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;
373 $self->_fileobj->begin_transaction;
378 my $self = shift->_get_self;
379 $self->_fileobj->end_transaction;
384 # my $self = shift->_get_self;
393 # Get access to the root structure
395 my $self = $_[0]->_get_self;
396 return $self->{fileobj};
401 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
403 my $self = $_[0]->_get_self;
404 return $self->{type};
409 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
411 my $self = $_[0]->_get_self;
412 return $self->{base_offset};
417 # Get access to the raw fh
419 my $self = $_[0]->_get_self;
420 return $self->_fileobj->{fh};
428 die "DBM::Deep: $_[1]\n";
433 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
438 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
443 # Store single hash key/value or array element in database.
445 my $self = shift->_get_self;
446 my ($key, $value) = @_;
448 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
449 $self->_throw_error( 'Cannot write to a readonly filehandle' );
453 # Request exclusive lock for writing
455 $self->lock( LOCK_EX );
457 my $md5 = $self->{engine}{digest}->($key);
459 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5, { create => 1 } );
461 # User may be storing a hash, in which case we do not want it run
462 # through the filtering system
463 if ( !ref($value) && $self->_fileobj->{filter_store_value} ) {
464 $value = $self->_fileobj->{filter_store_value}->( $value );
468 # Add key/value to bucket list
470 my $result = $self->{engine}->add_bucket( $tag, $md5, $key, $value );
479 # Fetch single value or element given plain key or array index
481 my $self = shift->_get_self;
484 my $md5 = $self->{engine}{digest}->($key);
487 # Request shared lock for reading
489 $self->lock( LOCK_SH );
491 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
498 # Get value from bucket list
500 my $result = $self->{engine}->get_bucket_value( $tag, $md5 );
504 # Filters only apply to scalar values, so the ref check is making
505 # sure the fetched bucket is a scalar, not a child hash or array.
506 return ($result && !ref($result) && $self->_fileobj->{filter_fetch_value})
507 ? $self->_fileobj->{filter_fetch_value}->($result)
513 # Delete single key/value pair or element given plain key or array index
515 my $self = $_[0]->_get_self;
518 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
519 $self->_throw_error( 'Cannot write to a readonly filehandle' );
523 # Request exclusive lock for writing
525 $self->lock( LOCK_EX );
527 my $md5 = $self->{engine}{digest}->($key);
529 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
538 my $value = $self->{engine}->get_bucket_value( $tag, $md5 );
540 if (defined $value && !ref($value) && $self->_fileobj->{filter_fetch_value}) {
541 $value = $self->_fileobj->{filter_fetch_value}->($value);
544 my $result = $self->{engine}->delete_bucket( $tag, $md5 );
547 # If this object is an array and the key deleted was on the end of the stack,
548 # decrement the length variable.
558 # Check if a single key or element exists given plain key or array index
560 my $self = $_[0]->_get_self;
563 my $md5 = $self->{engine}{digest}->($key);
566 # Request shared lock for reading
568 $self->lock( LOCK_SH );
570 my $tag = $self->{engine}->find_bucket_list( $self->_base_offset, $md5 );
575 # For some reason, the built-in exists() function returns '' for false
581 # Check if bucket exists and return 1 or ''
583 my $result = $self->{engine}->bucket_exists( $tag, $md5 ) || '';
592 # Clear all keys from hash, or all elements from array.
594 my $self = $_[0]->_get_self;
596 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
597 $self->_throw_error( 'Cannot write to a readonly filehandle' );
601 # Request exclusive lock for writing
603 $self->lock( LOCK_EX );
607 seek($fh, $self->_base_offset + $self->_fileobj->{file_offset}, SEEK_SET);
613 #XXX This needs updating to use _release_space
614 $self->{engine}->write_tag(
615 $self->_base_offset, $self->_type,
616 chr(0)x$self->{engine}{index_size},
625 # Public method aliases
627 sub put { (shift)->STORE( @_ ) }
628 sub store { (shift)->STORE( @_ ) }
629 sub get { (shift)->FETCH( @_ ) }
630 sub fetch { (shift)->FETCH( @_ ) }
631 sub delete { (shift)->DELETE( @_ ) }
632 sub exists { (shift)->EXISTS( @_ ) }
633 sub clear { (shift)->CLEAR( @_ ) }
640 DBM::Deep - A pure perl multi-level hash/array DBM
645 my $db = DBM::Deep->new( "foo.db" );
647 $db->{key} = 'value'; # tie() style
650 $db->put('key' => 'value'); # OO style
651 print $db->get('key');
653 # true multi-level support
654 $db->{my_complex} = [
655 'hello', { perl => 'rules' },
661 A unique flat-file database module, written in pure perl. True
662 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
663 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
664 handle millions of keys and unlimited hash levels without significant
665 slow-down. Written from the ground-up in pure perl -- this is NOT a
666 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
667 Mac OS X and Windows.
669 =head1 VERSION DIFFERENCES
671 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
672 before. While attempts have been made to be backwards compatible, no guarantees.
