7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use DBM::Deep::Engine;
40 use vars qw( $VERSION );
41 $VERSION = q(0.99_01);
44 # Setup constants for users to pass to new()
46 sub TYPE_HASH () { DBM::Deep::Engine->SIG_HASH }
47 sub TYPE_ARRAY () { DBM::Deep::Engine->SIG_ARRAY }
55 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
60 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
61 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
66 $args = { file => shift };
74 # Class constructor method for Perl OO interface.
75 # Calls tie() and returns blessed reference to tied hash or array,
76 # providing a hybrid OO/tie interface.
79 my $args = $class->_get_args( @_ );
82 # Check if we want a tied hash or array.
85 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
86 $class = 'DBM::Deep::Array';
87 require DBM::Deep::Array;
88 tie @$self, $class, %$args;
91 $class = 'DBM::Deep::Hash';
92 require DBM::Deep::Hash;
93 tie %$self, $class, %$args;
96 return bless $self, $class;
99 # This initializer is called from the various TIE* methods. new() calls tie(),
100 # which allows for a single point of entry.
105 # These are the defaults to be optionally overridden below
108 engine => DBM::Deep::Engine->new( $args ),
109 base_offset => undef,
112 # Grab the parameters we want to use
113 foreach my $param ( keys %$self ) {
114 next unless exists $args->{$param};
115 $self->{$param} = $args->{$param}
118 # locking implicitly enables autoflush
119 if ($args->{locking}) { $args->{autoflush} = 1; }
121 $self->{root} = exists $args->{root}
123 : DBM::Deep::_::Root->new( $args );
125 #XXX Right before this line, we have to set the physical parameters like
126 #XXX 2S vs. 4N vs. 8Q or max_buckets, etc.
127 $self->{engine}->setup_fh( $self );
134 require DBM::Deep::Hash;
135 return DBM::Deep::Hash->TIEHASH( @_ );
140 require DBM::Deep::Array;
141 return DBM::Deep::Array->TIEARRAY( @_ );
144 #XXX Unneeded now ...
150 # If db locking is set, flock() the db file. If called multiple
151 # times before unlock(), then the same number of unlocks() must
152 # be called before the lock is released.
154 my $self = shift->_get_self;
156 $type = LOCK_EX unless defined $type;
158 if (!defined($self->_fh)) { return; }
160 if ($self->_root->{locking}) {
161 if (!$self->_root->{locked}) {
162 flock($self->_fh, $type);
164 # refresh end counter in case file has changed size
165 my @stats = stat($self->_fh);
166 $self->_root->{end} = $stats[7];
168 # double-check file inode, in case another process
169 # has optimize()d our file while we were waiting.
170 if ($stats[1] != $self->_root->{inode}) {
171 $self->{engine}->close_fh( $self );
172 $self->{engine}->setup_fh( $self );
173 flock($self->_fh, $type); # re-lock
175 # This may not be necessary after re-opening
176 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
179 $self->_root->{locked}++;
189 # If db locking is set, unlock the db file. See note in lock()
190 # regarding calling lock() multiple times.
192 my $self = shift->_get_self;
194 if (!defined($self->_fh)) { return; }
196 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
197 $self->_root->{locked}--;
198 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
207 my $self = shift->_get_self;
208 my ($spot, $value) = @_;
213 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
214 ${$spot} = $value->_repr;
215 $value->_copy_node( ${$spot} );
218 my $r = Scalar::Util::reftype( $value );
219 my $c = Scalar::Util::blessed( $value );
220 if ( $r eq 'ARRAY' ) {
221 ${$spot} = [ @{$value} ];
224 ${$spot} = { %{$value} };
226 ${$spot} = bless ${$spot}, $c
234 die "Must be implemented in a child class\n";
238 die "Must be implemented in a child class\n";
243 # Recursively export into standard Perl hashes and arrays.
245 my $self = shift->_get_self;
247 my $temp = $self->_repr;
250 $self->_copy_node( $temp );
258 # Recursively import Perl hash/array structure
260 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
262 my $self = shift->_get_self;
265 # struct is not a reference, so just import based on our type
267 $struct = $self->_repr( @_ );
270 return $self->_import( $struct );
275 # Rebuild entire database into new file, then move
276 # it back on top of original.
278 my $self = shift->_get_self;
280 #XXX Need to create a new test for this
281 # if ($self->_root->{links} > 1) {
282 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
285 my $db_temp = DBM::Deep->new(
286 file => $self->_root->{file} . '.tmp',
291 $self->_copy_node( $db_temp );
295 # Attempt to copy user, group and permissions over to new file
297 my @stats = stat($self->_fh);
298 my $perms = $stats[2] & 07777;
301 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
302 chmod( $perms, $self->_root->{file} . '.tmp' );
304 # q.v. perlport for more information on this variable
305 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
307 # Potential race condition when optmizing on Win32 with locking.
