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,
109 base_offset => undef,
112 # Strip out the node-level parameters before passing $args to
113 # the root's constructor.
114 foreach my $param ( keys %$self ) {
115 next unless exists $args->{$param};
116 $self->{$param} = delete $args->{$param}
119 # locking implicitly enables autoflush
120 if ($args->{locking}) { $args->{autoflush} = 1; }
122 $self->{root} = exists $args->{root}
124 : DBM::Deep::_::Root->new( $args );
126 #XXX Right before this line, we have to set the physical parameters like
127 #XXX 2S vs. 4N vs. 8Q or max_buckets, etc.
128 $self->{engine}->setup_fh( $self );
135 require DBM::Deep::Hash;
136 return DBM::Deep::Hash->TIEHASH( @_ );
141 require DBM::Deep::Array;
142 return DBM::Deep::Array->TIEARRAY( @_ );
145 #XXX Unneeded now ...
151 # If db locking is set, flock() the db file. If called multiple
152 # times before unlock(), then the same number of unlocks() must
153 # be called before the lock is released.
155 my $self = shift->_get_self;
157 $type = LOCK_EX unless defined $type;
159 if (!defined($self->_fh)) { return; }
161 if ($self->_root->{locking}) {
162 if (!$self->_root->{locked}) {
163 flock($self->_fh, $type);
165 # refresh end counter in case file has changed size
166 my @stats = stat($self->_fh);
167 $self->_root->{end} = $stats[7];
169 # double-check file inode, in case another process
170 # has optimize()d our file while we were waiting.
171 if ($stats[1] != $self->_root->{inode}) {
172 $self->{engine}->close_fh( $self );
173 $self->{engine}->setup_fh( $self );
174 flock($self->_fh, $type); # re-lock
176 # This may not be necessary after re-opening
177 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
180 $self->_root->{locked}++;
190 # If db locking is set, unlock the db file. See note in lock()
191 # regarding calling lock() multiple times.
193 my $self = shift->_get_self;
195 if (!defined($self->_fh)) { return; }
197 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
198 $self->_root->{locked}--;
199 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
208 my $self = shift->_get_self;
209 my ($spot, $value) = @_;
214 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
215 ${$spot} = $value->_repr;
216 $value->_copy_node( ${$spot} );
219 my $r = Scalar::Util::reftype( $value );
220 my $c = Scalar::Util::blessed( $value );
221 if ( $r eq 'ARRAY' ) {
222 ${$spot} = [ @{$value} ];
225 ${$spot} = { %{$value} };
227 ${$spot} = bless ${$spot}, $c
235 die "Must be implemented in a child class\n";
239 die "Must be implemented in a child class\n";
244 # Recursively export into standard Perl hashes and arrays.
246 my $self = shift->_get_self;
248 my $temp = $self->_repr;
251 $self->_copy_node( $temp );
259 # Recursively import Perl hash/array structure
261 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
263 my $self = shift->_get_self;
266 # struct is not a reference, so just import based on our type
268 $struct = $self->_repr( @_ );
271 return $self->_import( $struct );
276 # Rebuild entire database into new file, then move
277 # it back on top of original.
279 my $self = shift->_get_self;
281 #XXX Need to create a new test for this
282 # if ($self->_root->{links} > 1) {
283 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
286 my $db_temp = DBM::Deep->new(
287 file => $self->_root->{file} . '.tmp',
292 $self->_copy_node( $db_temp );
296 # Attempt to copy user, group and permissions over to new file
298 my @stats = stat($self->_fh);
299 my $perms = $stats[2] & 07777;
302 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
303 chmod( $perms, $self->_root->{file} . '.tmp' );
305 # q.v. perlport for more information on this variable
306 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
308 # Potential race condition when optmizing on Win32 with locking.
309 # The Windows filesystem requires that the filehandle be closed
310 # before it is overwritten with rename(). This could be redone
314 $self->{engine}->close_fh( $self );
317 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
318 unlink $self->_root->{file} . '.tmp';
320 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
324 $self->{engine}->close_fh( $self );
325 $self->{engine}->setup_fh( $self );
332 # Make copy of object and return
334 my $self = shift->_get_self;
336 return DBM::Deep->new(
337 type => $self->_type,
338 base_offset => $self->_base_offset,
344 my %is_legal_filter = map {
347 store_key store_value
348 fetch_key fetch_value
353 # Setup filter function for storing or fetching the key or value
355 my $self = shift->_get_self;
359 if ( $is_legal_filter{$type} ) {
360 $self->_root->{"filter_$type"} = $func;
374 # Get access to the root structure
376 my $self = $_[0]->_get_self;
377 return $self->{root};
382 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
384 my $self = $_[0]->_get_self;
385 return $self->{type};
390 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
392 my $self = $_[0]->_get_self;
393 return $self->{base_offset};
398 # Get access to the raw fh
400 my $self = $_[0]->_get_self;
401 return $self->_root->{fh};
409 die "DBM::Deep: $_[1]\n";
414 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
419 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
424 # Store single hash key/value or array element in database.