676 Hopefully you are using Perl's excellent CPAN module, which will download
677 and install the module for you. If not, get the tarball, and run these
689 Construction can be done OO-style (which is the recommended way), or using
690 Perl's tie() function. Both are examined here.
692 =head2 OO CONSTRUCTION
694 The recommended way to construct a DBM::Deep object is to use the new()
695 method, which gets you a blessed, tied hash or array reference.
697 my $db = DBM::Deep->new( "foo.db" );
699 This opens a new database handle, mapped to the file "foo.db". If this
700 file does not exist, it will automatically be created. DB files are
701 opened in "r+" (read/write) mode, and the type of object returned is a
702 hash, unless otherwise specified (see L<OPTIONS> below).
704 You can pass a number of options to the constructor to specify things like
705 locking, autoflush, etc. This is done by passing an inline hash:
707 my $db = DBM::Deep->new(
713 Notice that the filename is now specified I<inside> the hash with
714 the "file" parameter, as opposed to being the sole argument to the
715 constructor. This is required if any options are specified.
716 See L<OPTIONS> below for the complete list.
720 You can also start with an array instead of a hash. For this, you must
721 specify the C<type> parameter:
723 my $db = DBM::Deep->new(
725 type => DBM::Deep->TYPE_ARRAY
728 B<Note:> Specifing the C<type> parameter only takes effect when beginning
729 a new DB file. If you create a DBM::Deep object with an existing file, the
730 C<type> will be loaded from the file header, and an error will be thrown if
731 the wrong type is passed in.
733 =head2 TIE CONSTRUCTION
735 Alternately, you can create a DBM::Deep handle by using Perl's built-in
736 tie() function. The object returned from tie() can be used to call methods,
737 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
738 file (as expected with most tie'd objects).
741 my $db = tie %hash, "DBM::Deep", "foo.db";
744 my $db = tie @array, "DBM::Deep", "bar.db";
746 As with the OO constructor, you can replace the DB filename parameter with
747 a hash containing one or more options (see L<OPTIONS> just below for the
750 tie %hash, "DBM::Deep", {
758 There are a number of options that can be passed in when constructing your
759 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
765 Filename of the DB file to link the handle to. You can pass a full absolute
766 filesystem path, partial path, or a plain filename if the file is in the
767 current working directory. This is a required parameter (though q.v. fh).
771 If you want, you can pass in the fh instead of the file. This is most useful for doing
774 my $db = DBM::Deep->new( { fh => \*DATA } );
776 You are responsible for making sure that the fh has been opened appropriately for your
777 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
778 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
779 needs to read from the fh.
783 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
784 not need to set this. However, it's there if you want it.
786 If you pass in fh and do not set this, it will be set appropriately.
790 This parameter specifies what type of object to create, a hash or array. Use
791 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
792 This only takes effect when beginning a new file. This is an optional
793 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
797 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
798 function to lock the database in exclusive mode for writes, and shared mode for
799 reads. Pass any true value to enable. This affects the base DB handle I<and
800 any child hashes or arrays> that use the same DB file. This is an optional
801 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
805 Specifies whether autoflush is to be enabled on the underlying filehandle.
806 This obviously slows down write operations, but is required if you may have
807 multiple processes accessing the same DB file (also consider enable I<locking>).
808 Pass any true value to enable. This is an optional parameter, and defaults to 0
813 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
814 restore them when fetched. This is an B<experimental> feature, and does have
815 side-effects. Basically, when hashes are re-blessed into their original
816 classes, they are no longer blessed into the DBM::Deep class! So you won't be
817 able to call any DBM::Deep methods on them. You have been warned.
818 This is an optional parameter, and defaults to 0 (disabled).
822 See L<FILTERS> below.
828 With DBM::Deep you can access your databases using Perl's standard hash/array
829 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
830 treat them as such. DBM::Deep will intercept all reads/writes and direct them
831 to the right place -- the DB file. This has nothing to do with the
832 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
833 using regular hashes and arrays, rather than calling functions like C<get()>
834 and C<put()> (although those work too). It is entirely up to you how to want
835 to access your databases.