308 # The Windows filesystem requires that the filehandle be closed
309 # before it is overwritten with rename(). This could be redone
313 $self->{engine}->close_fh( $self );
316 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
317 unlink $self->_root->{file} . '.tmp';
319 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
323 $self->{engine}->close_fh( $self );
324 $self->{engine}->setup_fh( $self );
331 # Make copy of object and return
333 my $self = shift->_get_self;
335 return DBM::Deep->new(
336 type => $self->_type,
337 base_offset => $self->_base_offset,
343 my %is_legal_filter = map {
346 store_key store_value
347 fetch_key fetch_value
352 # Setup filter function for storing or fetching the key or value
354 my $self = shift->_get_self;
358 if ( $is_legal_filter{$type} ) {
359 $self->_root->{"filter_$type"} = $func;
373 # Get access to the root structure
375 my $self = $_[0]->_get_self;
376 return $self->{root};
381 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
383 my $self = $_[0]->_get_self;
384 return $self->{type};
389 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
391 my $self = $_[0]->_get_self;
392 return $self->{base_offset};
397 # Get access to the raw fh
399 my $self = $_[0]->_get_self;
400 return $self->_root->{fh};
408 die "DBM::Deep: $_[1]\n";
413 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
418 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
423 # Store single hash key/value or array element in database.
425 my $self = shift->_get_self;
426 my ($key, $value) = @_;
428 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
429 $self->_throw_error( 'Cannot write to a readonly filehandle' );
433 # Request exclusive lock for writing
435 $self->lock( LOCK_EX );
437 my $md5 = $self->{engine}{digest}->($key);
439 my $tag = $self->{engine}->find_bucket_list( $self, $md5, { create => 1 } );
441 # User may be storing a hash, in which case we do not want it run
442 # through the filtering system
443 if ( !ref($value) && $self->_root->{filter_store_value} ) {
444 $value = $self->_root->{filter_store_value}->( $value );
448 # Add key/value to bucket list
450 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
459 # Fetch single value or element given plain key or array index
461 my $self = shift->_get_self;
464 my $md5 = $self->{engine}{digest}->($key);
467 # Request shared lock for reading
469 $self->lock( LOCK_SH );
471 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
478 # Get value from bucket list
480 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
484 # Filters only apply to scalar values, so the ref check is making
485 # sure the fetched bucket is a scalar, not a child hash or array.
486 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
487 ? $self->_root->{filter_fetch_value}->($result)
493 # Delete single key/value pair or element given plain key or array index
495 my $self = $_[0]->_get_self;
498 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
499 $self->_throw_error( 'Cannot write to a readonly filehandle' );
503 # Request exclusive lock for writing
505 $self->lock( LOCK_EX );
507 my $md5 = $self->{engine}{digest}->($key);
509 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
518 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
520 if (defined $value && !ref($value) && $self->_root->{filter_fetch_value}) {
521 $value = $self->_root->{filter_fetch_value}->($value);
524 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
527 # If this object is an array and the key deleted was on the end of the stack,
528 # decrement the length variable.
538 # Check if a single key or element exists given plain key or array index
540 my $self = $_[0]->_get_self;
543 my $md5 = $self->{engine}{digest}->($key);
546 # Request shared lock for reading
548 $self->lock( LOCK_SH );
550 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
555 # For some reason, the built-in exists() function returns '' for false
561 # Check if bucket exists and return 1 or ''
563 my $result = $self->{engine}->bucket_exists( $self, $tag, $md5 ) || '';
572 # Clear all keys from hash, or all elements from array.
574 my $self = $_[0]->_get_self;
576 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
577 $self->_throw_error( 'Cannot write to a readonly filehandle' );
581 # Request exclusive lock for writing
583 $self->lock( LOCK_EX );
587 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
593 #XXX This needs updating to use _release_space
594 $self->{engine}->write_tag(
595 $self, $self->_base_offset, $self->_type,
596 chr(0)x$self->{engine}{index_size},
605 # Public method aliases
607 sub put { (shift)->STORE( @_ ) }
608 sub store { (shift)->STORE( @_ ) }
609 sub get { (shift)->FETCH( @_ ) }
610 sub fetch { (shift)->FETCH( @_ ) }
611 sub delete { (shift)->DELETE( @_ ) }
612 sub exists { (shift)->EXISTS( @_ ) }
613 sub clear { (shift)->CLEAR( @_ ) }
615 package DBM::Deep::_::Root;
630 filter_store_key => undef,
631 filter_store_value => undef,
632 filter_fetch_key => undef,
633 filter_fetch_value => undef,
636 # Grab the parameters we want to use
637 foreach my $param ( keys %$self ) {
638 next unless exists $args->{$param};
639 $self->{$param} = $args->{$param}
642 if ( $self->{fh} && !$self->{file_offset} ) {
643 $self->{file_offset} = tell( $self->{fh} );
653 close $self->{fh} if $self->{fh};
663 DBM::Deep - A pure perl multi-level hash/array DBM
668 my $db = DBM::Deep->new( "foo.db" );
670 $db->{key} = 'value'; # tie() style
673 $db->put('key' => 'value'); # OO style
674 print $db->get('key');
676 # true multi-level support
677 $db->{my_complex} = [
678 'hello', { perl => 'rules' },
684 A unique flat-file database module, written in pure perl. True
685 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
686 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
687 handle millions of keys and unlimited hash levels without significant
688 slow-down. Written from the ground-up in pure perl -- this is NOT a
689 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
690 Mac OS X and Windows.
692 =head1 VERSION DIFFERENCES
694 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
695 before. While attempts have been made to be backwards compatible, no guarantees.
699 Hopefully you are using Perl's excellent CPAN module, which will download
700 and install the module for you. If not, get the tarball, and run these
712 Construction can be done OO-style (which is the recommended way), or using
713 Perl's tie() function. Both are examined here.
715 =head2 OO CONSTRUCTION
717 The recommended way to construct a DBM::Deep object is to use the new()
718 method, which gets you a blessed, tied hash or array reference.
