C<unshift()> and C<splice()>.
1024 Returns the number of elements in the array. Takes no arguments.
1026 my $len = $db->length();
1030 Adds one or more elements onto the end of the array. Accepts scalars, hash
1031 refs or array refs. No return value.
1033 $db->push("foo", "bar", {});
1037 Fetches the last element in the array, and deletes it. Takes no arguments.
1038 Returns undef if array is empty. Returns the element value.
1040 my $elem = $db->pop();
1044 Fetches the first element in the array, deletes it, then shifts all the
1045 remaining elements over to take up the space. Returns the element value. This
1046 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1049 my $elem = $db->shift();
1053 Inserts one or more elements onto the beginning of the array, shifting all
1054 existing elements over to make room. Accepts scalars, hash refs or array refs.
1055 No return value. This method is not recommended with large arrays -- see
1056 <LARGE ARRAYS> below for details.
1058 $db->unshift("foo", "bar", {});
1062 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1063 -f splice> for usage -- it is too complicated to document here. This method is
1064 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1068 Here are some examples of using arrays:
1070 my $db = DBM::Deep->new(
1072 type => DBM::Deep->TYPE_ARRAY
1075 $db->push("bar", "baz");
1076 $db->unshift("foo");
1079 my $len = $db->length();
1080 print "length: $len\n"; # 4
1082 for (my $k=0; $k<$len; $k++) {
1083 print "$k: " . $db->get($k) . "\n";
1086 $db->splice(1, 2, "biz", "baf");
1088 while (my $elem = shift @$db) {
1089 print "shifted: $elem\n";
1094 Enable automatic file locking by passing a true value to the C<locking>
1095 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1097 my $db = DBM::Deep->new(
1102 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1103 mode for writes, and shared mode for reads. This is required if you have
1104 multiple processes accessing the same database file, to avoid file corruption.
1105 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1106 NFS> below for more.
1108 =head2 EXPLICIT LOCKING
1110 You can explicitly lock a database, so it remains locked for multiple
1111 transactions. This is done by calling the C<lock()> method, and passing an
1112 optional lock mode argument (defaults to exclusive mode). This is particularly
1113 useful for things like counters, where the current value needs to be fetched,
1114 then incremented, then stored again.
1117 my $counter = $db->get("counter");
1119 $db->put("counter", $counter);
1128 You can pass C<lock()> an optional argument, which specifies which mode to use
1129 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1130 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1131 same as the constants defined in Perl's C<Fcntl> module.
1133 $db->lock( DBM::Deep->LOCK_SH );
1137 =head1 IMPORTING/EXPORTING
1139 You can import existing complex structures by calling the C<import()> method,
1140 and export an entire database into an in-memory structure using the C<export()>
1141 method. Both are examined here.
1145 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1146 walking the structure and adding keys/elements to the database as you go,
1147 simply pass a reference to the C<import()> method. This recursively adds
1148 everything to an existing DBM::Deep object for you. Here is an example:
1153 array1 => [ "elem0", "elem1", "elem2" ],
1155 subkey1 => "subvalue1",
1156 subkey2 => "subvalue2"
1160 my $db = DBM::Deep->new( "foo.db" );
1161 $db->import( $struct );
1163 print $db->{key1} . "\n"; # prints "value1"
1165 This recursively imports the entire C<$struct> object into C<$db>, including
1166 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1167 keys are merged with the existing ones, replacing if they already exist.
1168 The C<import()> method can be called on any database level (not just the base
1169 level), and works with both hash and array DB types.