839 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
840 or even nested hashes (or arrays) using standard Perl syntax:
842 my $db = DBM::Deep->new( "foo.db" );
844 $db->{mykey} = "myvalue";
846 $db->{myhash}->{subkey} = "subvalue";
848 print $db->{myhash}->{subkey} . "\n";
850 You can even step through hash keys using the normal Perl C<keys()> function:
852 foreach my $key (keys %$db) {
853 print "$key: " . $db->{$key} . "\n";
856 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
857 pushes them onto an array, all before the loop even begins. If you have an
858 extra large hash, this may exhaust Perl's memory. Instead, consider using
859 Perl's C<each()> function, which pulls keys/values one at a time, using very
862 while (my ($key, $value) = each %$db) {
863 print "$key: $value\n";
866 Please note that when using C<each()>, you should always pass a direct
867 hash reference, not a lookup. Meaning, you should B<never> do this:
870 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
872 This causes an infinite loop, because for each iteration, Perl is calling
873 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
874 it effectively keeps returning the first key over and over again. Instead,
875 assign a temporary variable to C<$db->{foo}>, then pass that to each().
879 As with hashes, you can treat any DBM::Deep object like a normal Perl array
880 reference. This includes inserting, removing and manipulating elements,
881 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
882 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
883 or simply be a nested array reference inside a hash. Example:
885 my $db = DBM::Deep->new(
886 file => "foo-array.db",
887 type => DBM::Deep->TYPE_ARRAY
891 push @$db, "bar", "baz";
894 my $last_elem = pop @$db; # baz
895 my $first_elem = shift @$db; # bah
896 my $second_elem = $db->[1]; # bar
898 my $num_elements = scalar @$db;
902 In addition to the I<tie()> interface, you can also use a standard OO interface
903 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
904 array) has its own methods, but both types share the following common methods:
905 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
909 =item * new() / clone()
911 These are the constructor and copy-functions.
913 =item * put() / store()
915 Stores a new hash key/value pair, or sets an array element value. Takes two
916 arguments, the hash key or array index, and the new value. The value can be
917 a scalar, hash ref or array ref. Returns true on success, false on failure.
919 $db->put("foo", "bar"); # for hashes
920 $db->put(1, "bar"); # for arrays
922 =item * get() / fetch()
924 Fetches the value of a hash key or array element. Takes one argument: the hash
925 key or array index. Returns a scalar, hash ref or array ref, depending on the
928 my $value = $db->get("foo"); # for hashes
929 my $value = $db->get(1); # for arrays
933 Checks if a hash key or array index exists. Takes one argument: the hash key
934 or array index. Returns true if it exists, false if not.
936 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
937 if ($db->exists(1)) { print "yay!\n"; } # for arrays
941 Deletes one hash key/value pair or array element. Takes one argument: the hash
942 key or array index. Returns true on success, false if not found. For arrays,
943 the remaining elements located after the deleted element are NOT moved over.
944 The deleted element is essentially just undefined, which is exactly how Perl's
945 internal arrays work. Please note that the space occupied by the deleted
946 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
947 below for details and workarounds.
949 $db->delete("foo"); # for hashes
950 $db->delete(1); # for arrays
954 Deletes B<all> hash keys or array elements. Takes no arguments. No return
955 value. Please note that the space occupied by the deleted keys/values or
956 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
957 details and workarounds.
959 $db->clear(); # hashes or arrays
961 =item * lock() / unlock()
967 Recover lost disk space.
969 =item * import() / export()
971 Data going in and out.
977 For hashes, DBM::Deep supports all the common methods described above, and the
978 following additional methods: C<first_key()> and C<next_key()>.
984 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
985 fetched in an undefined order (which appears random). Takes no arguments,
986 returns the key as a scalar value.
988 my $key = $db->first_key();
992 Returns the "next" key in the hash, given the previous one as the sole argument.
993 Returns undef if there are no more keys to be fetched.
995 $key = $db->next_key($key);
999 Here are some examples of using hashes:
1001 my $db = DBM::Deep->new( "foo.db" );
1003 $db->put("foo", "bar");
1004 print "foo: " . $db->get("foo") . "\n";
1006 $db->put("baz", {}); # new child hash ref
1007 $db->get("baz")->put("buz", "biz");
1008 print "buz: " . $db->get("baz")->get("buz") . "\n";
1010 my $key = $db->first_key();
1012 print "$key: " . $db->get($key) . "\n";
1013 $key = $db->next_key($key);
1016 if ($db->exists("foo")) { $db->delete("foo"); }
1020 For arrays, DBM::Deep supports all the common methods described above, and the
1021 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1022 C<unshift()> and C<splice()>.
1028 Returns the number of elements in the array. Takes no arguments.
1030 my $len = $db->length();
1034 Adds one or more elements onto the end of the array. Accepts scalars, hash
1035 refs or array refs. No return value.