426 my $self = shift->_get_self;
427 my ($key, $value) = @_;
429 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
430 $self->_throw_error( 'Cannot write to a readonly filehandle' );
434 # Request exclusive lock for writing
436 $self->lock( LOCK_EX );
438 my $md5 = $self->{engine}{digest}->($key);
440 my $tag = $self->{engine}->find_bucket_list( $self, $md5, { create => 1 } );
442 # User may be storing a hash, in which case we do not want it run
443 # through the filtering system
444 if ( !ref($value) && $self->_root->{filter_store_value} ) {
445 $value = $self->_root->{filter_store_value}->( $value );
449 # Add key/value to bucket list
451 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
460 # Fetch single value or element given plain key or array index
462 my $self = shift->_get_self;
465 my $md5 = $self->{engine}{digest}->($key);
468 # Request shared lock for reading
470 $self->lock( LOCK_SH );
472 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
479 # Get value from bucket list
481 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
485 # Filters only apply to scalar values, so the ref check is making
486 # sure the fetched bucket is a scalar, not a child hash or array.
487 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
488 ? $self->_root->{filter_fetch_value}->($result)
494 # Delete single key/value pair or element given plain key or array index
496 my $self = $_[0]->_get_self;
499 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
500 $self->_throw_error( 'Cannot write to a readonly filehandle' );
504 # Request exclusive lock for writing
506 $self->lock( LOCK_EX );
508 my $md5 = $self->{engine}{digest}->($key);
510 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
519 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
521 if (defined $value && !ref($value) && $self->_root->{filter_fetch_value}) {
522 $value = $self->_root->{filter_fetch_value}->($value);
525 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
528 # If this object is an array and the key deleted was on the end of the stack,
529 # decrement the length variable.
539 # Check if a single key or element exists given plain key or array index
541 my $self = $_[0]->_get_self;
544 my $md5 = $self->{engine}{digest}->($key);
547 # Request shared lock for reading
549 $self->lock( LOCK_SH );
551 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
556 # For some reason, the built-in exists() function returns '' for false
562 # Check if bucket exists and return 1 or ''
564 my $result = $self->{engine}->bucket_exists( $self, $tag, $md5 ) || '';
573 # Clear all keys from hash, or all elements from array.
575 my $self = $_[0]->_get_self;
577 if ( $^O ne 'MSWin32' && !_is_writable( $self->_fh ) ) {
578 $self->_throw_error( 'Cannot write to a readonly filehandle' );
582 # Request exclusive lock for writing
584 $self->lock( LOCK_EX );
588 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
594 #XXX This needs updating to use _release_space
595 $self->{engine}->write_tag(
596 $self, $self->_base_offset, $self->_type,
597 chr(0)x$self->{engine}{index_size},
606 # Public method aliases
608 sub put { (shift)->STORE( @_ ) }
609 sub store { (shift)->STORE( @_ ) }
610 sub get { (shift)->FETCH( @_ ) }
611 sub fetch { (shift)->FETCH( @_ ) }
612 sub delete { (shift)->DELETE( @_ ) }
613 sub exists { (shift)->EXISTS( @_ ) }
614 sub clear { (shift)->CLEAR( @_ ) }
616 package DBM::Deep::_::Root;
631 filter_store_key => undef,
632 filter_store_value => undef,
633 filter_fetch_key => undef,
634 filter_fetch_value => undef,
638 if ( $self->{fh} && !$self->{file_offset} ) {
639 $self->{file_offset} = tell( $self->{fh} );
649 close $self->{fh} if $self->{fh};
659 DBM::Deep - A pure perl multi-level hash/array DBM
664 my $db = DBM::Deep->new( "foo.db" );
666 $db->{key} = 'value'; # tie() style
669 $db->put('key' => 'value'); # OO style
670 print $db->get('key');
672 # true multi-level support
673 $db->{my_complex} = [
674 'hello', { perl => 'rules' },
680 A unique flat-file database module, written in pure perl. True
681 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
682 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
683 handle millions of keys and unlimited hash levels without significant
684 slow-down. Written from the ground-up in pure perl -- this is NOT a
685 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
686 Mac OS X and Windows.
688 =head1 VERSION DIFFERENCES
690 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
691 before. While attempts have been made to be backwards compatible, no guarantees.
695 Hopefully you are using Perl's excellent CPAN module, which will download
696 and install the module for you. If not, get the tarball, and run these
708 Construction can be done OO-style (which is the recommended way), or using
709 Perl's tie() function. Both are examined here.
711 =head2 OO CONSTRUCTION
713 The recommended way to construct a DBM::Deep object is to use the new()
714 method, which gets you a blessed, tied hash or array reference.
716 my $db = DBM::Deep->new( "foo.db" );
718 This opens a new database handle, mapped to the file "foo.db". If this
719 file does not exist, it will automatically be created. DB files are
720 opened in "r+" (read/write) mode, and the type of object returned is a
721 hash, unless otherwise specified (see L<OPTIONS> below).
723 You can pass a number of options to the constructor to specify things like
724 locking, autoflush, etc. This is done by passing an inline hash:
726 my $db = DBM::Deep->new(
732 Notice that the filename is now specified I<inside> the hash with
733 the "file" parameter, as opposed to being the sole argument to the
734 constructor. This is required if any options are specified.