720 my $db = DBM::Deep->new( "foo.db" );
722 This opens a new database handle, mapped to the file "foo.db". If this
723 file does not exist, it will automatically be created. DB files are
724 opened in "r+" (read/write) mode, and the type of object returned is a
725 hash, unless otherwise specified (see L<OPTIONS> below).
727 You can pass a number of options to the constructor to specify things like
728 locking, autoflush, etc. This is done by passing an inline hash:
730 my $db = DBM::Deep->new(
736 Notice that the filename is now specified I<inside> the hash with
737 the "file" parameter, as opposed to being the sole argument to the
738 constructor. This is required if any options are specified.
739 See L<OPTIONS> below for the complete list.
743 You can also start with an array instead of a hash. For this, you must
744 specify the C<type> parameter:
746 my $db = DBM::Deep->new(
748 type => DBM::Deep->TYPE_ARRAY
751 B<Note:> Specifing the C<type> parameter only takes effect when beginning
752 a new DB file. If you create a DBM::Deep object with an existing file, the
753 C<type> will be loaded from the file header, and an error will be thrown if
754 the wrong type is passed in.
756 =head2 TIE CONSTRUCTION
758 Alternately, you can create a DBM::Deep handle by using Perl's built-in
759 tie() function. The object returned from tie() can be used to call methods,
760 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
761 file (as expected with most tie'd objects).
764 my $db = tie %hash, "DBM::Deep", "foo.db";
767 my $db = tie @array, "DBM::Deep", "bar.db";
769 As with the OO constructor, you can replace the DB filename parameter with
770 a hash containing one or more options (see L<OPTIONS> just below for the
773 tie %hash, "DBM::Deep", {
781 There are a number of options that can be passed in when constructing your
782 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
788 Filename of the DB file to link the handle to. You can pass a full absolute
789 filesystem path, partial path, or a plain filename if the file is in the
790 current working directory. This is a required parameter (though q.v. fh).
794 If you want, you can pass in the fh instead of the file. This is most useful for doing
797 my $db = DBM::Deep->new( { fh => \*DATA } );
799 You are responsible for making sure that the fh has been opened appropriately for your
800 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
801 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
802 needs to read from the fh.
806 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
807 not need to set this. However, it's there if you want it.
809 If you pass in fh and do not set this, it will be set appropriately.
813 This parameter specifies what type of object to create, a hash or array. Use
814 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
815 This only takes effect when beginning a new file. This is an optional
816 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
820 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
821 function to lock the database in exclusive mode for writes, and shared mode for
822 reads. Pass any true value to enable. This affects the base DB handle I<and
823 any child hashes or arrays> that use the same DB file. This is an optional
824 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
828 Specifies whether autoflush is to be enabled on the underlying filehandle.
829 This obviously slows down write operations, but is required if you may have
830 multiple processes accessing the same DB file (also consider enable I<locking>).
831 Pass any true value to enable. This is an optional parameter, and defaults to 0
836 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
837 restore them when fetched. This is an B<experimental> feature, and does have
838 side-effects. Basically, when hashes are re-blessed into their original
839 classes, they are no longer blessed into the DBM::Deep class! So you won't be
840 able to call any DBM::Deep methods on them. You have been warned.
841 This is an optional parameter, and defaults to 0 (disabled).
845 See L<FILTERS> below.
851 With DBM::Deep you can access your databases using Perl's standard hash/array
852 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
853 treat them as such. DBM::Deep will intercept all reads/writes and direct them
854 to the right place -- the DB file. This has nothing to do with the
855 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
856 using regular hashes and arrays, rather than calling functions like C<get()>
857 and C<put()> (although those work too). It is entirely up to you how to want
858 to access your databases.
862 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
863 or even nested hashes (or arrays) using standard Perl syntax:
865 my $db = DBM::Deep->new( "foo.db" );
867 $db->{mykey} = "myvalue";
869 $db->{myhash}->{subkey} = "subvalue";
871 print $db->{myhash}->{subkey} . "\n";
873 You can even step through hash keys using the normal Perl C<keys()> function:
875 foreach my $key (keys %$db) {
876 print "$key: " . $db->{$key} . "\n";
879 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
880 pushes them onto an array, all before the loop even begins. If you have an
881 extra large hash, this may exhaust Perl's memory. Instead, consider using
882 Perl's C<each()> function, which pulls keys/values one at a time, using very
885 while (my ($key, $value) = each %$db) {
886 print "$key: $value\n";
889 Please note that when using C<each()>, you should always pass a direct
890 hash reference, not a lookup. Meaning, you should B<never> do this:
893 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
895 This causes an infinite loop, because for each iteration, Perl is calling
896 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
897 it effectively keeps returning the first key over and over again. Instead,
898 assign a temporary variable to C<$db->{foo}>, then pass that to each().