1171 B<Note:> Make sure your existing structure has no circular references in it.
1172 These will cause an infinite loop when importing.
1176 Calling the C<export()> method on an existing DBM::Deep object will return
1177 a reference to a new in-memory copy of the database. The export is done
1178 recursively, so all nested hashes/arrays are all exported to standard Perl
1179 objects. Here is an example:
1181 my $db = DBM::Deep->new( "foo.db" );
1183 $db->{key1} = "value1";
1184 $db->{key2} = "value2";
1186 $db->{hash1}->{subkey1} = "subvalue1";
1187 $db->{hash1}->{subkey2} = "subvalue2";
1189 my $struct = $db->export();
1191 print $struct->{key1} . "\n"; # prints "value1"
1193 This makes a complete copy of the database in memory, and returns a reference
1194 to it. The C<export()> method can be called on any database level (not just
1195 the base level), and works with both hash and array DB types. Be careful of
1196 large databases -- you can store a lot more data in a DBM::Deep object than an
1197 in-memory Perl structure.
1199 B<Note:> Make sure your database has no circular references in it.
1200 These will cause an infinite loop when exporting.
1204 DBM::Deep has a number of hooks where you can specify your own Perl function
1205 to perform filtering on incoming or outgoing data. This is a perfect
1206 way to extend the engine, and implement things like real-time compression or
1207 encryption. Filtering applies to the base DB level, and all child hashes /
1208 arrays. Filter hooks can be specified when your DBM::Deep object is first
1209 constructed, or by calling the C<set_filter()> method at any time. There are
1210 four available filter hooks, described below:
1214 =item * filter_store_key
1216 This filter is called whenever a hash key is stored. It
1217 is passed the incoming key, and expected to return a transformed key.
1219 =item * filter_store_value
1221 This filter is called whenever a hash key or array element is stored. It
1222 is passed the incoming value, and expected to return a transformed value.
1224 =item * filter_fetch_key
1226 This filter is called whenever a hash key is fetched (i.e. via
1227 C<first_key()> or C<next_key()>). It is passed the transformed key,
1228 and expected to return the plain key.
1230 =item * filter_fetch_value
1232 This filter is called whenever a hash key or array element is fetched.
1233 It is passed the transformed value, and expected to return the plain value.
1237 Here are the two ways to setup a filter hook:
1239 my $db = DBM::Deep->new(
1241 filter_store_value => \&my_filter_store,
1242 filter_fetch_value => \&my_filter_fetch
1247 $db->set_filter( "filter_store_value", \&my_filter_store );
1248 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1250 Your filter function will be called only when dealing with SCALAR keys or
1251 values. When nested hashes and arrays are being stored/fetched, filtering
1252 is bypassed. Filters are called as static functions, passed a single SCALAR
1253 argument, and expected to return a single SCALAR value. If you want to
1254 remove a filter, set the function reference to C<undef>:
1256 $db->set_filter( "filter_store_value", undef );
1258 =head2 REAL-TIME ENCRYPTION EXAMPLE
1260 Here is a working example that uses the I<Crypt::Blowfish> module to
1261 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1262 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1263 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1266 use Crypt::Blowfish;
1269 my $cipher = Crypt::CBC->new({
1270 'key' => 'my secret key',
1271 'cipher' => 'Blowfish',
1273 'regenerate_key' => 0,
1274 'padding' => 'space',
1278 my $db = DBM::Deep->new(
1279 file => "foo-encrypt.db",
1280 filter_store_key => \&my_encrypt,
1281 filter_store_value => \&my_encrypt,
1282 filter_fetch_key => \&my_decrypt,
1283 filter_fetch_value => \&my_decrypt,
1286 $db->{key1} = "value1";
1287 $db->{key2} = "value2";
1288 print "key1: " . $db->{key1} . "\n";
1289 print "key2: " . $db->{key2} . "\n";
1295 return $cipher->encrypt( $_[0] );
1298 return $cipher->decrypt( $_[0] );
1301 =head2 REAL-TIME COMPRESSION EXAMPLE
1303 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1304 compression / decompression of keys & values with DBM::Deep Filters.
1305 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1306 more on I<Compress::Zlib>.