735 See L<OPTIONS> below for the complete list.
739 You can also start with an array instead of a hash. For this, you must
740 specify the C<type> parameter:
742 my $db = DBM::Deep->new(
744 type => DBM::Deep->TYPE_ARRAY
747 B<Note:> Specifing the C<type> parameter only takes effect when beginning
748 a new DB file. If you create a DBM::Deep object with an existing file, the
749 C<type> will be loaded from the file header, and an error will be thrown if
750 the wrong type is passed in.
752 =head2 TIE CONSTRUCTION
754 Alternately, you can create a DBM::Deep handle by using Perl's built-in
755 tie() function. The object returned from tie() can be used to call methods,
756 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
757 file (as expected with most tie'd objects).
760 my $db = tie %hash, "DBM::Deep", "foo.db";
763 my $db = tie @array, "DBM::Deep", "bar.db";
765 As with the OO constructor, you can replace the DB filename parameter with
766 a hash containing one or more options (see L<OPTIONS> just below for the
769 tie %hash, "DBM::Deep", {
777 There are a number of options that can be passed in when constructing your
778 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
784 Filename of the DB file to link the handle to. You can pass a full absolute
785 filesystem path, partial path, or a plain filename if the file is in the
786 current working directory. This is a required parameter (though q.v. fh).
790 If you want, you can pass in the fh instead of the file. This is most useful for doing
793 my $db = DBM::Deep->new( { fh => \*DATA } );
795 You are responsible for making sure that the fh has been opened appropriately for your
796 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
797 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
798 needs to read from the fh.
802 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
803 not need to set this. However, it's there if you want it.
805 If you pass in fh and do not set this, it will be set appropriately.
809 This parameter specifies what type of object to create, a hash or array. Use
810 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
811 This only takes effect when beginning a new file. This is an optional
812 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
816 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
817 function to lock the database in exclusive mode for writes, and shared mode for
818 reads. Pass any true value to enable. This affects the base DB handle I<and
819 any child hashes or arrays> that use the same DB file. This is an optional
820 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
824 Specifies whether autoflush is to be enabled on the underlying filehandle.
825 This obviously slows down write operations, but is required if you may have
826 multiple processes accessing the same DB file (also consider enable I<locking>).
827 Pass any true value to enable. This is an optional parameter, and defaults to 0
832 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
833 restore them when fetched. This is an B<experimental> feature, and does have
834 side-effects. Basically, when hashes are re-blessed into their original
835 classes, they are no longer blessed into the DBM::Deep class! So you won't be
836 able to call any DBM::Deep methods on them. You have been warned.
837 This is an optional parameter, and defaults to 0 (disabled).
841 See L<FILTERS> below.
847 With DBM::Deep you can access your databases using Perl's standard hash/array
848 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
849 treat them as such. DBM::Deep will intercept all reads/writes and direct them
850 to the right place -- the DB file. This has nothing to do with the
851 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
852 using regular hashes and arrays, rather than calling functions like C<get()>
853 and C<put()> (although those work too). It is entirely up to you how to want
854 to access your databases.
858 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
859 or even nested hashes (or arrays) using standard Perl syntax:
861 my $db = DBM::Deep->new( "foo.db" );
863 $db->{mykey} = "myvalue";
865 $db->{myhash}->{subkey} = "subvalue";
867 print $db->{myhash}->{subkey} . "\n";
869 You can even step through hash keys using the normal Perl C<keys()> function:
871 foreach my $key (keys %$db) {
872 print "$key: " . $db->{$key} . "\n";
875 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
876 pushes them onto an array, all before the loop even begins. If you have an
877 extra large hash, this may exhaust Perl's memory. Instead, consider using
878 Perl's C<each()> function, which pulls keys/values one at a time, using very
881 while (my ($key, $value) = each %$db) {
882 print "$key: $value\n";
885 Please note that when using C<each()>, you should always pass a direct
886 hash reference, not a lookup. Meaning, you should B<never> do this:
889 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
891 This causes an infinite loop, because for each iteration, Perl is calling
892 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
893 it effectively keeps returning the first key over and over again. Instead,
894 assign a temporary variable to C<$db->{foo}>, then pass that to each().
898 As with hashes, you can treat any DBM::Deep object like a normal Perl array
899 reference. This includes inserting, removing and manipulating elements,
900 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
901 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
902 or simply be a nested array reference inside a hash. Example:
904 my $db = DBM::Deep->new(
905 file => "foo-array.db",
906 type => DBM::Deep->TYPE_ARRAY
910 push @$db, "bar", "baz";
913 my $last_elem = pop @$db; # baz
914 my $first_elem = shift @$db; # bah
915 my $second_elem = $db->[1]; # bar
917 my $num_elements = scalar @$db;
921 In addition to the I<tie()> interface, you can also use a standard OO interface
922 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
923 array) has its own methods, but both types share the following common methods:
924 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
928 =item * new() / clone()
930 These are the constructor and copy-functions.
932 =item * put() / store()
934 Stores a new hash key/value pair, or sets an array element value. Takes two
935 arguments, the hash key or array index, and the new value. The value can be
936 a scalar, hash ref or array ref. Returns true on success, false on failure.