1037 $db->push("foo", "bar", {});
1041 Fetches the last element in the array, and deletes it. Takes no arguments.
1042 Returns undef if array is empty. Returns the element value.
1044 my $elem = $db->pop();
1048 Fetches the first element in the array, deletes it, then shifts all the
1049 remaining elements over to take up the space. Returns the element value. This
1050 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1053 my $elem = $db->shift();
1057 Inserts one or more elements onto the beginning of the array, shifting all
1058 existing elements over to make room. Accepts scalars, hash refs or array refs.
1059 No return value. This method is not recommended with large arrays -- see
1060 <LARGE ARRAYS> below for details.
1062 $db->unshift("foo", "bar", {});
1066 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1067 -f splice> for usage -- it is too complicated to document here. This method is
1068 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1072 Here are some examples of using arrays:
1074 my $db = DBM::Deep->new(
1076 type => DBM::Deep->TYPE_ARRAY
1079 $db->push("bar", "baz");
1080 $db->unshift("foo");
1083 my $len = $db->length();
1084 print "length: $len\n"; # 4
1086 for (my $k=0; $k<$len; $k++) {
1087 print "$k: " . $db->get($k) . "\n";
1090 $db->splice(1, 2, "biz", "baf");
1092 while (my $elem = shift @$db) {
1093 print "shifted: $elem\n";
1098 Enable automatic file locking by passing a true value to the C<locking>
1099 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1101 my $db = DBM::Deep->new(
1106 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1107 mode for writes, and shared mode for reads. This is required if you have
1108 multiple processes accessing the same database file, to avoid file corruption.
1109 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1110 NFS> below for more.
1112 =head2 EXPLICIT LOCKING
1114 You can explicitly lock a database, so it remains locked for multiple
1115 transactions. This is done by calling the C<lock()> method, and passing an
1116 optional lock mode argument (defaults to exclusive mode). This is particularly
1117 useful for things like counters, where the current value needs to be fetched,
1118 then incremented, then stored again.
1121 my $counter = $db->get("counter");
1123 $db->put("counter", $counter);
1132 You can pass C<lock()> an optional argument, which specifies which mode to use
1133 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1134 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1135 same as the constants defined in Perl's C<Fcntl> module.
1137 $db->lock( DBM::Deep->LOCK_SH );
1141 =head1 IMPORTING/EXPORTING
1143 You can import existing complex structures by calling the C<import()> method,
1144 and export an entire database into an in-memory structure using the C<export()>
1145 method. Both are examined here.
1149 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1150 walking the structure and adding keys/elements to the database as you go,
1151 simply pass a reference to the C<import()> method. This recursively adds
1152 everything to an existing DBM::Deep object for you. Here is an example:
1157 array1 => [ "elem0", "elem1", "elem2" ],
1159 subkey1 => "subvalue1",
1160 subkey2 => "subvalue2"
1164 my $db = DBM::Deep->new( "foo.db" );
1165 $db->import( $struct );
1167 print $db->{key1} . "\n"; # prints "value1"
1169 This recursively imports the entire C<$struct> object into C<$db>, including
1170 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1171 keys are merged with the existing ones, replacing if they already exist.
1172 The C<import()> method can be called on any database level (not just the base
1173 level), and works with both hash and array DB types.
1175 B<Note:> Make sure your existing structure has no circular references in it.
1176 These will cause an infinite loop when importing.
1180 Calling the C<export()> method on an existing DBM::Deep object will return
1181 a reference to a new in-memory copy of the database. The export is done
1182 recursively, so all nested hashes/arrays are all exported to standard Perl
1183 objects. Here is an example:
1185 my $db = DBM::Deep->new( "foo.db" );
1187 $db->{key1} = "value1";
1188 $db->{key2} = "value2";
1190 $db->{hash1}->{subkey1} = "subvalue1";
1191 $db->{hash1}->{subkey2} = "subvalue2";
1193 my $struct = $db->export();
1195 print $struct->{key1} . "\n"; # prints "value1"
1197 This makes a complete copy of the database in memory, and returns a reference
1198 to it. The C<export()> method can be called on any database level (not just
1199 the base level), and works with both hash and array DB types. Be careful of
1200 large databases -- you can store a lot more data in a DBM::Deep object than an
1201 in-memory Perl structure.
1203 B<Note:> Make sure your database has no circular references in it.
1204 These will cause an infinite loop when exporting.
1208 DBM::Deep has a number of hooks where you can specify your own Perl function
1209 to perform filtering on incoming or outgoing data. This is a perfect
1210 way to extend the engine, and implement things like real-time compression or
1211 encryption. Filtering applies to the base DB level, and all child hashes /
1212 arrays. Filter hooks can be specified when your DBM::Deep object is first
1213 constructed, or by calling the C<set_filter()> method at any time. There are
1214 four available filter hooks, described below:
1218 =item * filter_store_key
1220 This filter is called whenever a hash key is stored. It
1221 is passed the incoming key, and expected to return a transformed key.