902 As with hashes, you can treat any DBM::Deep object like a normal Perl array
903 reference. This includes inserting, removing and manipulating elements,
904 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
905 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
906 or simply be a nested array reference inside a hash. Example:
908 my $db = DBM::Deep->new(
909 file => "foo-array.db",
910 type => DBM::Deep->TYPE_ARRAY
914 push @$db, "bar", "baz";
917 my $last_elem = pop @$db; # baz
918 my $first_elem = shift @$db; # bah
919 my $second_elem = $db->[1]; # bar
921 my $num_elements = scalar @$db;
925 In addition to the I<tie()> interface, you can also use a standard OO interface
926 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
927 array) has its own methods, but both types share the following common methods:
928 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
932 =item * new() / clone()
934 These are the constructor and copy-functions.
936 =item * put() / store()
938 Stores a new hash key/value pair, or sets an array element value. Takes two
939 arguments, the hash key or array index, and the new value. The value can be
940 a scalar, hash ref or array ref. Returns true on success, false on failure.
942 $db->put("foo", "bar"); # for hashes
943 $db->put(1, "bar"); # for arrays
945 =item * get() / fetch()
947 Fetches the value of a hash key or array element. Takes one argument: the hash
948 key or array index. Returns a scalar, hash ref or array ref, depending on the
951 my $value = $db->get("foo"); # for hashes
952 my $value = $db->get(1); # for arrays
956 Checks if a hash key or array index exists. Takes one argument: the hash key
957 or array index. Returns true if it exists, false if not.
959 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
960 if ($db->exists(1)) { print "yay!\n"; } # for arrays
964 Deletes one hash key/value pair or array element. Takes one argument: the hash
965 key or array index. Returns true on success, false if not found. For arrays,
966 the remaining elements located after the deleted element are NOT moved over.
967 The deleted element is essentially just undefined, which is exactly how Perl's
968 internal arrays work. Please note that the space occupied by the deleted
969 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
970 below for details and workarounds.
972 $db->delete("foo"); # for hashes
973 $db->delete(1); # for arrays
977 Deletes B<all> hash keys or array elements. Takes no arguments. No return
978 value. Please note that the space occupied by the deleted keys/values or
979 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
980 details and workarounds.
982 $db->clear(); # hashes or arrays
984 =item * lock() / unlock()
990 Recover lost disk space.
992 =item * import() / export()
994 Data going in and out.
996 =item * set_digest() / set_pack() / set_filter()
998 q.v. adjusting the interal parameters.
1004 For hashes, DBM::Deep supports all the common methods described above, and the
1005 following additional methods: C<first_key()> and C<next_key()>.
1011 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1012 fetched in an undefined order (which appears random). Takes no arguments,
1013 returns the key as a scalar value.
1015 my $key = $db->first_key();
1019 Returns the "next" key in the hash, given the previous one as the sole argument.
1020 Returns undef if there are no more keys to be fetched.
1022 $key = $db->next_key($key);
1026 Here are some examples of using hashes:
1028 my $db = DBM::Deep->new( "foo.db" );
1030 $db->put("foo", "bar");
1031 print "foo: " . $db->get("foo") . "\n";
1033 $db->put("baz", {}); # new child hash ref
1034 $db->get("baz")->put("buz", "biz");
1035 print "buz: " . $db->get("baz")->get("buz") . "\n";
1037 my $key = $db->first_key();
1039 print "$key: " . $db->get($key) . "\n";
1040 $key = $db->next_key($key);
1043 if ($db->exists("foo")) { $db->delete("foo"); }
1047 For arrays, DBM::Deep supports all the common methods described above, and the
1048 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1049 C<unshift()> and C<splice()>.
1055 Returns the number of elements in the array. Takes no arguments.
1057 my $len = $db->length();
1061 Adds one or more elements onto the end of the array. Accepts scalars, hash
1062 refs or array refs. No return value.
1064 $db->push("foo", "bar", {});
1068 Fetches the last element in the array, and deletes it. Takes no arguments.
1069 Returns undef if array is empty. Returns the element value.
1071 my $elem = $db->pop();
1075 Fetches the first element in the array, deletes it, then shifts all the
1076 remaining elements over to take up the space. Returns the element value. This
1077 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1080 my $elem = $db->shift();
1084 Inserts one or more elements onto the beginning of the array, shifting all
1085 existing elements over to make room. Accepts scalars, hash refs or array refs.
1086 No return value. This method is not recommended with large arrays -- see
1087 <LARGE ARRAYS> below for details.
1089 $db->unshift("foo", "bar", {});
1093 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1094 -f splice> for usage -- it is too complicated to document here. This method is
1095 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1099 Here are some examples of using arrays:
1101 my $db = DBM::Deep->new(
1103 type => DBM::Deep->TYPE_ARRAY
1106 $db->push("bar", "baz");
1107 $db->unshift("foo");
1110 my $len = $db->length();
1111 print "length: $len\n"; # 4
1113 for (my $k=0; $k<$len; $k++) {
1114 print "$k: " . $db->get($k) . "\n";
1117 $db->splice(1, 2, "biz", "baf");
1119 while (my $elem = shift @$db) {
1120 print "shifted: $elem\n";
1125 Enable automatic file locking by passing a true value to the C<locking>
1126 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1128 my $db = DBM::Deep->new(
1133 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1134 mode for writes, and shared mode for reads. This is required if you have
1135 multiple processes accessing the same database file, to avoid file corruption.
1136 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1137 NFS> below for more.
1139 =head2 EXPLICIT LOCKING
1141 You can explicitly lock a database, so it remains locked for multiple
1142 transactions. This is done by calling the C<lock()> method, and passing an
1143 optional lock mode argument (defaults to exclusive mode). This is particularly
1144 useful for things like counters, where the current value needs to be fetched,
1145 then incremented, then stored again.