938 $db->put("foo", "bar"); # for hashes
939 $db->put(1, "bar"); # for arrays
941 =item * get() / fetch()
943 Fetches the value of a hash key or array element. Takes one argument: the hash
944 key or array index. Returns a scalar, hash ref or array ref, depending on the
947 my $value = $db->get("foo"); # for hashes
948 my $value = $db->get(1); # for arrays
952 Checks if a hash key or array index exists. Takes one argument: the hash key
953 or array index. Returns true if it exists, false if not.
955 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
956 if ($db->exists(1)) { print "yay!\n"; } # for arrays
960 Deletes one hash key/value pair or array element. Takes one argument: the hash
961 key or array index. Returns true on success, false if not found. For arrays,
962 the remaining elements located after the deleted element are NOT moved over.
963 The deleted element is essentially just undefined, which is exactly how Perl's
964 internal arrays work. Please note that the space occupied by the deleted
965 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
966 below for details and workarounds.
968 $db->delete("foo"); # for hashes
969 $db->delete(1); # for arrays
973 Deletes B<all> hash keys or array elements. Takes no arguments. No return
974 value. Please note that the space occupied by the deleted keys/values or
975 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
976 details and workarounds.
978 $db->clear(); # hashes or arrays
980 =item * lock() / unlock()
986 Recover lost disk space.
988 =item * import() / export()
990 Data going in and out.
992 =item * set_digest() / set_pack() / set_filter()
994 q.v. adjusting the interal parameters.
1000 For hashes, DBM::Deep supports all the common methods described above, and the
1001 following additional methods: C<first_key()> and C<next_key()>.
1007 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1008 fetched in an undefined order (which appears random). Takes no arguments,
1009 returns the key as a scalar value.
1011 my $key = $db->first_key();
1015 Returns the "next" key in the hash, given the previous one as the sole argument.
1016 Returns undef if there are no more keys to be fetched.
1018 $key = $db->next_key($key);
1022 Here are some examples of using hashes:
1024 my $db = DBM::Deep->new( "foo.db" );
1026 $db->put("foo", "bar");
1027 print "foo: " . $db->get("foo") . "\n";
1029 $db->put("baz", {}); # new child hash ref
1030 $db->get("baz")->put("buz", "biz");
1031 print "buz: " . $db->get("baz")->get("buz") . "\n";
1033 my $key = $db->first_key();
1035 print "$key: " . $db->get($key) . "\n";
1036 $key = $db->next_key($key);
1039 if ($db->exists("foo")) { $db->delete("foo"); }
1043 For arrays, DBM::Deep supports all the common methods described above, and the
1044 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1045 C<unshift()> and C<splice()>.
1051 Returns the number of elements in the array. Takes no arguments.
1053 my $len = $db->length();
1057 Adds one or more elements onto the end of the array. Accepts scalars, hash
1058 refs or array refs. No return value.
1060 $db->push("foo", "bar", {});
1064 Fetches the last element in the array, and deletes it. Takes no arguments.
1065 Returns undef if array is empty. Returns the element value.
1067 my $elem = $db->pop();
1071 Fetches the first element in the array, deletes it, then shifts all the
1072 remaining elements over to take up the space. Returns the element value. This
1073 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1076 my $elem = $db->shift();
1080 Inserts one or more elements onto the beginning of the array, shifting all
1081 existing elements over to make room. Accepts scalars, hash refs or array refs.
1082 No return value. This method is not recommended with large arrays -- see
1083 <LARGE ARRAYS> below for details.
1085 $db->unshift("foo", "bar", {});
1089 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1090 -f splice> for usage -- it is too complicated to document here. This method is
1091 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1095 Here are some examples of using arrays:
1097 my $db = DBM::Deep->new(
1099 type => DBM::Deep->TYPE_ARRAY
1102 $db->push("bar", "baz");
1103 $db->unshift("foo");
1106 my $len = $db->length();
1107 print "length: $len\n"; # 4
1109 for (my $k=0; $k<$len; $k++) {
1110 print "$k: " . $db->get($k) . "\n";
1113 $db->splice(1, 2, "biz", "baf");
1115 while (my $elem = shift @$db) {
1116 print "shifted: $elem\n";
1121 Enable automatic file locking by passing a true value to the C<locking>
1122 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1124 my $db = DBM::Deep->new(
1129 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1130 mode for writes, and shared mode for reads. This is required if you have
1131 multiple processes accessing the same database file, to avoid file corruption.
1132 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1133 NFS> below for more.
1135 =head2 EXPLICIT LOCKING
1137 You can explicitly lock a database, so it remains locked for multiple
1138 transactions. This is done by calling the C<lock()> method, and passing an
1139 optional lock mode argument (defaults to exclusive mode). This is particularly
1140 useful for things like counters, where the current value needs to be fetched,
1141 then incremented, then stored again.
1144 my $counter = $db->get("counter");
1146 $db->put("counter", $counter);
1155 You can pass C<lock()> an optional argument, which specifies which mode to use
1156 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1157 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1158 same as the constants defined in Perl's C<Fcntl> module.
1160 $db->lock( DBM::Deep->LOCK_SH );
1164 =head1 IMPORTING/EXPORTING
1166 You can import existing complex structures by calling the C<import()> method,
1167 and export an entire database into an in-memory structure using the C<export()>
1168 method. Both are examined here.