1223 =item * filter_store_value
1225 This filter is called whenever a hash key or array element is stored. It
1226 is passed the incoming value, and expected to return a transformed value.
1228 =item * filter_fetch_key
1230 This filter is called whenever a hash key is fetched (i.e. via
1231 C<first_key()> or C<next_key()>). It is passed the transformed key,
1232 and expected to return the plain key.
1234 =item * filter_fetch_value
1236 This filter is called whenever a hash key or array element is fetched.
1237 It is passed the transformed value, and expected to return the plain value.
1241 Here are the two ways to setup a filter hook:
1243 my $db = DBM::Deep->new(
1245 filter_store_value => \&my_filter_store,
1246 filter_fetch_value => \&my_filter_fetch
1251 $db->set_filter( "filter_store_value", \&my_filter_store );
1252 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1254 Your filter function will be called only when dealing with SCALAR keys or
1255 values. When nested hashes and arrays are being stored/fetched, filtering
1256 is bypassed. Filters are called as static functions, passed a single SCALAR
1257 argument, and expected to return a single SCALAR value. If you want to
1258 remove a filter, set the function reference to C<undef>:
1260 $db->set_filter( "filter_store_value", undef );
1262 =head2 REAL-TIME ENCRYPTION EXAMPLE
1264 Here is a working example that uses the I<Crypt::Blowfish> module to
1265 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1266 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1267 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1270 use Crypt::Blowfish;
1273 my $cipher = Crypt::CBC->new({
1274 'key' => 'my secret key',
1275 'cipher' => 'Blowfish',
1277 'regenerate_key' => 0,
1278 'padding' => 'space',
1282 my $db = DBM::Deep->new(
1283 file => "foo-encrypt.db",
1284 filter_store_key => \&my_encrypt,
1285 filter_store_value => \&my_encrypt,
1286 filter_fetch_key => \&my_decrypt,
1287 filter_fetch_value => \&my_decrypt,
1290 $db->{key1} = "value1";
1291 $db->{key2} = "value2";
1292 print "key1: " . $db->{key1} . "\n";
1293 print "key2: " . $db->{key2} . "\n";
1299 return $cipher->encrypt( $_[0] );
1302 return $cipher->decrypt( $_[0] );
1305 =head2 REAL-TIME COMPRESSION EXAMPLE
1307 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1308 compression / decompression of keys & values with DBM::Deep Filters.
1309 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1310 more on I<Compress::Zlib>.
1315 my $db = DBM::Deep->new(
1316 file => "foo-compress.db",
1317 filter_store_key => \&my_compress,
1318 filter_store_value => \&my_compress,
1319 filter_fetch_key => \&my_decompress,
1320 filter_fetch_value => \&my_decompress,
1323 $db->{key1} = "value1";
1324 $db->{key2} = "value2";
1325 print "key1: " . $db->{key1} . "\n";
1326 print "key2: " . $db->{key2} . "\n";
1332 return Compress::Zlib::memGzip( $_[0] ) ;
1335 return Compress::Zlib::memGunzip( $_[0] ) ;
1338 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1339 actually numerical index numbers, and are not filtered.
1341 =head1 ERROR HANDLING
1343 Most DBM::Deep methods return a true value for success, and call die() on
1344 failure. You can wrap calls in an eval block to catch the die.
1346 my $db = DBM::Deep->new( "foo.db" ); # create hash
1347 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1349 print $@; # prints error message
1351 =head1 LARGEFILE SUPPORT
1353 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1354 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1355 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1356 by specifying the 'pack_size' parameter when constructing the file.
1359 filename => $filename,
1360 pack_size => 'large',
1363 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1364 instead of 32-bit longs. After setting these values your DB files have a
1365 theoretical maximum size of 16 XB (exabytes).
1367 You can also use C<pack_size =E<gt> 'small'> in order to use 16-bit file
1370 B<Note:> Changing these values will B<NOT> work for existing database files.
1371 Only change this for new files. Once the value has been set, it is stored in
1372 the file's header and cannot be changed for the life of the file. These
1373 parameters are per-file, meaning you can access 32-bit and 64-bit files, as
1376 B<Note:> We have not personally tested files larger than 2 GB -- all my
1377 systems have only a 32-bit Perl. However, I have received user reports that
1378 this does indeed work!
1380 =head1 LOW-LEVEL ACCESS
1382 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1383 you can call the C<_fh()> method, which returns the handle:
1385 my $fh = $db->_fh();
1387 This method can be called on the root level of the datbase, or any child
1388 hashes or arrays. All levels share a I<root> structure, which contains things
1389 like the filehandle, a reference counter, and all the options specified
1390 when you created the object. You can get access to this file object by
1391 calling the C<_fileobj()> method.