1311 my $db = DBM::Deep->new(
1312 file => "foo-compress.db",
1313 filter_store_key => \&my_compress,
1314 filter_store_value => \&my_compress,
1315 filter_fetch_key => \&my_decompress,
1316 filter_fetch_value => \&my_decompress,
1319 $db->{key1} = "value1";
1320 $db->{key2} = "value2";
1321 print "key1: " . $db->{key1} . "\n";
1322 print "key2: " . $db->{key2} . "\n";
1328 return Compress::Zlib::memGzip( $_[0] ) ;
1331 return Compress::Zlib::memGunzip( $_[0] ) ;
1334 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1335 actually numerical index numbers, and are not filtered.
1337 =head1 ERROR HANDLING
1339 Most DBM::Deep methods return a true value for success, and call die() on
1340 failure. You can wrap calls in an eval block to catch the die.
1342 my $db = DBM::Deep->new( "foo.db" ); # create hash
1343 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1345 print $@; # prints error message
1347 =head1 LARGEFILE SUPPORT
1349 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1350 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1351 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1352 by specifying the 'pack_size' parameter when constructing the file.
1355 filename => $filename,
1356 pack_size => 'large',
1359 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1360 instead of 32-bit longs. After setting these values your DB files have a
1361 theoretical maximum size of 16 XB (exabytes).
1363 You can also use C<pack_size =E<gt> 'small'> in order to use 16-bit file
1366 B<Note:> Changing these values will B<NOT> work for existing database files.
1367 Only change this for new files. Once the value has been set, it is stored in
1368 the file's header and cannot be changed for the life of the file. These
1369 parameters are per-file, meaning you can access 32-bit and 64-bit files, as
1372 B<Note:> We have not personally tested files larger than 2 GB -- all my
1373 systems have only a 32-bit Perl. However, I have received user reports that
1374 this does indeed work!
1376 =head1 LOW-LEVEL ACCESS
1378 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1379 you can call the C<_fh()> method, which returns the handle:
1381 my $fh = $db->_fh();
1383 This method can be called on the root level of the datbase, or any child
1384 hashes or arrays. All levels share a I<root> structure, which contains things
1385 like the filehandle, a reference counter, and all the options specified
1386 when you created the object. You can get access to this file object by
1387 calling the C<_fileobj()> method.
1389 my $file_obj = $db->_fileobj();
1391 This is useful for changing options after the object has already been created,
1392 such as enabling/disabling locking. You can also store your own temporary user
1393 data in this structure (be wary of name collision), which is then accessible from
1394 any child hash or array.
1396 =head1 CUSTOM DIGEST ALGORITHM
1398 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1399 keys. However you can override this, and use another algorithm (such as SHA-256)
1400 or even write your own. But please note that DBM::Deep currently expects zero
1401 collisions, so your algorithm has to be I<perfect>, so to speak. Collision
1402 detection may be introduced in a later version.
1404 You can specify a custom digest algorithm by passing it into the parameter
1405 list for new(), passing a reference to a subroutine as the 'digest' parameter,
1406 and the length of the algorithm's hashes (in bytes) as the 'hash_size'
1407 parameter. Here is a working example that uses a 256-bit hash from the
1408 I<Digest::SHA256> module. Please see
1409 L<http://search.cpan.org/search?module=Digest::SHA256> for more information.
1414 my $context = Digest::SHA256::new(256);
1416 my $db = DBM::Deep->new(
1417 filename => "foo-sha.db",
1418 digest => \&my_digest,
1422 $db->{key1} = "value1";
1423 $db->{key2} = "value2";
1424 print "key1: " . $db->{key1} . "\n";
1425 print "key2: " . $db->{key2} . "\n";
1431 return substr( $context->hash($_[0]), 0, 32 );
1434 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1435 of bytes you specify in the hash_size parameter (in this case 32).
1437 B<Note:> If you do choose to use a custom digest algorithm, you must set it
1438 every time you access this file. Otherwise, the default (MD5) will be used.
1440 =head1 CIRCULAR REFERENCES
1442 DBM::Deep has B<experimental> support for circular references. Meaning you
1443 can have a nested hash key or array element that points to a parent object.