1172 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1173 walking the structure and adding keys/elements to the database as you go,
1174 simply pass a reference to the C<import()> method. This recursively adds
1175 everything to an existing DBM::Deep object for you. Here is an example:
1180 array1 => [ "elem0", "elem1", "elem2" ],
1182 subkey1 => "subvalue1",
1183 subkey2 => "subvalue2"
1187 my $db = DBM::Deep->new( "foo.db" );
1188 $db->import( $struct );
1190 print $db->{key1} . "\n"; # prints "value1"
1192 This recursively imports the entire C<$struct> object into C<$db>, including
1193 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1194 keys are merged with the existing ones, replacing if they already exist.
1195 The C<import()> method can be called on any database level (not just the base
1196 level), and works with both hash and array DB types.
1198 B<Note:> Make sure your existing structure has no circular references in it.
1199 These will cause an infinite loop when importing.
1203 Calling the C<export()> method on an existing DBM::Deep object will return
1204 a reference to a new in-memory copy of the database. The export is done
1205 recursively, so all nested hashes/arrays are all exported to standard Perl
1206 objects. Here is an example:
1208 my $db = DBM::Deep->new( "foo.db" );
1210 $db->{key1} = "value1";
1211 $db->{key2} = "value2";
1213 $db->{hash1}->{subkey1} = "subvalue1";
1214 $db->{hash1}->{subkey2} = "subvalue2";
1216 my $struct = $db->export();
1218 print $struct->{key1} . "\n"; # prints "value1"
1220 This makes a complete copy of the database in memory, and returns a reference
1221 to it. The C<export()> method can be called on any database level (not just
1222 the base level), and works with both hash and array DB types. Be careful of
1223 large databases -- you can store a lot more data in a DBM::Deep object than an
1224 in-memory Perl structure.
1226 B<Note:> Make sure your database has no circular references in it.
1227 These will cause an infinite loop when exporting.
1231 DBM::Deep has a number of hooks where you can specify your own Perl function
1232 to perform filtering on incoming or outgoing data. This is a perfect
1233 way to extend the engine, and implement things like real-time compression or
1234 encryption. Filtering applies to the base DB level, and all child hashes /
1235 arrays. Filter hooks can be specified when your DBM::Deep object is first
1236 constructed, or by calling the C<set_filter()> method at any time. There are
1237 four available filter hooks, described below:
1241 =item * filter_store_key
1243 This filter is called whenever a hash key is stored. It
1244 is passed the incoming key, and expected to return a transformed key.
1246 =item * filter_store_value
1248 This filter is called whenever a hash key or array element is stored. It
1249 is passed the incoming value, and expected to return a transformed value.
1251 =item * filter_fetch_key
1253 This filter is called whenever a hash key is fetched (i.e. via
1254 C<first_key()> or C<next_key()>). It is passed the transformed key,
1255 and expected to return the plain key.
1257 =item * filter_fetch_value
1259 This filter is called whenever a hash key or array element is fetched.
1260 It is passed the transformed value, and expected to return the plain value.
1264 Here are the two ways to setup a filter hook:
1266 my $db = DBM::Deep->new(
1268 filter_store_value => \&my_filter_store,
1269 filter_fetch_value => \&my_filter_fetch
1274 $db->set_filter( "filter_store_value", \&my_filter_store );
1275 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1277 Your filter function will be called only when dealing with SCALAR keys or
1278 values. When nested hashes and arrays are being stored/fetched, filtering
1279 is bypassed. Filters are called as static functions, passed a single SCALAR
1280 argument, and expected to return a single SCALAR value. If you want to
1281 remove a filter, set the function reference to C<undef>:
1283 $db->set_filter( "filter_store_value", undef );
1285 =head2 REAL-TIME ENCRYPTION EXAMPLE
1287 Here is a working example that uses the I<Crypt::Blowfish> module to
1288 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1289 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1290 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1293 use Crypt::Blowfish;
1296 my $cipher = Crypt::CBC->new({
1297 'key' => 'my secret key',
1298 'cipher' => 'Blowfish',
1300 'regenerate_key' => 0,
1301 'padding' => 'space',
1305 my $db = DBM::Deep->new(
1306 file => "foo-encrypt.db",
1307 filter_store_key => \&my_encrypt,
1308 filter_store_value => \&my_encrypt,
1309 filter_fetch_key => \&my_decrypt,
1310 filter_fetch_value => \&my_decrypt,
1313 $db->{key1} = "value1";
1314 $db->{key2} = "value2";
1315 print "key1: " . $db->{key1} . "\n";
1316 print "key2: " . $db->{key2} . "\n";
1322 return $cipher->encrypt( $_[0] );
1325 return $cipher->decrypt( $_[0] );
1328 =head2 REAL-TIME COMPRESSION EXAMPLE
1330 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1331 compression / decompression of keys & values with DBM::Deep Filters.
1332 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1333 more on I<Compress::Zlib>.
1338 my $db = DBM::Deep->new(
1339 file => "foo-compress.db",
1340 filter_store_key => \&my_compress,
1341 filter_store_value => \&my_compress,
1342 filter_fetch_key => \&my_decompress,
1343 filter_fetch_value => \&my_decompress,
1346 $db->{key1} = "value1";
1347 $db->{key2} = "value2";
1348 print "key1: " . $db->{key1} . "\n";
1349 print "key2: " . $db->{key2} . "\n";
1355 return Compress::Zlib::memGzip( $_[0] ) ;
1358 return Compress::Zlib::memGunzip( $_[0] ) ;
1361 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1362 actually numerical index numbers, and are not filtered.