1393 my $file_obj = $db->_fileobj();
1395 This is useful for changing options after the object has already been created,
1396 such as enabling/disabling locking. You can also store your own temporary user
1397 data in this structure (be wary of name collision), which is then accessible from
1398 any child hash or array.
1400 =head1 CUSTOM DIGEST ALGORITHM
1402 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1403 keys. However you can override this, and use another algorithm (such as SHA-256)
1404 or even write your own. But please note that DBM::Deep currently expects zero
1405 collisions, so your algorithm has to be I<perfect>, so to speak. Collision
1406 detection may be introduced in a later version.
1408 You can specify a custom digest algorithm by passing it into the parameter
1409 list for new(), passing a reference to a subroutine as the 'digest' parameter,
1410 and the length of the algorithm's hashes (in bytes) as the 'hash_size'
1411 parameter. Here is a working example that uses a 256-bit hash from the
1412 I<Digest::SHA256> module. Please see
1413 L<http://search.cpan.org/search?module=Digest::SHA256> for more information.
1418 my $context = Digest::SHA256::new(256);
1420 my $db = DBM::Deep->new(
1421 filename => "foo-sha.db",
1422 digest => \&my_digest,
1426 $db->{key1} = "value1";
1427 $db->{key2} = "value2";
1428 print "key1: " . $db->{key1} . "\n";
1429 print "key2: " . $db->{key2} . "\n";
1435 return substr( $context->hash($_[0]), 0, 32 );
1438 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1439 of bytes you specify in the hash_size parameter (in this case 32).
1441 B<Note:> If you do choose to use a custom digest algorithm, you must set it
1442 every time you access this file. Otherwise, the default (MD5) will be used.
1444 =head1 CIRCULAR REFERENCES
1446 DBM::Deep has B<experimental> support for circular references. Meaning you
1447 can have a nested hash key or array element that points to a parent object.
1448 This relationship is stored in the DB file, and is preserved between sessions.
1451 my $db = DBM::Deep->new( "foo.db" );
1454 $db->{circle} = $db; # ref to self
1456 print $db->{foo} . "\n"; # prints "bar"
1457 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1459 B<Note>: Passing the object to a function that recursively walks the
1460 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1461 C<export()> methods) will result in an infinite loop. This will be fixed in
1464 =head1 CAVEATS / ISSUES / BUGS
1466 This section describes all the known issues with DBM::Deep. It you have found
1467 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1469 =head2 UNUSED SPACE RECOVERY
1471 One major caveat with DBM::Deep is that space occupied by existing keys and
1472 values is not recovered when they are deleted. Meaning if you keep deleting
1473 and adding new keys, your file will continuously grow. I am working on this,
1474 but in the meantime you can call the built-in C<optimize()> method from time to
1475 time (perhaps in a crontab or something) to recover all your unused space.
1477 $db->optimize(); # returns true on success
1479 This rebuilds the ENTIRE database into a new file, then moves it on top of
1480 the original. The new file will have no unused space, thus it will take up as
1481 little disk space as possible. Please note that this operation can take
1482 a long time for large files, and you need enough disk space to temporarily hold
1483 2 copies of your DB file. The temporary file is created in the same directory
1484 as the original, named with a ".tmp" extension, and is deleted when the
1485 operation completes. Oh, and if locking is enabled, the DB is automatically
1486 locked for the entire duration of the copy.
1488 B<WARNING:> Only call optimize() on the top-level node of the database, and
1489 make sure there are no child references lying around. DBM::Deep keeps a reference
1490 counter, and if it is greater than 1, optimize() will abort and return undef.
1494 (The reasons given assume a high level of Perl understanding, specifically of
1495 references. You can safely skip this section.)
1497 Currently, the only references supported are HASH and ARRAY. The other reference
1498 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1504 These are things like filehandles and other sockets. They can't be supported
1505 because it's completely unclear how DBM::Deep should serialize them.
1507 =item * SCALAR / REF
1509 The discussion here refers to the following type of example:
1516 # In some other process ...
1518 my $val = ${ $db->{key1} };
1520 is( $val, 50, "What actually gets stored in the DB file?" );
1522 The problem is one of synchronization. When the variable being referred to
1523 changes value, the reference isn't notified. This means that the new value won't
1524 be stored in the datafile for other processes to read. There is no TIEREF.
1526 It is theoretically possible to store references to values already within a
1527 DBM::Deep object because everything already is synchronized, but the change to
1528 the internals would be quite large. Specifically, DBM::Deep would have to tie
1529 every single value that is stored. This would bloat the RAM footprint of
1530 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1531 all to support a feature that has never been requested.