1148 my $counter = $db->get("counter");
1150 $db->put("counter", $counter);
1159 You can pass C<lock()> an optional argument, which specifies which mode to use
1160 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1161 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1162 same as the constants defined in Perl's C<Fcntl> module.
1164 $db->lock( DBM::Deep->LOCK_SH );
1168 =head1 IMPORTING/EXPORTING
1170 You can import existing complex structures by calling the C<import()> method,
1171 and export an entire database into an in-memory structure using the C<export()>
1172 method. Both are examined here.
1176 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1177 walking the structure and adding keys/elements to the database as you go,
1178 simply pass a reference to the C<import()> method. This recursively adds
1179 everything to an existing DBM::Deep object for you. Here is an example:
1184 array1 => [ "elem0", "elem1", "elem2" ],
1186 subkey1 => "subvalue1",
1187 subkey2 => "subvalue2"
1191 my $db = DBM::Deep->new( "foo.db" );
1192 $db->import( $struct );
1194 print $db->{key1} . "\n"; # prints "value1"
1196 This recursively imports the entire C<$struct> object into C<$db>, including
1197 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1198 keys are merged with the existing ones, replacing if they already exist.
1199 The C<import()> method can be called on any database level (not just the base
1200 level), and works with both hash and array DB types.
1202 B<Note:> Make sure your existing structure has no circular references in it.
1203 These will cause an infinite loop when importing.
1207 Calling the C<export()> method on an existing DBM::Deep object will return
1208 a reference to a new in-memory copy of the database. The export is done
1209 recursively, so all nested hashes/arrays are all exported to standard Perl
1210 objects. Here is an example:
1212 my $db = DBM::Deep->new( "foo.db" );
1214 $db->{key1} = "value1";
1215 $db->{key2} = "value2";
1217 $db->{hash1}->{subkey1} = "subvalue1";
1218 $db->{hash1}->{subkey2} = "subvalue2";
1220 my $struct = $db->export();
1222 print $struct->{key1} . "\n"; # prints "value1"
1224 This makes a complete copy of the database in memory, and returns a reference
1225 to it. The C<export()> method can be called on any database level (not just
1226 the base level), and works with both hash and array DB types. Be careful of
1227 large databases -- you can store a lot more data in a DBM::Deep object than an
1228 in-memory Perl structure.
1230 B<Note:> Make sure your database has no circular references in it.
1231 These will cause an infinite loop when exporting.
1235 DBM::Deep has a number of hooks where you can specify your own Perl function
1236 to perform filtering on incoming or outgoing data. This is a perfect
1237 way to extend the engine, and implement things like real-time compression or
1238 encryption. Filtering applies to the base DB level, and all child hashes /
1239 arrays. Filter hooks can be specified when your DBM::Deep object is first
1240 constructed, or by calling the C<set_filter()> method at any time. There are
1241 four available filter hooks, described below:
1245 =item * filter_store_key
1247 This filter is called whenever a hash key is stored. It
1248 is passed the incoming key, and expected to return a transformed key.
1250 =item * filter_store_value
1252 This filter is called whenever a hash key or array element is stored. It
1253 is passed the incoming value, and expected to return a transformed value.
1255 =item * filter_fetch_key
1257 This filter is called whenever a hash key is fetched (i.e. via
1258 C<first_key()> or C<next_key()>). It is passed the transformed key,
1259 and expected to return the plain key.
1261 =item * filter_fetch_value
1263 This filter is called whenever a hash key or array element is fetched.
1264 It is passed the transformed value, and expected to return the plain value.
1268 Here are the two ways to setup a filter hook:
1270 my $db = DBM::Deep->new(
1272 filter_store_value => \&my_filter_store,
1273 filter_fetch_value => \&my_filter_fetch
1278 $db->set_filter( "filter_store_value", \&my_filter_store );
1279 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1281 Your filter function will be called only when dealing with SCALAR keys or
1282 values. When nested hashes and arrays are being stored/fetched, filtering
1283 is bypassed. Filters are called as static functions, passed a single SCALAR
1284 argument, and expected to return a single SCALAR value. If you want to
1285 remove a filter, set the function reference to C<undef>:
1287 $db->set_filter( "filter_store_value", undef );
1289 =head2 REAL-TIME ENCRYPTION EXAMPLE
1291 Here is a working example that uses the I<Crypt::Blowfish> module to
1292 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1293 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1294 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1297 use Crypt::Blowfish;
1300 my $cipher = Crypt::CBC->new({
1301 'key' => 'my secret key',
1302 'cipher' => 'Blowfish',
1304 'regenerate_key' => 0,
1305 'padding' => 'space',
1309 my $db = DBM::Deep->new(
1310 file => "foo-encrypt.db",
1311 filter_store_key => \&my_encrypt,
1312 filter_store_value => \&my_encrypt,
1313 filter_fetch_key => \&my_decrypt,
1314 filter_fetch_value => \&my_decrypt,
1317 $db->{key1} = "value1";
1318 $db->{key2} = "value2";
1319 print "key1: " . $db->{key1} . "\n";
1320 print "key2: " . $db->{key2} . "\n";
1326 return $cipher->encrypt( $_[0] );
1329 return $cipher->decrypt( $_[0] );
1332 =head2 REAL-TIME COMPRESSION EXAMPLE
1334 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1335 compression / decompression of keys & values with DBM::Deep Filters.