1444 This relationship is stored in the DB file, and is preserved between sessions.
1447 my $db = DBM::Deep->new( "foo.db" );
1450 $db->{circle} = $db; # ref to self
1452 print $db->{foo} . "\n"; # prints "bar"
1453 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1455 B<Note>: Passing the object to a function that recursively walks the
1456 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1457 C<export()> methods) will result in an infinite loop. This will be fixed in
1460 =head1 CAVEATS / ISSUES / BUGS
1462 This section describes all the known issues with DBM::Deep. It you have found
1463 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1465 =head2 UNUSED SPACE RECOVERY
1467 One major caveat with DBM::Deep is that space occupied by existing keys and
1468 values is not recovered when they are deleted. Meaning if you keep deleting
1469 and adding new keys, your file will continuously grow. I am working on this,
1470 but in the meantime you can call the built-in C<optimize()> method from time to
1471 time (perhaps in a crontab or something) to recover all your unused space.
1473 $db->optimize(); # returns true on success
1475 This rebuilds the ENTIRE database into a new file, then moves it on top of
1476 the original. The new file will have no unused space, thus it will take up as
1477 little disk space as possible. Please note that this operation can take
1478 a long time for large files, and you need enough disk space to temporarily hold
1479 2 copies of your DB file. The temporary file is created in the same directory
1480 as the original, named with a ".tmp" extension, and is deleted when the
1481 operation completes. Oh, and if locking is enabled, the DB is automatically
1482 locked for the entire duration of the copy.
1484 B<WARNING:> Only call optimize() on the top-level node of the database, and
1485 make sure there are no child references lying around. DBM::Deep keeps a reference
1486 counter, and if it is greater than 1, optimize() will abort and return undef.
1490 (The reasons given assume a high level of Perl understanding, specifically of
1491 references. You can safely skip this section.)
1493 Currently, the only references supported are HASH and ARRAY. The other reference
1494 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1500 These are things like filehandles and other sockets. They can't be supported
1501 because it's completely unclear how DBM::Deep should serialize them.
1503 =item * SCALAR / REF
1505 The discussion here refers to the following type of example:
1512 # In some other process ...
1514 my $val = ${ $db->{key1} };
1516 is( $val, 50, "What actually gets stored in the DB file?" );
1518 The problem is one of synchronization. When the variable being referred to
1519 changes value, the reference isn't notified. This means that the new value won't
1520 be stored in the datafile for other processes to read. There is no TIEREF.
1522 It is theoretically possible to store references to values already within a
1523 DBM::Deep object because everything already is synchronized, but the change to
1524 the internals would be quite large. Specifically, DBM::Deep would have to tie
1525 every single value that is stored. This would bloat the RAM footprint of
1526 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1527 all to support a feature that has never been requested.
1531 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1532 mechanism for serializing coderefs, including saving off all closure state.
1533 However, just as for SCALAR and REF, that closure state may change without
1534 notifying the DBM::Deep object storing the reference.
1538 =head2 FILE CORRUPTION
1540 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1541 for a 32-bit signature when opened, but other corruption in files can cause
1542 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1543 stuck in an infinite loop depending on the level of corruption. File write
1544 operations are not checked for failure (for speed), so if you happen to run
1545 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1546 be addressed in a later version of DBM::Deep.
1550 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1551 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1552 about setting up your NFS server with a locking daemon, then using lockf() to
1553 lock your files, but your mileage may vary there as well. From what I
1554 understand, there is no real way to do it. However, if you need access to the
1555 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1556 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1558 =head2 COPYING OBJECTS
1560 Beware of copying tied objects in Perl. Very strange things can happen.
1561 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1562 returns a new, blessed, tied hash or array to the same level in the DB.
1564 my $copy = $db->clone();
1566 B<Note>: Since clone() here is cloning the object, not the database location, any
1567 modifications to either $db or $copy will be visible in both.
1571 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1572 These functions cause every element in the array to move, which can be murder
1573 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1574 a different location. This will be addressed in the forthcoming version 1.00.