1364 =head1 ERROR HANDLING
1366 Most DBM::Deep methods return a true value for success, and call die() on
1367 failure. You can wrap calls in an eval block to catch the die.
1369 my $db = DBM::Deep->new( "foo.db" ); # create hash
1370 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1372 print $@; # prints error message
1374 =head1 LARGEFILE SUPPORT
1376 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1377 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1378 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1379 by calling the static C<set_pack()> method before you do anything else.
1381 DBM::Deep::set_pack(8, 'Q');
1383 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1384 instead of 32-bit longs. After setting these values your DB files have a
1385 theoretical maximum size of 16 XB (exabytes).
1387 B<Note:> Changing these values will B<NOT> work for existing database files.
1388 Only change this for new files, and make sure it stays set consistently
1389 throughout the file's life. If you do set these values, you can no longer
1390 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1391 back to 32-bit mode.
1393 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1394 only a 32-bit Perl. However, I have received user reports that this does
1397 =head1 LOW-LEVEL ACCESS
1399 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1400 you can call the C<_fh()> method, which returns the handle:
1402 my $fh = $db->_fh();
1404 This method can be called on the root level of the datbase, or any child
1405 hashes or arrays. All levels share a I<root> structure, which contains things
1406 like the filehandle, a reference counter, and all the options specified
1407 when you created the object. You can get access to this root structure by
1408 calling the C<root()> method.
1410 my $root = $db->_root();
1412 This is useful for changing options after the object has already been created,
1413 such as enabling/disabling locking. You can also store your own temporary user
1414 data in this structure (be wary of name collision), which is then accessible from
1415 any child hash or array.
1417 =head1 CUSTOM DIGEST ALGORITHM
1419 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1420 keys. However you can override this, and use another algorithm (such as SHA-256)
1421 or even write your own. But please note that DBM::Deep currently expects zero
1422 collisions, so your algorithm has to be I<perfect>, so to speak.
1423 Collision detection may be introduced in a later version.
1427 You can specify a custom digest algorithm by calling the static C<set_digest()>
1428 function, passing a reference to a subroutine, and the length of the algorithm's
1429 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1430 objects. Here is a working example that uses a 256-bit hash from the
1431 I<Digest::SHA256> module. Please see
1432 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1437 my $context = Digest::SHA256::new(256);
1439 DBM::Deep::set_digest( \&my_digest, 32 );
1441 my $db = DBM::Deep->new( "foo-sha.db" );
1443 $db->{key1} = "value1";
1444 $db->{key2} = "value2";
1445 print "key1: " . $db->{key1} . "\n";
1446 print "key2: " . $db->{key2} . "\n";
1452 return substr( $context->hash($_[0]), 0, 32 );
1455 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1456 of bytes you specify in the C<set_digest()> function (in this case 32).
1458 =head1 CIRCULAR REFERENCES
1460 DBM::Deep has B<experimental> support for circular references. Meaning you
1461 can have a nested hash key or array element that points to a parent object.
1462 This relationship is stored in the DB file, and is preserved between sessions.
1465 my $db = DBM::Deep->new( "foo.db" );
1468 $db->{circle} = $db; # ref to self
1470 print $db->{foo} . "\n"; # prints "bar"
1471 print $db->{circle}->{foo} . "\n"; # prints "bar" again
1473 B<Note>: Passing the object to a function that recursively walks the
1474 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1475 C<export()> methods) will result in an infinite loop. This will be fixed in
1478 =head1 CAVEATS / ISSUES / BUGS
1480 This section describes all the known issues with DBM::Deep. It you have found
1481 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1483 =head2 UNUSED SPACE RECOVERY
1485 One major caveat with DBM::Deep is that space occupied by existing keys and
1486 values is not recovered when they are deleted. Meaning if you keep deleting
1487 and adding new keys, your file will continuously grow. I am working on this,
1488 but in the meantime you can call the built-in C<optimize()> method from time to
1489 time (perhaps in a crontab or something) to recover all your unused space.
1491 $db->optimize(); # returns true on success
1493 This rebuilds the ENTIRE database into a new file, then moves it on top of
1494 the original. The new file will have no unused space, thus it will take up as
1495 little disk space as possible. Please note that this operation can take
1496 a long time for large files, and you need enough disk space to temporarily hold
1497 2 copies of your DB file. The temporary file is created in the same directory
1498 as the original, named with a ".tmp" extension, and is deleted when the
1499 operation completes. Oh, and if locking is enabled, the DB is automatically
1500 locked for the entire duration of the copy.
1502 B<WARNING:> Only call optimize() on the top-level node of the database, and
1503 make sure there are no child references lying around. DBM::Deep keeps a reference
1504 counter, and if it is greater than 1, optimize() will abort and return undef.
1506 =head2 AUTOVIVIFICATION
1508 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1509 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1510 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1511 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1512 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1515 $db->{foo}->{bar} = "hello";
1517 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1518 being an empty hash. Try this instead, which works fine:
1520 $db->{foo} = { bar => "hello" };
1522 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1523 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1524 Probably a bug in Perl.