1336 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1337 more on I<Compress::Zlib>.
1342 my $db = DBM::Deep->new(
1343 file => "foo-compress.db",
1344 filter_store_key => \&my_compress,
1345 filter_store_value => \&my_compress,
1346 filter_fetch_key => \&my_decompress,
1347 filter_fetch_value => \&my_decompress,
1350 $db->{key1} = "value1";
1351 $db->{key2} = "value2";
1352 print "key1: " . $db->{key1} . "\n";
1353 print "key2: " . $db->{key2} . "\n";
1359 return Compress::Zlib::memGzip( $_[0] ) ;
1362 return Compress::Zlib::memGunzip( $_[0] ) ;
1365 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1366 actually numerical index numbers, and are not filtered.
1368 =head1 ERROR HANDLING
1370 Most DBM::Deep methods return a true value for success, and call die() on
1371 failure. You can wrap calls in an eval block to catch the die.
1373 my $db = DBM::Deep->new( "foo.db" ); # create hash
1374 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1376 print $@; # prints error message
1378 =head1 LARGEFILE SUPPORT
1380 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1381 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1382 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1383 by calling the static C<set_pack()> method before you do anything else.
1385 DBM::Deep::set_pack(8, 'Q');
1387 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1388 instead of 32-bit longs. After setting these values your DB files have a
1389 theoretical maximum size of 16 XB (exabytes).
1391 B<Note:> Changing these values will B<NOT> work for existing database files.
1392 Only change this for new files, and make sure it stays set consistently
1393 throughout the file's life. If you do set these values, you can no longer
1394 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1395 back to 32-bit mode.
1397 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1398 only a 32-bit Perl. However, I have received user reports that this does
1401 =head1 LOW-LEVEL ACCESS
1403 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1404 you can call the C<_fh()> method, which returns the handle:
1406 my $fh = $db->_fh();
1408 This method can be called on the root level of the datbase, or any child
1409 hashes or arrays. All levels share a I<root> structure, which contains things
1410 like the filehandle, a reference counter, and all the options specified
1411 when you created the object. You can get access to this root structure by
1412 calling the C<root()> method.
1414 my $root = $db->_root();
1416 This is useful for changing options after the object has already been created,
1417 such as enabling/disabling locking. You can also store your own temporary user
1418 data in this structure (be wary of name collision), which is then accessible from
1419 any child hash or array.
1421 =head1 CUSTOM DIGEST ALGORITHM
1423 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1424 keys. However you can override this, and use another algorithm (such as SHA-256)
1425 or even write your own. But please note that DBM::Deep currently expects zero
1426 collisions, so your algorithm has to be I<perfect>, so to speak.
1427 Collision detection may be introduced in a later version.
1431 You can specify a custom digest algorithm by calling the static C<set_digest()>
1432 function, passing a reference to a subroutine, and the length of the algorithm's
1433 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1434 objects. Here is a working example that uses a 256-bit hash from the
1435 I<Digest::SHA256> module. Please see
1436 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1441 my $context = Digest::SHA256::new(256);
1443 DBM::Deep::set_digest( \&my_digest, 32 );
1445 my $db = DBM::Deep->new( "foo-sha.db" );
1447 $db->{key1} = "value1";
1448 $db->{key2} = "value2";
1449 print "key1: " . $db->{key1} . "\n";
1450 print "key2: " . $db->{key2} . "\n";
1456 return substr( $context->hash($_[0]), 0, 32 );
1459 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1460 of bytes you specify in the C<set_digest()> function (in this case 32).
1462 =head1 CIRCULAR REFERENCES
1464 DBM::Deep has B<experimental> support for circular references. Meaning you
1465 can have a nested hash key or array element that points to a parent object.
1466 This relationship is stored in the DB file, and is preserved between sessions.
1469 my $db = DBM::Deep->new( "foo.db" );
1472 $db->{circle} = $db; # ref to self
1474 print $db->{foo} . "\n"; # prints "bar"
1475 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1477 B<Note>: Passing the object to a function that recursively walks the
1478 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1479 C<export()> methods) will result in an infinite loop. This will be fixed in
1482 =head1 CAVEATS / ISSUES / BUGS
1484 This section describes all the known issues with DBM::Deep. It you have found
1485 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1487 =head2 UNUSED SPACE RECOVERY
1489 One major caveat with DBM::Deep is that space occupied by existing keys and
1490 values is not recovered when they are deleted. Meaning if you keep deleting
1491 and adding new keys, your file will continuously grow. I am working on this,
1492 but in the meantime you can call the built-in C<optimize()> method from time to
1493 time (perhaps in a crontab or something) to recover all your unused space.
1495 $db->optimize(); # returns true on success
1497 This rebuilds the ENTIRE database into a new file, then moves it on top of
1498 the original. The new file will have no unused space, thus it will take up as
1499 little disk space as possible. Please note that this operation can take
1500 a long time for large files, and you need enough disk space to temporarily hold
1501 2 copies of your DB file. The temporary file is created in the same directory
1502 as the original, named with a ".tmp" extension, and is deleted when the
1503 operation completes. Oh, and if locking is enabled, the DB is automatically
1504 locked for the entire duration of the copy.
1506 B<WARNING:> Only call optimize() on the top-level node of the database, and
1507 make sure there are no child references lying around. DBM::Deep keeps a reference
1508 counter, and if it is greater than 1, optimize() will abort and return undef.