1535 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1536 mechanism for serializing coderefs, including saving off all closure state.
1537 However, just as for SCALAR and REF, that closure state may change without
1538 notifying the DBM::Deep object storing the reference.
1542 =head2 FILE CORRUPTION
1544 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1545 for a 32-bit signature when opened, but other corruption in files can cause
1546 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1547 stuck in an infinite loop depending on the level of corruption. File write
1548 operations are not checked for failure (for speed), so if you happen to run
1549 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1550 be addressed in a later version of DBM::Deep.
1554 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1555 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1556 about setting up your NFS server with a locking daemon, then using lockf() to
1557 lock your files, but your mileage may vary there as well. From what I
1558 understand, there is no real way to do it. However, if you need access to the
1559 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1560 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1562 =head2 COPYING OBJECTS
1564 Beware of copying tied objects in Perl. Very strange things can happen.
1565 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1566 returns a new, blessed, tied hash or array to the same level in the DB.
1568 my $copy = $db->clone();
1570 B<Note>: Since clone() here is cloning the object, not the database location, any
1571 modifications to either $db or $copy will be visible in both.
1575 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1576 These functions cause every element in the array to move, which can be murder
1577 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1578 a different location. This will be addressed in the forthcoming version 1.00.
1580 =head2 WRITEONLY FILES
1582 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1583 writeonly mode. STORE will verify that the filehandle is writable. However, there
1584 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1585 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1589 This section discusses DBM::Deep's speed and memory usage.
1593 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1594 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1595 multi-level hash/array support, and cross-platform FTPable files. Even so,
1596 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1597 with huge databases. Here is some test data:
1599 Adding 1,000,000 keys to new DB file...
1601 At 100 keys, avg. speed is 2,703 keys/sec
1602 At 200 keys, avg. speed is 2,642 keys/sec
1603 At 300 keys, avg. speed is 2,598 keys/sec
1604 At 400 keys, avg. speed is 2,578 keys/sec
1605 At 500 keys, avg. speed is 2,722 keys/sec
1606 At 600 keys, avg. speed is 2,628 keys/sec
1607 At 700 keys, avg. speed is 2,700 keys/sec
1608 At 800 keys, avg. speed is 2,607 keys/sec
1609 At 900 keys, avg. speed is 2,190 keys/sec
1610 At 1,000 keys, avg. speed is 2,570 keys/sec
1611 At 2,000 keys, avg. speed is 2,417 keys/sec
1612 At 3,000 keys, avg. speed is 1,982 keys/sec
1613 At 4,000 keys, avg. speed is 1,568 keys/sec
1614 At 5,000 keys, avg. speed is 1,533 keys/sec
1615 At 6,000 keys, avg. speed is 1,787 keys/sec
1616 At 7,000 keys, avg. speed is 1,977 keys/sec
1617 At 8,000 keys, avg. speed is 2,028 keys/sec
1618 At 9,000 keys, avg. speed is 2,077 keys/sec
1619 At 10,000 keys, avg. speed is 2,031 keys/sec
1620 At 20,000 keys, avg. speed is 1,970 keys/sec
1621 At 30,000 keys, avg. speed is 2,050 keys/sec
1622 At 40,000 keys, avg. speed is 2,073 keys/sec
1623 At 50,000 keys, avg. speed is 1,973 keys/sec
1624 At 60,000 keys, avg. speed is 1,914 keys/sec
1625 At 70,000 keys, avg. speed is 2,091 keys/sec
1626 At 80,000 keys, avg. speed is 2,103 keys/sec
1627 At 90,000 keys, avg. speed is 1,886 keys/sec
1628 At 100,000 keys, avg. speed is 1,970 keys/sec
1629 At 200,000 keys, avg. speed is 2,053 keys/sec
1630 At 300,000 keys, avg. speed is 1,697 keys/sec
1631 At 400,000 keys, avg. speed is 1,838 keys/sec
1632 At 500,000 keys, avg. speed is 1,941 keys/sec
1633 At 600,000 keys, avg. speed is 1,930 keys/sec
1634 At 700,000 keys, avg. speed is 1,735 keys/sec
1635 At 800,000 keys, avg. speed is 1,795 keys/sec
1636 At 900,000 keys, avg. speed is 1,221 keys/sec
1637 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1639 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1640 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1641 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1642 Run time was 12 min 3 sec.