1576 =head2 WRITEONLY FILES
1578 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1579 writeonly mode. STORE will verify that the filehandle is writable. However, there
1580 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1581 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1585 This section discusses DBM::Deep's speed and memory usage.
1589 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1590 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1591 multi-level hash/array support, and cross-platform FTPable files. Even so,
1592 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1593 with huge databases. Here is some test data:
1595 Adding 1,000,000 keys to new DB file...
1597 At 100 keys, avg. speed is 2,703 keys/sec
1598 At 200 keys, avg. speed is 2,642 keys/sec
1599 At 300 keys, avg. speed is 2,598 keys/sec
1600 At 400 keys, avg. speed is 2,578 keys/sec
1601 At 500 keys, avg. speed is 2,722 keys/sec
1602 At 600 keys, avg. speed is 2,628 keys/sec
1603 At 700 keys, avg. speed is 2,700 keys/sec
1604 At 800 keys, avg. speed is 2,607 keys/sec
1605 At 900 keys, avg. speed is 2,190 keys/sec
1606 At 1,000 keys, avg. speed is 2,570 keys/sec
1607 At 2,000 keys, avg. speed is 2,417 keys/sec
1608 At 3,000 keys, avg. speed is 1,982 keys/sec
1609 At 4,000 keys, avg. speed is 1,568 keys/sec
1610 At 5,000 keys, avg. speed is 1,533 keys/sec
1611 At 6,000 keys, avg. speed is 1,787 keys/sec
1612 At 7,000 keys, avg. speed is 1,977 keys/sec
1613 At 8,000 keys, avg. speed is 2,028 keys/sec
1614 At 9,000 keys, avg. speed is 2,077 keys/sec
1615 At 10,000 keys, avg. speed is 2,031 keys/sec
1616 At 20,000 keys, avg. speed is 1,970 keys/sec
1617 At 30,000 keys, avg. speed is 2,050 keys/sec
1618 At 40,000 keys, avg. speed is 2,073 keys/sec
1619 At 50,000 keys, avg. speed is 1,973 keys/sec
1620 At 60,000 keys, avg. speed is 1,914 keys/sec
1621 At 70,000 keys, avg. speed is 2,091 keys/sec
1622 At 80,000 keys, avg. speed is 2,103 keys/sec
1623 At 90,000 keys, avg. speed is 1,886 keys/sec
1624 At 100,000 keys, avg. speed is 1,970 keys/sec
1625 At 200,000 keys, avg. speed is 2,053 keys/sec
1626 At 300,000 keys, avg. speed is 1,697 keys/sec
1627 At 400,000 keys, avg. speed is 1,838 keys/sec
1628 At 500,000 keys, avg. speed is 1,941 keys/sec
1629 At 600,000 keys, avg. speed is 1,930 keys/sec
1630 At 700,000 keys, avg. speed is 1,735 keys/sec
1631 At 800,000 keys, avg. speed is 1,795 keys/sec
1632 At 900,000 keys, avg. speed is 1,221 keys/sec
1633 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1635 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1636 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1637 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1638 Run time was 12 min 3 sec.
1642 One of the great things about DBM::Deep is that it uses very little memory.
1643 Even with huge databases (1,000,000+ keys) you will not see much increased
1644 memory on your process. DBM::Deep relies solely on the filesystem for storing
1645 and fetching data. Here is output from I</usr/bin/top> before even opening a
1648 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1649 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1651 Basically the process is taking 2,716K of memory. And here is the same
1652 process after storing and fetching 1,000,000 keys:
1654 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1655 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1657 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1658 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1660 =head1 DB FILE FORMAT
1662 In case you were interested in the underlying DB file format, it is documented
1663 here in this section. You don't need to know this to use the module, it's just
1664 included for reference.
1668 DBM::Deep files always start with a 32-bit signature to identify the file type.
1669 This is at offset 0. The signature is "DPDB" in network byte order. This is
1670 checked for when the file is opened and an error will be thrown if it's not found.
1674 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1675 has a standard header containing the type of data, the length of data, and then
1676 the data itself. The type is a single character (1 byte), the length is a
1677 32-bit unsigned long in network byte order, and the data is, well, the data.