1510 =head2 AUTOVIVIFICATION
1512 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1513 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1514 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1515 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1516 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1519 $db->{foo}->{bar} = "hello";
1521 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1522 being an empty hash. Try this instead, which works fine:
1524 $db->{foo} = { bar => "hello" };
1526 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1527 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1528 Probably a bug in Perl.
1532 (The reasons given assume a high level of Perl understanding, specifically of
1533 references. You can safely skip this section.)
1535 Currently, the only references supported are HASH and ARRAY. The other reference
1536 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1542 These are things like filehandles and other sockets. They can't be supported
1543 because it's completely unclear how DBM::Deep should serialize them.
1545 =item * SCALAR / REF
1547 The discussion here refers to the following type of example:
1554 # In some other process ...
1556 my $val = ${ $db->{key1} };
1558 is( $val, 50, "What actually gets stored in the DB file?" );
1560 The problem is one of synchronization. When the variable being referred to
1561 changes value, the reference isn't notified. This means that the new value won't
1562 be stored in the datafile for other processes to read. There is no TIEREF.
1564 It is theoretically possible to store references to values already within a
1565 DBM::Deep object because everything already is synchronized, but the change to
1566 the internals would be quite large. Specifically, DBM::Deep would have to tie
1567 every single value that is stored. This would bloat the RAM footprint of
1568 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1569 all to support a feature that has never been requested.
1573 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1574 mechanism for serializing coderefs, including saving off all closure state.
1575 However, just as for SCALAR and REF, that closure state may change without
1576 notifying the DBM::Deep object storing the reference.
1580 =head2 FILE CORRUPTION
1582 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1583 for a 32-bit signature when opened, but other corruption in files can cause
1584 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1585 stuck in an infinite loop depending on the level of corruption. File write
1586 operations are not checked for failure (for speed), so if you happen to run
1587 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1588 be addressed in a later version of DBM::Deep.
1592 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1593 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1594 about setting up your NFS server with a locking daemon, then using lockf() to
1595 lock your files, but your mileage may vary there as well. From what I
1596 understand, there is no real way to do it. However, if you need access to the
1597 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1598 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1600 =head2 COPYING OBJECTS
1602 Beware of copying tied objects in Perl. Very strange things can happen.
1603 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1604 returns a new, blessed, tied hash or array to the same level in the DB.
1606 my $copy = $db->clone();
1608 B<Note>: Since clone() here is cloning the object, not the database location, any
1609 modifications to either $db or $copy will be visible in both.
1613 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1614 These functions cause every element in the array to move, which can be murder
1615 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1616 a different location. This will be addressed in the forthcoming version 1.00.
1618 =head2 WRITEONLY FILES
1620 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1621 writeonly mode. STORE will verify that the filehandle is writable. However, there
1622 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1623 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1627 This section discusses DBM::Deep's speed and memory usage.
1631 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1632 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1633 multi-level hash/array support, and cross-platform FTPable files. Even so,
1634 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1635 with huge databases. Here is some test data:
1637 Adding 1,000,000 keys to new DB file...
1639 At 100 keys, avg. speed is 2,703 keys/sec
1640 At 200 keys, avg. speed is 2,642 keys/sec
1641 At 300 keys, avg. speed is 2,598 keys/sec
1642 At 400 keys, avg. speed is 2,578 keys/sec
1643 At 500 keys, avg. speed is 2,722 keys/sec
1644 At 600 keys, avg. speed is 2,628 keys/sec
1645 At 700 keys, avg. speed is 2,700 keys/sec
1646 At 800 keys, avg. speed is 2,607 keys/sec
1647 At 900 keys, avg. speed is 2,190 keys/sec
1648 At 1,000 keys, avg. speed is 2,570 keys/sec
1649 At 2,000 keys, avg. speed is 2,417 keys/sec
1650 At 3,000 keys, avg. speed is 1,982 keys/sec
1651 At 4,000 keys, avg. speed is 1,568 keys/sec
1652 At 5,000 keys, avg. speed is 1,533 keys/sec
1653 At 6,000 keys, avg. speed is 1,787 keys/sec
1654 At 7,000 keys, avg. speed is 1,977 keys/sec
1655 At 8,000 keys, avg. speed is 2,028 keys/sec
1656 At 9,000 keys, avg. speed is 2,077 keys/sec
1657 At 10,000 keys, avg. speed is 2,031 keys/sec
1658 At 20,000 keys, avg. speed is 1,970 keys/sec
1659 At 30,000 keys, avg. speed is 2,050 keys/sec
1660 At 40,000 keys, avg. speed is 2,073 keys/sec
1661 At 50,000 keys, avg. speed is 1,973 keys/sec
1662 At 60,000 keys, avg. speed is 1,914 keys/sec
1663 At 70,000 keys, avg. speed is 2,091 keys/sec
1664 At 80,000 keys, avg. speed is 2,103 keys/sec
1665 At 90,000 keys, avg. speed is 1,886 keys/sec
1666 At 100,000 keys, avg. speed is 1,970 keys/sec
1667 At 200,000 keys, avg. speed is 2,053 keys/sec
1668 At 300,000 keys, avg. speed is 1,697 keys/sec
1669 At 400,000 keys, avg. speed is 1,838 keys/sec
1670 At 500,000 keys, avg. speed is 1,941 keys/sec
1671 At 600,000 keys, avg. speed is 1,930 keys/sec
1672 At 700,000 keys, avg. speed is 1,735 keys/sec
1673 At 800,000 keys, avg. speed is 1,795 keys/sec
1674 At 900,000 keys, avg. speed is 1,221 keys/sec
1675 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1677 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1678 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1679 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1680 Run time was 12 min 3 sec.