1528 (The reasons given assume a high level of Perl understanding, specifically of
1529 references. You can safely skip this section.)
1531 Currently, the only references supported are HASH and ARRAY. The other reference
1532 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1538 These are things like filehandles and other sockets. They can't be supported
1539 because it's completely unclear how DBM::Deep should serialize them.
1541 =item * SCALAR / REF
1543 The discussion here refers to the following type of example:
1550 # In some other process ...
1552 my $val = ${ $db->{key1} };
1554 is( $val, 50, "What actually gets stored in the DB file?" );
1556 The problem is one of synchronization. When the variable being referred to
1557 changes value, the reference isn't notified. This means that the new value won't
1558 be stored in the datafile for other processes to read. There is no TIEREF.
1560 It is theoretically possible to store references to values already within a
1561 DBM::Deep object because everything already is synchronized, but the change to
1562 the internals would be quite large. Specifically, DBM::Deep would have to tie
1563 every single value that is stored. This would bloat the RAM footprint of
1564 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1565 all to support a feature that has never been requested.
1569 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1570 mechanism for serializing coderefs, including saving off all closure state.
1571 However, just as for SCALAR and REF, that closure state may change without
1572 notifying the DBM::Deep object storing the reference.
1576 =head2 FILE CORRUPTION
1578 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1579 for a 32-bit signature when opened, but other corruption in files can cause
1580 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1581 stuck in an infinite loop depending on the level of corruption. File write
1582 operations are not checked for failure (for speed), so if you happen to run
1583 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1584 be addressed in a later version of DBM::Deep.
1588 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1589 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1590 about setting up your NFS server with a locking daemon, then using lockf() to
1591 lock your files, but your mileage may vary there as well. From what I
1592 understand, there is no real way to do it. However, if you need access to the
1593 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1594 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1596 =head2 COPYING OBJECTS
1598 Beware of copying tied objects in Perl. Very strange things can happen.
1599 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1600 returns a new, blessed, tied hash or array to the same level in the DB.
1602 my $copy = $db->clone();
1604 B<Note>: Since clone() here is cloning the object, not the database location, any
1605 modifications to either $db or $copy will be visible in both.
1609 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1610 These functions cause every element in the array to move, which can be murder
1611 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1612 a different location. This will be addressed in the forthcoming version 1.00.
1614 =head2 WRITEONLY FILES
1616 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1617 writeonly mode. STORE will verify that the filehandle is writable. However, there
1618 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1619 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1623 This section discusses DBM::Deep's speed and memory usage.
1627 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1628 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1629 multi-level hash/array support, and cross-platform FTPable files. Even so,
1630 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1631 with huge databases. Here is some test data:
1633 Adding 1,000,000 keys to new DB file...
1635 At 100 keys, avg. speed is 2,703 keys/sec
1636 At 200 keys, avg. speed is 2,642 keys/sec
1637 At 300 keys, avg. speed is 2,598 keys/sec
1638 At 400 keys, avg. speed is 2,578 keys/sec
1639 At 500 keys, avg. speed is 2,722 keys/sec
1640 At 600 keys, avg. speed is 2,628 keys/sec
1641 At 700 keys, avg. speed is 2,700 keys/sec
1642 At 800 keys, avg. speed is 2,607 keys/sec
1643 At 900 keys, avg. speed is 2,190 keys/sec
1644 At 1,000 keys, avg. speed is 2,570 keys/sec
1645 At 2,000 keys, avg. speed is 2,417 keys/sec
1646 At 3,000 keys, avg. speed is 1,982 keys/sec
1647 At 4,000 keys, avg. speed is 1,568 keys/sec
1648 At 5,000 keys, avg. speed is 1,533 keys/sec
1649 At 6,000 keys, avg. speed is 1,787 keys/sec
1650 At 7,000 keys, avg. speed is 1,977 keys/sec
1651 At 8,000 keys, avg. speed is 2,028 keys/sec
1652 At 9,000 keys, avg. speed is 2,077 keys/sec
1653 At 10,000 keys, avg. speed is 2,031 keys/sec
1654 At 20,000 keys, avg. speed is 1,970 keys/sec
1655 At 30,000 keys, avg. speed is 2,050 keys/sec
1656 At 40,000 keys, avg. speed is 2,073 keys/sec
1657 At 50,000 keys, avg. speed is 1,973 keys/sec
1658 At 60,000 keys, avg. speed is 1,914 keys/sec
1659 At 70,000 keys, avg. speed is 2,091 keys/sec
1660 At 80,000 keys, avg. speed is 2,103 keys/sec
1661 At 90,000 keys, avg. speed is 1,886 keys/sec
1662 At 100,000 keys, avg. speed is 1,970 keys/sec
1663 At 200,000 keys, avg. speed is 2,053 keys/sec
1664 At 300,000 keys, avg. speed is 1,697 keys/sec
1665 At 400,000 keys, avg. speed is 1,838 keys/sec
1666 At 500,000 keys, avg. speed is 1,941 keys/sec
1667 At 600,000 keys, avg. speed is 1,930 keys/sec
1668 At 700,000 keys, avg. speed is 1,735 keys/sec
1669 At 800,000 keys, avg. speed is 1,795 keys/sec
1670 At 900,000 keys, avg. speed is 1,221 keys/sec
1671 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1673 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1674 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1675 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1676 Run time was 12 min 3 sec.