1678 Here is how it unfolds:
1682 Immediately after the 32-bit file signature is the I<Master Index> record.
1683 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1684 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1685 depending on how the DBM::Deep object was constructed.
1687 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1688 number). The first 8-bit char of the MD5 signature is the offset into the
1689 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1690 index element is a file offset of the next tag for the key/element in question,
1691 which is usually a I<Bucket List> tag (see below).
1693 The next tag I<could> be another index, depending on how many keys/elements
1694 exist. See L<RE-INDEXING> below for details.
1698 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1699 file offsets to where the actual data is stored. It starts with a standard
1700 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1701 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1702 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1703 When the list fills up, a I<Re-Index> operation is performed (See
1704 L<RE-INDEXING> below).
1708 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1709 index/value pair (in array mode). It starts with a standard tag header with
1710 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1711 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1712 header. The size reported in the tag header is only for the value, but then,
1713 just after the value is another size (32-bit unsigned long) and then the plain
1714 key itself. Since the value is likely to be fetched more often than the plain
1715 key, I figured it would be I<slightly> faster to store the value first.
1717 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1718 record for the nested structure, where the process begins all over again.
1722 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1723 exhausted. Then, when another key/element comes in, the list is converted to a
1724 new index record. However, this index will look at the next char in the MD5
1725 hash, and arrange new Bucket List pointers accordingly. This process is called
1726 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1727 17 (16 + new one) keys/elements are removed from the old Bucket List and
1728 inserted into the new index. Several new Bucket Lists are created in the
1729 process, as a new MD5 char from the key is being examined (it is unlikely that
1730 the keys will all share the same next char of their MD5s).
1732 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1733 when the Bucket Lists will turn into indexes, but the first round tends to
1734 happen right around 4,000 keys. You will see a I<slight> decrease in
1735 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1736 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1737 right around 900,000 keys. This process can continue nearly indefinitely --
1738 right up until the point the I<MD5> signatures start colliding with each other,
1739 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1740 getting struck by lightning while you are walking to cash in your tickets.
1741 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1742 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1743 this is 340 unodecillion, but don't quote me).
1747 When a new key/element is stored, the key (or index number) is first run through
1748 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1749 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1750 for the first char of the signature (in this case I<b0>). If it does not exist,
1751 a new I<Bucket List> is created for our key (and the next 15 future keys that
1752 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1753 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1754 this point, unless we are replacing an existing I<Bucket>), where the actual
1755 data will be stored.
1759 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1760 (or index number), then walking along the indexes. If there are enough
1761 keys/elements in this DB level, there might be nested indexes, each linked to
1762 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1763 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1764 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1765 plain key are stored.
1767 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1768 methods. In this process the indexes are walked systematically, and each key
1769 fetched in increasing MD5 order (which is why it appears random). Once the
1770 I<Bucket> is found, the value is skipped and the plain key returned instead.
1771 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1772 alphabetically sorted. This only happens on an index-level -- as soon as the
1773 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1774 so it's pretty much undefined how the keys will come out -- just like Perl's
1777 =head1 CODE COVERAGE
1779 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1780 B<Devel::Cover> report on this module's test suite.
1782 ----------------------------------- ------ ------ ------ ------ ------ ------
1783 File stmt bran cond sub time total
1784 ----------------------------------- ------ ------ ------ ------ ------ ------
1785 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1786 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1787 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1788 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1789 Total 97.9 85.9 79.7 100.0 100.0 94.3
1790 ----------------------------------- ------ ------ ------ ------ ------ ------
1792 =head1 MORE INFORMATION
1794 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1795 or send email to L<DBM-Deep@googlegroups.com>.
1799 Joseph Huckaby, L<jhuckaby@cpan.org>
1801 Rob Kinyon, L<rkinyon@cpan.org>
1803 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1807 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1808 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1812 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1813 This is free software, you may use it and distribute it under the
1814 same terms as Perl itself.