1684 One of the great things about DBM::Deep is that it uses very little memory.
1685 Even with huge databases (1,000,000+ keys) you will not see much increased
1686 memory on your process. DBM::Deep relies solely on the filesystem for storing
1687 and fetching data. Here is output from I</usr/bin/top> before even opening a
1690 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1691 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1693 Basically the process is taking 2,716K of memory. And here is the same
1694 process after storing and fetching 1,000,000 keys:
1696 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1697 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1699 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1700 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1702 =head1 DB FILE FORMAT
1704 In case you were interested in the underlying DB file format, it is documented
1705 here in this section. You don't need to know this to use the module, it's just
1706 included for reference.
1710 DBM::Deep files always start with a 32-bit signature to identify the file type.
1711 This is at offset 0. The signature is "DPDB" in network byte order. This is
1712 checked for when the file is opened and an error will be thrown if it's not found.
1716 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1717 has a standard header containing the type of data, the length of data, and then
1718 the data itself. The type is a single character (1 byte), the length is a
1719 32-bit unsigned long in network byte order, and the data is, well, the data.
1720 Here is how it unfolds:
1724 Immediately after the 32-bit file signature is the I<Master Index> record.
1725 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1726 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1727 depending on how the DBM::Deep object was constructed.
1729 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1730 number). The first 8-bit char of the MD5 signature is the offset into the
1731 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1732 index element is a file offset of the next tag for the key/element in question,
1733 which is usually a I<Bucket List> tag (see below).
1735 The next tag I<could> be another index, depending on how many keys/elements
1736 exist. See L<RE-INDEXING> below for details.
1740 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1741 file offsets to where the actual data is stored. It starts with a standard
1742 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1743 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1744 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1745 When the list fills up, a I<Re-Index> operation is performed (See
1746 L<RE-INDEXING> below).
1750 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1751 index/value pair (in array mode). It starts with a standard tag header with
1752 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1753 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1754 header. The size reported in the tag header is only for the value, but then,
1755 just after the value is another size (32-bit unsigned long) and then the plain
1756 key itself. Since the value is likely to be fetched more often than the plain
1757 key, I figured it would be I<slightly> faster to store the value first.
1759 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1760 record for the nested structure, where the process begins all over again.
1764 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1765 exhausted. Then, when another key/element comes in, the list is converted to a
1766 new index record. However, this index will look at the next char in the MD5
1767 hash, and arrange new Bucket List pointers accordingly. This process is called
1768 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1769 17 (16 + new one) keys/elements are removed from the old Bucket List and
1770 inserted into the new index. Several new Bucket Lists are created in the
1771 process, as a new MD5 char from the key is being examined (it is unlikely that
1772 the keys will all share the same next char of their MD5s).
1774 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1775 when the Bucket Lists will turn into indexes, but the first round tends to
1776 happen right around 4,000 keys. You will see a I<slight> decrease in
1777 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1778 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1779 right around 900,000 keys. This process can continue nearly indefinitely --
1780 right up until the point the I<MD5> signatures start colliding with each other,
1781 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1782 getting struck by lightning while you are walking to cash in your tickets.
1783 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1784 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1785 this is 340 unodecillion, but don't quote me).
1789 When a new key/element is stored, the key (or index number) is first run through
1790 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1791 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1792 for the first char of the signature (in this case I<b0>). If it does not exist,
1793 a new I<Bucket List> is created for our key (and the next 15 future keys that
1794 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1795 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1796 this point, unless we are replacing an existing I<Bucket>), where the actual
1797 data will be stored.
1801 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1802 (or index number), then walking along the indexes. If there are enough
1803 keys/elements in this DB level, there might be nested indexes, each linked to
1804 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1805 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1806 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1807 plain key are stored.
1809 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1810 methods. In this process the indexes are walked systematically, and each key
1811 fetched in increasing MD5 order (which is why it appears random). Once the
1812 I<Bucket> is found, the value is skipped and the plain key returned instead.
1813 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1814 alphabetically sorted. This only happens on an index-level -- as soon as the
1815 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1816 so it's pretty much undefined how the keys will come out -- just like Perl's
1819 =head1 CODE COVERAGE
1821 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1822 B<Devel::Cover> report on this module's test suite.
1824 ----------------------------------- ------ ------ ------ ------ ------ ------
1825 File stmt bran cond sub time total
1826 ----------------------------------- ------ ------ ------ ------ ------ ------
1827 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1828 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1829 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1830 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1831 Total 97.9 85.9 79.7 100.0 100.0 94.3
1832 ----------------------------------- ------ ------ ------ ------ ------ ------
1834 =head1 MORE INFORMATION
1836 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1837 or send email to L<DBM-Deep@googlegroups.com>.
1841 Joseph Huckaby, L<jhuckaby@cpan.org>
1843 Rob Kinyon, L<rkinyon@cpan.org>
1845 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1849 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1850 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1854 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1855 This is free software, you may use it and distribute it under the
1856 same terms as Perl itself.