1680 One of the great things about DBM::Deep is that it uses very little memory.
1681 Even with huge databases (1,000,000+ keys) you will not see much increased
1682 memory on your process. DBM::Deep relies solely on the filesystem for storing
1683 and fetching data. Here is output from I</usr/bin/top> before even opening a
1686 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1687 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1689 Basically the process is taking 2,716K of memory. And here is the same
1690 process after storing and fetching 1,000,000 keys:
1692 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1693 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1695 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1696 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1698 =head1 DB FILE FORMAT
1700 In case you were interested in the underlying DB file format, it is documented
1701 here in this section. You don't need to know this to use the module, it's just
1702 included for reference.
1706 DBM::Deep files always start with a 32-bit signature to identify the file type.
1707 This is at offset 0. The signature is "DPDB" in network byte order. This is
1708 checked for when the file is opened and an error will be thrown if it's not found.
1712 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1713 has a standard header containing the type of data, the length of data, and then
1714 the data itself. The type is a single character (1 byte), the length is a
1715 32-bit unsigned long in network byte order, and the data is, well, the data.
1716 Here is how it unfolds:
1720 Immediately after the 32-bit file signature is the I<Master Index> record.
1721 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1722 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1723 depending on how the DBM::Deep object was constructed.
1725 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1726 number). The first 8-bit char of the MD5 signature is the offset into the
1727 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1728 index element is a file offset of the next tag for the key/element in question,
1729 which is usually a I<Bucket List> tag (see below).
1731 The next tag I<could> be another index, depending on how many keys/elements
1732 exist. See L<RE-INDEXING> below for details.
1736 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1737 file offsets to where the actual data is stored. It starts with a standard
1738 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1739 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1740 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1741 When the list fills up, a I<Re-Index> operation is performed (See
1742 L<RE-INDEXING> below).
1746 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1747 index/value pair (in array mode). It starts with a standard tag header with
1748 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1749 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1750 header. The size reported in the tag header is only for the value, but then,
1751 just after the value is another size (32-bit unsigned long) and then the plain
1752 key itself. Since the value is likely to be fetched more often than the plain
1753 key, I figured it would be I<slightly> faster to store the value first.
1755 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1756 record for the nested structure, where the process begins all over again.
1760 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1761 exhausted. Then, when another key/element comes in, the list is converted to a
1762 new index record. However, this index will look at the next char in the MD5
1763 hash, and arrange new Bucket List pointers accordingly. This process is called
1764 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1765 17 (16 + new one) keys/elements are removed from the old Bucket List and
1766 inserted into the new index. Several new Bucket Lists are created in the
1767 process, as a new MD5 char from the key is being examined (it is unlikely that
1768 the keys will all share the same next char of their MD5s).
1770 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1771 when the Bucket Lists will turn into indexes, but the first round tends to
1772 happen right around 4,000 keys. You will see a I<slight> decrease in
1773 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1774 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1775 right around 900,000 keys. This process can continue nearly indefinitely --
1776 right up until the point the I<MD5> signatures start colliding with each other,
1777 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1778 getting struck by lightning while you are walking to cash in your tickets.
1779 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1780 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1781 this is 340 unodecillion, but don't quote me).
1785 When a new key/element is stored, the key (or index number) is first run through
1786 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1787 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1788 for the first char of the signature (in this case I<b0>). If it does not exist,
1789 a new I<Bucket List> is created for our key (and the next 15 future keys that
1790 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1791 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1792 this point, unless we are replacing an existing I<Bucket>), where the actual
1793 data will be stored.
1797 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1798 (or index number), then walking along the indexes. If there are enough
1799 keys/elements in this DB level, there might be nested indexes, each linked to
1800 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1801 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1802 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1803 plain key are stored.
1805 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1806 methods. In this process the indexes are walked systematically, and each key
1807 fetched in increasing MD5 order (which is why it appears random). Once the
1808 I<Bucket> is found, the value is skipped and the plain key returned instead.
1809 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1810 alphabetically sorted. This only happens on an index-level -- as soon as the
1811 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1812 so it's pretty much undefined how the keys will come out -- just like Perl's
1815 =head1 CODE COVERAGE
1817 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1818 B<Devel::Cover> report on this module's test suite.
1820 ----------------------------------- ------ ------ ------ ------ ------ ------
1821 File stmt bran cond sub time total
1822 ----------------------------------- ------ ------ ------ ------ ------ ------
1823 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1824 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1825 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1826 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1827 Total 97.9 85.9 79.7 100.0 100.0 94.3
1828 ----------------------------------- ------ ------ ------ ------ ------ ------
1830 =head1 MORE INFORMATION
1832 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1833 or send email to L<DBM-Deep@googlegroups.com>.
1837 Joseph Huckaby, L<jhuckaby@cpan.org>
1839 Rob Kinyon, L<rkinyon@cpan.org>
1841 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1845 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1846 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1850 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1851 This is free software, you may use it and distribute it under the
1852 same terms as Perl itself.