1646 One of the great things about DBM::Deep is that it uses very little memory.
1647 Even with huge databases (1,000,000+ keys) you will not see much increased
1648 memory on your process. DBM::Deep relies solely on the filesystem for storing
1649 and fetching data. Here is output from I</usr/bin/top> before even opening a
1652 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1653 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1655 Basically the process is taking 2,716K of memory. And here is the same
1656 process after storing and fetching 1,000,000 keys:
1658 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1659 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1661 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1662 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1664 =head1 DB FILE FORMAT
1666 In case you were interested in the underlying DB file format, it is documented
1667 here in this section. You don't need to know this to use the module, it's just
1668 included for reference.
1672 DBM::Deep files always start with a 32-bit signature to identify the file type.
1673 This is at offset 0. The signature is "DPDB" in network byte order. This is
1674 checked for when the file is opened and an error will be thrown if it's not found.
1678 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1679 has a standard header containing the type of data, the length of data, and then
1680 the data itself. The type is a single character (1 byte), the length is a
1681 32-bit unsigned long in network byte order, and the data is, well, the data.
1682 Here is how it unfolds:
1686 Immediately after the 32-bit file signature is the I<Master Index> record.
1687 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1688 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1689 depending on how the DBM::Deep object was constructed.
1691 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1692 number). The first 8-bit char of the MD5 signature is the offset into the
1693 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1694 index element is a file offset of the next tag for the key/element in question,
1695 which is usually a I<Bucket List> tag (see below).
1697 The next tag I<could> be another index, depending on how many keys/elements
1698 exist. See L<RE-INDEXING> below for details.
1702 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1703 file offsets to where the actual data is stored. It starts with a standard
1704 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1705 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1706 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1707 When the list fills up, a I<Re-Index> operation is performed (See
1708 L<RE-INDEXING> below).
1712 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1713 index/value pair (in array mode). It starts with a standard tag header with
1714 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1715 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1716 header. The size reported in the tag header is only for the value, but then,
1717 just after the value is another size (32-bit unsigned long) and then the plain
1718 key itself. Since the value is likely to be fetched more often than the plain
1719 key, I figured it would be I<slightly> faster to store the value first.
1721 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1722 record for the nested structure, where the process begins all over again.
1726 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1727 exhausted. Then, when another key/element comes in, the list is converted to a
1728 new index record. However, this index will look at the next char in the MD5
1729 hash, and arrange new Bucket List pointers accordingly. This process is called
1730 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1731 17 (16 + new one) keys/elements are removed from the old Bucket List and
1732 inserted into the new index. Several new Bucket Lists are created in the
1733 process, as a new MD5 char from the key is being examined (it is unlikely that
1734 the keys will all share the same next char of their MD5s).
1736 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1737 when the Bucket Lists will turn into indexes, but the first round tends to
1738 happen right around 4,000 keys. You will see a I<slight> decrease in
1739 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1740 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1741 right around 900,000 keys. This process can continue nearly indefinitely --
1742 right up until the point the I<MD5> signatures start colliding with each other,
1743 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1744 getting struck by lightning while you are walking to cash in your tickets.
1745 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1746 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1747 this is 340 unodecillion, but don't quote me).
1751 When a new key/element is stored, the key (or index number) is first run through
1752 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1753 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1754 for the first char of the signature (in this case I<b0>). If it does not exist,
1755 a new I<Bucket List> is created for our key (and the next 15 future keys that
1756 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1757 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1758 this point, unless we are replacing an existing I<Bucket>), where the actual
1759 data will be stored.
1763 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1764 (or index number), then walking along the indexes. If there are enough
1765 keys/elements in this DB level, there might be nested indexes, each linked to
1766 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1767 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1768 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1769 plain key are stored.
1771 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1772 methods. In this process the indexes are walked systematically, and each key
1773 fetched in increasing MD5 order (which is why it appears random). Once the
1774 I<Bucket> is found, the value is skipped and the plain key returned instead.
1775 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1776 alphabetically sorted. This only happens on an index-level -- as soon as the
1777 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1778 so it's pretty much undefined how the keys will come out -- just like Perl's
1781 =head1 CODE COVERAGE
1783 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1784 B<Devel::Cover> report on this module's test suite.
1786 ----------------------------------- ------ ------ ------ ------ ------ ------
1787 File stmt bran cond sub time total
1788 ----------------------------------- ------ ------ ------ ------ ------ ------
1789 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1790 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1791 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1792 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1793 Total 97.9 85.9 79.7 100.0 100.0 94.3
1794 ----------------------------------- ------ ------ ------ ------ ------ ------
1796 =head1 MORE INFORMATION
1798 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1799 or send email to L<DBM-Deep@googlegroups.com>.
1803 Joseph Huckaby, L<jhuckaby@cpan.org>
1805 Rob Kinyon, L<rkinyon@cpan.org>
1807 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1811 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1812 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1816 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1817 This is free software, you may use it and distribute it under the
1818 same terms as Perl itself.